IESL Transaction Partb2017 Technical Paper PDF

“To promote the acquisition and interchange of technical knowledge,
advance the science and practice of engineering in all its branches and
regulate professional activities in Sri Lanka.”
THE INSTITUTION OF ENGINEERS

SRI LANKA
Transactions – 2017
PART B
Technical Papers
This Institution does not, as a body, hold itself responsible for statements made or opinions expressed
either in the Papers read or the discussions which have occurred at the Meetings.
CEO / Executive Secretary

The articles published in the Transactions are based on Engineering Research and some are
of Professional Interest. All published papers have been refereed in anonymity by at least
two subject specialists.
Published by:
The Institution of Engineers, Sri Lanka

120/15, Wijerama Mawatha
Colombo 07
Tel : 2 698426, 2685490, 2 699210

Fax : 2 699202
E-mail :iesl@slt.lk
Website : http://www.iesl.lk
Printed at:
Karunaratne & Sons (Pvt) Ltd.

65 C, Midellamulahena,
Thalgahawila Road, Horana.
Info@karusons.com
ii
CONTENTS
Editor’s note xi
1. Rainfall Thresholds for Initiation of Landslides in Ratnapura District

By: Eng. L R Abhayarathne 1
2. Experimental Study on the Effect of Presence of Deep Embedded Retaining Walls on

the Settlement of Shallow Foundations
By : Eng. M P Amarasinghe and Eng. (Dr.) L I N De Silva 11
3. Shrinkage Behavior of a Liner Material with Expansive Properties for use in an

Engineered Landfill
By : Eng. (Ms) V Kayaththiry 17
4. Identification of Clayey Properties for Cricket Pitch Surfacing and Analysing the
Variation of Surface Crack Densities in different Pitches
By : Eng. W S U Perera, Eng. (Dr.)U P Nawagamuwa and Mr. G N U Thilakarathna 25
5. Design of Soil–Cement–Bentonite (SCB) backfill and construction of SCB slurry walls

along the earthen dam at Iranamadu Reservoir in Kilinochchi
By: Eng. (Dr.) P N Wickramarachchi, Eng. S Weerakoon, Mr. S Semage,
Eng. N Madushanka and Eng. T Rajagobu 33
6. Ballast Mats Made of Recycled Tyre Waste to Enhance the Performance of Rail Track
Substructure under Cyclic Load
By: Eng. (Dr.) S K Navaratnarajah and Eng. (Prof.) B Indraratna 41
7. Analysis of Energy Dissipation in Post-Tensioned Hybrid Precast Concrete Frames

By: Eng. K L P S Perera, Eng. W D C R Wanniarachchi and Dr. P Thammarak 51
8. Study on Filler Slab Construction

By: Eng. (Dr.) K Baskaran, Eng. A Gopikrishna and Eng. P Thiruvarankan 59
9. Energy Dissipater for Post-Tensioned Steel Frame System

By: Eng. Ms. W D C R Wanniarachchi, Eng. K L P S Perera and Dr. P Thammarak 67
10. Simplified Approach for Predicting Fatigue Life of Fillet Welded Joints with FEM
By: Eng. (Ms.) S Varakini, Eng. K Ubamanyu and
Eng. (Dr.) H M Y C Mallikarachchi 75
11. Growing Colombo: Some Concerns from Wind Engineering Point of View
By: Dr. A U Weerasuriya and Mr. K A B Weerasinghe 81
12. Contributory factors for Cost and Time Overruns of Construction Projects
By: Eng. Leelanath Daluwatte and Eng. (Prof.) Malik Ranasinghe 89
13. Sub-Contracting and Accident Risk in Construction Industry : A Case Study in

Singapore
By: Eng. W A Asanka and Eng.(Prof.) Malik Ranasinghe 97
iii
14. Techno-Economic Analysis of Building Energy System with Net Meter Solar PV in Sri
Lanka
By: Eng. (Dr.) K T M U Hemapala and Eng.( Ms.) T M S C Jayasinghe 105
15. Multi-Functional Floor Cleaning Robot for Domestic Environment

By: Eng. H G T Milinda and Eng. B G D A Madhusanka 115
16. Development of CZTS thin Film from Spin Coating for Solar Cell Application
By: H A D I Dilshan, W D D Weerasinghe, Eng. N L W Wijewantha,
Eng. W P S Wijesinghe and Dr. D Atygalle 121
17. Characterization of Fuel Briquettes Made for Household Cooking from Sawdust and
Rice Husk
By: M Y A R Ahmed, U M A Devaraja, K A R Mihiranga and
Prof. B A J K Premachandra 125
18. Assessing the Performance of a Destructive Measurement System: A Case of an

Industrial Glove Manufacturing Environment
By: Mrs. H A D Perera and Dr. P Gamage 135
19. Development of ANN Based IMC for Biomass Pack Bed Gasification Plant with
Enhanced Efficiency
By : Eng. G V C Rasanga, Eng. (Dr.) K T M U Hemapala and
Eng. (Dr.) A G Buddhika P Jayasekara 143
20. Organic Rankine Cycle (ORC): Performance of Working Fluids and Energy Recovery
Potential in Sri Lankan Thermal Power Plants
By : Eng. B S R Fernando, Dr. R A C P Ranasinghe and Dr. H K G Punchihewa 151
21. Mechanized Fertilizer Applicator Operated by Two-wheel Tractor for Coconut

Plantation Industry in Sri Lanka
By: Eng. S A P S Silva 161
22. Water Loss Reduction by Understanding the Reticulation A Case Study

By : Eng. S G G Rajkumar 167
23. Implementation of Water Safety Plans for Rural Water Supply Schemes Managed By
CBOs in Sri Lanka – Common Risks and Analysis
By: Mr. T R Samaraweera and Eng. (Prof.) (Mrs.) Niranjanie Ratnayake 173
24. A Study on Recontamination Risks in Water Distribution Systems, using A Network

Model
By : Mr. M H V Lasantha and Eng. (Prof.) (Mrs.) Niranjanie Ratnayake 181
25. Review of Irrigation Ordinance and Identifying the Gaps to fulfil the Current Needs
of the Society
By: Eng. (Mrs.) Badra Kamaladasa, Eng. Lalith De Alwis and Eng. G D F U Perera 189
26. Analysis of Long-term Trends of Climatological Parameters in Kigali, Rwanda

By: Eng. (Dr.) D P C Laknath, Eng. N R Josiah and Eng. T A J G Sirisena 197
iv
27. Development of a Flood Forecasting And Data Dissemination System For Kalu River
Basin in Sri Lanka
By: Eng. S Kokularamanan, Eng. (Dr.) A W M Rasmy, Eng. (Dr.) Duminda Perera
and Eng. (Prof.) Toshio Koike 205
28. A GIS Based Asset Management Approach to Upgrade the Level of Service to Pipe
Borne Water Consumers in Western North Region
By: Eng. A M H K Abeykoon and Eng. B U J Perera 211
29. Application of 3D Physical Modelling for Verification and Optimization of Hydraulic

Designs for Salinity Barriers in Sri Lanka
By: Eng. (Mrs.) S M C K Subasinghe, Eng. (Ms.) K P S T Pathirana,
Eng. (Ms.) N L Engiliyage and Eng. G H N Rathnasiri 219
30. Numerical Model Investigations on Dispersion of Industrial Effluents in East Coast of

Sri Lanka
By: Eng. (Dr.) S Wickramaratne and Eng. (Ms.) K P S T Pathirana 227
31. Numerical Modelling Study for RahfalhuHuraa Resort Development in Maldives

By: Eng. M S L Fernando, Eng. I L Abeygoonasekara and
Eng. S M D E Suriyamudali 239
32. Wave Tranquillity Study for a Fishery Harbour using Boussinesq Type Wave Model
By: Eng. (Dr.) D P C Laknath, Eng. D E N Senarathne and Eng. K K P P Ranaweera 247
33. Verification of Wave Transformation to Establish the Nearshore Wave Climate

By: Eng. (Dr.) D P L Ranasinghe, Eng. N L Engiliyage, Eng. H P G M Caldera and
Eng. I G I Kumara 255
34. Fabrication and Mechanical Property Analysis of Microcrystalline Cellulose Based

Polymer Composite for Engineering Applications
By: Ms. H A D Saumyadi, Mr. L D Rajapaksha, Eng. A M P B Samarasekara,
Dr. D A S Amarasinghe and Prof. L Karunanayake 263
35. Synthesis and Characterization of Nanoparticle-Incorporated Antimicrobial Coating

for Food Packaging Applications
By: Ms. S PA Madushani, Ms. D P Egodage, Eng. A M P B Samarasekara,
Dr. D A S Amarasinghe, Mr. H T S Jayalath and Mr. S M N S Senerath 271
36. Assessing the Impact on Environment, Health and Safety due to Releases from
Chemical Process Routes
By: Eng. H B B Anuradha, Eng. (Dr.) Manisha Gunasekera and Dr. Olga Gunapala 277
37. Antecedents of Work Engagement : Case Study of Outsourced Employees in

Telecommunication Service Sector of Sri Lanka
By: Eng. K A Chathura S Kumarasinghe 285
38. The Structure of Land Prices in Colombo and its Implications

By : Eng. (Dr.) P I A Gomes and R K Hashan 305
v
39. Knowledge Management System for Mobile Telecommunication Sector in Sri Lanka
By: Eng. Jeewantha Nadeera, Eng. (Ms.) Shalanika Dayarathna and
Eng. Eranda Ihalagedara 311
40. Effect of Technical and Non-Technical competence on the Job Performance of Private
Sector Engineers in Sri Lanka
With special reference to a group of large scale engineering companies
By : Eng. J R Ekanayayake 317
41. Artificial Bone Substitutes and Bone Tissue Engineering as Alternative Methods for
Natural Bone Repair and Regeneration in Severe Traumas and Complex Clinical
Conditions of Bones.
By : Eng. (Miss.) B R M G W U B Balagalla 327
42. Condition Assessment of Power Transformers using Dissolved Gas Analysis

By : Eng. W T D Samarasingha, Eng. (Dr.) J R S S Kumara and
Eng. (Prof.) M A R M Fernando 333
43. Battery - Supercapacitor Hybrid Energy Storage Control for Electric Vehicle
By: Mr. W M J Madhusankha, Mr. G A T P Madhuranga, Mr. S A S Indika,
Eng. (Dr.) B G L T Samaranayake, Eng. (Dr.) P J Binduhewa and
Eng. (Prof.) J B Ekanayake 341
44. Pitch Control of a Twin Rotor System using Error Dynamics Based Nonlinear
Controller
By: Mr. Yasitha Liyanage, Eng. Jagannath Wijekoon, Eng. Shirantha Welikala and
Eng. (Dr.) Lilantha Samaranayake 349
45. Museum Navigation System with Beacons

By: Eng. N D C D Kannangara and Eng. W D S S Bandara 355
46. Adaptation of SCASA Technique for surge Protective Devices for Equatorial Belt
Countries
By : Eng. J P D S Athuraliya, Eng. N L Rathnayake, Eng. N A A N Dilrukshi, Eng. P
Mahadevan, Eng. R Gihara, Mr. T M S Dias, Mr. S R J S Bandara, Ms. H Wijesooriya
and Eng. (Prof.) Nihal Kularatna 363
47. Detecting Key Actors in Multiple SIM Networks Via Centrality Measures and Mobile
Service usage Behaviors
By: Eng. (Dr.) A J M Korale and Mr. N S Weeraman 373
48. Intelligent Pest Repellent System for Sri Lankan Farming Industry
By : Eng. P N Karunanayake, Mr. W H H N de Soysa, Mr. J A C L C Jayasundara,
Mr. Y G Wanniarachchi and Mr. A P I Karunarathne 379
49. Land Cover Classification and Sub Component Analysis Using Hyperspectral
Imagery
By: Ms. A M R Abeysekara, Mr. T S J Oorloff, Ms. S S P Vithana,
Eng. (Dr.) H M V R Herath, Eng. (Dr.) G M R I Godaliyadda and
Eng. (Dr.) M P B Ekanayake 385
vi
50. Spatial Modulation for Indoor MIMO Visible Light Communication System
By : Mr. U G R K W S Palitharathna, Mr. A G D U K Samarasinghe,
Mr. Y M L D Wijayarathna, Eng. (Dr.) S A H A Suraweera and
Eng. (Dr.) G M R I Godaliyadda 391
51. Energy Management System of a Solar Driven DC System

By : Mr. H M A A B Herath, Mr. K K S U Kodithuwakku, Eng. Mrs. D R D S
Dasanayaka, Dr. P J Binduhewa, Eng. (Prof.) J B Ekanayake and
Eng. (Dr.) S M K B Samarakoon 401
52. Laser Ranging Based Intelligent System for Unknown Environment Mapping
By: Mr. T H M N C Thelasingha, Mr. U V B L Udugama, Ms. E M S P Ekanayake,
Eng. (Dr.) G M R I Godaliyadda, Eng. (Dr.) M P B Ekanayake, Eng. (Dr.) B G L T
Samaranayake and Eng. (Dr.) J V Wijayakulasooriya 411
53. A Novel Handover Mechanism for Visible Light Communication Network

By : Mr. M A N Perera, Mr. N G K R Wijewardhana, Ms. A A D T Nissanka,
Dr. S A H A Suraweera and Eng. (Dr.) G M R I Godaliyadda 421
54. Technical, Economic, Operational and Dispatch Analysis of Optimum Integration of

Renewable Energy in to the Sri Lankan Power system
By : Eng. Buddhika Samarasekara, Eng. Randika Wijekoon and
Eng. Lathika Attanayake 429
55. Optimum Use of Solar Inverter by Feeding Reactive Power at the Night
By: Eng. (Ms.) K Kanchanee Navoda and Eng. (Dr.) Asanka S Rodrigo 439
56. Smart Electronic Analyzer for Determination of Chicken Meat Spoilage

By: Eng. M P U Isuranga, Mrs. H Pasqual and Dr. N S Weerakkody 447
57. Power Electronic Interface of a Solar Driven DC System

By : Mr. W W M D B Wijesinghe, Mr. H M K C B Wijerathne,
Ms. B H T S Gunaratne, Dr. P J Binduhewa and Eng. (Prof.) J B Ekanayake 455
58. Indoor Navigation System with RFID Tags for Virtually Impaired Individuals
By : Eng. J M K V Jayalath, Mr. D M G C Jayathilake, Mr. M A C I U Yapa,
Dr. S A H A Suraweera and Eng. (Dr.) G M R I Godaliyadda 463
59. Energy Efficiency Benchmarking of Pumps in Water and Wastewater

Industries in Sri Lanka
By: Eng. B G A Dinesh 469
60. Feasibility of Hybrid Reactors for Removal of Heavy Metals in Wastewater

By: G S Nallaperuma and Dr. W M K R T W Bandara 475
61. Feasibility of UASB Reactor for Removal of Heavy Metal in Wastewater

By: U L N C Jayathileke and Dr. W M K R T W Bandara 485
62. GIS-Based Site Suitability Analysis for Wastewater Treatment Plants

By: N A Nusky, S I S Wijesingha, Dr. W M K R T W Bandara,
B M L A Basanayake and Dr. Terrance M Rengarasu 493
vii
63. Air Pollution Monitoring Through Crowdsourcing
By : Eng. M I N P Munasinghe, Eng. (Dr.) G I U S Perera, Mr. J K W D B
Karunathilaka, Mr. B C S Cooray and Eng. K G D Manupriyal 499
64. A Conceptual Tool for Assessing Building Life Cycle Carbon Emissions in the
Context of Sri Lanka
By: Eng. R P Kumanayake, Prof. H B Luo and Eng. R M P S Bandara 507
65. Upgrading of the Existing Sedimentation Tank at Hemmathagama Water Supply

Scheme
By: Eng. L Wijesinghe and Eng. (Dr.) D R I B Werellagama 517
66. Risk Assessment of the Active Ingredient, Glyphosate of the Herbicide, ROUNDUP®
Present in Water Sources of CKDu Prevalent Areas for Humans
By: Eng. (Ms.) S Gunarathna, Eng. (Dr.) (Ms.) B Gunawardana,
Eng. (Prof.) M Jayaweera and Mr. K Zoysa 525
67. Field-Scale Study on Treating Landfill Leachate Using Anaerobic Filters Packed with
Low-Cost Filter Media
By : Eng. (Dr.) W K C N Dayanthi, Eng. (Mrs.) B S R Nanayakkara,
Prof. K Kawamoto, Eng.(Mrs.) M D B A Premathilaka and Mr. O V N Dulvin 531
68. Variations to Key Chemical Parameters during Dried Faecal Sludge and Municipal
Solid Waste Co-Composting
By: Eng. H M N Jayawardena, Eng. (Dr.) S C Fernando, Eng. (Dr.) S H P
Gunawardena and Eng. N Jayathilake 537
69. Life Cycle Assessment Based Tool to Visualize Environmental Impact of

Construction Materials & Products
By : Mr. H M M M Jayawickrama, Mrs. K G N H Weerasinghe, Eng. R L Peiris,
Eng. (Dr.) A K Kulatunga 545
70. Safety Analysis of Anaerobic Digestion Plant Operation

By : Eng. (Dr.) Deshai Botheju and Eng. (Dr.) P G Rathnasiri 553
71. Simulating the Effect of Modifying Reversed A2O Process on Combined Removal of
Nitrogen, Phosphorous and Organic Matter in Municipal Wastewater
By: Eng. W B Roshen P S Fernando and Eng. (Dr.) P G Rathnasiri 565
72. Extraction of Cellulose from Sri Lankan Agricultural Waste for Engineering
Applications
By: Ms. M P A Nanayakkara, Ms. W G A Pabasara, Eng. A M P B Samarasekara,
Dr. D A S Amarasinghe and Prof. L Karunanayake 571
73. Environmental Impacts of Organic Fertilizer Produced from Industrial Zone Central
Effluent Treatment Plant Sludge using Delta-D Technology
By : Eng. M F H M Aadhil and Eng. (Prof.) S A S Perera 579
viii
74. Field Experiments on Low-Cost Combined-Media an Anaerobic Filter System for the
Treatment of Landfill-Leachate
By: Eng. (Ms.) D M T P Fernando, Eng. (Dr.) W K C N Dayanthi,
Eng. (Ms.) K G D Chamenuka, Eng. (Mrs.) B S R Nanayakkara and
Prof. K Kawamoto 587
75. Low-Cost drinking Water Filter

By: Mr. U A C Perera and Eng. (Dr.) (Ms.) S C S Karunaratne 595
76. Monitoring Carbon Monoxide (CO) Levels at Five Major Schools in Kandy City using
Real Time Air Quality Monitoring Sensor Network
By: Eng. R G R N K Ariyaratne, Eng. (Dr.) D G G P Karunaratne,
Eng. (Dr.) M A Elangasinghe, Eng. (Dr.) A Manipura, Dr. S P Kodithuwakku,
Dr. A D Siribaddana and Mr. K H N Abayalath 605
77. Dense Graded Asphalt for Sri Lanka : Considerations related to Climate and
Different Traffic Conditions
By: Eng. W W Bandara and Eng. R P M M Premathilaka 617
78. Suitability of Recycled Pet Fiber and Carbonized Wood Ash in Hot Mix Asphalt
Concrete
By : Eng. N Jegatheesan, Eng. (Dr.) T M Rengarasu and
Eng. (Dr.) W M K R T Wasala Bandara 623
79. Evaluating Road user Cost for Highway Work Zones – Case Study for Urban
Upgrading Projects in Sri Lanka
By: M H M Hamsath and Eng. (Dr.) H R Pasindu 631
80. Development of a Methodology for Road Maintenance Planning of Low Volume

Roads Based on Roughness Data
By: Mr. M A D De Silva and Eng. (Dr.) H R Pasindu 641
81. Travel Time Estimation Based on Dynamic Traffic Data and Machine Learning
Principles
By: Mr. Sakitha P Kumarage, Mr. V M Jayawardana, Dr. Dimantha De Silva
and Eng. (Prof.) J M S J Bandara 649
82. Infrastructure Feasibility Analysis of Inland Water Transportation from Peliyagoda to

Hanwella along Kelani River
By: Mr. K P H S Kalahe and Eng. (Dr.) R M N T Sirisoma 659
83. Hydroplaning Risk Evaluation in High Speed Low Traffic Volume Highways
By: Eng. (Ms.) M D M Sheranie and Eng. (Mrs.) S Senevirathne 667
ix
x
Editor’s Note………
--------------------------------------------------------------------------------------------------------------------------
Dear Readers of Transactions:
It is my pleasure to introduce Transactions (Part B): Technical Papers presented at the 111th
Annual Sessions of the Institution of Engineers, Sri Lanka. I am honoured to edit this year’s
Technical Papers submitted to be presented at the 111th Annual Sessions and sincerely hope
that the papers included in here are of value and interest to Engineers and make
contributions to the advancement of the Profession of Engineering. As the Editor of the
Annual Sessions, I am proud to say that the vision of the IESL in providing our Engineers
with a platform to share their research and educational experiences is progressively gaining
attention and popularity.
As an engineer, the ability to conduct thorough and accurate research while clearly
communicating the results is essential for facilitating the advancement of the acquired
knowledge base. Knowing where and how to find different types of information will help
solve engineering problems, in both academic as well as the professional career. Lack of
investigation into engineering guidelines, standards, and best practices can result in failure
with severe repercussions. Traditionally, engineering research has yielded knowledge
essential to translating scientific advances into technologies that affect everyday life. Most
certainly, engineering research has resulted in the vast progress made by the humanity
everywhere in the world. Engineering research, in the future, will generate technological
innovations to address enormous challenges in a multitude of areas of the socio-economic
systems, in most cases, they will definitely be multi-disciplinary in nature.
All the manuscripts submitted for the Annual Sessions have been reviewed by at least two
prominent persons in order to select the highest quality and pertinent research papers that
are worthwhile in publishing in the Transactions. Many people were involved in this process,
beginning with the authors who submit their papers and the reviewers by unrelentingly
providing their expertise and time, giving their opinions on the technical and scientific merit
of the contents, mathematical and experimental analysis, as well as the scope and
contribution to the advancement of the respective field. The reviewers’ comments were
immensely useful in helping the Editor in making decisions regarding acceptance of a given
paper. Their comments helped the authors too, by providing constructive criticisms that
enabled them to improve the papers as well as maintaining the quality and higher standards
of the Transactions. While the Annual Sessions have become a traditional meeting point and
forum for engineers, researchers, and professionals, where the discussion topics are recorded
as proceeding papers, the Transactions is considered a serious reference because the
published papers provide a more detailed description of the research work, mainly focused
on the current trends in engineering theory and practice. Their validity, relevance,
importance, and originality are scrutinised by a peer reviewing process.
I am very grateful to the Council of the IESL for inviting me to be the Editor of this year’s
Annual Sessions. I extend my sincere appreciations to the staff of the IESL for all their
support and help. Finally, once again, I am most appreciative of the efforts of the authors
and reviewers, and I am deeply indebted for their commitment, dedication, and creativity.
I wish you all very fruitful deliberations during the Annual Sessions.
Eng. (Professor) Jagath Manatunge
B.Sc. Eng. Hons. (Moratuwa), D.I.C., M.Sc. (Lond.), Ph.D. (Saitama), C. Eng., M.I.E. (SL),
MIEPSL
Editor, IESL Transactions (Part B) 2017
xi
AnnualSessions
Annual SessionsofofIESL,
IESL,pp.
pp.[1 [1 - 10],
- 9], 20172017
© The
© The Institution
InstitutionofofEngineers,
Engineers,SriSri Lanka
Lanka
Rainfall Thresholds for Initiation of Landslides in

Ratnapura District
L. R. Abhayarathne
Abstract: Ratnapura is a district frequently affected by landslides in Sri Lanka and its
vulnerability is increased due to its location in the wet zone. The purpose of this study was to find the
rainfall thresholds for initiation of landslides in Ratnapura district and the thresholds were based on
one day to seven days cumulative rainfall. 225 landslide incidents which occurred from May 1978 to
June 2014 were used for the analysis. Probability tables were developed by combining the landslide
data with rainfall data. Then, through SPSS a curvilinear Regression model was created for each
threshold value using the probability, as the independent variable and the rainfall linked to it as the
dependent variable. The correlation of daily rainfall to the cumulative rainfall was calculated up to
seven days. The thresholds obtained from the analysis were as follows, for one day threshold (
), for two day cumulative threshold and for three day cumulative
threshold where „R‟ is the rainfall in millimetres and„t‟ is the probability of
landslide occurrence. The thresholds found for four to seven day cumulative rainfall were not valid
due to outliers they contain. When the current NBRO thresholds were compared with the one day
threshold value found in the study, the probability received for landslide occurrence was 58.47% for
alert level, 66.46% for warning and 77.72% for evacuation level. By looking at the correlation values
only the rainfall of the previous day had a significant effect on the cumulative rainfall for occurrence
of landslides. In conclusion, the study was able to establish rainfall thresholds for the prediction of
landslides based on one day to three day cumulative rainfall.
Keywords: Disaster Management, Landslide, Rainfall Threshold, Ratnapura District
1. Introduction created frequent landslide events in over 10

districts including Ratnapura district.
Sri Lanka is a country prone to handful of Therefore, the minimum amount of rainfall that
natural disasters and one of them is Landslides. is needed to trigger a landslide would fall
A landslide may include a variety of processes within the interest of disaster analysts,
that result downward and outward movement researchers, investigators, disaster responders
of slope forming material that may comprise of and the community living in these landslide
rock, artificial fill, soil or a combination of them. hazard prone areas.
They can be very destructive like the one
recently happened in Afghanistan which buried Currently (August 2017) only a general
more than 2000 people [26]with their property threshold has been established by NBRO for
due to the rapid onset that they possess. hourly and daily rainfall which ignores the
Similarly in Sri Lanka the landslide event on antecedent rainfall but studies have proven
29thof October 2014 at Meeriyabeddawatta, in with more antecedent rainfall the current
Haldumulla area at Badulla District, shocked threshold to trigger landslides may reduce [6].
the country affecting nearly 1875 people of 521 The recent scenario occurred at Meeriyabedda
families, killing 13 people and destroying 66 had the highest cumulative rainfall over a
houses in the area.[32]. period of more than one week proving us the
Rainfall is one of the recognized triggers of need of having a warning method based on the
landslides and the fact that Sri Lanka is a antecedent rainfall as well.
tropical country with a mean annual rainfall
varying from under 900mm in the dry zone to
over 5000mm in the wet zone[28]increases the
susceptibility for landslides occurrence. With
heavy rainfall the ground water level is
increased through infiltration [5]and as a result Eng. L. R. Abhayarathne, AMIE(SL ), AMSSE(SL),
it increases the pore water pressure thus B.Sc. Eng. (Moratuwa), M.Sc. (Colombo), Civil Engineer,
State Engineering Corporation of Sri Lanka
leading to more landslides[5].This high and
unpredictable precipitation in the wet zone has
1 ENGINEER
Current (August 2017) regional landslide  Steady state model ([7])
thresholds established by NBRO considers only  Transient “piston-flow” model ( [14])
the daily rainfall as follows,  Transient diffusive model ([10])
 Alert Level 75mm/day
 Warning Level 100mm/day These Process-based models can be used to
 Evacuation Level 75mm/hour or determine the amount of rainfall needed to
150mm/day generate slope failures, the location of the slope
failure and the time of occurrence but with
These thresholds have been defined mainly limitations,
based on the practical experience supported by  They need detailed spatial information
statistical analysis and are helpful to issue on the morphology, soil characteristics,
regional/general warnings to the whole hydrological and lithological data that
country. However, they do not consider the will control the initiation of landslides
influence of antecedent rainfall thus making  The difficulty of collecting this type of
them neither adequate nor accurate to provide information accurately over large areas
a proper warning. Therefore a study to define and its rarity because of the lack of
antecedent rainfall thresholds for local testing equipment
situations is immensely helpful for precise  These models can be used in such
landslide predictions. events where location and time with
precipitation measurements are known
2. Literature Review of slope failures but such information is
not commonly available and costly to
A threshold could be defined as the minimum obtain
or maximum value needed for a certain
phenomenon, condition, result or action to As a result of such limitations, physically based
occur. The minimum threshold will indicate models are more suitable to forecast shallow
the lowest level while the maximum threshold landslides (debris flows and soil slides), but not
will indicate the highest level of the quantity deep-seated landslides[6].
needed and for a rainfall induced landslide a A different approach was adopted in researches
rainfall threshold will indicate the minimum done by Crozier (1999)[14] and Glade et al.
rainfall required to trigger a landslide. (2000)[15] where an antecedent soil water status
There were two main types of thresholds (ASWS) model was developed and linked with
defined in literature as physical based and the soil moisture conditions to predict the
empirical based. occurrence of landslides. It was a simplified
 Physical based (process-based, conceptual model which could estimate soil
conceptual) ([7]; [8]; [9]; [11]) moisture daily. In addition the model
 Empirical (historical, statistical) based ( performed a soil water balance which took
[12]; [1]) excess precipitation over a time period up to
the day before the landslide occurred and
2.1 Physical Based Models calculated the drainage through the decay
This type of thresholds are based on slope function for the evapotranspiration and loss of
stability models and used extensively in water through drainage e.g. by analysing
geotechnical engineering ([9]).In order to link hydrograph recession curves (Glade et al.,
the past slope stability data with the rainfall 2000)[15]. Besides this Crozier (1999)[14]by
patterns process based models use infiltration using a calibrated ASWS model was able to
models (e.g. [13]). For an example the “leaky successfully forecast days with landslides and
barrel” model proposed by Wilson (1989)[2] days without landslides for a period of 8
was used to forecast the accumulation of months in 1996 though it was not implemented
infiltrated water into the ground. as a landslide warning system [1].
Further this model was used by Wilson and 2.2 Empirical Based Models
Wieczorek (1995)[8] to predict debris flow This type of rainfall thresholds are defined by
occurrence at La Honda, in the San Francisco collecting rainfall data of landslide events and
Bay region. Apart from this, Crosta and Frattini drawing lower bound lines to rainfall events
(2003)[11] also compared three infiltration that resulted the landslides in semi-logarithmic,
models given below to forecast the geography logarithmic or Cartesian coordinates [6]. The
and time of debris flows in the Lecco Province, thresholds are drawn manually with lesser
in northern Italy, calculations. When information on rainfall
ENGINEER 2
conditions that did not triggerlandslides were Among them the intensity-duration thresholds
available (e.g. [22]), the thresholds would act as were the most popular type of threshold used
a clear cut separators for rainfall events that in literature and they had the form of,
resulted and did not resulted in landslides.
Where: I (mean) rainfall intensity, D is rainfall
Apart from this since the variables change like duration, and c ≥ 0, α, and ß are parameters.
rainfall, soil type, slope angles etc. with
geography, the same threshold couldn‟t be 2.2.2 Thresholds that consider the
taken to all locations, therefore thresholds antecedent conditions
would differ with geography.
Soil Moisture and Ground water conditions
When rainfall threshold is defined, the key affect the occurrence of landslides ([5]).
indicator associated with it is the rainfall However, the above parameters are hard to
intensity. The rainfall intensity is the rainfall measure or predict since they will change
accumulated over a period of time which is abruptly with various factors such as
usually measured in mm/h. However hourly, geological, morphological, soil condition and
daily or even longer periods of rainfall had properties, and temperature variations.
been used depending on the time period Antecedent precipitation has a direct impact on
considered for the analysis. When the the ground water levels and soil moisture;
observation period is long it only gives us the therefore it can be used to predict the initiation
average rainfall for that period, neglecting the of landslides. Here, the simplest way to
peak intensities and most of the studies use this establish a threshold is to consider the
mean rainfall rather than the peak intensities. antecedent rainfall. For an example Govi et al.
(1985)[17] found that the minimum 60 day
Further the empirical thresholds can be cumulative rainfall amount required to initiate
categorized in accordance to the geological landslides in Piedmont region of Italy was
boundaries considered such as the local, 300mm and when the daily rainfall was
regional or global thresholds. All these compared within the 60 days, at least 140mm
thresholds will give us the minimum level of rainfall was required to initiate a slope failure.
rainfall below which landslides do not occur.
. In literature there are a number of periods
Based on the rainfall measurements, the considered,
empirical thresholds can be divided in to three  3 day period - Kim et al. (1991)[18]
groups [6],  4day period - Heyerdahl et al.
a) Thresholds that use event rainfall (2003)[23]
measurements  10day period Crozier (1999)[14]
b) Thresholds that consider the antecedent  7, 10 and 15 days period Aleotti (2004)
conditions [12]
c) Other thresholds  18day period (3-day event rainfall and
15-day antecedent rainfall) Chleborad
2.2.1 Thresholds that use event rainfall
(2003)[3]
measurements
 2, 5, 15 and 25 day periods Terlien
In event rainfall measurements rainfall from a (1998)[19]
single or several events were used even though  1 to 59 day period De Vita (2000)[20]
they did not cause landslides. This threshold
While many authors and researchers accept the
can be subdivided in to four groups,
fact, a handful of authors have challenged the
1. Rainfall intensity – duration (ID)
importance of the antecedent rainfall for the
thresholds
initiation of landslides. Among them,
2. Rainfall event – duration (ED)
followings are prominent.
thresholds
3. Rainfall event – intensity (EI)  Brand et al. (1984)[24] could not find a
thresholds correlation in the antecedent rainfall
4. Thresholds based on the total event and slope failure events due to high
rainfall (“Total event rainfall rainfall intensity in Hong Kong
thresholds” or “Rainfall Total event  Corominas and Moya (1999) [16]found
thresholds”) that due to large inter-particle voids in
slopes covered by coarse debris had a
tendency to create debris flow without
3 ENGINEER
the influence of antecedent rainfall in However at the National Building Research
the Pyrenees Organization (NBRO)web site there were fewer
 Corominas (2000)[22] investigated the incidents reported when compared to the
possibility of triggering of shallow Ministry of Disaster Management website.
landslides due to the existence of large Even in this data the type of landslide was not
macro pores on slopes mantled by clearly mentioned in all of the entries. Therefore
impervious soils without the effect of the landslide details were extracted from the
antecedent rainfall Ministry of Disaster Management website.
 Aleotti (2004)[12] could not establish a
relationship between cumulative and After collecting the landslide data the rainfall
critical rainfall and the initiating of data for the relevant locations and dates were
landslides in the Piedmont region in collected from the Meteorological Department
Italy of Sri Lanka. The Meteorological Department
has stations of its own plus independent
2.2.3 Other thresholds stations functioning in estates as well. Out of
Apart from the above two main threshold types 225 landslide data collected, 51 landslides
there has been other threshold types defined, occurred at Kalawana Divisional Secretariat
 Reichenbach et al. (1998) [11] related Division of Ratnapura (22.67%), 32 landslides
the discharge measurements of occurred at Imbulpe Divisional Secretariat
different gauging stations in the Tiber Division (14.22%) and in Ratnapura Divisional
River basin, in central Italy to the Secretariat Division of Ratnapura District had
occurrence (or non-occurrence) of only 21 events (9.33%).
landslides and relevant relationships
for the flood volume and maximum
mean daily discharge Table 1 - Landslide Events in Divisional Level
 Ayalew (1999)[21] investigated the of Ratnapura
possibility of landslides in Ethiopian
Plateau and found that it had a Divisional
correlation with the number of days Secretariat Landslide
with rainfall more than 5mm and the Division Events Percentage %
antecedent rainfall up to the date of Kalawana 51 22.67
landslide occurrence Imbulpe 32 14.22
Ratnapura 21 9.33
3. Research Methodology
Balangoda 17 7.56
3.1 Data Collection Eheliyagoda 17 7.56
In order to find the rainfall threshold for the Kolonna 17 7.56
initiation of landslides in Ratnapura district a
Opanayaka 11 4.89
database was built of rainfall events that
resulted landslides in the Ratnapura district. Nivithigala 10 4.44
Weligepola 9 4.00
For this purpose landslide data was obtained Elapatha 8 3.56
from the database maintained by Ministry of
Kiriella 8 3.56
Disaster Management (Disinventar.lk) where
225 landslide incidents were obtained which Pelmadulla 8 3.56
had occurred in Ratnapura district from Kuruvita 6 2.67
13/05/1978 to 02/06/2014. The problem was Ayagama 5 2.22
the difficulty in finding the exact location of the
Kahawatta 3 1.33
landslides in the database but it provided the
division which the landslide occurred in the Godakawela 2 0.89
Ratnapura district. From literature and by
using the database the number of deaths, The rainfall of the exact location the landslide
injured and missing people, houses destroyed occurred was not available since most of the
and damaged, victims and people affected, locations were not under observation as
families affected and evacuated were found but landslide prone areas at the moment of
Geology and type of landslide was not landslide so automatic rain gauges were not
mentioned in the detail. available. Therefore the rainfall measurements
of stations close to the landslide event were
used.
ENGINEER 4
Apart from this rainfall data could not be Table 2 - Summary of Thresholds to be
obtained up to the exact moment that the analysed
landslides occurred. The rainfall obtained was
the daily rainfall on the date that the landslides Threshold Data Used Code
happened. Hourly rainfall was available for the One Day Rainfall data of the day
landslides occurred near the Ratnapura Threshold landslide occurred E1
Meteorological office but the daily rainfall was Cumulative rainfall Data
used so a common basis could be maintained of the day and the
for all the landslide events. Two Day previous day the
Threshold landslide occurred E2
3.2 Data Analysis Cumulative rainfall Data
The 225 landslide incidents were used as seven of the day and the
subsets to find the 7 day cumulative threshold Three Day previous 2 days the
Value, 6 day cumulative threshold Value, 5 day Threshold landslide occurred E3
cumulative threshold Value, 4 day cumulative Cumulative rainfall Data
threshold Value, 3 day cumulative threshold of the day and the
Value, 2 day cumulative threshold Value and 1 Four Day previous 3 days the
day threshold Value. Threshold landslide occurred E4
Cumulative rainfall Data
Landslide incidents where rainfall could not be of the day and the
found and the landslide events that did not Five Day previous 4 days the
result from rainfall were neglected. This was Threshold landslide occurred E5
achieved by ignoring the addition of rainfall Cumulative rainfall Data
events less than or equal 5mm to the of the day and the
cumulative rainfall and ignoring the rainfall Six Day previous 5 days the
events with a cumulative rainfall less than or Threshold landslide occurred E6
equal to 5mm. Cumulative rainfall Data
of the day and the
Afterwards the calculation of cumulative Seven Day previous 6 days the
rainfall for the threshold events was completed Threshold landslide occurred E7
as below,
Then the cumulative rainfall data was arranged
in the ascending order and a frequency
distribution was prepared to find the rainfall
linked with the probability. After the
probability was found the correlation of each
daily rainfall with the cumulative rainfall up to
that particular day was investigated. Then
using SPSS (IBM Statistic Software) a
curvilinear Regression model was created for
each threshold value using the probability
found as the independent variable and the
rainfall linked to it as the dependent variable in
R = α eβt ...(1)
Where „R‟ denotes rainfall in millimetres, „t‟
denotes probability of landslide occurrence and
„α‟ and „β‟ are constants.
4. Results and Discussion
The curvilinear Regression Model obtained

using SPSS for one day threshold value is
shown in Figure 1 and two day threshold value
is shown in Figure 2. Where „R‟ is the rainfall in
millimetres and „t‟ is the probability (%). The
summary of all the curvilinear regression
models developed is shown in Figure 3.
5 ENGINEER
Rainfall (mm)
Rainfall (mm)
Figure 1 - Probability Distribution Curve for E1 Figure 2 - Probability Distribution Curve for E2
600
E1
500
E2
E3
400
E4
Rainfall (mm)
300
E5
E6
200
E7
100
0
0 20 40 60 80 100
Probability %
Figure 3 - Summary of Thresholds
ENGINEER 6
Table 3 - Summary of Thresholds For the determination of acceptable
thresholds, the acceptable level of risk which
Threshold Probabilistic
was linked to the probability of the occurrence
Threshold For Rainfall
of landslides was needed to be found.
One day threshold
According to UNDP definitions the acceptable
(E1)
amount of risk was defined as “the level of
Two day threshold
losses that is acceptable without destroying
(E2)
lives, national economy or personal
Three day
finances”[4]. Since the financial losses
threshold (E3)
occurred due to all the landslide events was
Four day threshold not available, the loss of human life due to
(E4) landslide in Ratnapura was referred. The
Five day threshold number of landslide events that caused loss of
(E5) life was 36 and out of them 10 events occurred
Six day threshold due to rainfall less than 75mm on the date of
(E6) the landslide. If the existing threshold was
Seven day lowered to at least to 40% probability then the
threshold (E7) number of events would have dropped to 7 in
The rainfall thresholds for Ratnapura District addition to this the consideration of the
found were built by building a curvilinear antecedent rainfall threshold is important
regression model between the probability since from the study it is very clear that there
(found through frequency distribution) and is a probability of landslide occurrence in the
the respective rainfall. Since the curves next few days due to rainfall occurred on the
showed the shape of an exponential previous day therefore the results of the 40%
distribution the curvilinear regression model probability are given in Table 5 below but
was adapted to interpret the results. there is a practical problem in lowering the
threshold due to the issuing of alerts more
However from E4 to E7 the thresholds were frequently. This could be a nuisance to the
decreasing this could not happen because the public making the alerts less effective.
thresholds should increase when the rainfalls Therefore the thresholds established by NBRO
were cumulating and the reason behind this can be agreed upon and further the new
was the outliers that add up to the frequency thresholds E2 and E3 can be utilized to give
distribution when number of rainy days were warning based on antecedent rainfall.
increasing. In the analysis only events less
than 5mm or cumulative events less than 5mm Table 5 - Proposed Alert Levels for 40%
were ignored but this was not sufficient when Probability
for 4 day cumulative rainfall and above. Rainfall in (mm)
Therefore the results in this analysis are not E1 E2 E3
valid from four day cumulative threshold (E4) Alert Level 38.57 62.59 68.88
to seven day cumulative threshold (E7). Warning Level 66.19 105.81 116.43
Comparison of results with the current
Evacuation
thresholds established by NBRO is shown in
Level 94.87 150.15 165.23
Table 4.
Table 4 - Comparison with Current Following this antecedent rainfall, the effect of
Thresholds it on the future landslides falls under the
interest of this analysis therefore the
Probability from
correlation of the rainfall received on each day
NBRO Warning established models
(previous day, 2 days back, 3 days back etc.)
and Alert Levels
(E2) (E3) with the cumulative rainfall was investigated.
(E1)
The summary of the results are given in „Table
Alert 75mm 58.4 6‟ and it was evident that only the rainfall of
45.17 42.43
Level /day 7 the previous day had a significant effect on the
100m cumulative rainfall of the landslide. The
Warning 66.4
m/da 53.39 50.65 reason behind this observation was the high
Level 6
y intensity rainfall received due to the
150m prevailing tropical climate.
77.7
Evacuatio m/da 64.97 62.24
2
n Level y
7 7 ENGINEER
Table 6 - Correlation Result Summary evacuation messages based on hourly basis
would be possible for the public. Furthermore
Pearson‟s hourly rainfall data would enable us to create
Coefficient hourly thresholds which would be effective
Correlation 'r' against high intensity rainfalls. Recently
Cumulative Rainfall in NBRO has started collecting hourly rainfall
rainfall up to in the data with automated rain gauges which
previous day of previous would pave the way to establish better
landslide day 0.65 thresholds on hourly basis. In turn great
Cumulative Rainfall improvement can be expected when
rainfall up to 2 prior to 2 responding to high intensity rainfall events
days prior to days before which occur frequently in this region.
landslide landslide 0.22
Cumulative Rainfall The fact that continuous rainfall over a period
rainfall up to 3 prior to 3 of time increases the antecedent moisture
days prior to days before content in soil which leads to a higher
landslide landslide 0.21 probability of landslides could not be ignored.
Cumulative Rainfall Therefore continuous monitoring is required
rainfall up to 4 prior to 4 for this type of study since new data inputs
days prior to days before will give more accurate results for future
landslide landslide 0.15 predictions. With the ongoing worldwide
Cumulative Rainfall climate change future studies are essential to
rainfall up to 5 prior to 5 establish better thresholds.
days prior to days before
landslide landslide 0.34 Acknowledgement
Cumulative Rainfall
I wish to thank my supervisor Dr. Gamini
rainfall up to 6 prior to 6
Jayathissa for his guidance and support to
days prior to days before
successfully complete my research work. Also
landslide landslide 0.13
I would take this opportunity to thank all the
staff members who helped me and guided me
Similar results were observed by Brand et al.
to complete my research work as well as
(1984)[25] for Hong Kong region where
academic work specially Dr. Rohan Perera, Dr.
establishment of a correlation between the
Mahesh Edirisinghe, Mr. Niranga Alahakoon,
antecedent rainfall and the occurrence of slope
Mr. N. C. Perera and Ms. Kushani De Silva.
failures failed.
5. Conclusions References
1. Wieczorek, G. F. and Glade, T., “Climatic
It could be agreed that the one day rainfall Factors Influencing Occurrence of Debris
thresholds ( ) that found in this Flows,” in Debris-flow Hazards and Related
analysis was matching with the practical Phenomena, Berlin, Heidelberg: Springer Berlin
thresholds established by the NBRO and in Heidelberg, 2005, pp. 325–362.
addition to that the study was successful in
2. Wilson, R. C., Torikai, J. D. and Ellen, S. D.,
producing thresholds for two day cumulative
“Development of Rainfall Warning Thresholds
rainfall and three day for Debris Flows in the Honolulu District,
cumulative rainfall . In the Oahu,” Dept. of the Interior, U.S. Geological
analysis it was evident that when it comes to Survey, 1992.
tropical countries like Sri Lanka most
3. A. F. Geological Survey (U.S.), “Preliminary
landslides were triggered by the high intensity
Evaluation of a Precipitation Threshold for
rainfall despite rare incidents like Anticipating the Occurrence of Landslides in
Meeriyabedda therefore only the rainfall of the Seattle, Washington, area,” U.S. Geological
the previous day has a significant effect on the Survey Publication, 2003.
cumulative rainfall for initiation of landslides.
4. UNDP, “Disaster Risk Assessment,” New
In addition to issuing of warning messages
York, 2010.
based on probability, a mechanism to report
the hourly rainfall using automated rain 5. Wieczorek, G., “Landslide Triggering
gauges would be very useful. Only then the Mechanisms,” Landslides Investig. Mitig., no.
process of issuing alert, warning and 247, pp. 76–90, 1996.
ENGINEER 8
6. Campbell, R. H., Soil Slips, Debris Flows, and 19. Terlien, M. T. J., “The Determination of
Rainstorms in the Santa Monica Mountains and Statistical and Deterministic Hydrological
Vicinity, Southern California, U.S. GEOLO. Landslide-Triggering Thresholds,” Environ.
Washington DC: U.S. Government Printing Geol., vol. 35, no. 2–3, pp. 124–130, Aug. 1998.
Office, 1975.
20. De Vita, P., “Fenomeni di
7. Guzzetti, F., Peruccacci, S., Rossi, M., and instabilitàdellecoperturepiroclastichedeiMonti
Stark, C. P., “Rainfall Thresholds for the Lattari, di Sarno e di Salerno (Campania)
Initiation of Landslides in Central and edanalisideglieventipluviometricideterminanti,
Southern Europe,” Meteorol. Atmos. Phys., vol. ” Quad. di Geol. Appl., vol. 7, no. 2, pp. 213–235,
98, no. 3, pp. 239–267, 2007. 2000.
8. Montgomery, D. R., and Dietrich, W. E., “A 21. Ayalew, L., “The Effect of Seasonal Rainfall on
Physically Based Model for the Topographic Landslides in the Highlands of Ethiopia,” Bull.
Control on Shallow Landsliding,” Water Eng. Geol. Environ., vol. 58, no. 1, pp. 9–19, Aug.
Resour. Res., vol. 30, no. 4, pp. 1153–1171, Apr. 1999.
1994.
22. Corominas, J., “Landslides and climate,” in
9. Wu, W. and Sidle, R. C., “A Distributed Slope VIII international symposium on landslides,
Stability Model for Steep Forested Basins,” Cardiff, UK, 2000, p. CD_ROM. 2000.
Water Resour. Res., vol. 31, no. 8, pp. 2097–2110,
Aug. 1995. 23. Heyerdahl, H., Harbitz, C., Domaas, U., and
Sandersen, F., “Rainfall Induced Lahars in
10. Iverson, R. M., “Landslide Triggering by Rain Volcanic Debris in Nicaragua and El Salvador:
Infiltration,” Water Resour. Res., vol. 36, no. 7, practical mitigation,” in Proceedings of
pp. 1897–1910, Jul. 2000. international conference on fast slope movements—
prediction and prevention for risk mitigation, 2003,
11. Crosta, G. B., and Frattini, P., “Distributed pp. 275–282.
Modelling of Shallow Landslides Triggered by
Intense Rainfall,” Nat. Hazards Earth Syst. Sci., 24. Brand, E. W., Premchitt, J., and Phillipson, H.
vol. 3, no. 1/2, pp. 81–93, 2003. B., “Relationship between Rainfall and
Landslides in Hong Kong,” in Proceedings of the
12. Aleotti, P., “A Warning System for Rainfall- 4th International Symposium on Landslides, 1984,
Induced Shallow Failures,” Eng. Geol., vol. 73, vol. 1, pp. 377–384.
no. 3, pp. 247–265, 2004.
25. Brand, E., “Landslides in Southeast Asia: A
13. Heber Green, W. and Ampt, G. A., “Studies on State-of-the-Art Report,” in Proceedings of the
Soil Phyics.,” J. Agric. Sci., vol. 4, no. 01, p. 1, 4th International Symposium on landslides, 1984.
May 1911.
26. BBC, “BBC News - Afghanistan Landslide Day
14. Crozier, M. J., “Prediction of Rainfall-Triggered of Mourning declared,” 2014. [Online].
Landslides: A Test of the Antecedent Water Available: http://www.bbc.com/news/world-
Status Model,” Earth Surf. Process. Landforms, asia-27273684. [Accessed: 10-Oct-2014].
vol. 24, no. 9, pp. 825–833, Aug. 1999.
27. Ministry of Disaster Management, “DMC-SL
15. Glade, T., Crozier, M., and Smith, P., (Disaster Management Center - Sri Lanka),”
“Applying Probability Determination to Refine 2014. [Online]. Available:
Landslide-Triggering Rainfall Thresholds http://www.dmc.gov.lk/index_english.htm.
Using an Empirical „Antecedent Daily Rainfall
Model,‟” Pure Appl. Geophys., vol. 157, no. 6–8, 28. Department of Meteorology Sri Lanka,
pp. 1059–1079, Aug. 2000. “Climate in Sri Lanka,” 2012. [Online].
Available:
16. Corominas, J., and Moya, J., “Reconstructing http://www.meteo.gov.lk/index.php?option=
Recent Landslide Activity in Relation to com_content&view=article&id=106&Itemid=8
Rainfall in the Llobregat River basin, Eastern 1&lang=en. [Accessed: 10-Oct-2014].
Pyrenees, Spain,” Geomorphology, vol. 30, no. 1,
pp. 79–93, 1999.
17. Cannell, R. Q., Belford, R. K., Blackwell, P. S.,

Govi, G., and Thomson, R. J., “Effects of
Waterlogging on Soil Aeration and on Root
and Shoot Growth and Yield of Winter Oats
(Avenasativa L.),” Plant Soil, vol. 85, no. 3, pp.
361–373, Oct. 1985.
18. Kim, S., Hong, W., and Kim, Y., “Prediction of

Rainfall-Triggered Landslides in Korea,”
Landslides, vol. 2, pp. 989–994, 1991.
9 9 ENGINEER
10
AnnualSessions
IESL,pp.
pp.[11[1- -16],
10],2017
2017
© The
© The Institution
Lanka
Experimental Study on the Effect of Presence of Deep

Embedded Retaining Walls on the Settlement of
Shallow Foundations
M.P. Amarasinghe and L.I.N. De Silva
Abstract: Most of the foundation designs are conservative, as the failure in foundation will
result in failure of the structure. The effects of presence of deep embedded retaining walls on the
settlement characteristics of shallow foundations have been investigated to find out whether the effect
from embedded retaining walls can be used for an economical design. With the intention to
investigate on that behaviour, an experiment on a scaled model was carried out as the field
experiments are expensive, complex and time consuming. The behaviour of load with the depth of
embedment and the distance to the retaining wall from footing edge was investigated. In the scaled
model the load at 25mm settlement increased with the depth of embedment and decreased with the
increase of the gap. It was concluded that in shallow footings confined with a deep embedded
retaining wall the capacity against settlement failure of the soil can be increased. But when the gap
between footing edge and retaining wall is high this effect was minimized.
Keywords: Shallow foundations, Deep embedded retaining walls, Settlement
1. Introduction are time consuming, expensive and complex.

When designing a shallow foundation the
Foundations are used to transfer the loads of bearing capacity and the settlement of it has to
structures to the ground. A good foundation be considered. Settlements can be either
design can extend the lifetime of a structure. immediate or consolidation[3].
Raft foundation is a concrete slab extended over
a large area reducing the contact pressure Not much is known on the effect of deep
supporting many concrete columns, beams and embedded structures on settlement of shallow
slabs, widely used in constructing buildings foundations. It is planned to analyse and
with basements[1]. Often excavation of great identify whether there is a relationship between
depths is required for raft foundation load immediately below the footing and
construction therefore support from embedded settlement of footing with the change of depth
retaining walls is required. of embedment of retaining wall. And the
relationship between load and the gap between
Embedded retaining walls are used in many retaining structure and foundation if footing is
civil engineering applications for embankment confined by a deep embedded retaining wall.
stabilization, basement constructions, in
underground passes and for waterfront 2. Literature Review
projects. Embedded retaining walls obtain
lateral support from penetrating into the It has been found that the “Thickness of the raft
ground and may also be supported by has a little influence on the load settlement
structural members. During early days aspects. However increasing the raft thickness
embedded retaining walls were constructed reduces the differential settlement and imposed
with either soldier piles or steel sheet piles. load”. From a 3D numerical analysis it was
Later concrete walls were formed using various found that most critical total vertical settlement
techniques such as slurry trenches, contiguous has occurred at the centre of the raft followed
or secant piles and diaphragm walls[2]. by corner point and at centre of the edge
respectively [1].
The purpose of this research is to analyse the
effect of presence of deep embedded retaining
Eng. M.P. Amarasinghe, B.Sc.Eng(Moratuwa)
structures for the settlement characteristics of a
shallow rigid foundation using experimental Eng. (Dr.) L.I.N. De Silva, B.Sc.Eng(Moratuwa), M.Eng
(Tokyo), Ph.D. (Tokyo), MIE(Sri Lanka), Senior Lecturer,
techniques. Actual applications of deep Department of Civil Engineering, University of Moratuwa.
excavations, shallow foundation construction
confined with deep embedded retaining walls
1
11
It was found that there is an effect from the
installation of diaphragm walls on the 300mm
surrounding and adjacent buildings. “Panel
length is the most affecting factor of ground
movement and lateral stress reduction during
panel installation”[4].
According to [5]“earth retaining structures are

influenced by the construction procedure,
rigidity of support system and the sequence of
excavation. Attributes of the wall, such as
height and depth of penetration have a greater
influence on the wall movements.”
Figure 1 –Timber boxes
Maximum ground surface settlement takes
A large container was filled with sand up to
place near the centre of the excavation and
100mm. In filling the container the falling
settlements at the corners are smaller.Wall
height of sand was kept constant to keep the
deformation and ground movements are
density of sand uniform.
smaller close to the wall corner than around the
wall centre[6].
It can be seen that the load–settlement

behaviour of an unpiled raft is almost linear
and the stiffness of the raft on loose sand is
much lower than that on dense sand[7].
3. Objectives
 Experimental investigation on the effect of

deep embedded retaining structures on the
settlement of shallow foundation using a
physical model. Figure 2 - Timber box placed inside the
 Propose recommendations on the strength container
enhancement and settlement reduction due
to the presence of deep embedded retaining Then the 300mmx300mm timber box was kept
structure. inside, levelled and filled with sand for another
500mm from outside and 300mm from inside.
4. Methodology After that timber plate was kept at the centre of
After doing a thorough literature review an the box. Because of the practical difficulty in
experimental procedure is adopted using a placing the dial gauge on the timber plate few
model shallow foundation confined by a deep iron plates were first kept on the plate as loads
embedded retaining wall. and then using hydraulic jack those plates were
loaded.
5. Experimental Procedure
An experiment using a large scale physical

model is very difficult. Therefore, an
experimental procedure using a scaled model
of a shallow foundation confined with a deep 2 0 0 mm
embedded retaining wall was used. Two timber
boxes of 300mm*300mm and 500mm*500mm of
height 600mm were used as the retaining walls.
A timber plate of 200mm*200mmwas used as
the footing.
Figure 3 - Timber plate of 200mm*200mm
placed at centre
2
12
the timber box in 100mm steps. The same
procedure was followed for the 500mm*500mm
timber box and readings were obtained
changing the depth of embedment in 100mm
step increments starting from a depth of
300mm. The depth of excavation was kept at a
constant value of 200mm in all the cases. Finally
readings were obtained for the case where
there’s no embedded retaining wall present.
Table 1- Summary of the experimental

Figure 4 - Filling the container at a constant procedure
height
Size of the Depth of Depth of
timber box embedment (d) excavation
(h)
100mm
300mmx300mm 200mm 200mm
300mm
100mm
500mmx500mm 200mm 200mm
300mm
Figure 5 - Placing the steel plate loads
Figure 7 – Experimental setup
Table 1 summarizes the scenarios in the

Figure 6- Final arrangement of the experiment experimental setup where,
h- Depth of excavation
Number of divisions proving ring deflected for d- Depth of embedment
every 2mm deflection in dial gauge was B- Width of the timber plate
measured and converted to a force using L- Width of the timber box
calibration data. As there was an initial
settlement due to the load from iron plates that
6. Experiment results
load and settlement must be added to the load
and deflection. Therefore measured the mass of
After carrying out the experiment on a small
the steel plates, hydraulic jack and timber plate
scale model results were obtained. In the
and graphically calculated the load that would
experiment for 2mm settlements the load
result in a 25mm settlement. The soil unit
applied was calculated from the proving ring
weight was calculated using the proctor mould.
readings using calibration charts. Here it was
Proctor mould was filled the same way in
assumed that the soil is homogeneous and
which the container was filled and from same
drained. Settlement that was considering is the
height. Then the weight of sand that occupy the
immediate or elastic settlement as sand was
proctor mould was found and specific weight
used where pore water dissipation is
was calculated.
immediate.
Similarly a series of experiments were carried
out changing the depth of embedment by lifting
3
13
Results were obtained changing the depth of Then at 25mm settlement, the value
embedment and the distance to the wall from corresponding to the load was obtained.
plate edge (i.e. size of the timber boxes). There The results of the experiment for the two timber
was an initial settlement due to the steel plates boxes of size 300mmx300mm and
loaded on the timber plate. Therefore in order 500mmx500mm are shown in Figure 8 and
to get the actual settlement the settlement due Figure 9 respectively with the change of depth
to the load from steel plates and hydraulic jack of embedment of the two timber boxes. Similar
must be added. This settlement values were pattern was observed in the settlement-stress
calculated using a graphical approach curves in both the cases of L=300mm and
assuming the immediate settlement have a L=500mm. Settlement increasing linearly as the
linear relationship with the load applied. This stress applied increased and when the depth of
linear relationship is identified in[1],[7],[8]. embedment increased the stress required to
settle by 25mm was high.
0
0 50 100 150 200 250 300 350 400
-5
-10
Depth of
-15 embedment
Settlement (mm)
300mm
-20
200mm
-25
100mm
-30 0mm
-35
-40
-45
Stress (kN/m2
Figure 8 - Settlement- Load chart for L=300 with the variation in depth of embedment
0
0 50 100 150 200 250 300 350 400
-5
-10
-15 Depth of
Settlement (mm)
embedment
-20 300mm
200mm
-25
100mm
-30
0
-35
-40
-45
Stress (kN/m2)
Figure 9 - Settlement- Load chart for L=500 with the variation in depth of embedment
4
14
400
350
300
250 Timber box

Stress (kN/m2)
dimention
200
300mm*300mm
150 500mm*500mm
100
50
0
0 50 100 150 200 250 300 350
Depth of embedment (mm)
Figure 10 - Stress- depth of embedment for different timber box dimensions
400
350
300
Depth of
250 embedent
Stress (kN/m2)
0mm
200
100mm
150 200mm
300mm
100
50
0
0 20 40 60 80 100 120 140 160
Distance to the wall from plate edge (mm)
Figure 11 - Stress- distance to the wall from plate edge for various depth of embedment
According to the small scale experimental the size of the timber box used. That means as
model the stress required to settle the plate by the distance from wall to plate edge increases
25mm is increasing as the depth of embedment the required stress to cause a settlement of
increased as discussed. It can be seen that the 25mm is low.
stress required increases with the decrease in
5
15
7. Conclusion 8. Terzaghi, K. Peck, R. B. and Mesri, G. “Soil
Mechanics in Engineering Practice, Third
In this study, the effects of presence of deep Edition,” Wiley-Interscience Publ. John Wiley Sons,
embedded retaining walls on the settlement Inc., p. 664 pp., 1996.
characteristics of rigid shallow foundations
have been investigated. An experiment was
carried out on a scaled model and results were
obtained. It was observed that the load at
25mm settlement increases with the depth of
embedment as well as when the wall is closer to
the shallow foundation.
The reason for the decrease in stress at 25mm

settlement as the gap increase is the effect of
embedded wall on the slip surface and on stress
bulb. As the gap increased the effect was
minimized.
It can be concluded that for rigid shallow

footings confined with a deep embedded
retaining wall, the capacity against settlement
failure of the soil can be increased but as the
distance between the retaining wall and the
footing edge increases the capacity against
settlement decreases.
References
1. Leong T. K. and Huat, C. S. “Sustainable design

for unpiled-raft foundation structure,” in
Procedia Engineering, 2013, vol. 54, pp. 353–364.
2. Simpson, W. Brian; Powrie, “Embedded

retaining walls: theory, practice and
understanding,” in 15th International Conference
on Soil Mechanics and Geotechnical Engineering,
2001.
3. Rajapakse, R.Geotechnical Engineering Calculations

and Rules of Thumb. 2008.
4.. Comodromos, E. M. Papadopoulou, M. C.and

Konstantinidis, G. K. “Effects from diaphragm
wall installation to surrounding soil and
adjacent buildings,” Comput. Geotech., vol. 53,
pp. 106–121, 2013.
5. Vaziri, H. H. “A simple numerical model for

analysis of propped embedded retaining walls,”
Int. J. Solids Struct., vol. 33, no. 16, pp. 2357–2376,
1996.
6. Wu, C. H. Ou, C. Y. and Tung, N. “Corner

effects in deep excavations-establishment of a
forecast model for taipei basin T2 zone,” J. Mar.
Sci. Technol., vol. 18, no. 1, pp. 1–11, 2010.
7. Nguyen, D. D. C. Jo, S. B. and Kim, D. S.

“Design method of piled-raft foundations under
vertical load considering interaction effects,”
Comput. Geotech., vol. 47, pp. 16–27, 2013.
6
16
AnnualSessions
IESL,pp.
pp.[17[1- -23],
10],2017
2017
© The
© The Institution
Lanka
Shrinkage Behavior of a Liner Material with Expansive

Properties for use in an Engineered Landfill
V. Kayaththiry
Abstract: Leachate from municipal solid waste landfills contains wide range hazardous
compounds; these hazardous compounds have been shown to migrate and contaminate the surrounding
environment and impair the use of groundwater. The use of modern landfill liner systems minimizes the
migration of hazardous material, it employs a composite liner consisting of a geomembrane overlying a
compacted clay liner or a geosynthetic clay liner, but the cost of the geomembrane in Sri Lanka is not
affordable. Therefore instead of geomembrane the usage of bentonite-amended expansive soil was
considered here and characteristics of the expansive soil were examined. Compared to geosynthetic clay
liners (GCL) and high density Polyethylene (HDPE) liners, compacted clay liners (CCL) are less
expensive, if the materials are locally available. It is possible to reduce the thickness by using less
permeable soil such soil obtained from Moragahakanda, Sri Lanka. This soil is found to be an expansive
soil and its hydraulic properties and Shrinkage behavior could be further improved by amending with
Bentonite. In addition to the hydraulic conductivity criteria, it is necessary to examine its shrinkage
behaviour during drying to investigate the sustainability of its use as a liner. In this examination three
different types bentonite-amended expansive soil with following percentage of bentonite content 0, 5 and
10% were analyzed and compared the results. Here analyze was done mainly for variation of moisture
contents, volume changes and characteristics of crack with time. For the above purpose following tests
were performed, specific gravity test, mechanical analysis, Atterberg limit test, compaction test and
shrinkage behavior test. Soil shrinkage and soil compaction represent the dynamics of soil structure with
the hydraulic and mechanical stress. An image processing method using ArcGIS 10.2 software was used
to determine area reduction and cracked area.
Keywords: Expansive soil, Landfill liner, Shrinkage, Image processing
1. Introduction major and long term issue is the generation of

leachate which is a liquid produced when water
or other liquids come into contact with waste or
In Sri Lanka, there was no any engineered
due to release of moisture present within itself
landfills have not been established until the
creates one of the major challenges in designing
establishment of “PILISARU” for disposal of
municipal solid waste. However, the need for and maintaining an engineered landfill.
engineered landfills cannot be ignored with the
increasing quantity of solid waste generation
due to rapid urbanization. Municipal solid waste
management is a major issue in urban areas. The
solid disposal collection is done properly but the
final disposal is a major problem in cities. In
Colombo disposal of waste is causing major
social and political issues. All existing disposal
sites are “dumps” as opposed to sanitary landfill
sites. As all know the incident of Figure 1 - Meethotaramulla landfill slide
Meethotaramulla landfill slide occurred in April
2017. The people were affected by the incident
who were living in the surrounded areas. The Eng. (Ms) V. Kayaththiry, B.Sc. Eng. (University of
Peradeniya), Civil Engineer at Central Engineering
officers who are related with waste management
Consultancy Bureau (CECB),
have been not worked properly. Hence the
17
Therefore, an essential component of an an earlier study as reported by Wanigaratne
engineered landfill is the liner system which acts et.al. ([2], 2012) in which the hydraulic
as a barrier for leakage of leachate and prevents conductivity was determined on specimens
the transportation of contaminants to the nearby prepared by consolidating one dimensionally to
water resources. Synthetic liners consisting of pressures of 46, 92and 184 kPa. Initially, soil
geomembranes or geosynthetic clay liners (GCL) classification tests such as liquid limit, plastic
are widely in use as liner materials of engineered limit, particle size distribution and specific
landfills constructed in developed and gravity tests were carried out to identify the
developing countries. However, as geosynthetic basic physical soil properties.
materials are not manufactured in Sri Lanka,
In this study, the shrinkage characteristics were
Compacted Clay Liners (CCL) is an alternative
investigated on the specimens having the same.
To be considered as a material for a liner.
However, in addition to meeting hydraulic Specimens of diameter 150 mm and thicknesses
conductivity criteria of a liner, one major of 5, 10and 20 mm were prepared using
challenge of using CCL as a liner material is the Moragahakanda expansive soil and they were
possibility of developing shrinkage cracks amended by 5 and 10% bentonite and
during the dry season. In this regard, use of consolidated to 46.2 and 92.4 kPa. Trimming of
expansive soils in CCL is advantageous as they compacted specimens to above thicknesses was
exhibit a very high plasticity index (PI) which made possible by using templates of ring shape
prolong the initiation of cracking upon drying of required heights (Figure 2) into which the
and have the characteristic of self-healing upon compacted samples were inserted during
wetting. Kurukulasuriya et.al. ([1], 2013) has extraction from the compaction mould.
investigated expansive soil obtained from Altogether, 18 specimens were prepared and
Moragahakanda in the Central Province of Sri placed on glass plates and left for air drying
Lanka and that hasamended by 5 and 10% (Figure 3).
bentonite (Bentofix®, NAUE GmbH & Co. KG,
Germany) by weight as candidate materials for a The change in area of flat surface exposed to the
liner and found that they satisfy short and long atmosphere and the development of cracks were
term hydraulic criteria. In this study, the monitored daily during the drying process. For
shrinkage behaviour of the same candidate this purpose, digital photographs were taken
materials obtained from Moragahakanda is using a 14 megapixel camera from a fixed height
investigated. under almost constant light and ambient
temperature conditions. This was facilitated by
2. Materials and Methods using a frame as shown in Figure 4. The
moisture content too was computed at the time
The expansive soil for the study was obtained of taking the photograph using the weight of the
from the same location (07˚ 35'32.80 N specimen obtained immediately after taking the
80˚49’59.90 E) from which soil was obtained for photograph.
Figure 2 - Circular shaped templates to prepare specimens of thickness (i) 5 mm, (ii) 10 mm and (iii) 20
mm
18
Figure 3 - Compacted specimens of candidate materials left for air drying
Figure 4 - Wooden frame used for photographing the compacted samples during air drying
These images were fed into GIS (ArcMap height. Therefore, it is necessary to calibrate the
version 10.2) software and the “IsoCluster output result given by the software with
Unsupervised Classification” tool was used to reference to the height at which the photograph
multivariate the image. Then, the “Extract by was taken. This was done by calculating the
Attributes” tool was used to select the sample areas of a square and rectangular shapes placed
from the image and the area was calculated by within a larger circle and photographing at the
converting the raster dataset to polygon features same height the where 5, 10and 20 mm thick
using the “Raster to Polygon” tool. specimens had been photographed (Figure 5).
However, the output result of the GIS software

will be valid for an image taken at a particular
A1 A1
A2 A2
Figure 5 - Image used for calibration of the GIS software computation (a) Object of known area
(b) Classification image
19
3. Results
3.1 Physical and Compaction Properties
Table 1 - Physical and Compaction Properties of Candidate Soils
Bentonite Soil M Soil M+5% Soil M+10%

Liquid limit (%) 600 * 66 49 * 70 *
Plastic limit (%) 55* 29 21 * 25 *
Plasticity index (%) 545* 37 28 * 50 *
Particle density 2.68* 2.59 2.57 2.55
Swell pressure (kPa) - 126.1* 267.7 * 357 *
Maximum Dry Density (Mg/m3) - 1.65 1.48 1.71
Optimum Moisture Content (%) - 18.0 26.8 14.0
*- after Wanigaratne et.al, ([2], 2012)
The results of Atterberg limits, specific gravity

and swelling pressure tests are given in Table 1 3.2 Shrinkage Characteristics
for Moragahakanda expansive soil (Soil M) and
An example of the transformation of digital
that amended by 5% (Soil M+5%) and 10% (Soil
images taken during the drying process by the
M+10%) Bentonite. The Moragahakanda soil was
GIS software to compute the surface and crack
classified as sandy Clay of Intermediate
area (where applicable) is shown in Figure 6.
Plasticity (CIS).
The computed value given by the GIS software
The preparation of compacted candidate soil was multiplied by the calibration coefficient
samples was carried out so that the void ratio of given in Table 3, which was obtained by the
the compacted soil sample corresponds to procedure described in the earlier section. As
consolidated under 46.2 and 92.4 kPa pressure. expected, the greatest calibration coefficient was
Table 2 gives the void ratio of the consolidated obtained when the object was furthest away
soil samples (Wanigaratne et.al. [2], 2012) and from the camera lens.
the corresponding values of dry density that
should be achieved on the compacted soil
samples used in this study.
Table 2 - Target Dry Density of Compacted Candidate Soil Samples
Soil M Soil M+5% Soil M+10%
Consolidation Pressure (kPa) 46.2 92.4 46.2 92.4 46.2 92.4
Void Ratio 0.74 0.612 0.78 0.729 0.576 0.502
Dry density (Mg/m3) 1.53 1.65 1.44 1.48 1.62 1.69
iv
20
(a) (b) (c) (d)
Figure 6 - Steps of image processing in ArcGIS Software, (a) Sample, (b) Classification image, (c)
Extracted image and (d) Rastered image
Table 3 - Calibration coefficients used to compute surface areas of compacted specimens
Calibration
Thickness of specimen (mm)
Coefficient
20 1.12
10 1.16
5 1.20
The reduction of moisture content is rapid in shrinkage take place during the drying stage of
thin specimens as well as in specimens whose the specimens does not reduce the area
initial void ratio is high as those correspond with significantly and once the equilibrium moisture
samples consolidated to 46.2 kPa. This is content is reached, the specimens continue to
understandable as thin specimens and those undergo contraction at the equilibrium water
specimens that correspond with low content. As the drying takes place, the large
consolidation pressure contained low amount of amount of water absorbed into the diffuse
moisture and all the specimens had more or less negatively charged layer in montmorillonite clay
same area of surface exposed to the atmosphere mineral present in Moragahakanda expansive
through which the exchange of water molecules soil and in Bentonite would be released, thus
and air could take place during the drying causing the reduction in the distance between
process. the layers.
Figure 7 shows the variations of percentage area Figure 8 shows that for all the 18 specimens,
reduction with moisture content of the although the reduction of surface area occurred,
specimens with an initial voids ratio no shrinkage cracks were developed. This
corresponding to that of specimens consolidated observation isvery important as during dry
under a pressure of 46.2 and 92.6 kPa. It is seen season the liner will lose moisture as the leachate
that although the moisture content is reduced generated is minimal, and it is very
gradually during the drying process of advantageous for the liner not to develop
specimens, most of the reduction of area took shrinkage cracks, although expansive soil has
place once the specimens reached the the characteristic of self-healing the cracks when
equilibrium moisture content. It appears that the wetted.
21
Figure 7 - Variations of percentage area reduction with moisture content of the specimens with an
initial voids ratio corresponding to that of specimens consolidated under a pressure of 46.2 kPa and
92.4kPa.
22
Figure 8 - One of the first day and last day same specimen images.
4. Conclusions to IESLAS2017, Reviewers for their assistance to

release this paper successfully.
This study was carried out to investigate the
shrinkage behaviour of candidate liner materials Reference
prepared using expansive soils from
Moragahakanda and that was amended by 5 and 1. Kurukulasuriya, L. C., Wanigarathna, D.,
10% of Bentonite. The above candidate materials Kawamoto, K., (2013). “Sustainability of liner
were found to conform to hydraulic criteria of materials in a municipal landfill constructed using
locally available expansive soils”. Proceedings of
liner materials by a previous study and
14th International Waste Management and
therefore, the specimens in this study was
Landfill Symposium. Sardinia, Italy.
prepared to the same degree of compaction
having the same voids ratio. Based on the 2. Darshika Wanigarathna, Chandana
outcome of this study, the following conclusions Kurukulasuriya, Shoichiro Hamamoto, Ken
can be made. Kawamoto (2012) “Locally Available Expansive Soils
as a Liner Material for Municipal Landfills”.
(1)Area reduction takes place, even after Proceedings, 2nd International Conference on
reaching the equilibrium water content Sustainable Built Environment (ICSBE 2012),
during drying. Special session on Water and Waste
Management, SBE/12/231. Kandy, Sri Lanka.
(2)No shrinkage cracks within the surface of the
3. Chao-Sheng Tang;, Yu-Jun Cui; , Anh-Minh Tang;
specimens were visible, thus making the
, Bin Shi (2010). Experiment evidence on the
materials highly suitable as a liner material, temperature dependence of desiccation cracking.
provided that other criteria to determine the Elsevier, Issue Engineering Geology, p. 261–266.
suitability as a liner material such as
hydraulic conductivity and durability issues 4. Abeyrathne, W. K. A. P. et al.(2012). Development
of a landfill clay liner using locally available
are satisfied.
expansive soil. Civil Engineering Research
(3) Accordingly 10% adding bentonite has been Exchange Symposium, Issue Faculty of
found as most effectively. Noted that the Engineering,University of Ruhuna, pp. 48-52.
adding percentage more than 10% will work 5. Atique, A. & Sanchez, M.(2011). Analysis of
other way as increasing the hydraulic Cracking Behavior of Drying Soil. Singapore, 2nd
conductivity from 1X10-7 cm/sec. International Conference on Environmental
Science and Technology.
Acknowledgement
I have done this research under the supervision

of Dr. L. C. Kurukulasuriya (lecturer, University
of Peradeniya) and a great appreciation is given
23
24
AnnualSessions
IESL,pp.
pp.[25[1- -32],
10],2017
2017
© The
© The Institution
Lanka
Identification of Clayey Properties for Cricket Pitch

Surfacing and Analysing the Variation of Surface
Crack Densities in different Pitches
W.S.U. Perera, U.P. Nawagamuwa and G.N.U. Thilakarathna
Abstract: Clayey top surface of a cricket pitch is the most critical factor which governs the pace
and bounce during a match. Previous studies on the same have found out that the soil grading, clay
and sand fraction, organic matter content, moisture content (MC%), clay mineralogy and the grass
coverage of the top 125mm layer of a cricket pitch are the major governing factors of the playing
character of different pitches. This research mainly focuses on the soil grading, clay and sand content
and clay mineralogy of the top clayey top surfacing. Laboratory tests were done on three clay samples
namely “Grumusol” from Murunkan (MU), a highly expansive soil sample from Kotawehera (KO)
and the currently used clay in cricket pitch (TY) preparation by Sri Lanka Cricket. This paper
compares the results obtained through the conventional hydrometer test and the laser particle
analyser (LPA) method and the clay content is eventually determined. X-Ray diffraction test was done
on three samples to investigate their clay mineralogy and the results are compared with the indirect
methods used to predict the clay mineralogy. Sample cricket pitches were prepared and the variation
of the surface crack densities of cricket pitches made of MU and TY was compared to propose a
relationship between clay mineralogy and crack density percentage. The most suitable clay fulfilling
the required properties for a fast and bouncy cricket pitch surfacing is identified as MU.
Keywords: Clay content, Crack density, Cricket pitch, Laser particle analyser, X-Ray diffraction
1. Introduction carried out to find the soil properties of local

and foreign soils and the purpose of these
Cricket is an interesting sporting event which is efforts was to find a locally available soil which
popular in every nook and corner in Asia. matches with the soil properties of fast pitch
Though it seems like a combat between bat and material.
ball, the cricket pitch plays a huge hidden role In these research studies, most of the soils
in deciding the day‟s winning team where the tested consist of low clay contents and the clay
ball behaviour is significantly influenced by mineralogy of the soils were predicted by
several factors inherited by the cricket pitch indirect methods (Nawagamuwa et al. [4]),
itself. (Perera and Nawagamuwa [7]). (Perera et al.
Ball behaviour after the impact is unpredictable [8]).
due to several reasons. Studies on this matter Prevailing method used in those studies to
have found out that the ball behaviour after the determine the finer particle content was the
impact is caused by the properties of top most conventional hydrometer test in which the
clay layer of the turf. Among these factors clay output was depended on several assumptions.
content, sand content, clay mineralogy, Assumptions made in hydrometer test have no
Moisture content (MC %) and top surface grass relation with laser particle analysis (LPA), thus
and cracking conditions are dominant. could give a better perspective in determining
This unpredictable ball behaviour of the cricket the clay contents.
pitches in different venues in different
continents has made the life uneasy for the
Eng. W.S.U. Perera, B.Sc. Eng. (Hons) (Moratuwa), AMIE
batsmen over decades. Most of the elegant
(SL), M.Sc. (Moratuwa), Lecturer on contract, Department of
batsmen in Asian cricket arena who have Civil Engineering, University of Moratuwa.
legendary records in home grounds fail to score
big numbers in Australian fast and bouncy Eng.(Dr) U.P. Nawagamuwa, B.Sc. Eng. (Hons)
(Moratuwa), MEng(AIT), DrEng(YNU), CEng, MIE(SL)
wickets. Therefore, development of fast and Senior Lecturer/Chartered Engineer (Civil)Department of
bouncy cricket pitches is a timely need in Asia. Civil Engineering, University of Moratuwa.
Several studies had been carried out in Sri
Mr. G.N.U. Thilakarathna, B.Sc. Eng. (Hons) (Moratuwa)
Lanka regarding the development of fast and Department of Civil Engineering, University of Moratuwa.
bouncy cricket wickets. Laboratory tests were
25
Moreover, indirect methods such as high conventional hydrometer tests and wet sieve
activity and high plasticity characteristics were analysis. Table below shows the clay and silt
used to predict the clay mineralogy in the contents obtained from Figure 1.
previous research studies. Besides, crack
density on the pitch surface is a critical factor
which influences the ball behaviour. Crack
density on the pitch surface may depend on
several factors such as MC%.
This paper compares and contrasts the clay
contents obtained by conventional hydrometer
and LPA tests in the process of developing fast
and bouncy cricket pitches in Sri Lanka. Also
the validity of predicting the clay mineralogy
by indirect methods was checked by doing X-
Ray Diffraction tests. The effect of MC% and
the clay mineralogy on the surface crack
density was studied and several correlations
were identified. Figure 1 - Particle size distribution graph
(Perera & Nawagamuwa, 2015)
2. Literature Review
Table 1 - Clay and Silt contents, Plastic
characteristics [2]
2.1 Different types of cricket pitches
% < % <
Soil PL LL PI
0.075 0.002
Cricket pitches have been categorized into two type mm mm
(%) (%) (%)
main categories according to the ball behaviour.
MU 2 80 70.9 28.1 43.1
The most common categories are “fast” and
“slow” pitches (James et al. [3]) Bang
15 83 45.6 18.6 26.9
alore
“Fast” pitches quite commonly are “Bouncy” TY 17 64 49.8 30.5 18.3
pitches while “slow” pitches are “low and
spinning” pitches. Therefore, the pitches are of “Grumusol” soil (MU) is a highly active soil
two kinds which are known as fast and bouncy (CH range), hence belongs to the Smectite clay
pitches, slow and low pitches group. MU is known for a soil with high clay
(Nawagamuwa et al. [4]). content [7]. In contrast the above results
showed less clay content in MU when
2.2 Effect of cricket pitch surface compared to other two soils. In addition the
clay content was very much less in all three
Stewart and Adams [6] had found that the tested soils.
playing characteristics of a cricket pitch were
primarily a function of the top 25mm of the soil. 2.3.2 Perera et al., (2016) [8]
This may be an over simplification as the sub
base and the sub grade too play an important Particle Size distribution
part in ensuring good bounce but most modern Seven clayey soils from different locations in Sri
experts believe that the top 100mm-150mm can Lanka were tested in this research for their
be considered to be the most crucial [4]. Particle size distribution, plastic characteristics
and degree of colloidal activity etc.
2.3 Previous researches regarding the soil Summary of the particle size distribution is
properties of cricket pitches in Sri shown in Table 2.
Lanka Hydrometer test and the wet sieve analysis
were used to obtain the above results.
2.3.1 Perera and Nawagamuwa, (2015) [7] Batticaloa A sample gave the highest clay
content of 24% but all the tested soil samples
Soil sample taken from Mannar area which is gave comparatively low clay and higher silt
named as “Murunkan soil sample” had been percentages. When the particles are bonded
tested for its ability to be used as a fast and together, it will become a larger colloid.
bouncy cricket pitch soil. Therefore, dry sieve may not provide accurate
Figure 1 shows the particle size distribution results in this situation where the accuracy will
curves obtained by Perera et al. [7] using depend on the clay and silt content to produce
26
the pace and bounce of a wicket. Hence wet Particle size distribution
sieve analysis could be treated better. Particle size distribution curves were obtained
by hydrometer analysis tests. Results were
Table 2 - Comparison of Finer particle tabulated in Table 3.
percentages [8]
Clay Type % finer than 0.075 % finer Table 3 - Clay percentages [4]
mm than Soil sample Clay percentage
0.002 mm SSC bulk clay bag 28
Tyronne 64 15 SSC pitch 38
Murunkan 80 2 Clay from Supplier 16
Kotawehera 73 8 Modified sample 42
Bangalore 83 17
Batticaloa A 99 22
Fairly higher values for the clay percentages
Batticaloa B 97 24 were obtained by this research by using
Batticaloa C 95 6 hydrometer test.
Plastic characteristics and Activity

Plastic characteristics
Plasticity Index was 14 and Activity was 0.42 in
Murunkan sample gave the highest LL (74%), the currently used soil in SSC cricket pitch,
PI (47%) from among all tested clays which which implied that the predominant clay
makes MU an extremely plastic clay. High mineral was Kaolinite. Authors have predicted
plasticity is one of the main characteristics of the currently used soil for pitch preparation in
Smectite clays (Herath et al. [1]). Therefore, MU Sri Lanka contains Kaolinite clay mineralogy
has a high possibility of being a Smectite clay. since the activity was less than 1.
Authors recommend the use of expansive
Degree of Colloidal activity clayey soils that can be found in the dry zone of
Sri Lanka (Herath et al. [1]) whose primary clay
Degree of Colloidal activity is one of the mineralogy is Smectite.
indirect methods used to predict the clay
mineralogy. Crack analysis
Activity is related to the mineralogy and
geological history of clays. (Skempton et al. [8]) The crack pattern of the pitch was observed and
the maximum crack widths were measured.
Activity = Plasticity Index / Clay Fraction--- (1)
Calculated Degree of Colloidal activity values

are shown in Table 3.
Error! Not a valid link.

Having a value higher than 1 for the activity
implies that soil belongs to Smectite clay Modified Model Control Model
mineralogy. Activity (Calculated by equation
(1)) was increased due to the low clay contents Figure 2 - Crack patterns in two soil models [4]
obtained by the hydrometer. However, MU
gave comparatively high value for activity. The maximum crack width observed at a MC%
Authors had predicted both MU and KO to be of 7.5% for the control model was 1.5mm. In
Montmorillonite clays which belong to Smectite this research, authors had considered only the
clay group since MU and KO had maximum crack width. Therefore, the factors
comparatively high activity values. affecting the crack propagation on the cricket
pitch surface should be further studied.
2.3.3 Nawagamuwa et al., ( 2009) [4]
2.4 Comparison of clay mineralogies [1]
Authors had tested two soils, a soil sample
taken from SSC cricket stadium Sri Lanka Following conclusions were made observing
(Control model) and the same soil mixed with the results made by Herath et al. [1]
50% of bentonite (Modified model).
27
Table 4 - Comparison of the properties of 4. Methodology
Kaolinite and Smectite clays, [1]
 Background of the research was studied
Kaolinite Smectite through literature review.
Low plasticity Very high plasticity  Selection and preparation of samples for
Low drying shrinkage High drying the LPA test.
shrinkage  Obtaining the results of LPA test.
Less cracking density High cracking density  Comparing the soil grading curves of LPA
Low compressive High compressive test with conventional hydrometer test.
strength strength  Determining the new clay contents of three
clays.
2.5 Significance of Surface cracking  Calculating the “Activity of clays” and
density comparing with early results.
 Validating the indirect approaches for
Surface cracking of the compacted clayey soil in predicting clay mineralogy by doing XRD
natural turf pitches is an inherent quality and it test.
adds variations to the behaviour of the cricket  Preparing model cricket pitches using TY
ball within different phases of an international and MU soils.
test match. Sometimes surface cracks influence  Studying the variation of surface crack
the ball to change its trajectory in an densities of two model pitches
unpredictable way which causes difficulties for  Finding a relationship between Clay
batsmen and thereby effecting dismissals. mineralogy and the variation of crack
Therefore, crack density is one of the dominant density.
characteristics of a cricket pitch which
influences the ball behaviour itself. The 5. Selection of Samples
maximum crack width can be around 10mm
however, the playing ability is decided by the Tests were done using 3 samples.
umpires. [10].
 Murunkan (MU) - “Grumusol” clay
sample from Murunkan, Mannar, Sri
2.6 Laser Particle Analysis
Lanka. The soil plasticity index was
more than 60 and the clay activity was
Laser diffraction method depends upon the
higher than one. This soil group has the
diffracted light produced when a laser beam
highest clay content and highest
passes through a dispersion of particles in air or
amount of 2:1 clay minerals among Sri
in liquid. The angle of diffraction increases as
Lankan soils classified so far. (Sobana
the particle size increases. Therefore, this
et al. [9])
method is reliable and accurate when
 Tyronne (TY)- Sample from “Tyronne
measuring the particle sizes distributions
Fernando Stadium” cricket pitch
between 0.1 and 192 µm (Rathnayake et al. [5]).
Katubedda which has the same clay
Also, this method is relatively fast and can be
used by Sri Lanka Cricket (SLC) for all
performed on very small sized samples.
other cricket pitches in Sri Lanka[7]
 Kotawehera (KO)- Sample from
3. Objectives Kotawehera in Nikaweratiya region, Sri
Lanka [7]
Three objectives of this research could be
formulated as follows.
6. Experimental Work
 Comparing the soil grading curves
given by hydrometer test and LPA tests
6.1 Laser Particle analysis (LPA)
and obtaining corrected clay contents.
 Checking the accuracy of the indirect
LPA test was done according to ASTM B822-92
methods such as “Activity of Clays”,
standards by using SHIN-KIGYO PRO OP2.
used to predict the clay mineralogy by
comparing the results with XRD test Researches done before gave lower clay
results. percentages due to the conventional
 Identifying a correlation between clay hydrometer test. Charged clay particles tend to
mineralogy and the crack density. clog together making heavier particles. Even
though charges of clay particles can be
28
neutralized by the dispersing agent used in Table 5 - New Clay and Silt contents of the
Hydrometer test, the organic matters which clays
binds the clay grains together cannot be Particle MU (%) TY (%) KO (%)
Size
removed by it. Still clay grains are attached
2 µm 62.1 45.5 64.2
together by organic matter and form larger 75 µm 77.2 91.3 76.8
colloid particles and sedimentation process 2 -75 µm 15.1 45.8 12.0
becomes quick giving lower finer particle
percentages in hydrometer test.
6.3 Modified degree of colloidal activity
However, organic matter inside the finer
particles collected from wet sieve analysis was Since the clay content of the above soils was
removed by adding H2O2 and the charges were changed, degree of colloidal activity was
neutralized by further adding the dispersing calculated as follows in Table 6.
agent during the LPA test (Rathnayake, et al.,
Table 6 - Modified degree of colloidal activity
2005).
Clay type PI (%) CF (%) Activity
Without organic matter in presence the clay MU 47.0 62.07 0.75
grains get separated without forming heavy TY 19.3 45.50 0.42
particles. Therefore higher clay contents could KO 34.7 64.25 0.54
be obtained for each tested soils.
6.2 Particle size distribution by LPA test 6.4 USCS Classification
Grading curve of the particle sizes >0.075mm Following abbreviations were used in Figure .
was produced by wet sieve analysis while the Murunkan (MU), Kotavehera (KO), Bangalore
LPA was done for collected finer particles, (BAN), Batticaloa A{B(A)}, Batticaloa B{B(B)},
passed through the 0.075mm sieve. Grading
Batticaloa C{B(C)}, Soil from Tyronne Fernando
curve obtained by LPA test was combined with
the grading of wet sieve analysis by stadium(TY).
overlapping 75µm point of LPA test on the
exact point given by wet sieve analysis. Rest of
the values of LPA test was modified by
multiplying with the ratio obtained at 0.075mm
point. Combined particle size distribution was
shown in figure 3.
Tabulated result of figure 3 is shown in Table 5.
Figure 4 - USCS classification chart for tested

soils by Perera , et al., 2016 [8]
6.5 XRD Test
XRD test was adopted in this research to find

the clay mineralogy. Powder XRD provides
detailed information on the crystallographic
Figure 3 - Modified particle size distribution structure and physical properties of materials.
graphs XRD graphs were given as an output by the
XRD test done by BRUKER D8 ADVANCE eco
Figure 3Figure 4 - USCS classification chart for
workstation. Summary of the results from the
tested soils by Perera , et al., 2016 [8]
XRD test was summarized in Tables 7, 8 and 9.
29
Table 7- Summary of the XRD Test for MU Table 10 - Surface crack density percentage
2 Theta Description Relative Day MU% TY%
Angle intensity (%) 1 2.79 3.13
5.378˚ Montmorillonite 30.60
2 5.35 3.14
5.627˚ Notronite 27.30
20.880˚ Notronite 27.60 3 7.45 3.94
26.666˚ Notronite 100.00 4 8.75 4.35
5 8.71 5.05
Table 8 - Summary of the XRD Test for KO 6 8.13 9.63
2 Theta Description Relative
Angle intensity
Crack density depends on the clay mineralogy
(%)
and MC%. Smectite is a type of an expansive
5.613˚ Montmorillonite 100.00
clay which tends to excessively shrink in low
19.670˚ Sauconite 9.90 moisture environments. Hence it shows a
significant cracking when subjected to drying
Table 9 - Summary of the XRD Test for TY conditions [1].
2 Theta Description Relative
Angle intensity (%)
12.357˚ Kaolinite 14.60 7. Results and Discussion
18.321˚ Triazine 50.80
7.1 Particle size distribution test
26.674˚ Triazine 100
20.874˚ Quartz 14.10
Particle size distribution curves can be obtained
24.927 ˚ Kaolinite 7.4
by the results of sieve analysis test and
50.159 ˚ Quartz 6.4
conventional Hydrometer test.
Hydrometer test is based on number of
assumptions where the accuracy of the results
6.6 Surface crack density analysis
can be less but the LPA test can be more
accurate since it is based on digital technology
Cracking of the pitch was quantitatively
that utilizes diffraction patterns of a laser beam
measured during this experiment using an passed through any object ranging from 0.02µm
image processing method. Model pitch was to 2000µm in size.
prepared in University of Moratuwa grounds Laser particle analysis gave a higher value for
and digital photographs were taken on each the clay fraction when compared with the
day of testing and a square of 300mmX300mm conventional hydrometer test. Both MU and
was photographed with clear boundaries and KO had above 60% of clay content while TY
had 45.5% of the same. Clearly the two
the images were processed using MATLAB
alternative soils (MU and KO) have higher clay
software to calculate the percentage pixel content when compared to conventional soil
density which showed cracks. which is used in current cricket pitches. High
clay content may lead to proper bonding of
pitch material and reduce the crumbling and
deteriorating the pitch surface. Therefore, the
strength in between the particles of the pitch
surface of MU and KO may be higher when
compared to TY.
7.2 USCS Classification

According to the Figure , MU, KO and B (A)
could be identified as high plasticity clays while
BAN and B (B) was medium plasticity clays.
Figure 5 - Actual image of the pitch and image
High plasticity is one of the main characteristics
analysis results by MATLAB software.
of Smectite clays. Therefore, according to the
indirect way of predicting clay mineralogy by
Model pitches were made from MU and TY
high plasticity MU, KO and B (A) could be
soils and the variation of the surface crack
Smectite clays [1].
patterns were observed for six days. Crack
density percentages were calculated by the
software and summarized in Table 10.
30
7.3 Degree of colloidal activity Variation of crack density percentages were
After inserting the corrected value obtained for plotted against days and shown in Figure 6.
the clay fraction by LPA, OM content was less
than 1% as shown in Table 6. Highest Activity
was recorded from MU as 0.75. Since the value
is less than “one”, MU cannot be directly
categorized in to Smectite clay group. Therefore
X Ray Diffraction test should be carried out for
further analysis of the clay mineralogy.
7.4 XRD test

Referring to the results given in Table 7,
Montmorillonite is a very soft phyllosilicate
group of minerals which is a member of the Figure 6 - Crack density percentage vs days
Smectite group. Besides Nontronite is the iron
(III) rich member of the Smectite group of clay According to Figure 6, Crack density
minerals (Anthony et al. [11]). Majority of the percentage of MU increased with time than that
MU sample is in Smectite clay group. of TY. XRD test revealed that KO and MU
Referring to the results given in Table 8, belongs to Smectite clay group. Smectite is an
Sauconite is a complex phyllosilicate mineral of expansive clay which tends to shrink in low
the Smectite clay group [11]. Since the majority moisture conditions [1]. Hence it shows a
of the sample are in Smectite group KO could significant cracking when subjected to drying
be named as a Smectite clay. conditions. This would have caused for the
Referring to the results given in Table 9, higher crack density on the surface of the MU
Triazine is an organic chemical compound with model pitch than TY.
the formula (HCN)3. Majority of the TY soil are MC is another parameter which governs the
organic compounds, quartz or kaolinite clays. crack density [10]. However, MC of the top soil
Hence the currently used soil for the cricket in both pitches reduced with time due to drying
pitch preparation in SL is a typical Kaolinite caused under direct sun light.
clay with high percentage of organic
compounds. As expected, crack density percentage has a
Indirect methods which were used to predict significant negative relationship with moisture
the clay mineralogy gave some close answers. content for both pitches. However, the gradient
KO and MU gave high PI value which was an of MU is higher than TY which implies MU
indication of a fat clay. Normally clays that generates higher crack densities than TY with
belong to Smectite clay groups are high the reduction of MC.
plasticity clays. Therefore, the prediction
matches with results given by XRD results.
Activity of clays was determined using the clay
content given by the hydrometer test in the
previous approach. In that case MU and KO
gave higher values for the activity which
implied promising results of being a Smectite
soil which was later confirmed by the results of
XRD test.
Activity of clays was recalculated by corrected
clay content obtained from LPA test. All the
results were less than 1 and it was an indication Figure 7 - Crack density% vs MC%
of Kaolinite clays. However, MU gave the
highest activity value of 0.75 and it was In order to find the relationship between clay
somewhat closer to „one‟. Majority of this result mineralogy and crack density the effect of MC
did not match with the results of XRD test. should be removed. To remove the effect
caused by MC, normalized crack density by
7.5 Surface crack density analysis MC was plotted against days as shown in
Crack density variation on the top surface was Figure 8.
photographed daily and images processed by
MATLAB software were further analysed.
31
References
1. Herath, J., 1973. Industrial clays of Sri
Lanka; Geology, Minerology,and Appraisal
for Ceramics and other Industries.
Economic Bulletin, Geological survey,Sri
Lanka, Volume No.1, pp. 1-113.
2. Skempton, A. W., 1953. The Colloidal
Activity of Clays. Zurich, s.n., pp. 57-61.
3. James, D., Carré, M. & Haake, S., 2005.
Predicting the playing character of cricket
Figure 8 - Crack density (%)/ MC% vs Days pitches. Sports Engineering, Volume 8, pp.
193-207.
According to Figure 8, Crack density % / MC% 4. Nawagamuwa, U., Senanayake, A., Silva, S.
values of MU was higher than for TY after the & Sanjeewa, D., 2009. Improvement of Local
2nd day of the test. Therefore crack density has Soils in Order to make “Fast & Bouncy”
a significant correlation with the clay cricket. Engineer,Institution of Engineers,
mineralogy. This implies that MU which Sri Lanka, Volume 42(4), pp. 46-55.
belongs to Smectite clay group is highly 5. Rathnayake, N. et al., 2005. Anthropogenic
expansive clay and it matches with the Impacts Recorded in the Sediments of
characteristics of fast and bouncy pitches in Lunawa, a Small Tropical Estuary, Sri
Western Australian Cricket Association Lanka. Environmental Geology, 48(2), pp.
grounds [4]. 139-148.
6. Stewart, V. I. & Adams, W. A., 1968. County
8. Conclusions Cricket Wickets. Journal of Sports Turf
Research Institute, Volume 44, pp. 49-60.
In this research clay contents were determined 7. Perera, W. S. U. & Nawagamuwa, U. P.,
by Laser particle analysis since the results given 2015. Identification of Local Soil for
by conventional hydrometer depend on Development Cricket pitches. s.l.,
assumptions which may have effect on the Department of Civil Engineering,
lower clay contents. Results by LPA test gave University of Moratuwa.
higher clay contents for the tested clays. MU 8. Perera, W., Nawagamuwa, U. & Wijerathna,
and KO had high values for the clay contents H., 2016. Study on Properties of Locally
and matches with the soil properties of a fast available Clays to be used in Fast and
and bouncy clay. Bouncy Cricket Pitches. s.l., Institute of
Activity values which were calculated by the Engineers' Sri Lanka.
clay contents given by the conventional 9. Sobana, A., Mapa, R. & Gowthamy, P., 2014.
hydrometer matched with the results given by CHRACTARIZATION OF GRUMUSOLS
XRD test where KO and MU with high activity IN MANNAR DISTRICT IN SRI LANKA
were considered as Smectite clays. AND THEIR APPLICABILITY TO
AGRICULTURE. Proceedings of the
Activity values were recalculated with the
Peradeniya University, p. 527.
results given by LPA test and the results were
less than one for each tested soil and slightly 10. Tainton, N. et al., 1998. Principles and
deviated from XRD test results. Practice of Pitch Preparation. [Online]
According to the results of crack density Available at:
http://static.espncricinfo.com/db/ABOUT
percentage variation MU displayed its highly
_CRICKET/PITCHES/PREP_OF_PITCHES
expansive nature than TY, which is an
.html
indication of Smectite clay. [Accessed 2 June 2016].
Moreover Crack density percentage showed a
significant negative relationship with Moisture 11. Anthony, J. W., Bideaux, R. A., Bladh, K. W.
content for both soils. & Nichols, M. C., 2010. Handbook of
Mineralogy. [Online]
Effect of MC% was removed by normalising the
Available at:
crack density percentage by MC% and plotted http://www.handbookofmineralogy.org/
against the day. Results showed that Crack [Accessed June 2016].
density percentage depends on the clay
mineralogy where Smectite clay showed higher
surface crack densities than Kaolinite clay.
32
AnnualSessions
IESL,pp.
pp.[33[1- -39],
10],2017
2017
© The
© The Institution
Lanka
Design of Soil–Cement–Bentonite (SCB) Backfill and

Construction of SCB Slurry Walls along the Earthen
Dam at Iranamadu Reservoir in Kilinochchi
P.N. Wickramarachchi, S. Weerakoon, S. Semage, N. Madushanka and T. Rajagobu
Abstract: Soil-Cement-Bentonite (SCB) slurry walls have been used with increasing frequency
in recent years in United States, Canada and European countries to provide barriers to the lateral flow
of groundwater or seepage in situations where the strength of a normal soil-bentonite wall would be
inadequate to carry foundation loads. However, this technique has been used with less frequency in
the Asian region up to date. The addition of cement to the backfill blend allows the backfill to set and
form a more rigid system that can support additional overlying loads which the earthen dams
undergo.
Construction procedure and quality control for the SCB wall are more demanding than that needed
for conventional soil-bentonite slurry walls. Backfill mix design to meet the given strict test
parameters, sampling and testing of this type of wall involve more exacting procedures. Designing the
SCB backfill is a complex issue involving conflicting actions of the various materials involved. While
the SCB wall provides additional strength, permeability is one property that suffers in comparison to
soil-bentonite slurry walls. The most demanding tests were a uniaxial compressive strength (UCS) and
permeability in this project, and specified values were not less than 100 KPa and not more than 200
KPa for seven days and 1*10-7 cm/sec, respectively. The backfill was designed to get the average UCS
as 160 KPa and permeability as 1*10-8 cm/sec.
The most challenging work involved with this are, mixing in large quantities, transportation and
placing the SCB slurry within the specified period to maintain the quality of mixture consistently
throughout the project period. However, with greater care and having engineering input, it was
possible to successfully construct the SCB slurry wall, 1700 m in length along the earthen dam for the
1st time in Sri Lanka.
Keywords: Permeability, Soil – Cement – Bentonite Slurry, Uniaxial Compressive Strength
1. Introduction supposed to be. The cultivation of paddy was

badly affected, and so does the farmers.
Rehabilitation and improvement of water and Therefore, the relevant authorities have
irrigation networks are a main strategic identified the urgency and burning desire to
movement of any government in order to rehabilitate the project and comfort the life of
provide better infrastructure facilities which common people. At present, the tanks bund is
can be heavily affected by rapid development. not provided with the required strength to
The same phenomena were applied during the store water up to the full flood level.
last two decades in Sri Lanka, and as a result,
many areas were developed and facilitating Eng.(Dr.)P.N.Wickramarachchi, C.Env.Eng., AMIE(SL),
the farmers and villagers. After peace prevails B.Sc.Eng.(Peradeniya), M.Eng.(Tahiland), Dr.
Eng.(Saitama), Manger Engineering, Access Engineering
in Northern Province, many such projects PLC.
were implemented by local and foreign
Eng.S.Weerakoon, AMIE(SL), B.Sc.Eng.(Peradeniya),
funding agencies. Project Manager, Access Engineering PLC.
Mr. S. Semage, NCT(SL) Senior Material Engineer, Access
The Iranamadu tank (the length of the earthen Engineering PLC.
dam is around 2750 m) and its main feature of Eng. N. Madushanka, AMIE(SL), B.Sc.Eng.(Ruhuna),
earth bund have not been rehabilitated in a Site Engineer, Access Engineering PLC.
technically sound manner during last few Eng. T. Rajagobu, C.Eng, MIE(SL), B.Sc.Eng.
decades due to many constraints. As a result, (Peradeniya), Deputy Project Director, Provincial
the tank was not able to function as it is Irrigation Department, Killinochchi.
1
33 ENGINEER
Soil-Cement-Bentonite (SCB) slurry walls have Figure 1 below shows the basic features of
often been used in recent years in United proposed rehabilitation including the SCB wall.
States, Canada and European countries to The SCB wall was coloured in blue in Figure 1.
provide barriers to the lateral flow of
groundwater or seepage. However, this The designers have given the dimension of SCB
technique has rarely been used in the Asian wall, and the contractor was asked to design the
region. The addition of cement to the backfill SCB backfill mixture and complete construction
blend allows the backfill to set and form a more works. The construction work includes the design
rigid system that can support additional of backfill mix and verifies for the given criteria by
overlying loads which the earthen dams are the contracts especially for the uniaxial
undergone. compressive strength (UCS) and permeability for
the 7 and 28 days, respectively.
While the SCB wall provides additional
strength, permeability is one property that
generally suffers in comparison to soil- 3. Materials and Method
bentonite slurry walls. The most demanding
tests are a uniaxial compressive strength (UCS) Basically, the significant materials for the SCB wall
and permeability. were bentonite, Portland cement and soil. The
preliminary investigation showed that the
2. Project Information excavated soil of slurry wall trench could be used
as soil for the mix (OMC 11.6 %, LL 33, PI 14.3,
The Iranamadu reservoir is the one of most Clayey loam, MDD 2.024 g/cm3). Even the normal
important water body in Northern Province, Portland cement was able to utilise with the
and the requirement for the new development supplier’s source approval. However, the special
was a key to develop the area. Though there consideration has to be paid for the bentonite to be
was a number of main features to rehabilitate used. The project specification clearly mentioned
and or develop, the SCB wall gained elevated that the proposed bentonite should exhibit the
interest among them. The main reason for that properties as per the API Spec 13 A [1].
one was, it was considered as the very first
construction in Sri Lanka.
The earthen dam was 2750 m long, and it was

decided to construct only 1700 m length of SCB
wall along the dam towards the upstream. The
width of SCB wall was 600 mm, and depths
were varied from 4 m to 6 m depends on the
location and especially the location of existing
puddle core.
Figure 1 – Basic features of proposed rehabilitation works

ENGINEER 2
34
3.2 SCB backfill mixture preparation at site
Table 1 – Properties of Bentonite Powder
Specification
Property Requirment
YP/PV Ratio >3
Viscometer at 600 rpm > 30
Filtrate Loss < 15 cm3
Moisture Content < 10 %
Residue > 75 micrometer < 4 %
Certification Meets API 13 A, (4)
3.1 SCB backfill design

Figure 2 – Measuring slump of backfill
Designing the backfill was a major challenge.
The key parameter was to maintain the Once the lab scale tests are passed, the optimum
permeability in the range of 10-7 cm/s range. mixture was decided to carry out the
Rayn and Day [2] have investigated and found construction works at the site. Proper detail plan
that to reach 10-6 cm/s is also somewhat difficult was prepared considering the required amount
in SCB backfill mix, however, with the strict per day and construction time was given and
quality control procedure, and proper site characteristics. Initially, it was planned to
engineering input reaching 10-7 cm/s is feasible, carry out 20 m length in the backfilling the slurry
and they have gotten it in their practical work. trench. However, the required ponds for mixing
bentonite slurry and filling slurry were prepared
Initially, three types of mixes were considered for more lengths. The investigation revealed that
and performed the required tests. The details are the mixture to be backfilled within 3 hrs after
shown in Table 2. The initial composition was mixing. Based on those details, the number of
based on the literature. It was found that less tippers and excavator were selected. The mixing
number of literature and all studies were based ponds and required machines were placed in the
on the USA and European region. Owaidat et al., middle of the whole length so that the logistics
[3] studied several mixes and ranged the mix and transportation activities were easy and
proportions as 94 %, 2 % and 4 %, soil, bentonite comfortable. Based on the calculation for the
and cement, respectively. mix, it was added to the mixing pad, the
calculated soil mass, dehydrated bentonite
Table 2 – Backfill Mixes slurry and the calculated amount of cement
slurry by 1000 L centrifugal mixer.
Compositions (%)
Type of mix
Soil Cement Bentonite
A 93.8 4.0 2.2
B 94.0 3.5 2.5
C 93.0 4.0 2.5
Another important aspect of backfill was the

working slump. The accepted range was 120 –
180 mm.
Figure 3 – Mixing of soil, cement and

bentonite at mixing pad using excavator
3
35 ENGINEER
3.2 Trench excavation and backfilling deviates from specified properties. The
excavated trench was filled by the
A 600 mm width and 1700 m length trench dehydrated bentonite slurry.
were excavated to the depths ranging from 4
m to 6 m depending on the location of puddle
core along the dam. Initially, borehole
investigation was done to identify and locate
the existing core paddle. Once it is located
then, excavation was commenced stage wise.
The trench was filled with bentonite slurry to
the depth of 60 cm to the top of the surface to
avoid any collapsing the banks because of
deep excavation. Approximately 30 - 40 m of
the section was excavated each working day.
Backfill mixture was transported using the

three cube tippers to the excavated trench and
placed with the assistance of skid steer loaders
(Bob Cat). The mixture was placed in such a
way that in the opposite direction to the
excavation. At the initial stage, it was assisted
Figure 5 – Trench filled with bentonite
by tremie pipe to place exactly to the place
where necessary. Before the transportation, the
The SCB construction was performed as continued
all required tests were performed including
work patterns in every day. To continue the
the slump with the presence of Engineer and
previous section, the joint face was scraped up to 100
the client. The approximate mixing time was
mm thickness section and face was prepared as fresh
range from 1 hr to 1.5 hrs depending on the
section and then continued to backfill new material
quantity of soil volume.
on top of that face. Once the backfill was done, the
top surface was covered by 30 cm thick soil layer by
every day.
4. Quality control procedure
The complete performances of SCB wall depend on

the states of quality which is maintained at the site.
Therefore, very strict quality control procedure was
adopted in every aspect to keep the super
performance of construction and aftercare. In the
site lab with constant supervision by the Engineer
and client, the tests were carried out as specified in
the contract documents. All tests were performed by
experienced lab staff.
The recommendation made by the literature and

specification such as API Spec 13 A, ASTM D 5084,
ASTM D 4380 [5] and so on.
Figure 4 – Trench excavation
The whole process was carried out as

night shifts. It helped the site management
not to hinder the daytime progress and
work in comparably lower temperatures.
During the daytime, the average
temperature was over 350 C, and it might
cause to lose the calculated moisture and
ENGINEER 4
36
nearly 6 -7 days. Then the UCS and
Table 3 – Tests performed permeability tests were commenced in both 7
days and 28 days.
Tests Frequency Backfill mixing was carried out in such a way

that, pre-determined quantities are met exactly
Initial Bentonite Slurry
and thorough mixing procedure was done to
Viscosity 2 /shift get the homogeneous mixture. Slump tests
Density 2 /shift were performed to confirm that workability
Filtrate Loss 2 /shift had attained so that filling to continue. If the
Bentonite Content 1 /project slump was not gained, based on the reading
either bentonite slurry was added or further
mixing was done. However, the greater care
Trench Slurry was done to place the backfill with in the
Unit Weight 2 /shift specified 3 hr time frame.
Viscosity 2 /shift

Backfill Material
Slump 1 / shift
Gradation 1 / 2000 m3 5.1 Results of initial mix design
Density 1 / shift
UCS 1/ 500 m
As per the Table 3, initial 3 mixes were chosen
Permeability 1/ 500 m to carry out the testing, namely, type A, B and
C. Initial idea was generated through the
Table 4 – Test parameters of initial mixes literature as discussed. Once the mix designs
are carefully were done, it was easy to finalize
the appropriate design for the project. The
Permeability 10 -8
Compositions (%) UCS (kPa)
Type of mix (cm/s)
Soil Cement Bentonite 7D 28 D 7D 28 D
A 93.8 4 2.2 145 225 1.72 1.5
B 94.0 3.5 2.5 125 210 2.93 1.6
C 93.0 4 2.5 155 245 2.88 2.2
Special care was taken to carry out the UCS main and key parameter was always the
and permeability tests since these were done permeability, and Unconfined Compressive
at independent third party lab. Sampling was Strength (UCS) then followed. Based on the
done in a very careful manner and as per the results of Table 3, it is clearly demonstrated
recommendation by Ryan and Day [2] and that UCS values are increasing with cement
Cermark and Law [4] and project content. The main purpose of adding cement
specifications and standards. Sample was to gain additional strength for the mix. On
preparation and transportation to the labs the other hand, the UCS has increased with the
were done under the special attention on setting. That is well-understood phenomena in
trench specialists and material engineer. Site concrete technology, and here also it has been
collected samples were treated not to loose the valid. The specification to achieve was 7 days
moisture and kept minimum 4 days at site UCS in between 100 to 200 kPa. All the mixes,
before transportation to the lab. Once that value have been achieved. Figure 6 shows
delivered to the lab, then the saturation of the failure pattern for the one sample.
samples was carried out, the time period was
5
37 ENGINEER
Table 5 – UCS and Permeability (k)
values for mix B
Permeability 10 -8
UCS (kPa)
Type of mix (cm/s)
7D 28 D 7D 28 D
B (Mix design) 125 210 2.93 1.60
B (1st set) 180.3 270 2.89 1.63
B (2nd set) 197.2 277.4 2.78 2.90
B (3rd set) 120.2 238.6 2.65 2.14
Samples and even preservation until to reach

to the lab is also very key areas to keep the
close eye. In addition, the basic tests as listed in
Table 2 are also very vital in order to get the
better UCS and k throughout the project.
Figure 6 – Failure pattern for a sample

6. Encountered difficulties and
the solution given
The most important and interesting parameter
was the permeability. Measuring the
Many practical difficulties were encountered in
permeability is also a challengeable task for
both design and construction aspects during
the lab staff. Experience and knowledgeable
the project.
personnel are required to carry out the tests. A To initiate a mix design, there are various
tiny error or negligence can be caused to get
aspects to consider. During that stage, the
erroneous values, so that whole work and
considerable time period was consumed to
effort can be wasted. Therefore, permeability
reach a final design, mainly because, to get the
experiment should get more close attention.
28 days permeability, it had to wait nearly 2
All mixes have exhibited elevated values for
months from the initial working. In practice, in
the permeability values ranging 1.72 to 2.88
Sri Lanka, very few competent labs are located.
*10-8 at 7 days and from 1.5 to 2.2 *10-8 at 28
As a result, further time consumption can
days, respectively. The values were more than
occur. So, better planning and early
what was expected in the specifications. As commencement of such time-consuming tests
expected the permeability has reduced with
are highly recommended.
the time as the material matrix get set.
However, the values were not decreased by
Proper planning and logistics arrangement is
order, so the permeability values remained in
also a somewhat problematic area, and it is
same order.
required constant attention to plan the work
smoothly. It is very costly to prepare several
Based on the satisfactory results, it was
mixing ponds and mixing pads along the dam
decided t move to the type B, where better
or vicinity. Probably one location to cover
results with lower cement contents. Therefore, approximately 2 km is recommended based on
the rest of work carried out with mix B
our finding and night shift can be more
throughout the project.
effective in such work.
The figures in the table clearly showed both
Mixing the backfill material is more
UCS and permeability values are in proper
challengeable in large scale and will be again
order so that the construction activities were
time-consuming if necessary steps are not
able to move smoothly. It is our understanding
being taken. As a result, the productivity and
that continuous engineering input and strict
efficiency can be lowered drastically and
quality control procedures were the most creating unfavourable condition at the site in
important aspect to get much better values.
terms of finance, time and quality. Mixing time
and the way excavator moves are very
important aspects. If mixing time passes more
ENGINEER 6
38
than 2 hr, it is unlikely to get a homogeneous 2. Ryan, C., R, and Day, S. R, Soil-Cement-
and good slump mixture. Another significant Bentonite Slurry Walls, Proceedings of Deep
aspect is to select the soil quantity for a batch Foundations 2002: An International Perspective on
Theory, Design, Construction, and Performance.,
to mix. Higher quantities can cause to extended
American Society of Civil Engineers, Orlando,
time to missing, and non-homogenous mixture CO, 2002, pp. 713-727.
and lower quantity can cause the productivity
issues. 3. Owaidat, L. M., Andromalos, K. B, and Sisley,
J.L, Construction of a Soil-Cement-Bentonite
Placing backfill material is also an area to keep Slurry Wall for a Levee Strengthening program,
a close eye. Improper placing can lead to Proceeding of pan-Am CGS Geotechnical
creating the non-uniform layers in the trench Conferences., 2009, pp 512-520.
so that some path can be created and water can
move through them once the tank is at full 4. Cermak, J., and Law, M. T, Cut off performances
level. of a Soil – Cement – Bentonite Wall, Proceedings of
Pan-Am CGS Geotechnical Conferences., 2011, pp.
713-727.
7. Conclusions 5. ASTM, Compressive Strength of Molded

Soil-Cement Cylinders, Vol. 4.08, West
For the first time in Sri Lanka, the design and Conshohocken, PA, 2000.
construction work of SCB wall was carried out
successfully using only local engineering know
how.
Though some literature suggested reaching the

permeability values to 10-7 cm/s, in this project
it has been achieved up to the 10-8 cm/s in all
case with material composition as 94 %, 3.5 %
and 2.5 %, the soil, cement and bentonite,
respectively.
The corresponding UCS values were ranged

between 100 to 200 kPa at 7 days and over 200
kPa after the 28 days.
Productivity can be increased up to four folds

with the same machinery with proper
planning. Site observation and feeling are vital
aspects in order to increase the productivity
and even value engineering.
Acknowledgements
The Authors wish to express their ultimate

gratitude for Provincial Irrigation Department
in Kilinochchi for their great support given.
Further, the laboratory staffs of Access
Engineering PLC (PMD 1) are also
commended for their valuable services. In
addition, the Department of Civil Engineering,
University of Moratuwa is highly appreciated
for the extended support for the lab services.
References
1. API, Specification for Oil-Well Drilling – Fluids
Materials, Specification 13A-4, Washington, DC,
1990,
7
39 ENGINEER
AnnualSessions
IESL,pp.
pp.[41[1- -49],
10],2017
2017
© The
© The Institution
Lanka
Ballast Mats Made of Recycled Tyre Waste to Enhance

the Performance of Rail Track Substructure
under Cyclic Load
S.K. Navaratnarajah and B. Indraratna
Abstract: The rail transportation system in many countries plays a significant role in the
conveyance of bulk freight and passengers. Increasing demand for high-speed rail and fast heavy haul
poses a serious challenge for stability of tracks on the problematic ground. Ballast is a key foundation
material placed underneath the sleepers which provides structural support against high cyclically
induced stresses imparted by moving trains. Degradation of ballast contributes to a large percentage
of maintenance costs, apart from affecting the longevity and stability of track. The problem is more
critical in isolated rail track locations such as bridges and tunnels, where the substructure is much
stiffer than the surface track assembly. In recent years, use of rubber mats to stabilise the stiffer track
foundation has become increasingly popular. These rubber mats placed under the ballast layer is
called as Under Ballast Mats (UBMs). Currently, there is a lack of comprehensive assessment on the
geotechnical behaviour of ballast using these artificial inclusions under cyclic loading. In this study, a
series of large-scale laboratory tests were carried out to verify the performance of ballasted track and
understand the use of energy absorbing mats placed on top of a stiff foundation in the attenuation of
dynamic stresses and subsequent mitigation of ballast degradation. Cyclic loads were simulated using
a large-scale process simulation prismoidal triaxial apparatus (PSPTA). The study shows that UBM
distributes the stress applied from moving trains more uniformly by increasing effective contact area,
which then reduces the dynamic amplification of applied vertical stress and leads to much less ballast
degradation. The UBM used in this study was made from recycled tyre waste is a better solution to
turn waste into value.
Keywords: Railroad ballast; Cyclic Load; Under Ballast Mat; Deformation; Degradation
1. Introduction The ballast provides adequate resiliency when

Railways are one of the major modes of subjected to low-frequency cyclic loading
transportation in the conveyance of passengers
and bulk commodities. The railway industries However, when track vibrations are excessive
are emphasiziing the implementation of high (for example, high-frequency loading), the
train speed corridors and heavier freight resiliency of track diminishes, hence additional
operations to achieve efficient and cost-effective substructure inclusions become imperative to
services. However, increased train speeds and restore the original (fresh track) resiliency.
higher freight loads accelerate the deterioration Compared to the ballast, the high ductile nature
of track foundation, which in turn increases the of resilient materials such as rubber mats can
maintenance costs. A major proportion of track absorb a significant amount of energy during
maintenance funds is spent on the geotechnical cyclic loading at relatively high train speed.
related problems of the substructure, including Rubber mats placed beneath the ballast layer
the ballast layer [1-3]. The engineering actively decreases the amount of strain energy
behaviour of ballast is one of the most transferred to the ballast layer, thereby
important aspects governing the stability and reducing particle deformation and degradation.
performance of a railway track because its
Eng. (Dr.) S.K. Navaratnarajah, Ph.D.(Aus), MSc (USA),
progressive degradation under heavy cyclic
B.Sc. Eng.(Hons) (Peradeniya), Senior Lecturer, Department
loads from freight and passenger trains of Civil Engineering, University of Peradeniya.
contributes largely to overall track deformation Eng. (Prof.) B. Indraratna, Ph.D.(Alberta), MSc and BSc-
[4]. This, therefore, necessitates a thorough Hons (London), FTSE, FIEAust, FASCE, FGS, FAusIMM,
understanding of the geotechnical behaviour of FIES, DIC, C.Eng., CPEng, Distinguished Professor of Civil
the ballast layer, if the costs of track Engineering and Research Director, Centre for
Geomechanics and Railway Engineering; Program Leader,
maintenance are to be reduced. ARC Centre of Excellence for Geotechnical Science and
Engineering; University of Wollongong, Australia.
1
41 ENGINEER
This in effect, improves the overall resiliency, 2. Large-Scale Laboratory Testing
and the longevity of the track.
To assess the role of under ballast mats in more
Under Ballast Mats (UBM) eliminates the hard detail, a series of cyclic loading tests were
interface between the ballast and underlying conducted using the state-of-the-art large-scale
hard layers, which enables the ballast to bed Process Simulation Prismoidal Triaxial
into a relatively softer pad and increase its Apparatus (PSPTA) at the University of
contact area at the interfaces, which helps to Wollongong, Australia. The ballast material
distribute the cyclic load over a larger contact commonly used in New South Wales (NSW)
area (Fig. 1). As a consequence, the excessive rail tracks was used for laboratory testing.
interface and inter-particle contact forces of Details of the test materials and their
ballast are eliminated, which in turn reduces specifications and the cyclic load test
settlement and ballast breakage [5-9]. However, procedures are discussed below.
studies conducted to analyse the effectiveness
of UBMs in minimising ballast deformation and 2.1 Test Materials
degradation under cyclic loading are limited.
The materials used in this study are fresh
ballast and recycled tyre waste UBMs. The fresh
ballast tested is latite (volcanic) basalt, a
common igneous rock found along the south
coast of NSW, and obtained from Bombo
quarry, near Wollongong, NSW. The ballast
samples were prepared in accordance with
current Australian practices [10]. The
laboratory samples were cleaned by water to
remove any dust and clay adhering to the
surface of the aggregates and dried before
screening through selected sieves and then
mixed in desired proportions to obtain the
required particle size distribution (PSD). The
PSD and grain size characteristics of the ballast
material are shown in Figure 2. To simulate a
hard subgrade condition such as a rail track
over a concrete bridge, a reinforced concrete
base was placed underneath the ballast layer.
Figure 1 - Cyclic Axle Load Distribution:

(a) Without UBM; (b) With UBM (modified
after [8])
The primary objectives of this research study

are to assess the stress-strain and degradation
behaviour of ballasted rail track and verify its
performance when energy absorbing ballast
mats are used in track foundation. Large-scale
process simulation triaxial testing was
conducted to evaluate the stress-strain and
degradation characteristics of ballast with and
without ballast mats. Empirical models were
developed based on the laboratory finding to
evaluate the resilient modulus of ballast and
ballast breakage index.
Figure 2 - Particle Size Distribution (PSD) and
Grain Size Characteristics of Fresh Ballast
ENGINEER 2
42
Table 1- Mechanical Properties of UBM
Thickness 10 mm
Weight 9.2 kg/m2
Young's Modulus 6.12 MPa
Tensile Strength 600 kPa
Tensile Strain at Failure 80%
Static Stiffness, Cstat (DIN 2010) 0.2 N/mm3
Dynamic Stiffness, Cdyn1 (5-30 Hz) (DIN 0.46-0.59
2010) N/mm3
The mechanical properties of the UBM used in

this study, along with a photograph of the UBM
samples are shown in Table 1. The UBM was 10
mm thick and was made from recycled rubber
granulates encapsulated within a polyurethane
elastomer compound. The mat was placed on
top of the concrete base during cyclic loading
tests. Even though these polymeric materials
deteriorate when exposed to the environment
(thermal, oxidation, photolytic, and hydrolysis),
they have a handy service life of approximately
20 years [11, 12].
2.2 Cyclic Load Test Procedure

The repeated cyclic loading from wheels
moving along a rail track was simulated by the
PSPTA shown in Figures. 3a and b. The cubical
triaxial chamber is 0.8 m long by 0.6 m wide by
0.6 m high. This apparatus can realistically
simulate the stress and boundary conditions mm thick layers, compacted by a rubber
found in ballasted track. In a real rail track, padded vibratory hammer to a typical field
lateral movement of the ballast is not fully density of approximately 1560 kg/m3. The rail-
restrained, particularly parallel to the sleepers sleeper assembly was placed on top of the
[13]. Therefore, the four vertical walls of the compacted ballast, and the space around the
PSPTA chamber were built to allow free concrete sleeper was filled with 150 mm thick
movement under applied cyclic loading. In this compacted crib ballast.
study, only two walls parallel to the sleeper
were allowed to move under low confinement, Some photographs of sample preparation for
the other two walls were locked in position to testing are shown in Figures. 4 (a)-(f). The cyclic
simulate a typical straight track (longitudinal loading was the equivalent of a 25t axle load to
deformation parallel to the rail is negligible). simulate a passenger train, and a 35t axle load
for a coal train with a frequency of 15 and 20
The test specimen consisted of 150 mm thick Hz; this simulated a train traveling at about 110
oncrete base overlain by a 300 mm thick layer and 145 km/h, respectively [14]. The
of ballast to simulate a hard subgrade condition corresponding cyclic loading in a sinusoidal
of a rail track. When testing with UBM, a wave form is shown in Fig. 5. The amplitude of
rubber mat was placed on top of the concrete cyclic loads was determined using American
base. Ballast aggregates were placed on top of Railway Engineering Association (AREA)
the concrete base/rubber mat in three equal 100 method [15].
3
43 ENGINEER
Lateral confinement was applied by hydraulic were recorded by data loggers connected to
jack connected to movable vertical walls that computers. A total of 500,000 load cycles were
simulated a low confining stress of 10-15 kPa. applied for each cyclic load test.
The walls were locked for movements in the
longitudinal direction to ensure plane strain 3. Results and Discussion
conditions, while the pressure exerted on the
walls was measured. Electronic potentiometers Eight cyclic load tests were carried out with
and settlement plates were used to measure the and without UBM at varying frequencies (15
deformation of the ballast layer. The vertical Hz and 20 Hz) and axle loads (25t and 35t). The
and lateral loads and associated deformation stress-strain response and degradation of
ENGINEER 4
44
ballast observed in these tests and a comparison vertical walls were restrained (plane strain
with and without the use UBM are explained in condition, 2=0), shear (q) and volumetric (p)
the following sections. strains were calculated from the following
expression by Timoshenko and Goodier [16]:
3.1 Cyclic Deformation and Strain
Responses
The vertical and lateral deformations of ballast q =
2
3
 2
1   32  1 3  ….(1)
under cyclic loading with and without shock
mats were measured at selected number of load  p =1   3 …..(2)
cycles (N=1, 100, 500, 1000, 5000, 10000, 50000,
100000,…..500000). Since the longitudinal where 1, 2 and 3 – major, intermediate, and
minor principal strains, respectively.
5
45 ENGINEER
Vertical and lateral plastic deformation (and elements in the track can cause excessive plastic
strains) with and without the UBM are shown and elastic deformations of ballast, as well as
in Figures. 6(a) and (b), and the corresponding fatigue damage to other track components.
volumetric and shear strains are shown in Figs. Therefore, a proper design of track stabilised
7(a) and (b), respectively. As expected, with resilient material with respect to their
deformation was high due to the hard-concrete thickness and stiffness relative to the stiffness of
base at the bottom of the ballast and the use of the track substructure is vital.
UBM on top of the concrete base significantly
reduced the ballast deformation. The overall 3.2 Ballast Pressure Variation
shear and volumetric strains were decreased in
In this study, the pressure cells placed at the
the order of 10 to 25% for the applied cyclic
sleeper-ballast interface, in the middle of the
loads and frequencies. The test results also
ballast layer, and at the ballast-concrete base
indicated that the ballast deformed rapidly up
interface were used to measure the dynamically
to around 10,000 cycles due to its initial
induced pressure variation along the depth of
densification and further packing after the
the ballast mass. Figure 8 shows the variations
corners of the sharp angular aggregates began
in pressure with the depth of the ballast layer. It
to break, but once the ballast stabilised after
is observed that the dynamically induced
around 100,000 cycles, the rate of ballast strain
stresses in the ballast increase with the axle
decreased as the load cycles increased.
load and train speed (loading frequency). These
results compare well with the limited field data
Hunt [17] reported that increasing the elastic
ENGINEER 6
46
obtained for a train with a 25 t axle load significantly by the UBM. On average, when all
traveling at a speed of 73km/h (10 Hz) by three layers are considered, UBM reduced the
Nimbalkar and Indraratna [18] at a rail bridge degradation by more than 40% when it was
site at the Singleton, NSW, Australia as shown used on top of a concrete base.
in Figure 8. When a rubber mat is placed on top
of a concrete base, there is a notable reduction 3.4 Resilient Modulus and BBI Models
in stress in the ballast mass, especially at the
Multiple regression models were developed
bottom (ballast-concrete base interface) and
using MATLAB to predict the Resilient
middle layers. Stress amplification due to a stiff
Modulus (MR) and BBI based on the laboratory
base is now curtailed by the softer interface
test results. The MR model is based on
introduced by the rubber mat between the
Equation (3) and is being adapted from NCHRP
ballast and the concrete.
[20]. The BBI model is a function of bulk stress
(), and the damping ratio (D) of the ballast as
3.3 Ballast Degradation
shown in Equation (4).
Progressive particle degradation occurs when a
repetitive cyclic load is applied to the ballast     
2 k
 
k3
…..(3)
mass. Initially, grinding and breakage of MR k1 Pa    oct   1
 Pa   Pa  
corners of the angular ballast at the inter-
particle contacts take place, followed by BBI aMRb  c  D 

d
…..(4)
complete fracture across the body of the
particles depending on the strength of the where, k1, k2, k3, a, b, c, d – Multiple regression
parent rock and the level of load increment. constants; Pa– atmospheric pressure (101.3 kPa);
This breakage of ballast particles contributes to   1   2   3 ; oct–octahedral shear
increased axial and lateral strains and causes
( 1 (1   2 )2  ( 2   3 )2  ( 3   1 )2 ) ; and σ1, σ2
differential track settlement. Ballast oct 3
degradation in this study was quantified using and σ3 – major, intermediate, and minor
the Ballast Breakage Index (BBI) method principal stresses, respectively
proposed by Indraratna, et al. [19]. To quantify
particle breakage by this method, particle size The actual and predicted MR and BBI are
analysis was carried out before and after each shown in Figures. 10 (a) and (b), respectively.
cyclic triaxial test. To better estimate ballast The coefficient of determination (R2) is greater
breakage with depth, the ballast layer was than 0.9, which shows that the MR and BBI are
colour coded and divided into three equal predicting well with the variables used in the
thicknesses of 100 mm, and then the BBI values model.
were assessed separately for each layer. The BBI
calculated for each layer are presented in Figure 4. Conclusions
9. As expected, ballast breakage was higher at
the top layer due to higher inter-particle contact The performance of ballasted rail track under
stress at the top, followed by the middle, and cyclic loads with under ballast mats (UBM)
bottom layers. The results shown in Figure 9 placed on top of a hard foundation (concrete
indicate that ballast degradation is reduced base) was studied through large-scale
7
47 ENGINEER
laboratory testing. The test results confirmed 5. Bolmsvik, R., "Influence of USP on Track
that UBM reduced the ballast deformation, Response - A Literature Survey," Abetong Teknik
stress and degradation quite considerably. The ABVäxjö, Sweden, 2005.
overall shear and volumetric strains decreased
6. Loy, H. (2008) Under Sleeper Pads: Improving
in the order of 10-25% when UBM was used.
Track Quality While Reducing Operational Costs.
The interparticle stress and the interface stress European Railway Review.
at the ballast-concrete base interface were
significantly reduced by the use of UBM. The 7. Dahlberg, T., "Railway Track Stiffness Variations
reduction in ballast degradation was more than - Consequences and Countermeasures,"
40% when UBM was used on top of a hard International Journal of Civil Engineering, vol. 8, no.
track foundation. It is evident from this study 1, pp. 1-12, 2010.
that by placing UBM, loads on the ballast bed
can be distributed more homogeneously by the 8. Indraratna, B., Navaratnarajah, S. K., Nimbalkar,
hard base and thereby improves track stability. S., and Rujikiatkamjorn, C., "Use of Shock Mats
for Enhanced Stability of Railroad Track
This study confirmed that the recycled UBM
Foundation," Australian Geomechanics Journal,
used herein had an adequate damping ratio to Special Edition: ARC Centre of Excellence for
absorb the energy transmitted to the track, thus Geotechnical Science and Engineering, vol. 49, no. 4,
preventing excessive ballast degradation. The pp. 101-111, 2014.
benefits of reducing degradation with UBM
potentially minimize the number of 9. Navaratnarajah, S. K., Indraratna, B., and
maintenance operations and reduce the life Nimbalkar, S., "Performance of Rail Ballast
cycle costs of track foundations. Especially, the Stabilized with Resilient Rubber Pads under
rubber mats can be made of recycled material Cyclic and Impact Loading," presented at the
such as tyre waste gives more value to the Proceedings of the International Conference on
Geotechnical Engineering, Colombo, Sri Lanka,
waste materials.
10-11, August, 2015.
Acknowledgement 10. AS 2758.7 (2015) "Aggregates and Rock for

Engineering Purposes, Part 7: Railway Ballast."
Authors wish to acknowledge the Australian Standard Australia, Sydney, New South Wales,
Research Council (ARC) for its financial Australia.
support. The assistance provided by senior
technical officers, Alan Grant, Cameron Neilson 11. Ito, M. and Nagai, K., "Degradation Issues of
and Ritchie Mclean at the University Polymer Materials used in Railway Field,"
Polymer Degradation and Stability, vol. 93, no. 10,
Wollongong, Australia is also much
pp. 1723-1735, 10, 2008.
appreciated.
12. Lakuši, S., Ahac, M., and Haladin, I.,
References "Experimental Investigation of Railway Track
with under Sleeper Pad," presented at the 10th
1. Navaratnarajah, S. K. and Indraratna, B., "Use of Slovenian Road and Transportation Congress,
Rubber Mats to Improve the Deformation and Portoroz, Slovenija, 2010.
Degradation Behavior of Rail Ballast under Cyclic
Loading," Journal of Geotechnical and 13. Indraratna, B., Salim, W., Ionescu, D., and
Geoenvironmental Engineering, vol. 143, no. 6, p. Christie, D., "Stress-Strain and Degradation
04017015, 2017. Behaviour of Railway Ballast under Static and
Dynamic Loading, Based on Large-Scale Triaxial
2. Ionescu, D., Indraratna, B., and Christie, H. D., Testing," in Proc. 15th Int. Conference on Soil
"Behaviour of Railway Ballast under Dynamic Mechanics and Getechnical Engineering, Istanbul,
Loads," in Proc. 13th South East Asian Geotechnical 2001, vol. 3, pp. 2093-2096.
Conference, Taipei, 1998, pp. 69-74.
14. Indraratna, B., Biabani, M., and Nimbalkar, S.,
3. Indraratna, B., Salim, W., and Christie, D., "Behavior of Geocell-Reinforced Subballast
"Improvement of Recycled Ballast using Subjected to Cyclic Loading in Plane-Strain
Geosynthetics," in Proc. 7th International Condition," Journal of Geotechnical and
Conference on Geosynthetics, Nice, France, 2002, pp. Geoenvironmental Engineering, vol. 141, no. 1, pp.
1177-1182. 1-16, 2015.
4. Indraratna, B., Salim, W., and Rujikiatkamjorn, 15. Li, D. and Selig, E., "Method for Railroad Track
C., Advanced Rail Geotechnology: Ballasted Track. Foundation Design. I: Development," Journal of
Rotterdam, Netherlands: CRC Press/Balkema, Geotechnical and Geoenvironmental Engineering, vol.
2011. 124, no. 4, pp. 316-322, 1998.
ENGINEER 8
48
16. Timoshenko, S. and Goodier, J. N., Theory of
Elasticity, 3rd Edition ed. Singapore: McGraw-
Hill 1970.
17. Hunt, G., "Review of the Effect of Track Stiffness

on Track Performance," RSSB, Research Project,
vol. 372, 2005.
18. Nimbalkar, S. and Indraratna, B., "Improved

Performance of Ballasted Rail Track Using
Geosynthetics and Rubber Shockmat," Journal of
Geotechnical and Geoenvironmental Engineering, vol.
142, no. 8, p. 04016031, 2016.
19. Indraratna, B., Lackenby, J., and Christie, D.,

"Effect of Confining Pressure on the Degradation
of Ballast under Cyclic Loading," (in En),
Géotechnique, vol. 55, no. 4, pp. 325-328, 2005.
20. NCHRP, "Guide for Mechanistic-Empirical

Design of New and Rehabilitated Pavement
Structures - 1-37A Final Report," National
Cooperative Highway Research Program,
Transportation Research Board, National
Research Council, Washington, DC, USA.2004.
9
49 ENGINEER
AnnualSessions
IESL,pp.
pp.[51[1- -58],
10],2017
2017
© The
© The Institution
Lanka
Analysis of Energy Dissipation in Post-Tensioned

Hybrid Precast Concrete Frames
K.L.P.S. Perera, W.D.C.R. Wanniarachchi and P. Thammarak
Abstract: This paper introduces the analysis of a post-tensioned precast concrete hybrid
framing system for providing increased ductility in seismic prone zones in following a damage
avoidance design philosophy. A Compression Free Device (CFD) being an externally mounted
replaceable steel yielding coupon mounted between a set of grips allowing free movement during
compression, is to be used as an energy dissipater to counter the hybrid framing system‟s inherent
lack of energy dissipative capacity. The compression free behaviour allows a lighter, slender energy
dissipater allowing higher energy dissipation through tensile yielding without buckling. The analysis
was performed using SAP2000 and further component to be performed using ABAQUS, through
modelling of a subassembly consisting of a single frame connection with a beam and column
cruciform assembly. The analyses were performed as a series of non-linear time history analyses and a
non-linear reversed cyclic pushover analysis for the subassembly with the CFD modelled using multi-
linear plastic link elements. Due to the compression free behavior, to provide energy dissipation for
both directions, two sets of CFDs are required for each frame connection fixed on each column face.
The analyses were performed for three models, with one being a monolithic frame connection, and the
other two for the hybrid frame system with and without the CFD. Comparing the performance of the
three systems, it was observed that CFD was successfully able to reduce the movement of the
structure during seismic excitation and dissipate energy successfully. In comparison to the monolithic
frame system, showed the ability of the hybrid frame system to allow elastic behaviour up to
significant drift ratios through the ability to gap-open at the beam column interface. These indicate its
feasibility for smart buildings for earthquake prone zones, allowing greater versatility by allowing
faster reoccupation after seismic events through replacement of the yielded coupons.
Keywords: Hybrid Precast Systems, Post-Tensioned Hybrid Systems, Energy Dissipaters, Seismic
Design, Finite Element Modelling, Damage Avoidance Design, Compression Free Device
1. Introduction construction time, higher concrete strength, and

greater quality control and facilitates ease of
Designing of structures resistant to seismic construction. According to the Precast Seismic
loading is a critical component of the design Structural Systems (PRESSS) research program,
phase of any high-rise building. With the recent “To a considerable extent, the reason for the
boom in the construction of high rises lack of advancement of precast structural
throughout the globe and the importance of systems in the United States can be attributed
considering the effect of dynamic loading on to uncertainty about their seismic
these structures, new advancements to provide performance…The codes require confirmation
economic and practical advantages over of the suitability of new designs by satisfaction
conventional systems have become a necessity. of performance criteria provided by expensive
A structural system which is able to dissipate and time consuming structural testing. This
the large energy of a seismic event while requirement inhibits innovation and has forced
avoiding damage to the critical structural research and design practice into a narrow
components would provide large economical focus of reinforced [monolithic] concrete
savings, while also enabling faster emulation instead of expanding the scope to
rehabilitation avoiding complex, costly repairs. take advantage of the strengths and differences
of precast construction.” [1]
Precast construction being a relatively newer
Eng. K.L.P.S. Perera, B.Sc. Eng.Hons. (AIT), Assistant
technique and owing to its reduced adoption
Lecturer, Department of Civil Engineering, SLIIT
especially in seismically active areas, the
amount of research performed on this Eng. W.D.C.R. Wanniarachchi, B.Sc.Eng. Hons. (AIT), Civil
technology for such applications has been Engineer(structural), MAGA Engineering (pvt) Ltd.
limited. Precast concrete technology when Dr. P. Thammarak, Ph.D.(Texas) Senior Lecturer, Dept. of
compared to conventional CIP concrete Structural Engineering, AIT
possesses several advantages such as reduced
1
51 ENGINEER
1.1 Hybrid Framing System 1.3 Objectives
The hybrid connection is a type of gap-opening The main objective of this paper is to analyse
system whose behaviour upon being acted on and compare the behaviour of a precast frame
by a lateral force is shown in Figure 1. The under selected loading conditions where the
opening of the beam-column interface allows precast frame uses a pure hybrid connection;
drift in the frame while causing minimal and as an alternative, the hybrid connection
damage to the main structural elements. modified with an external energy dissipater. A
sacrificial member which can be readily
replaced after an earthquake which aims to
prolong the viable lifetime of the building as
well as reduce repair costs and retrofitting costs
by reducing damage sustained by the core
structural elements. Also reducing time for re-
occupancy after seismic activity.
Figure 1 – Gap-Opening Behaviour
The pure hybrid frame system uses a
combination of mild-steel and post-tensioning In designing and construction of precast
(PT) strands which gives it the hybrid naming. systems, for determination of optimum seismic
During a seismic event, the PT strands are performance of a given precast structure, the
designed so that they remain elastic and selection of the type and the accurate detailing
provide a restorative force which will be able to and specification of the beam-column
move the frame back to its original position connections is of paramount importance. The
after being acted on by lateral forces, which two bodies, which have played a significant
gives it a self-re-centering behaviour[2]. The role in the process of the development of
mild steel is designed to dissipate the energy precast systems having improved seismic
during the event by yielding. The compressive performance, have been the programs initiated
forces imposed on the joint by the action of the by the National Institute of Standards and
PT strands gives rise to shear resistance at the Testing (NIST) and the Precast Seismic
joint through friction at the interface, which Structural Systems (PRESSS). The PRESSS
will aid in the shear force transfer. This is the research program was formed by the United
energy dissipation mechanism which in States and Japan as a joint effort between the
contrast to the typically preferred beam hinge two countries for the “large scale testing for
mechanism where the plastic hinge is formed seismic response of precast concrete buildings”.
on the beam, the plastic hinge instead forms in [1]
the connection region and in controlled
elements which gives it a damage avoidance
design concept allowing the main structural
elements to avoid being damaged during
seismic events.A successful implementation of
the hybrid precast connection system is in „The
Paramount‟ a 39 story 128 m high building; the
tallest precast concrete framed building in
seismic Zone 4.
1.2 Statement of the Problem
The problem in most systems is that the energy

dissipater is situated inside the connection. This Figure 2 – Comparison of the normalized
will lead to being unable to repair or replace the cyclic energy[3]
energy dissipater after it has yielded during an
earthquake. This will make the structure not As seen in Figure 2 testing of various hybrid-
able to resist further seismic events unless framing system connections with different
costly, labour intensive and time-consuming energy dissipaters provided varying levels of
repairs are done. energy dissipation notably being most of them
being better than the monolithic framing
system at drift ratios below 1.5%. The
ENGINEER 2
52
advantage seen by the NPZ4 specimen using The Cam Grip Energy Dissipation Device
stainless steel was that the connection was still (CGDD) shown in Figure 4 consists of a
functional even at an extreme drift ratio of 6%. sacrificial steel member between two cam-grips.
All the hybrid systems managed to be in a When under tension, the grips will lock and
serviceable condition at higher drift ratios allow force to be carried by the member and
compared to the monolithic connection yield. When a compression occurs, the grips
indicating their versatility.[3] release and allows the member to slide freely
preventing buckling. This behaviour allows the
2.1 Energy Dissipaters for Hybrid Frame use of a thin steel member due to only tensile
Systems loads being carried. Also the external
The typical energy dissipater in HFS is the mild installation allows observation and faster
reinforcing bars inserted at the beam-column replacement of the yielded member after
interface which yield and provide the nonlinear seismic activity.
inelastic energy dissipation of the seismic
energy. The disadvantage of the typical 3. Methodology
methods was that the energy dissipater is
For analysis of the proposed structural systems
situated within the connection, hence difficult
a series of Non-Linear Static and Dynamic
to access for maintenance, inspection or
Analyses were performed to determine and
replacement, which leads to potential problems.
examine the behaviour and performance of the
structures to determine the efficiency of
utilising the CFD for the Hybrid Framing
Connection.
The following are the key procedures followed

during the study:
1. Selection of the subassemblies for the
analyses, loading criteria, and other
Figure 3 – Energy Dissipation Behaviour required inputs.
2. Determine method of modelling and
In the Figure 3 it can be observed the analysis for each system.
components that make up the specific „flag 3. Modelling and Analysis of each system.
shaped‟ hysteretic backbone curve for a hybrid 4. Evaluation and comparison of the
connection with an energy dissipater installed. obtained results.
The PT tendons provides the self-centering
force without yielding, which allows it to The selected subassembly for the analyses was
continuously provide the recentering behavior a cruciform shaped beam and column joint,
to the connection. The energy dissipater which was used for the research conducted by
provides the energy dissipation in this Thien, 2015);[4] an experimental test on the
connection as depicted. The combination of the efficiency of using a CFD. The beam-column
two provides the desired behaviour of the subassembly was selected instead of a full
hybrid connection which allows it to both be a frame to be analysed since during the literature
highly ductile connection while dissipating review it was determined that the modelling of
adequate energy and preventing residual a full frame system having a hybrid frame
displacements in the structure. system incorporating gap-opening behaviour,
post tensioning strands and external energy
dissipating devices is a much complicated task
several magnitudes more cumbersome than the
modelling of a simple steel or concrete frame
structure. Also the required large number of
calculations for the design of individual
connections at different joints of the frame
structure due to varying section sizes, capacity
demands and the complexity involved in
incorporating these into a full frame model
Figure 4 – Behaviour of CGDD system further justified the selection of the cruciform
shaped subassembly. Another determining
factor was the very high computational
3
53 ENGINEER
demand involved in the analysis of a full frame For the modelling of the subassembly, the
structure. literature review revealed three approaches for
The properties of the test specimen beam was modelling the hybrid frame connection.
length of 1200mm, width of 170mm and 270mm
depth as shown in Figure 5. The column was The discrete springs approach as shown in
2000mm tall with a square section of 320mm x Figure 7,where a number of linear and non-
320mm. The beam was post tensioned with two linear link elements are used to emulate the
7-wire strands in PVC ducts each tensioned up behaviour at the joint was a considered method
to 8.4 Tons. since it was tested by previous researches as
being substantially accurate over the single
spring method where a single non-linear
rotational spring emulating the combined
behaviours of the springs in the discrete
model[5]. The complexity in carrying out the
large number of calculations for determining
the behaviour of the joint in the single
rotational spring model indicated that the
discrete spring model provided the best
approach with suitablefidelity in the
computational analysis. Also the level of detail
provided by using the discrete spring model
allowed finer control over the modelling
Figure 5 – Test Specimen Details process with adjustments to the parameters of
the individual elements at the connection not
3.1 Loading and Design Criteria
resulting in changes to the whole model as with
The support conditions are as shown in Figure a single spring model. The discrete spring
6 with the base pinned and the other ends model used also allowed the depiction of the
supported on rollers. Site location was selected gap-opening behaviour at the beam-column
as Los Angeles, California due to its high interface for the lateral loading.
seismicity where Hybrid Frame Systems have
been considered for construction. Seismic The first model attempted was using simple
loading will be carried for Northridge, Kobe frame elements to model the beam and column
and Loma Prieta earthquakes, of which all three and using a similar approach as (Ricles J. , et
have had significant effects on structural al.). [6] for discretising the connection into
designs, and all three sets being used having master and slave nodes connected through
determined as having 2% probability of rigid body constraints to keep plane section
exceedence of occurring in 50 years for the Los hypothesis and enforce the boundary
Angeles area. conditions at the joint and through the use of
multiple link elements to model the connection
region. This model was developed further to
incorporate the CFD but the difficulty of
incorporating the effect of the gap-opening on
the behaviour of the beam led to selection of a
new method for the analysis.
Figure 6 – Support Conditions

A set of Non-Linear Pushover Analyses was
also performed. The pushover analysis was
done as a reversed cyclic loading based on a
drift monitoring approach. For each loading
cycle, 30 sub steps were used as a minimum for
the loading to improve the data accuracy in
reporting and to improve convergence.
Figure 7 – Discrete springs model [7] (left) and
3.2 Determine method of modelling and single rotational spring model [8] (right)
analysis
The technique used to simulate the gap-
opening behaviour at the joint and the non-
ENGINEER 4
54
linear property of the CFD was a discrete The design for the compression free device was
springs approach combined with using of shell also conducted through a nonlinear FEA
elements to model the frame elements. through using „nonlinear‟ layered shell
elements in the software to obtain the force vs.
The frame elements were modelled using thick displacement plot for the steel coupon in
shell elements using 4000psi (27MPa) concrete tension through the application of a monitored
with a non-linear mander model utilized for displacement force.
simulating the non-linear behaviour of
concrete. Thick shell area property was selected 4. Results and Discussion
since it provides higher accuracy for shear
deformation and since the thickness of each This section presents and examines the results
shell element is considerably large compared to obtained from the various non-linear time
its other dimensions. history analyses and the non-linear static
pushover analysis for reversed cyclic loading.
The gap opening behaviour at the connection The resulting behaviours were compared to
interface was emulated by the use of link determine if the Hybrid Frame System (HFS)
elements having the “gap” property with very using the CFD provides performance for
high axial stiffness in compression, but zero consideration for practical implementation in
stiffness in tension, allowing the beam and real world design and construction of seismic
column to separate freely, but emulating the resistant building structures.
actual interaction of compressive crushing
when the gap closes. For each structure that was analysed, a set of
predetermined analysis outputs were obtained.
These were then compared to the outputs from
the other models to determine the efficiency of
each system, specifically to determine the
efficiency provided by the CFD, to check if the
device is working as intended by dissipating
sufficient energy as observed from the energy
plots and the hysteresis diagrams for the CFDs.
4.1 Performance of the Hybrid Frame

Connection with the Compression Free Device
Upon analysis of each structure, the behaviour

Figure 8– Analysis model of HFC with CFD of each was observed to be in line with those
expected of the structures through the literature
The CFD (Compression Free Device) was review. The gap-opening behaviour at the
modelled using a link property using a multi- connection as well as the effect of post
linear plastic property link element. The tensioning was seen in the analysis results as
material selected for evaluating the properties can be seen in Figure 10.
of the steel coupon was A36 steel with a yield
strength (Fy) of 36MPa and ultimate strength
(Fu) of 58Mpa and a Poisson‟s ratio (ν) of 0.26.
The force-displacement (F-Δ) behaviour of the
sacrificial rod was manually calculated,
optimized, and given for the properties of the
link as shown in Figure 9. The compression free
behaviour was given by providing the link with
negligible stiffness when under negative
displacement.
Figure 10- Stresses at 6% drift for HFC
Figure 9 – Force-Displacement graph of CFD
5
55 ENGINEER
For the Nonlinear Time History Analyses 4.2 Performance of the Hybrid Frame
performed on the subassembly with the post- Connection with the Compression Free Device
tensioning device installed, results were for NSP
obtained for the three ground motions that
were applied. For the Tab as ground motion the Figure 15 shows the hysteretic behaviour for
following graphs depicts some of the obtained the CFD obtained from the Non-Linear Static
results for the structural behaviour such as the Pushover Analysis conducted on the hybrid
drifts at the top, and the hysteretic behaviour of precast frame model with the compression free
the CFD during the loading, including the device installed. The recorded drift at the top of
energy plots for the various energy components the column with each step is shown in Figure
through the time period; also in Table 1. 14:
Table 1 - Energy Components(kJ)of HFC-CFD

Input Kineti ModalDam Link
c p Hystrtc
TABAS 0.4256 0.0211 0.3837 0.0415
KOBE 0.0291 0.0009 0.0209 0.0041
N.RIDG 0.1322 0.0190 0.1026 0.0146
E
Figure 14 - Imposed drift vstime step for NSP
It was observed from the analysis that the

behaviour of the CFD was following the
idealized compression free behaviour through
the range of motion and was dissipating energy
according to the provided parameters though
fracture strain was not realized. The Figure 15
depicts the hysteresis behaviour of the east side
CFD during the NSP testing.
Figure 11 - Drift vs time for LA30 (Tabas)
Figure 15 - Hysteresis behaviour of CFD on

East side for NSP
Figure 12 - Hysteresis for CFD for LA30
4.3 Comparison of the Hybrid Frame
System with the CFD and without the CFD
For the tested dynamic analyses, a comparison

of the parameters such as the energy
components and the drifts was done to identify
the efficacy of introducing the CFD to the
already versatile Hybrid Framing System. The
following figures indicates the comparison of
the two systems.
Figure 13 - Energy plots for LA30
ENGINEER 6
56
Table 2 - Drift at top of column comparison
NO CFD % Avg.
CFD redu %
Tabas Max 0.00321 0.00275 85.75
75.31
Min -0.00349 -0.00226 64.86
Kobe Max 0.00304 0.00133 43.65
53.39
Min -0.00256 -0.00162 63.13
North Max 0.00144 0.00125 86.75
Ridge 89.68
Min -0.00177 -0.00164 92.60
Figure 16 - Energy comparison for Kobe-LA22
5. Conclusion
From the literature review performed of hybrid
precast frame systems, it was expected to have
higher drift values for the post tensioned
hybrid connection, and to observe reduction of
the drifts of the hybrid frame connection when
using the compression free device. From the
inspection of the results of the analyses, the
results expected from the literature review were
observed successfully. The CFD was observed
Figure 17 - Energy comparison for Northridge-
to be successful at providing energy dissipation
LA28
to the tested model as observed by the
reduction in the drift of the system during the
earthquakes applied to the models. The
yielding behaviour during tension and the free
movement of the CFD during compression and
it‟s energy dissipation was successfully
observed in the hysteresis diagrams of the
CFDs although the energy dissipation capacity
was not as good as expected to make it
performance efficient.
Figure 18 - Energy comparison for Tabas-LA30 From the experimental research conducted by
Thien [4] it was observed that by using a single
Figures 16to 18shows the energy component CFD passing through the column resulted in
plots for the HFC with and without the CFD.It the device not being able to provide any energy
is observed that in all the ground motions, the dissipation due to failure of the steel coupon to
presence of the CFD has led to less energy reach the yielding strain due to relative
being dissipated through modal damping in the movement of the connection elements. The use
structure, which indicates that less damage will of two CFDs on either face of the column as
occur to the structure and that the drifts will be tested during this study would give a possible
minimized. However, from the results we can solution to the problem faced when using a
observe that a majority of damping occurs single CFD, as well as providing better energy
through modal damping. In the Kobe ground dissipation though being more cumbersome
motion, it manages to dissipate a larger amount due to requiring two sets of devices for each
of energy compared to that occurring in the connection.
other ground motions; this can be attributed to From this study, it was confirmed that using
the fact that the Kobe ground motion has a the CFDs in a hybrid frame connection can
greater period with less intense shocks, which make the hybrid connection dissipate sufficient
allows the link to dissipate energy properly. energy to make it a viable system and
combined with the ease of inspection and
When comparing the drifts as in Table 2, it can replacement of the steel coupon makes the use
be seen that the CFD device manages to reduce of the CFD even more appealing.
the drift at the top of the column for all the
ground motions, especially in the Kobe ground The major drawback faced with this system was
motion. in the modelling phase, the highly complex and
7
57 ENGINEER
computationally expensive procedure required system could provide advantages in the Sri
to model the Hybrid Frame System would pose Lankan construction industry context over the
a critical challenge to the widespread adoption CIP system currently used.
of any form of Hybrid Frame System unless
significant advances in computing or an Acknowledgement
alternative method to capture the behaviour of
the connection is discovered. The current The authors‟ wishes to acknowledge the
methods used for the analysis of Hybrid Frame assistance given by the Asian Institute of
Systems and other Self Centering Moment Technology, Thailand, Dr.Punchet Thammarak
Resisting Frames in concrete and steel requires and associated masters students for providing
expert knowledge in Finite Element Modelling the valuable opportunity to pursue this
or other alternative unconvential analysis research and for providing guidance and
methods, which are typically not used in knowledge.
current mainstream design processes. This
makes Hybrid Systems a controversial system References
with lots of benefits but with limited adoption.
1. Priestley, M. N., “Overview of PRESSS Research
Program”, PCI Journal, Vol. 36, No. 04, 1991, pp.
5.1 Recommendations
50-57.
From the results obtained from the analysis and 2. El-Sheikh, M., Sause, R., Pessiki, S. and L. Lu,
the methodology followed in this study, it was "Seismic behaviour and design of unbonded
observed that the analysis and design post-tensioned precast concrete frames," PCI
procedure for hybrid frame systems is Journal, vol. 44, no. 3, 1999, pp. 54-71.
computationally expensive in terms of the time 3. Cheok, G., and Stone, W., "Performance of 1/3-
and effort required for modelling and analysis. scale model precast concrete beam-column
Therefore it would be beneficial to future connections subjected to cyclic inelastic loads—
researches and also would aid in the Report No. 4.," National Institute of Standards
widespread adoption of the Hybrid Frame and Technology, Gaithersburg, 1994.
System, to develop a commercial software
package designed to aid in the analysis and 4. Thammarak, P., Thien, M., Htun, L., and Khine,
M., "The Development of a Compression-Free
design of Hybrid Frame Systems for both
Energy Dissipative Brace," in Proceedings of the
concrete and steel moment resisting frames. Tenth Pacific Conference on Earthquake
This would massively reduce the workload Engineering, Sydney, 2015.
involved in the analysis and design procedure
5. Christopoulos, C., Filiatrualt, A., Uang, C. and B.
for this type of system which is a highly
and Folz, "Post-tensioned Energy Dissipating
suitable structural systems solution to highly Connections for Moment-Resisting Steel
earthquake prone areas while reducing the Frames," Journal of Structural Engineering, vol.
post-earthquake damage to the structures. 128, no. 9, 2002, pp. 1111-1120.
Future researchers could study how the 6. Ricles, J., Sause, R., Garlock, M., and Zhao, C.,
placement of the device in the system such as "Posttensioned Seismic-Resistant Connections
the mounting location and the number of for Steel Frames," Journal of Structural
Engineering, vol. 127, no. 2, pp. 113-121, 2001.
devices could affect the performance of the
connection where a larger number of devices
7. Amaris, A., Pampanin, S., Bull, D., & Carr, A.
with elements yielding at certain loadings “Experimental Performance OfHybrid Frames
would be effective. Higher yield devices for Systems With Non-Tearing Floor Connections”
reducing damage at large displacements and 14th WorldConference in Earthquake
lesser yield strength devices for Engineering (14WCEE). Beijing, China:
controllingdrifts for smaller imposed WorldConference in Earthquake Engineering,
displacements. 2008.
8. Dobossy, M., Garlock, M., & VanMarcke, E.

For countries like Sri Lanka having recently
“Comparison Of Two Self-centeringSteel
started on major construction projects such as
Moment Frame Modeling Techniques: Explicit
significant high rises; considering the Gap Models, And Non-linearRotational Spring
importance of faster construction, higher Models”. 4th International Conference on
quality required, cost effectiveness, the EarthquakeEngineering, (p. Paper No.101). Taipei,
feasibility of precast concrete hybrid frames Taiwan, 2006.
could be researched upon to evaluate if the
ENGINEER 8
58
Annual Sessions of IESL, pp. [1 - 10], 2017
© The Institution of Engineers, Sri Lanka
Study on Filler Slab Construction

K. Baskaran, A. Gopikrishna and P. Thiruvarankan
Abstract: During the design process while fulfilling the functional requirements engineers need
to consider ways to reduce the cost of construction. Filler slab construction, which reduces the self-
weight of slabs and has reduced loads along the rest of the load path up to foundation, is an effective
way for cost reduction. However, application of this technology is found to be limited in Sri Lanka
due to the vacuum in knowledge. In the post war resettling program, in the Northern Province of Sri
Lanka, roof slabs of few houses were constructed with filler slab technology, similar to those adopted
in India. In the present study, filler slab construction is revisited, with the sole objective to throw some
light on this approach. Literature survey, field survey, desk study, physical modelling and finite
element modelling were some methods adopted in the present study. The purpose of this paper is to
share the findings of the present study with the industry. This paper includes brief introduction on
filler slab technology, state of the art found from literature and field surveys, design methods adopted
in the present study (solid slab approach and flange beam approach), details of two series of
experiments on filler slab (specimen with burnt clay tiles as filler material) and solid slab (control
specimen), the observations made and subsequent finite element modelling details. The purpose of the
two series of experiments is to study the behaviour of filler slab under distributed load and patch
load. Flexural failure and post flexural punching shear failure were observed during the physical
model testing. The paper briefs on cost comparison and construction difficulties before concluding
with summary of findings in the present study.
Keywords: RC solid slab, Flexure, Punching shear
1. Introduction rooms, lecture halls, conference halls and

auditoriums.
While transferring the applied loads in a safe
manner, reducing the amount of materials Although the filler slab technology was
needed and controlling the escalating cost are adopted in few buildings constructed in the last
the challenges ahead of structural engineers. As century in Sri Lanka, this technology is not
commonly practiced, concrete in tension is practiced in the recent past. Recently for the
considered as cracked and not contribute for post war resettling programme, in the Northern
flexural load carrying action in a simply province of Sri Lanka roof slabs of few houses
supported reinforced concrete slab subjected to were constructed with filler slab technology,
transverse loading. Therefore, not much after looking at similar construction in India.
concrete is needed to the bottom part; except to However, theoretical or experimental evidence
hold the reinforcement together, provide cover are not there for adoption. Hence the aim of the
to the reinforcement and shear resistance. present study is to throw some light on filler
Therefore significant amount of the concrete slab technology. The objectives of the present
can be replaced by the lightweight materials study are to
such as mud blocks, bricks, hollow blocks, semi  Identify a design approach
earthen pots, PVC pipes, coconut husks,  Find the behaviour up to failure
autoclaved solid blocks and calicut tiles. The (flexure or punching shear)
amount of replaced concrete depends on the  Identify any difficulties during
thickness, shape of the filler material and the
construction and compare cost
amount of concrete filled between the voids of
the filler materials. 2. State of the art
According to the literature review and field
Use of filler materials reduces the self-weight of survey, calicut tile was selected as a suitable
the slab and results in savings in the required filler material. The parameters considered in
amount of reinforcement and concrete [5]. By the selection process are dimension of the slab,
using suitable filler materials, aesthetic
appearance and thermal comfort inside the Eng. (Dr.) K.Baskaran, B.Sc.Eng.(Peradeniya), Dr.Eng.
building can be improved [6]. According to (Cambridge), Senior lecturer of Civil Engineering,
Department of Civil Engineering, University of Moratuwa.
Jaisingh & Singh [7], sound absorption of the
filler slab is greater than the normal reinforced Eng. A.Gopikrishna, B.Sc.Eng.(Hons.)(Moratuwa),
QA/QC Engineer, Access Engineering PLC.
concrete slab therefore it is suitable for class
Eng. P.Thiruvarankan, B.Sc.Eng.(Hons.)(Moratuwa),
AMIE(SL).
59
1
material cost, fire resistance, water leakage Then the shear resistance and the deflection
through the filler materials, aesthetic view, check were done according to clause 3.4.5 and
thermal comfort and availability. Due to lack of clause 3.4.6 of BS 8110: part 1: 1985 ([2]) to
fire resistance, coconut husk and the PVC pipes ensure that the filler slab will not fail in other
are not suitable materials for filler slab modes.
construction. When considering the dimension 3.2 Flange beam approach
of the slab, hollow blocks, semi earthen pots Consider a flange beam for which the web
and mud blocks are not suitable in Sri Lanka. thickness is bw. The equation for total tensile
Aesthetic view of the slab with calicut tiles is force in the steel is same as above. To find the
better than other filler materials. Hence, calicut total compressive force, the equation (2) is
tiles were chosen as the filler material for this slightly modified since “b” is replaced by “bf”
study. (Figure 3). Here also shear resistance and
3. Theoretical study
In a theoretical study of slabs, both flexure and
punching shear are needed to be considered at
ULS as they are the governing failure modes.
Section 3.1 and 3.2 discuss the flexure theory
and the section 3.3 discusses the punching
shear theory. Two different approaches were Figure 3 - Flange beam approach diagram
used to determine the flexural resistance of the
filler slab and the RC solid slab according to deflection check were done according to clauses
clause 3.4 of BS 8110: part 1: 1985. The mentioned in solid slab approach (section 3.1).
considered cross section is shown in Figure 1.
3.3 Punching shear check
a) ACI 318:1995 code ([3])
Shear stress caused around the columns in slabs

due to the direct shear and moment is given by,
(a) Section 11.12.6.2, vu = Vu/A + γ*Mu*c …(4)

(b)
Figure 1 – Cross section of filler slab (a) and Allowable stress
RC solid slab (b) Section 11.12.2.1, vc = minimum ( 0.17(1+2/
βc)*√fc, 0.083(2+ αs *d/u) *√fc, 0.33*√fc) …(5)
3.1 Solid slab approach
Consider a strip of slab with width “b”. The If vu <φvc,
No punching shear reinforcement is required
b) BS EN 1992-1-1:2004(E) code ([4])
Equation 6.38, vEd= βVEd/uid …(6)
Equation 6.9, VRd,max= αcwbw zv1fck/γ(cotθ+tanθ)

… (7)
Figure 2 - Solid slab approach diagram
Section 6.4.4, VRd,c = CRd,c k (100ρl fck)1/3 + k1σcp
total compressive force in concrete is C. The ≥ (vmin + k1σcp) …(8)

total tensile force in steel is T (See Figure 2).
C = 0.45fcu*b*0.9x …(1)
T = 0.87fy*As …(2)
4. Experimental study
4.1 Materials
Where, x is the depth of the neutral axis. Two series of experiments were carried out
By equating forces x was obtained. Hence (one was subjected to line load to check the
moment resistance (M) was found. flexure and other one was subjected to
concentrated load to check punching shear
M = 0.87fyAsz …(3) capacity). Ordinary Portland cement was used.
10mm diameter ribbed steel bars were used for
main bars and distribution bars. The Calicut tile
60
2
was used as the filler material. Grade 20
concrete was used and the mix design was done
according to DoE method. 10mm and 20mm
maximum aggregate sizes were used for Series
one experiment and Series two experiment
respectively. Here, Series one experiment
represents the flexure model (aggregate size
immaterial) and Series two experiment represents
the punching shear model (aggregate size
matters).
4.2 Casting procedure
1.51m length and 1.05m width flexure model
slabs and 1.46m length and 0.965m width
punching shear model slabs were casted with
concrete having w/c ratio of 0.52 and tested in
the laboratory. Thickness of the both slabs used
in Series one and two experiment were 125mm
and the bottom and side cover to reinforcement
was 25mm. The dimensions of the both slabs
were selected with respect to the size of the test
rig. The filler slab and RC solid slab model
design was carried out with the concern of the
ribbed slab approach which is given in clause
3.6 of BS 8110: part 1: 1985 and clause 3.4.4.4 of Figure 4.2 – Details of filler slab (a) and RC
BS 8110: part 1: 1985 respectively. Spacing of solid slab (b) for Series two experiment
the main and secondary bars are 330mm and
480mm respectively in Series one experiments
and 470mm and 305mm in Series two
experiments. Both main bars and distribution
bars were tied together and reinforcement mesh
was formed. Then it was placed in the timber
formwork. Two calicut tiles (one laid on top of
the other) were stacked within the rectangular
space formed by the reinforcement which is
showed in the Figures 4.1 and 4.2.Before
placing the tiles, tiles were soaked in water for
ten minutes for the saturation purpose to avoid
the absorption of water from the concrete.
Concrete with high workability was used to
spread well between the tiles and
reinforcement. Manual compaction method was
used because the tiles might be broken by the
poker vibrator. The casted slab models are
shown in Figure 5.
Figure 5 - Casted filler slab (a) and RCsolid

slab (b)
4.3 Testing arrangement

The RC solid slab and filler slab were tested
after 28 days, after ensuring the control cube
samples have achieved the desired compressive
strength value (Average cubes strength of series
Figure 4.1 – Details of filler slab for Series one one experiment & series two experiment are
experiment 21.2 MPa & 26.38 MPa respectively). The slab
was simply supported at two edges and
61
3
subjected to an increasing concentrated load for
punching shear model and distributed load for
flexure model at middle. The increasing point
load, distributed load was applied with the
help of a hydraulic jack to the punching shear
model and flexure model respectively.
A proving ring was used in the loading

arrangement to measure the applied loads. Dial
gauges were fixed at the edge span of the slab
for punching shear model and fixed at three
points (middle and two edges) of the slab for
flexure model as shown in Figure 6, to measure
(a) (b)
the deflection. The applied load and the
Figure 7.1 - Observed crack pattern of filler
corresponding deflections were recorded.
slab (a) and RC solid slab (b) for Series one
experiment
(c) (d)
Figure 7.2 – Observed crack pattern of filler
(a)
slab (c) and RC solid slab (d) for Series two
experiment
90
80
70
60
(b)
Load (kN)
50
Figure 6 - Testing Arrangement for Series one
experiment (a) and Series two experiment (b)
40
4.4 Test results
30
The effective length between the supports was
865 mm and 1420 mm for punching shear 20
model and flexure model, respectively. The
crack was observed at bottom of the middle 10
span of the slab first and propagates diagonally
for punching shear model and bottom of the 0
middle span of the slab for flexure model as
0 5 10
shown in figures 7.1 and 7.2 respectively. The
failure load was considered as ultimate load. Deflection (mm)
Then the serviceability load and the
corresponding deflection were determined Filler slab RCC slab
from Figures 8.1, 8.2 and ensured that the
deflection was within the allowable limit.
Figure 8.1 - Load Vs Deflection graph for
Series one experiment
62
4
6. Construction difficulties and
100
feasibility of filler slab
90
80 During the casting of the filler slab, manual
compaction method was used. Because poker
70 vibrator might break and disturb the alignment
60 of the filler material. In big projects, using
Load (kN)
manual compaction method is not possible. So,

50 using self-compacting concrete may be a
40 solution to overcome from this issue. High
workable concrete for the filler slab was used
30
due to the adequate spreading of concrete
20 between the calicut tiles. Super plasticizers can
be used to overcome this problem. Time
10
consumption for the filler slab casting is higher
0 than the reinforced concrete solid slab casting
0 5 10 15 due to time taken for calicut tile laying. It is
Deflection (mm) possible to reduce the time consumption by
regular practice in filler slab construction.
Filler slab RCC slab 7. Finite element modelling (FEM)

Figure 8.2 - Load Vs Deflection graph for ANSYS 12.1 software was used to analyse the
Series two experiment results of load, deflection and crack pattern of
5. Cost comparison both filler and RC slab, whether it is similar to
the results of Series two experiment model or
Both filler slab and RC solid slab of punching not. For concrete and Calicut tiles, solid 65
shear model was considered for this element and for reinforcement bars link 8
calculation. Table 1 describes the cost of two element was used. Loading setup, properties of
types of slabs. the materials and boundary conditions were
same as from experimental study. Non-linear
Table 1 - Cost comparison between filler slab analysis was done to this model. Load also
and reinforced concrete slab applied as step by step by using load step
option 9 ([1]). Results from ANSYS software are
Unit Total
shown in Figures 9& 10.From ANSYS software
Slab Material cost Quantity cost
analysis, RC solid slab failed at ultimate load of
type (Rs) (Rs)
98.56kN with 9.2mm maximum deflection and
Concrete
14000 0.11 m3 filler solid slab failed at 92.51kN ultimate load
(m3)
with 6.9mm maximum deflection.
Steel
Filler 120 5.77 kg
(kg) 2952
slab
Calicut
tile (one 40 18
tile)
Concrete
RC 14000 0.20 m3
(m3)
solid 3492
Steel
slab 120 5.77 kg
(kg)
Above total cost was calculated for 1460mm *

965mmslab panel of the casted filler slab and
the unit cost of the materials were as of
December 2016 ([8], [9]). Cost of the filler
material was not considered because unwanted
materials with low cost or no cost can be used.
According to the table 1, 16% cost can be saved (a)
by using filler slab as an alternative.
63
5
8. Theoretical, experimental and FEM
results comparison
Theoretical, experimental and FEM results of
Series one and two experiments were compared
and checked. Comparison of the results are
given in Tables 2 and 3 for Series one and Series
two, respectively.
Table 2 – Theoretical and experimental results

of moment capacity, deflection, shear for
Series one experiment
Theoretical
Experimental
Flange beam
Solid slab
(b)
approach
approach
Figure 9 - Load Vs deflection of filler slab (a) Analysis
and the RC solid slab (b) for Series two
experiment
In above graphs, „VALU‟ represents the Flexural

deflection in millimetres and „TIME‟ represents failure load
52.23 40.29 39.99
the load in newtons. of filler slab
(kN)
Flexural
failure load
62.56 39.29 38.99
of RCC solid
slab (kN)
Shear
resistance of
- 77.9 118.8
filler slab
(kN)
Shear
resistance of
- 76.9 76.9
RCC solid
slab (kN)
(a) Deflection of Actual deflection <
filler slab 7.05 Allowable
(mm) deflection
Deflection of Actual deflection <
RCC solid 7.45 Allowable
slab (mm) deflection
In Series one experiment, experimental flexural

failure load of filler slab and RC solid slab are
higher than the theoretical value. Reason
behind the different values of theoretical and
experimental is yield stress of the steel may be
increased after the strain hardening. And the
flange beam approach shear resistance of the
filler slab is higher than the solid beam
approach. But the filler slab was considered as
(b) set of flange beam is an effective way to
analyse. Reason is the reduction in shear area of
Figure 10 - Crack pattern of filler slab (a) and concrete is considered in this method when
the RCC solid slab (b) for Series two analysing for shear stress. Both slabs failed
experiment under flexure during the experiment, so that no
values in experimental values for shear for
Series one experiment. Both experimental and
theoretical deflections are in allowable limit.
64
6
Table 3 – Theoretical, Experimental and FEM the experimental and theoretical of both slabs
results of moment capacity, deflection, shear are in safe limit and FEM final deflection at
and punching shear for series two experiment failure also approximately same as
experimental values. Theoretical punching
Theoretical shear capacity from two different codes (ACI
FEM
Experimental
and BS EN 1992) is lower than the experimental
Flange beam
Solid slab
approach
approach
values.
Analysis
9. Yield line Analysis
Main concept behind the upper bound theory is
Flexural actual collapse load is less than or equal to
failure mechanism load. And also, yield line patterns
load of 47.04 47.89 48.59 - also can be checked for each boundary and load
filler slab condition. Yield line analysis was carried out
(kN) for the punching shear model. For the moment
Flexural capacity, core cutter test and tensile strength of
failure the steel were measured. The results of the
load of 43.4 46.89 47.59 - yield line analysis are given in Table 4,
RCC solid
slab (kN) Table 4 - Results of yield line analysis
Shear
resistance Failure Actual
- 70.5 116.1 -
of filler mechanism load collapse load
slab (kN) (kN) (kN)
Shear
RC slab 102.1 93.2
resistance
of RCC - 69.5 69.5 - Filler slab 86.34 84.15
solid slab
(kN) Failure mechanism load > Collapse load
Actual (Yield line analysis is satisfied)
Deflection
deflection <
of filler 7.8 6.9 10. Conclusions
Allowable
slab (mm)
deflection
Deflection Actual In the present study only a limited numbers of
model slabs were tested in the laboratory.
of RCC deflection <
9.8 9.2 However they revealed both solid slab
solid slab Allowable
approach and flange beam approach can be
(mm) deflection
used safely to design filler slabs. In Series one,
Punching
41.6 (ACI) flexural failures were observed and in Series
shear two, post flexural punching shear failures were
failure observed. For punching shear capacity the
84.15 92.51
load of 68.6 (BS EN prediction from ACI and BS EN are
filler slab 1992) conservative. However use of upper bound
(kN) yield line analysis gave reasonably close
Punching predictions.
40.6 (ACI)
shear In the present study even under patch load post
failure flexural punching shear failures have
93.2 98.56
load of 67.6 (BS EN happened. Mix design and use of compaction
RCC slab 1992) need extra care with filler slab approach. Also
(kN) filler slab construction needs more labour
contribution compared to RC solid slab
In Series two experiment, experimental flexural approach. Hence it is recommended to study
failure load values of both slabs are nearly same the whole life cost analysis for filler slab
as theoretical values. Flange beam approach incorporating the thermal, acoustic advantages
shear resistance values are higher than the solid too.
approach shear resistance like Series one
experiment. Here also, shear failure was not
noticed during the experiment. Deflection of
65
7
References
1. ANSYS Theory reference. ANSYS, Inc.

Southpointe, 275 Technology Drive, Canonsbury,
1999.
2. British Standard Insititution. BS 8110: part 1:

Structural use of design and construction, 1985,28-30
p.
3. American concrete Institute. ACI 318: Building

Code Requirements for Reinforced Concrete, 1995,
169-176 p.
4. British Standard Insititution Euro code 2: Design of

concrete structures - Part 1-1: General rules and rules
for buildings, 2004, 97-106 p.
5. Ayush, S., Cost Effective and Innovative Housing

Technology. International Journal for Scientific
Research & Development, Vol. 2, 2014, pp. 27-29.
6. Harrison, S. W. & Sinha, B. P. A study of

alternative building materials and technologies
for housing in Bangalore, India. Construction and
building materials, Vol. 9, No. 4, 1995,pp. 211-217.
7. Jaisingh, M.P, Jaisingh, L. & Singh, B., n.d. “A RC

filler slab with non-autoclaved cellular concrete”.
Central building research institute, Rotterdam, 1998.
8. http://www.siamcitycement.com/, Visited 10th

December 2016
9. http://www.ceylonsteel.com/, Visited 10th

December 2016
66
8
AnnualSessions
IESL,pp.
pp.[67[1- -74],
10],2017
2017
© The
© The Institution
Lanka
Energy Dissipater for Post-Tensioned Steel Frame

System
W.D.C.R. Wanniarachchi, K.L.P.S. Perera and P. Thammarak
Abstract: Following the Northridge earthquake in 1994, connection failures were found in steel
moment resisting structures with welded connections. Brittle fractures initiated within the connections
at a low level of plastic demand lead to the development of improved welded connections focusing on
the rotational ductile capacity. However, due to the limitation of these connections, steel moment-
resisting frames(MRF) with post-tensioned(PT) strands running parallel to the beam, which is capable
of providing adequate stiffness, ductility and strength during a seismic event are introduced. In this
paper, Compression Free Device (CFD), a more advanced alternative to the top and seat angle and
friction damper is introduced as the energy dissipating device for the PT MRFs. The CFD device with
two major components, sacrificial steel rod and grip is designed with the mechanism of dissipating
energy only in tension yielding and free movement during compression to prevent buckling of the
steel rod. In this study, detailed shell models of PT beam-column connection, PT connection with CFD
and welded moment resisting connection are developed and analysed under cyclic loading using a
commercial software and further development of the model to be done using an advanced finite
element analysis software. The model, Beam-Column cruciform assembly with two sets of CFDs
which is modelled using multi-linear plastic link elements, fixed on each face of the column flange is
analysed by performing the nonlinear-dynamic (Non-Linear Time History Analysis) and non-linear
static (pushover) analyses to compare the behaviour of three models on the basis of energy
dissipation, displacement and hysteresis. Time history analysis results indicate that the PT connection
with CFD exceeds the seismic performance of a welded moment-resisting connection subjected to the
same earthquake. Efficacy of the device in dissipating energy and ease of repairing and replacement
after an earthquake provides pronounced viability for seismic events.
Keywords: Post-Tensioned, Self-Centering Moment-Resisting Frames, Shell Modelling, Energy

Dissipaters, Welded Connections, Finite Element Modelling
1. Introduction considered the most seismic resistant detail of

any beam-column connection based on
The sudden release of energy during an experimental work since it can withstand high
earthquake due to the sudden rupture of the plastic rotation‖ [2], [3]. The top and bottom
fault within Earth‘s crust causes seismic waves, flanges of the beam are welded directly to the
low-frequency sound wave to propagate column by full penetration groove welds to
through the Earth's crust or along its surface. transfer moment. The beam web is bolted or
This phenomenon of seismic waves reaching welded to a shear plate, which is attached to the
the Earth‘s surface is the main reason for the column by welding to transfer shear [2].
ground to vibrate at certain frequencies. When
However, 1994 Northridge earthquake caused
the structure and ground share the structure‘s
unprecedented damage to conventional welded
natural frequency (resonance), it amplifies the
beam-to-column connections in steel MRFs.
effects of an earthquake, causing buildings to
Many brittle failures were reported. A common
suffer more damage. Steel MRFs are designed
failure pattern of beam-to column joints is
to resist earthquakes while the building should
cracking of the bottom beam flange to column
accommodate extensive yielding and plastic
weld. The main causes of this type of failure are
deformations and dissipate large seismic
weld defects at the root of the weld (downward
energy. Taking advantage of the ductility of
welding) and high stresses at the exterior of the
steel material, steel buildings are expected to
beam flange.
provide adequate ductility—the ability to
withstand large inelastic deformations without Eng. Ms. W.D.C.R Wanniarachchi, B.Sc.Eng. Hons. (AIT),
significant strength degradation and without Civil Engineer (Structural), MAGA Engineering (pvt) Ltd.
the development of collapse and instability [1]. Eng. K.L.P.S Perera, B.Sc. Eng. Hons. (AIT), Assistant
Lecturer, Department of Civil Engineering, SLIIT.
―Prior to the major earthquake of Northridge,
welded-flange-bolted-web connections were Dr. P. Thammarak, Ph.D.(Texas) Senior Lecturer, Dept. of
Structural Engineering, AIT.
67
In most of the tested ‗pre-Northridge‘ joint column flange. In each connection, either yield-
typologies were observed reduced values of based or friction-based energy dissipation
beam plastic rotation (<0.01 rad). Connection devices are installed.
failure should be avoided before the member
failure because as found in Northridge
earthquake the connection failure is not ductile
as that of steel member failure.
After evaluation of Northridge earthquake,

many efforts have been made find suitable
solutions for observed damage in pre-
Northridge steel beam-column connections and
to improve steel MRFs. Improved welded
unreinforced flange connection, Reduced beam Figure 1 - PT steel connection (a) Energy
section connection, Connection with welded dissipating bars [4] (b) Top & seat angles [5]
haunches: top and bottom haunch and bottom Beam depth, the PT force, strength and the
haunch were few suggestions. Nonetheless, elastic stiffness of the PT strands are the
earthquake-induced damage in steel MRFs is properties affecting the flexural behavior of a
still inevitable under moderate-to-severe post-tensioned connection. There will not be
earthquakes. The repair of damaged buildings any damage to the beam and beam will not
with high residual deformations may not be yield significantly till the PT strands remain
economically feasible; thus, demolishing may elastic.
be necessary for such intensively damaged
buildings. Researchers have proposed several types of
energy dissipating devices for post-tensioned
1.1 PT steel beam-column connection connections, such as replaceable energy
As an alternative to the welded connections, dissipating bars, top and seat angle and friction
researchers came up with the idea of using self- dampers. Post-tensioned connections were
centring systems to avoid structural failures characterized by their flag-shaped hysteretic
and performance weaknesses resulting from behavior and behaved in a similar manner
local residual drift at connection level and regardless of the energy dissipating device
global residual drift at the structural level. Self- installed in the connection [4]. But
centring behaviour was characterised by nearly discrepancies in the hysteretic behavior, the
zero residual deformation at the end of the resistance of the shear force and bending
cyclic loading tests, in which the system also moment depended on the type of the energy
dissipate the energy. The only type of dissipater.
behaviour which ensured these two abilities
was the flag-shaped hysteresis behaviour. Two
main self-centring strategies have been
proposed by researchers: using shape memory
alloys (SMAs) or using post-tensioning forces.
In the strategy of using post-tensioning force,
the technique used to recover the inelastic
deformations induced on the structure is pre-
stressing the structure before imposing loads.
This is considered a very reliable method to
pull the structure back to its original position
after a seismic event.
A PT connection consists of two main Figure 2 - Hysteretic behavior of PT

components; post-tensioned strands which connection with replaceable energy
provide the self-centering capacity and energy dissipating bars
dissipation devices installed in the connection
As illustrated in Figure 2, the PT connection
to dissipate energy.
initially behaved as a fully rigid welded
The connection utilizes high strength post- connection experiencing no relative rotation
tensioned strands which pass through the between the beam and the column when the
frame columns and run parallel to the beam applied moment was less than the moment
web along the span are anchored against the provided by the PT strands. A gap started
68
opening between the beam and the column deformations. Hence, more durable energy
when the applied moment is more than the PT dissipater based on friction was proposed [8].
strands moment. Energy dissipating device
starts contributing to the system at this instant PT strands and the friction forces in the friction
to resist the moment of the connection. device provided the moment resistance of the
connection. If the friction device is installed at
1.2 Energy dissipaters for PT steel frames the beam web, it is used to provide shear
resistance when the gap opens, and the contact
1.2.1 Energy Dissipating Bars area between the beam and the column flange
In the PT connection with energy dissipating is reduced.
bars, replaceable steel bars were installed in the
1.3 Compression free energy dissipater (CFD
connection. In order to avoid buckling energy
CAM-GRIP device)
dissipating bars were confined with steel
In this research CFD device, passive energy
cylinders [6]. These bars start to elongate and dissipating device which increases the energy
yield when the gap opens. Post-tensioned dissipating capacity and reduces the seismic
strand and the energy dissipating bars provide demand on the structure will be used [9].
the resistance to the bending moment [4]. Even
though shear resistance was provided by As illustrated in Figure 4, the device consists of
friction at the intersection of the beam and 2 main parts, one is a sacrificial member, and
the other one is grips. When the load is applied
column flange, it dropped substantially upon
on the specimen (or building) the sacrificial
gap opening. Energy dissipating bars
member (or steel core) will be in tension, thus,
contributed to the moment resisting in two the grips will generate a force to hold the steel
stages; first stiffness prior to yielding and core, allow it to yield, then energy is dissipated.
second stiffness in the post yielding stage. Reversely, when the load is removed, the steel
core will be in compression but two grips will
1.2.2 Top and Seat Angles
open up, allow steel core to freely slide in, so no
When the gap starts to open, top and seat compression force is taken by steel core. The
angles formed plastic hinges at the top and main advantage of CGDD is it will allow the
bottom [7]. Since three plastic hinges are steel core to get rid of compression and its
formed during the decompression and gap buckling. External installation of the device will
opening of the connection, the hysteretic provide the ease of repairing or replacement
behavior took a curved shape. once it‘s damaged after cycles of loading. The
zero internal force is expected in steel coupon
when it moves towards the compression free
device and energy will only loss in tension
yielding. The downward displacement of the
steel rod is resisted by inclined forces produced
by two cams.
Figure 3 - PT connection with seat angles as

energy dissipaters (a) Moment-rotation
behavior (b) Configuration [5]
1.2.3 PT Connections with Friction Damper
Even though both yield based energy

dissipaters, energy dissipating bars and seat
angles have high reliability and energy
dissipating capacity, they were required to be Figure 4 - Cam-Grip device with steel coupon
replaced after the seismic event due to plastic under motion towards CFD and away from CFD
69
2. Methodology 2.1 Mesh Design & Modelling
In order to investigate the behavior of steel Even though discrete spring model, integrated
frames with PT connection when subjected to a single rotational spring model were considered
ground motion, Ricles [10] designed a six-story as modelling techniques for this research, due
six bay moment-resisting frame using DRAIN- to technical difficulties and complicated
2DX. In Ricles research, the structure was calculations combination of the discrete spring
assumed to be an office building situated in model and shell modelling approach is adopted
seismic design category D. Hence seismic to model the welded beam-column connection
hazard exposure Group III is assumed. In this and the post-tensioned beam-column
research cross-section of beam and column was connection with Cam-Grip energy dissipating
based on the Ricles [10] research paper which device.
focuses on the comparison of the post-
In the welded beam-column model, effect of
tensioned moment-resisting frame with pre-
welding was considered by constraining all
Northridge welded beam-column connection.
degrees of freedom in the beam-column
A36 and A572 Gr.50 steel were used for the interface. The gap opening behaviour of the
designing of the beams and columns, and the post-tensioned beam-column connection is
floor dead load of 4.79 kPa, live load of 2.40 kPa modelled by non-linear link element with one
are considered. degree of freedom, axial. Self-centring
mechanism of high axial stiffness in
compression and zero stiffness in tension is
illustrated by link element in the software.
Four post-tensioned strands running parallel to

the beam web were designed using the cable
element. The cable elements were modelled
without being bound to the beam web shell
elements to simulate their external placement.
Each strand had an area of 280 mm², since the
referred 3D model had 8 post-tensioning
strands, each strand having an area of 140 mm²
.The material A416Gr270 is assigned for the
defined cable element. Post-tensioning force of
Figure 5 - Selected sub-assembly in elevation 729 kN is allocated per a cable element.
of Prototype Six-Story, Six-Bay MRF [10]
Multi-linear plastic link was used for modelling
Each PT connection had four levels of strands, of the compression free energy dissipating
with two strands, one on each side of the beam device (CFD). The properties of the link were
web, in each level. A post-tensioned strand calculated by a manual method using A36 steel
designed in this research had an area of 140 as the material of the steel coupon and
mm², a Young‘s modulus of 200 GPa, and an calculating the force – displacement behaviour
ultimate strength of 1,860 MPa. Post-tensioning of the link element with it having negligible
force of 729 kN was used. stiffness in compression. Gripping device
which used to fix the CFD to the beam flange is
depicted by a stiffened steel shell section.
Figure 6 - Schematic of a PT connection

Figure 7 - Cross section of the CFD
70
2.2 Analysis of the models Results obtained from the Non-Linear Time
History Analyses were taken for the drifts, base
The behavior of the post-tensioned beam- shears and energy plots. The following figures
column test assembly, PT beam-column with depicts the results of the system using the CFD
compression free energy dissipating device under Northridge EQ loading.
model and conventional welded moment-
resisting connection having the same geometry
and element properties were compared with
each other. The comparison was based on the
date obtained from the non-linear static
analysis (Pushover) and non-linear dynamic
analysis. Displacement, residual drift and
energy dissipating capacity were characteristics
observed for the comparison.
Time history analysis was done for nonlinear

evaluation of dynamic structural response
under loading. The lateral loads will be
provided as a time history loading of a selected
earthquake. According to the code ASCE/SEI 7-
10 [11], minimum 3 time history analysis are to Figure 8 - Gap opening behavior of the PT
be selected. Hence following seismic events connection with CFD
were selected from the Pacific Earthquake
Engineering Research (PEER) database;
 Northridge (LA28) (PGA: 1.33g, M: 6.69)

 Loma Prieta (LA21) (PGA: 0.472g, M: 6.93)
 Kobe (LA24) (PGA: 1.282g, M: 6.9)
The design response spectrum was constructed

and the selected time histories were converted
into the response spectrum using the PRISM
software. The scaled time histories were
converted from the response spectra and
applied to the model, which was analyzed to
obtain the maximum results. Pushover analysis
will be done by giving displacement controlled
force at the top in the both direction. Figure 9 – Drift (m) vs time (s) for LA28
(Northridge)
3. Results & Discussion
In this chapter the results obtained from the

Nonlinear Static Pushover analysis and
Nonlinear Time History Analyses of the three
tested models are depicted and the results are
analyzed to compare the differences between
the three systems to evaluate their benefits and
disadvantages.
3.1 Performance of PT connection with CFD
Evaluation of the analysis results of each Figure 10 – Energy Plots for LA28 (Northridge)
system was verified with the predicted [Energy in kJ vs. time (s)]
behavior of each system from the literature
From Figure 10 it can be observed that the CFD
review. The gap-opening behavior (Figure 8) at
does manage to dissipate some of the input
the connection and the energy dissipation
energy from the ground motion and reduces
through yielding of the CFDs were clearly
the modal damping of the structure. In
observed in the outputs.
71
addition, the hysteresis of the CFD can be
observed in Figure 11.
Figure 12 – Force (kN) vs displacement (m)

behaviour for NSP
3.3 Comparison of PT system with/without

CFD
Plots for LA28 (Northridge) for CFD Input energy, kinetic energy, potential energy,
modal damping of PT connection with CFD
Table 1 shows the energy components into device is compared with results of the post-
which the input energy was converted to. The tensioned connection without energy
Link Hysteretic column indicates the energy
dissipating device. Below figures depict the
dissipated through the yielding of the CFD.
comparison of the results for the three seismic
The energy dissipation through the CFDs was events.
observed to be plateauing at a certain value as
observed in Figure 10 indicating that after a
certain point, the energy dissipation through
the CFDs did not occur. This can be attributed
to the resulting significant drift occurring at the
plateauing time as seen in Figure 9. This large
motion yields the CFD significantly and
dissipates energy but afterwards the CFD will
not be effective for further smaller movements
since it has yielded. This is in comparison to a
typical damper, which will provide continuous
damping. This is observed as a downside to the Figure 13 – Energy (kJ) components for Kobe
CFD system. (LA21)
Table 1 - Comparison of Energy (kJ)

Output Modal Link
Case Input Kinetic Potential Damp Hystrtc
LA21 0.2043 0.0329 0.1247 0.1646 0.0392
LA28 0.4005 0.0905 0.2274 0.3365 0.0623
LA24 0.1431 0.0264 0.0376 0.115 0.0194
3.2 Performance of PT connection for NSP
Upon the evaluation of the behaviour of the Figure 14 – Energy (kJ) components for
structure during the non-linear static pushover, Northridge (LA28)
the obtained force-displacement behaviour is
shown in figure 12. The re-centring behaviour
at low drifts and the energy dissipation
occurring due to the yielding of the CFD can be
seen.
72
Table 2 – Comparison of Energy (kJ) in
Welded Frame
Otput Modal
Case Input Kinetic Potential Damp
LA21 0.7603 0.0827 0.3518 0.7603
LA28 1.7843 0.2611 0.5524 1.7843
LA24 0.3723 0.0481 0.0724 0.3656
For the Non-Linear Static Pushover test it was

Figure 15 – Energy (kJ) components for Loma observed the expected behaviour of a welded
Prieta (LA24) frame connection with the hysteretic behaviour
After evaluating the figures 13,14 and 15 it is of the connection showing the yielding of the
observed that when the CFD is introduced connection as seen in Figure 17. This shows that
there is over a 20 percent decrease in the modal the energy dissipation occurs throughout the
energy component showing that the energy range of motion and that upon force removal
dissipation through modal damping is less for the residual deformations will be present, as the
PT connection with CFD device for all three plot does not pass through the origin as seen in
ground motions. This prevents the permanent Figure 12 for the PT frame.
damage to the structure and make the structure
more stable after the earthquake. Also the
kinetic energy can be observed to have
decreased with the addition of the CFDs as
shown in Figures 13 and 14 meaning less
movement of the structure. Link hysteretic
illustrate the energy dissipating capacity of the
CFD.
3.4 Performance of Welded Frame Connection
Non-linear dynamic time history analysis was

performed for the conventional welded beam-
plot
column subassembly to evaluate the behaviour
of the connection under the three ground
4. Conclusion & Recommendation
motions. An example of the largely modal
damping occurring in welded connections can
Post-tensioned steel connection with Cam-Grip
be observed in the energy plots where almost
device offers seismic response that is superior
all the energy is dissipated through modal to the seismic response of the conventional
damping as in Figure 16 and Table 2. moment resisting welded connection in terms
of reducing the residual displacements and
damage in the beam elements but not in the
amount of energy dissipation. Conventional
welded connection can dissipate significantly
higher energy in the early stages of the
earthquake which results in damping of the
response in the later stages. The dissipated
energy in welded connection however is a
result of the permanent damage in the frame
elements (plastic hinges). In passive PT
connection with CFD acceptable levels of
response (deformations) can be achieved
Figure 16 – Energy plot for Northridge (LA28) without severe damage (without plastic
(Energy in kJ vs. time (s)) hinges).
Seismic response of passive post-tensioned steel

frames is heavily dependent on the connection
73
parameters Fpt (post-tensioning force) and β and providing the great opportunity to conduct
(energy dissipation factor). When these the research
parameters are selected correctly, post-
tensioned steel frames exhibit very good References
dynamic behaviour under seismic actions. It is
expected that increasing the initial post-
tensioning force in the connection would result 1 FEMA 335D, ―State of the Art Report on
in lower energy dissipating capability for the Connection Performance,‖ Program to Reduce
same loading and performance level. In further the Eartquake Hazards of Steel Moment-Frame
research this should be considered and should Structures, California, 2000.
run the analysis using different post-tensioning
forces. 2 Popov, E. P., T.-S. Yang and S.-P. Chang, ―Design
of steel MRF connections before and after 1994
Further advanced research should focus on the Northridge earthquake,‖ 1998.
energy dissipating behaviour of the
compression free device (CFD) relative to the 3 Popov, E., and Stephen, R., ―Cyclic Loading of
existing energy dissipation devices discussed in Full-Size Steel Connections,‖ American Iron and
the literature review. The energy dissipating Steel Institute,, Washington, D.C., 1972.
factor (β) which is the ratio of moment
provided by energy dissipation device installed 4 Christopoulos, C., Filiatrault, A., Uang, C. M.,
and Folz, B., ―Posttensioned energy dissipating
in the connection and moment provided by PT
connections for moment-resisting,‖ J. Struct. Eng.,
strands before gap opening, should be
vol. 9, no. 128, p. 1111–1120, 2002.
calculated and compared, since it have a high
influence on ductility, moment capacity, energy
5 Bruneau, M., ―Seismic Retrofit of Steel
dissipating capacity and self-centring
Structures,‖ Tlaxcala, Mexico, 2004.
properties of the post-tensioned connection.
6 Christopoulos, C., Filiatrualt, A., and Folz, B.,
Due to certain limitations, only 2D beam-
―Seismic Response of Self-centring Hysteretic
column subassembly was modelled and
SDOF Systems,‖ Earthquake Engineering and
analysed to evaluate the seismic performance.
Structural Dynamics, vol. 5, no. 31, pp. 1131-1150,
To develop this research further, it‘s quite
2002a.
important to consider an entire structure.
Three-dimensional 6 storey frame would be an 7 Garlock, M., Ricles, J., and Sause, R., ―Cyclic load
ideal model for analysis of the PT steel tests and analysis of bolted top-and-seat angle
moment-resisting frame since they are connections.,‖ J. Struct. Eng., vol. 12, no. 129, pp.
proposed as a replacement of conventional 1615-1625, 2003.
MRFs, which are usually designed for low to
medium high structures. CFD device installed 8 Rojas, P., Ricles, J., and Sause, R., ―Seismic
in the post-tensioned steel MRF model will Performance of Post-tensioned Steel Moment
provide a broader idea of the energy Resisting Frames with Friction Devices,‖ Journal
dissipating capacity of the CFD and it would of Structural Engineering, vol. 131, no. 4, pp. 529-
also help to determine the optimum connection 540, 2005.
location to place the CFD. This determination is
quite important since it would be cost effective 9 Thammarak, P., Thien, M., Htun, L., and K. M.T,
and more practical than installing the CFD at ―The Development of a Compression-Free
each joint. Energy Dissipative Brace,‖ Sydney, Australia ,
2015.
The experimental study should be conducted
on this topic and compare the experimental 10 Ricles, J., Sause, R., Garlock, M., and Zhao, C.,
values with analytical values. The yielding ―Posttensioned seismic-resistant connections for
behaviour during tension and the free steel frames,‖ J. Struct. Eng, vol. 128, no. 7, pp.
movement of the CFD during compression and 113-121, 2001.
its energy dissipation was successfully
observed in the hysteresis diagrams of the 11 ASCE/SEI 7-10, Minimum Design Loads for
CFDs. Buildings and other Structures, American Society
of Civil Engineering, Structural Engineering
Acknowledgement Institute, 2010.
The authors wish to acknowledge Asian

Institute of Technology (AIT) for the guidance
74
AnnualSessions
IESL,pp.
pp.[75[1- -79],
10],2017
2017
© The
© The Institution
Lanka
Simplified Approach for Predicting Fatigue Life of

Fillet Welded Joints with FEM
S. Varakini, K. Ubamanyu and H.M.Y.C. Mallikarachchi
Abstract: Subjected to cyclic loading, many materials fail well below their yield strength due to
the phenomenon of fatigue. Stress concentrating factors, such as sharp changes in geometry, are
especially conducive for fatigue crack initiation. Owing to the inherently complicated nature of the
welding process, a weld presents many such sites favourable for initiation and therefore becomes a
point of weakness to the whole structure, especially when subjected to a cyclic loading. However,
fatigue testing is expensive and time-consuming, and at the same time, computer simulations
comprising of the complex details of the welding process can be computationally expensive when
combined with fatigue analysis. In this paper, a simplified fatigue crack simulation technique using a
two-dimensional model has been explored. A specimen of fillet welded joint was modelled using the
commercial finite element software package Abaqus, and it was analysed under various stress
conditions in order to obtain the energy release rate at the crack tip (by the J-integral method) at
different crack lengths and thereby predict its fatigue life for each stress level.
Keywords: Fatigue life, crack initiation and propagation, J-integral, fillet weld
1. Introduction In order to incorporate the complex factors

involved in welding, 3-dimensional finite
Fatigue is a phenomenon related to fracture element models are generally found to have
that causes the material to fail well below their been used. However, processing such complex
yield strength when a cyclic load is applied. models can be time-consuming, especially due
Fatigue life comprises of three main stages: to a large number of load cycles involved. This
crack initiation, crack propagation and sudden paper investigates the suitability of using a
fracture. The initiation and propagation depend simple 2-dimensional finite element model
on a number of complex factors such as developed using Abaqus, commercial finite
geometric shape, material properties, exposure element package.
conditions, surface condition, loading type and
production variables like residual stress. The 1.2 Theoretical Background
interconnectivity of these factors is still a topic
of extensive research. Total fatigue life can be expressed as follows.
1.1 Fatigue in Welded Joints NT = N i + N p … (1)

where Ni, Np and NT refers to the number of
Welds represent an area of weakness with cycles for crack initiation, propagation and total
regards to many of the above-mentioned fatigue life respectively.
influences. Sharp changes in geometry at the
toe of a fillet weld, residual stresses introduced One of the most common approaches in
as a result of the welding process and the determining crack initiation is given by
change of base metal property in the heat Equation 2 and Tricoteaux, Fardoun, Degallaix,
affected zone (HAZ) are of major concern in & Sauvage [2] have successfully used it in a
determining the fatigue performance of welds. local approach to deal with Ni of welded joints.
Furthermore, workmanship is an important
Eng. (Ms.) S. Varakini, B.Sc. Eng. (Hons.) (Moratuwa),
factor in deciding the degree of stress Department of Civil Engineering, University of Moratuwa.
concentrators present in a weld. Otegui, Kerr,
Eng. K. Ubamanyu, B.Sc.Eng.(Hons.) (Moratuwa), AMIE
Burns, & Mohaupt [1] have identified
(SL), Department of Civil Engineering, University of
undercuts, slag inclusions and spatter as the Moratuwa.
main stress concentrators in welded
Eng. (Dr.) H.M.Y.C. Mallikarachchi, C.Eng., MIE(SL),
connections. B.Sc.Eng.(Hons.)(Moratuwa), Ph.D.(Cantab), Senior
Lecturer, Department of Civil Engineering, University of
Moratuwa.
1
75 ENGINEER
available in finite element software packages.
However, it must be noted that the cyclic
J-integral ΔJ should be calculated using
Equation 4 presented by Ochensberger &
Kolednik [5], and it does not denote the range
between maximum and minimum values, Jmax
and Jmin.
√ ….(4)
Figure 1 - Crack tip region and the contour According to the Paris Law [6], crack growth is
used for J-Integral given by Equation 5 in LEFM. As the effect of
the plastic region near the crack tip is of
concern in this work, a modification was used.
+ …(2) El Haddad, Dowling, Topper, & Smith [7] have
shown that the relationship given in Equation 6
where ∆ε/2 is the local total strain amplitude, gives a good approximation of crack behaviour
σ’f and ε’f are fatigue strength and ductility in this zone for plane stress conditions. This
coefficients, σo is the initial local mean stress, E was used to calculate the stress intensity factor
is the Young’s Modulus and b and c are fatigue ∆K required in the Paris Law relationship.
strength and ductility exponents.
da/dN = C (ΔK)m ….(5)
Another commonly accepted approach for ΔJ = (ΔK)2/E ….(6)
crack initiation period is to assume that a micro
crack exists at the location of high-stress The number of load cycles (N) for a crack to
concentration and it attains macro crack status propagate from an initial crack length ai to a
when it reaches a depth of 0.3 mm [1]. In this final length af is given by Equation 7.
study, this depth is taken as the threshold for
crack initiation and the time taken to achieve it ∫ ….(7)
is calculated using fracture mechanics models
This paper presents a numerical model that was
describing micro crack growth up to 0.3 mm. developed using Abaqus finite element analysis
package to determine J-integral. Using
Linear Elastic Fracture Mechanics (LEFM) Equation 7 the number of cycles required for a
cannot be used for the initiation phase since the micro crack to reach the 0.3 mm threshold depth
size of the plastic region surrounding the crack was determined in an attempt to predict the
tip is significant at these short crack lengths as time taken for fatigue crack initiation.
illustrated in Figure 1. Effect of plasticity can be
addressed by adopting an Elastic Plastic
Fracture Mechanics (EPFM) approach. The 2. Simulation of Fillet Weld
J-integral based fatigue crack growth rate
prediction provides a good approximation of A 2-dimensional half-model of a fillet weld was
short crack behaviour [3]. created using symmetry boundary conditions
along the vertical line of symmetry shown in
Rice [4] initially developed this integral as a Figure 2. Ideally, the shape of the weld should
path independent approximate method to correspond to that of the section-cut at the
determine strain concentrations by cracks and location of minimum toe radius of the weld as
notches. A closed contour integral along an this is the critical location for fatigue crack
arbitrary path around the crack tip is defined as initiation [1]. However, this geometric detail
the J-integral, and a formal definition is given has not been covered in this study. To verify
by Equation 3. this method, the results from an experimental
work carried out by Otegui et al. [1], as
∮[ ] …..(3) illustrated in Figure 2(a) and 2(b) was used.
Material properties are listed in Table 1.
where W is the strain energy density, Ti is Residual stress distribution was obtained by a
traction vector, ui,k is displacement vector, and welding process involving a heat input of
n is the outward unit normal to the integration 2.9x106 J/m as used in [1]. The temperature
path Γ. This could be determined using the distribution at the end of the welding process.
contour integral evaluation capabilities
ENGINEER 2
76
y
(a)
Figure 3 - Meshing scheme used in the

Line of symmetry numerical model and rosette mesh in the
crack tip region
material and 1/r singularity of perfectly plastic
material. These formed a rosette pattern in the
crack tip region, and the rest was meshed using
a relatively coarser mesh as shown in Figure 3.
Seam cracks were introduced at different
depths below 0.3 mm, and the J-integral value
(b)
was studied. The number of cycles required for
crack growth between each depth interval was
Figure 2 - Fatigue testing by Otegui et al. [1] (a)
Geometry of the specimen (b) Experimental set-up
determined, and the summation was taken as
(dimensions are in millimetres) the fatigue crack initiation period, Ni.
To avoid convergence issues, stress was
was simulated by means of an uncoupled simulated by applying a cyclic displacement at
thermal analysis and this was applied as the the free end. Stress conditions, similar to that
initial condition in the successive fatigue shown in Figure 4, were achieved for each load
analysis. cycle.
In the fatigue analysis, the cyclic load was 3. Results and Discussion
applied at a minimum to maximum load ratio, The J-integral was recorded for each stress
R = 0.1. Using the contour integral nodal amplitude over crack lengths of 0.05 to 0.3 mm,
enrichment and direct cyclic analysis capability with 0.05 mm increments. Jmin and Jmax values
of Abaqus, the J-integral was determined. Ten were extracted from the J-integral plot
contour integral were requested, and the produced by Abaqus for each crack length. ΔK
stabilized value was taken as the J-integral at was calculated using Equations 4 and 6. The
that particular load. procedure is illustrated for an applied nominal
A special mesh pattern was used in the crack stress amplitude of 190 MPa. The ΔK vs a curve
initiation region to avoid the stress singularity was fitted to a power law relationship as shown
occurring at the crack tip. According to the in Figure 5. Using experimental data available
findings presented in [8] and the
recommendations of Abaqus user manual [9],
quadrilateral quarter-point elements of type
CPS8R belonging to the quadratic order were
used to capture the combined effect of 1/√r
singularity characteristic of the linear elastic
Table 1 - Material Properties of Test Specimen

Property
Yield Strength (MPa) 421 Figure 4 - Stress distribution and
UTS (MPa) 509 indeformed shape
the literature [1] the values of C and m were
Elongation (%) 53 determined to be 6 x 10-10 and 8.8071
Young’s Modulus (GPa) 210 respectively.
3
77 ENGINEER
The relationship presented in Figure 5 was
applied to the Paris Law and integrated to Simulation
obtain the number of cycles needed for short  Experiment [1]
crack propagation between each length
interval. Results are presented in Table 3 for an
applied nominal stress amplitude of 190 MPa.
Table 2 - Calculation of ΔJ and ΔK

a Jmin Jmax ΔJ ΔK
(mm) (W/m2) (W/m2) (W/m2) MPa√
0.05 0.010 0.058 0.019834 2.04
0.10 0.010 0.066 0.024619 2.27 Figure 6 - Comparison of simulated results
with experimental results presented by Otegui
0.15 0.012 0.077 0.028205 2.43
et al. [1]
0.20 0.014 0.087 0.031200 2.56
0.25 0.014 0.090 0.033007 2.63 particular stress the number of cycles given by
the trend line of simulation data is lower than
0.30 0.015 0.095 0.034502 2.69
that of experimental data. This could be due to
the following reasons.
• The actual welding parameters and cooling

conditions were not known, and reasonable
boundary conditions and interactions were
assumed in the model
• The residual stress distribution was not

verified. Detrimental tensile stress
distribution was simulated in this study.
However, if compressive residual stresses
had developed during the actual welding
process it would have had a beneficial effect
on the fatigue crack initiation life and
Figure 5 - Variation of ΔK with crack length
effectively increased it
Table 3 - Number of Cycles for Short Crack 4. Conclusions

Propagation
A technique was developed using elastic plastic
fracture mechanics concepts to predict crack
Crack Length (mm) N (kilocycles)
initiation life of fatigue cracks at the toe of a
welded joint. EPFM parameter ΔJ was obtained
0.05-0.10 93.8 using a simple 2D simulation and was used to
determine the number of load cycles for small
0.10-0.15 44.3 crack growth until a critical 0.3 mm length was
reached.
0.15-0.20 27.5
0.20-0.25 19.3
A graphical comparison of experimental and
simulation results shows that the numerical
0.25-0.30 14.6 results follow a similar trend line to that
exhibited by the experimental results. In the
Total time calculated for crack initiation = 199.5 overlapping region of both curves between 150
kilo-cycles to 200 MPa, experimental data points show
close correlation with the simulation results.
The procedure was repeated for different However, there is room for future work in
nominal stress amplitudes, and the results are order to improve the accuracy of the crack
presented as an S-N Curve shown in Figure 6. initiation life predictions. Special consideration
Note that two trend lines were drawn by fitting must be given to the effect of residual stresses
experimental and simulated data, separately which can be beneficial or detrimental
using power law. From this, it is seen that for a depending on whether it is tensile or
ENGINEER 4
78
compressive. Also, means should be explored
to fully incorporate the welding parameters
and conditions into a two-dimensional model.
This technique proposes a relatively simple and

computationally efficient procedure for the
determination of fatigue crack initiation period,
Ni .
Acknowledgement
The authors are thankful to the National

Research Council, Sri Lanka and Senate
Research Committee, University of Moratuwa
for financial assistance.
References
1. Otegui, J. L., Kerr, H. W., Burns, D. J., &

Mohaupt, U. H., “Fatigue Crack Initiation from
Defects at Weld Toes in Steel,” Int. J. Press. Vessel.
Pip., Vol. 38, No. 05, 1989, pp. 385–417.
2. Tricoteaux A., Fardoun, F., Degallaix, S., &

Sauvage, F., “Fatigue Crack Initiation Life
Prediction in High Strength Structural Steel
Welded Joints,” Fatigue Fract. Eng. Mater. Struct.,
Vol. 18, No. 02, February, 1995, pp. 189–200.
3. Suresh, S., Ritchie, R. O., “Propagation of Short

Fatigue Cracks,” Int. Met. Rev., Vol. 29, No. 01,
1984, pp. 445–475.
4. Rice, J. R., “A Path Independent Integral and the

Approximate Analysis of Strain Concentration by
Notches and Cracks,” J. Appl. Mech., Vol. 35, No.
02, 1968, pp. 379–386.
5. Ochensberger, W., Kolednik O., “A New Basis for

the Application of the J-integral for Cyclically
Loaded Cracks in Elastic–plastic Materials,” Int. J.
Fract., Vol. 189, No. 01, September, 2014, pp. 77–
101.
6. Paris, P. C., Gomez, M. P., & Anderson, W. E.,“A

Rational Analytical Theory of Fatigue,” Trend
Eng., Vol. 13, 1961, pp. 9–14.
7. El Haddad, M. H., Dowling, N. E., Topper, T. H.,

& Smith, K. N., “J-integral Applications for Short
Fatigue Cracks at Notches,” Int. J. Fract., Vol. 16,
No. 01, February, 1980, pp. 15–30.
8. Dhondt, G., “General Behaviour of Collapsed 8-

node 2-d and 20-node 3-d Isoparametric
Elements,” Vol. 36, November, 1993, pp. 1223–
1243.
9. Dassault Systems Simulia Corp., Abaqus 6.14:

Abaqus Analysis User’s Guide. Providence, RI,
2013.
5
79 ENGINEER
AnnualSessions
IESL,pp.
pp.[81[1- -88],
10],2017
2017
© The
© The Institution
Lanka
Growing Colombo: Some Concerns from Wind

Engineering Point of View
A. U. Weerasuriya and K.A.B. Weerasinghe
Abstract: Colombo is rapidly developing as a major business centre in the South Asia region. As
a result, the city is transforming its urban environment at a greatrate of growing numbers in high-rise
buildings and the population. The construction of high-rise buildings would induce two types of wind
nuisance: high wind speed and low wind speed those are not experienced by citizens before. The
impacts of these wind nuisance can be predicted using the knowledge on wind engineering in the
absence of the field measurements and simulation data for Colombo. This article aims to demonstrate
probable wind nuisances that would be occurred in the Colombo city future using similar experiences
in wind-related issues reported in other countries. A number of design strategies and long-term
policies are proposed to minimise the anticipated wind nuisances resulting from future development
projects in Colombo city. In addition, a set of recommendations is given in this article to increase the
awareness of the urban wind climate among the stakeholders of the development projects and is to
encourage employing wind data for future building designs and urban planning in Sri Lanka.
Keywords: Building Form, Colombo city, Wind Environment, Wind Nuisance, Urban Planning
1. Introduction development process are presented rationally.

A number of solutions for the wind-related
Colombo, the commercial capital of Sri Lanka, issues are proposed and anticipated difficulties,
undergoes significant modifications on its required skills and resources pursuing are
historical form and the city environment. The discussed in detailed to increase the awareness
Colombo port city project and a number of tall of the urban wind climate in the development
building constructions are two such of the Colombo city.
modifications that could influence the
environmental conditions of Colombo. 2. The Colombo City
Although the urbanisation is an inevitable and
an inherent feature of the modern world, it 2.1 Geography and the Urban Climate
should be managed well such that it can lead to
the sustainable growth of a city with preserving The Colombo city (6°54′N, 79°52′E) bounds
the environment, liveability, and many other to the Indian Ocean in the west and
factors [1]. The liveability has become a major Muthurajawela marshland in the east by
measure of a sustainable city because of its streching the city in the north-south direction
immense influences on health, life expectancy, parallel to the coastline. The Colombo city, the
and even psychological conditions of citizens. main focus of this article, is at as the centre of
the Colombo Metropolitan Region (CMR),
The wind is one of the major climate factors which extends about 100 km in the north–south
that determine the city‘s liveability. Wind direction and 40 km in the east–west direction.
contributes many aspects of the liveability The permanent population of the city was
including air quality, outdoor thermal comfort, about 0.56 million in 2012 while more than 1
pedestrian wind comfort, natural ventilation, million people daily travel to Colombo from
and for recreational activities. Therefore, a city other areas of Sri Lanka [2].
with its properly designed wind environment
can easily in achieve a higher level of liveability
compared with a city with improper wind Dr. A.U.Weerasuriya, B.Sc. Eng. (Moratuwa), M.Sc.
conditions. (Moratuwa), Ph.D.(HKUST), Post-Doctoral Fellow,
Department of Civil and Environmental Engineering, , The
Hong Kong University of Science and Technology, Hong
In this article, the main focus and discussion Kong, China.
direct on how Colombo city can achieve a
liveable environment in its growth phase from Mr. K.A.B. Weerasinghe, B.Sc. Eng. (Moratuwa), M.Sc.
(Moratuwa), Lecturer, Institute of Technology (ITUM),
the wind engineering point of view. The University of Moratuwa.
anticipated wind-related issues due to the city
Institute of Technology, University of Moratuwa (ITUM)
1
81
The climate of Colombo can be classified as the averaged mean wind speed or 26 ms-1of the 3-
tropical climate with hot and humid conditions second gust wind speeds with50-years return
that prevail throughout the year. The Asiatic period as the design wind speeds for Colombo.
monsoons induce variations in air temperature Due to, difficulties in setting up measurement
and humidity such a way that the annual air locations and conduct field measurement
temperature and humidity are in the range of campaigns and the lack of interest create near
23-32 oC, and 80%-85%, respectively [3]. The ground level wind speed data scarce for
monsoons define the urban wind climate of Colombo. Nevertheless, a field study conducted
Colombo with the prevailing wind direction in by Johansson and Emmanuel[8] to measure
southeast and northeast the city as shown in wind speeds close to the ground shows that 5-
Figure 1. minute average wind speed is about 1.9 ms-1
with the minimum wind speed of 0.6 ms-1 at the
congested city centre of Colombo. It should be
noted that some of these measurements were
taken at a height of 10 m or above, thus it is
reasonable to assume that the wind speed near
the groundis substantially lower than 1.9 ms-1
due to the wind slowdown from the friction
between wind and rough surfaces of ground
buildings, trees, and vehicles.
2.2 Urban Form of Colombo
The vernacular architecture of Colombo city,

which consists of wide eaves, open verandas,
and courtyards to cope up with hot and humid
climate, has been gradually replaced by the
modern architecture. The buildings designed
according to the modern architecture are in
Figure 1 - Annual wind rose of Colombo theshape of a solid vertical slab. These modern
(radial distances show the number of hours buildings are low to medium rise and are
that wind speeds prevailed)[4]
concentrated along traffic arteries in the city.
Particularly, the medium-rise buildings located
Due to the lack of available data in wind
along the Colombo-Galle highway (A2), which
speeds, several researchers made crude
a main entrance to the city, play a major role in
estimations of the magnitude of average wind
controlling the sea-breeze penetration into the
speed in Colombo. The Design Manual ―Design
city. At the city centre, buildings are congested
building for highwinds, 1978‖[5] proposed 2
along narrow streets and in the outskirt of the
types of 3-second gust wind speeds: 38 ms-1 and
city, buildings are separated by sufficient
33 ms-1 for the post-disaster structures and the
distances between them. Compared to other
general structures built in zone 3 where
major cities worldwide, Colombo has only a
Colombo is located. Although these wind
few high-rise buildings, which were also built
speeds are for the purpose of structural design,
in the recent past. The Bank of Ceylon
the details of wind speed profile or the
Headquarters building (105 m height), World
measurement heights are not presented in the
Trade Centre tower (152 m height), Hilton
manual. To compensate the absence of
tower apartments (129 m height), and the
background information onthe design manual,
Crescat office and Condominium (113 m
Holmes and Waller [6] have suggested to
height) are the buildings taller than 100 m and
assume the design wind speeds are based on
built before theyear 2000. The scarcity of open
50-year return period and are measured at 10 m
spaces and green areas (i.e., public parks)
height above the ground with the roughness
within the city is another noticeable difference
length of 0.02 m. With these two parameters, a
between Colombo and other major cities
designer can calculate the wind speed at any
worldwide. Most of the major cities are keen to
height using empirical wind profile models, the
maintain green areas within the city limit such
logarithmic or the power-law model. Nandalal
as the central park in New York, USA and
and Abeyruwan [7] have fitted the
several public parks around London, England
measurements of 3-minute averaged wind
because of the parks act as ‗lungs of the city‘
speeds taken over 16 years to the Gumbel
providing clean air for breathing. In Colombo,
distribution and have found 20 ms-1 of 3-minute
2
82
only Viharamahadevi Park is available as a certainly challenge wind engineers when
large greenery area that may be a vital factor in estimating wind loads and predicting building
controlling microclimate conditions of the city. responses. However, this article does not intend
to discuss the challenges imposed by high-rise
2.3 Studies on Urban Climate of Colombo buildings on structural engineers but mainly
focuses on the possible environmental issues
Most of the wind-related research done in Sri that can be caused by to the construction of
Lanka had focused on structural wind high-rise buildings in Colombo city.
engineering aspects such as calculating wind
loading on high-rise buildings [9] or assessing
the adoptability of international wind loading (a) (b)
standards to the Sri Lankan context [10-12]. An
exception is a series of studies done by
Emmanuel and his research group [3, 8, 13-16]
in which the focuses were on several aspects of
the urban climate of Colombo including
outdoor thermal comfort, urban heat island
(UHI) effect, and air pollution. Although none
of these studies directly relate to wind
engineering, Emmanuel and his research group
were keen to identify wind as a critical factor in
the urban climate of Colombo. For example,
Emmanuel‘s research group has identified the
importance of sea breeze penetration in
achieving outdoor thermal comfort thus,
recommended astaggered building Figure 3 - Proposed two high-rise buildings in
arrangement in order to enhance the air Colombo; (a) Altair (240 m) (source:
circulation in the city [8]. Moreover, their effort www.altair.lk), and (b) Krish towers (Tower 1
to measure wind speeds at the street level is - 420 m height) (source:
www.skyscrapercenter.com)
invaluable as these numbers indicate the
influence of compact building arrangement on
The possible wind nuisance in future Colombo
the prevailing low wind speeds in Colombo.
city will depend on the way high-rise buildings
arranged in the city. The future high-rise
3. Wind Nuisance in Future buildings can be arranged as closely-spaced
Colombo City buildings as similar to the skyline in Hong
Kong (Figure 4(a)) or as a group of high-rise
buildings in the centre of the city forming the
central business district (CBD) as in San
Francisco, USA (Figure 4(b)) or as the
combination of both. A skyline is a possibility
in the coastline parallel to the Colombo-Galle
road when developers attempt to provide the
scenic view of the Indian Ocean for many
tenants. On the other hand, the centre of
Colombo city including the Fort area is a
suitable area to develop as the CBD in Colombo
with a number of high-rise buildings built
Figure 2 - Details of the completed and together. Depend on the development type,
proposed high-rise buildings in Colombo[17]. these two building arrangements will define
two types of wind nuisances: (1) unpleasant or
The trend of constructing high-rise buildings, dangerous high wind speeds, and (2)
as shown in Figure 2, would be the main cause undesirable low wind speeds. In the following
of wind-related issues in the future Colombo subsections, causes and repercussions of these
city. The proposed and under-construction two wind nuisances are discussed in detail.
high-rise buildings are not only being gigantic
but also with atypical shapes and complex
building forms. (Figure 3). The unorthodox
shapes and superior building heights would
3
83
owners, design engineers, architects, and city
(a) authorities for mismanaging the wind
environment near high-rise buildings. Windy
conditions can also cause negative economic
impacts for shop owners by preventing
customers coming to shopping precincts [19].
3.2 Undesirable Low Wind Speeds
High land prices and limited availability of

lands in cities force to construct closely-spaced
(b) high-rise buildings, which subsequently induce
excessive sheltering causing inadequate city
ventilation. Particularly, high-rise buildings
built adjoining each other, act like a wall
against incoming wind flows and drastically
cut down the wind penetration into the city.
This phenomenon is referred to as the ‗wall
effect of buildings‘ [20]and is commonly found
in waterfront areas of HongKong, which are
similar to the coastal sites in Colombo.
Figure 4 - (a) Skyline of Hong Kong (source: A number of issues including outdoor thermal
www.wikimedia.org), and (b) high-rise discomfort exaggerated UHI effect,
buildings in CBD of San Francisco (source: accumulation of air pollutants near the ground
www.xpressmagazine.org) level and favourable conditions to spread
airborne pathogens are related to the weak air
3.1 Unpleasant or Dangerous High Wind circulation in a city. None of the above-
Speed mentioned issues are standalone problems but
they are interconnected with each other causing
The windy areas around tall buildings are severe detrimental effects on both citizens and
formed as a result of deflected high-speed the urban environment. For example, a 3 oC of
winds from high altitudes to the pedestrian UHI effect in Hong Kong induces high thermal
level. There are two pronounced windy areas stresses on inhabitants of the city almost every
formed near a high-rise building, first, from the day and night in summer and consequently,
downwash of the windward side, and second, acquires more energy to maintain comfortable
from separated wind flows at the windward indoor temperature, thereby increasing the
corners. The properties of these windy areas energy use and CO2emissions [21]. Therefore,
(i.e. the intensity and the area) vary with the cities suffering from low wind speeds setup
dimensions of the buildings those flow features regulations to promote the air circulation at
are attached[18]. Therefore, properly designed ground level in urban areas.
building scan easily control the windy areas
formed in their vicinity. Another mitigation 4. Design strategies, and long-
method is to use passive controllers (discussed
in Section 4)but they are less successful against
term policies
violent high-speed winds and can be applied on
case basis rather than universally. The following subsections present strategies
and long-term policies that Colombo can adopt
The possible adverse effects of the windy areas to mitigate anticipated issues arisen from
are greater difficulty in walking and operating high/low wind speeds.
doors, unable to use umbrellas on rainy days,
blown down of light vehicles such as bicycles, 4.1 Design Strategies for High Wind
and in the worst case, people can be injured Speeds
from falls caused by wind gusts. Due to the low
ambient wind speed, accidents to the The design strategies on high wind speeds
pedestrians would be unlikely happend in Sri intend to protect people by shielding the
Laka, nevertheless such an unfortunate incident activity areas (e.g. walkways) from dangerous
would lead legal actions against the building high winds or by assigning suitable areas for
certain activities according to prevailing wind
4
84
conditions (e.g. threshold wind speeds for In the city scale, the design strategies against
shopping areas and lingering areas). high wind speeds are controlled by city
As a design strategy, certain building forms authorities than by individual designer. An
such as buildings with large openings near the example is that the authorities will design the
ground are not recommended for the locations orientation of street grid, which is generally in
with high wind speeds. To protect inhabitants the diagonal direction to the prevailing winds
from windy areas near high-rise buildings to provide the maximum sheltering effect to
Architectural designs and building features are pedestrians. Another city scale design strategy
effectively employed, for example, building‘s to reduce high wind speeds at the pedestrian
main doors are located away from the level isbuilding narrow streets with deep street
windward corners to avoid occupants suddenly canyons to encourage the skimming flow
exposed to windy areas and podium structures regime, in which the wind flows over the
are adopted to deflect the downwash flow buildings rather than entering to the canyon.
away from the building (Figure 4).
Aerodynamically favourable building shapes 4.2 Design Strategies for Low Wind
can also be helpful in minimising windy areas Speeds
forming near high-rise buildings.
Design strategies for low wind speeds are
completely opposite to the strategies used to
(a) (b) mitigate high wind speeds. More specifically,
design strategies for low wind speeds promote
the air circulation near the ground by opining
‗the more wind the better‘ instead restricting
the wind circulation. In order to achieve high
air circulation rate, some forbidden building
forms and architectural features in windy cities
are embraced by cities with low wind speeds.
Buildings with an open space at the ground
level or commonly known as ‗lift-up‘ buildings
Figure 5 - (a) Intense downwash of a building are such a favourable building form in cities
without a podium structure, and (b) deflected with low wind speeds [22]. And certain
downwash of a building with a podium building forms, such as podium structures are
structure not recommended, or only being adopted
following special guidelines such as a zero
Economic concerns over the reduced rentable setback for podiums to deflect more high-speed
floor area and unattractive appearance are winds from higher altitudes to the street level.
some factors curtail the flexibility of It is also recommended to have voids at
modifications of an architectural design. In different heights of buildings, greenery areas at
such case, designers can use active or passive the ground level and intermediate levels, and
mitigation methods to protect occupants and maintain sufficient separation distances from
pedestrians from high wind speeds created by adjacent buildings in the locations with low
the building or the nearby buildings. ambient wind speeds.
Penwarden and Wise[19] demonstrated several
practical solutions including the construction of In the city scale planning, major highway
an additional roof and an installation of glass arteries are built parallel to the prevailing wind
screens that they had proposed for ill- directions creating main breezeways of the city.
conditioned shopping precincts in the United The secondary streets are designed as wide
Kingdom.But they also reported that these roads connecting to the main highways by
mitigation methods were costly and difficult to forming minor breezeways and air paths
implement on the completed buildings and (Figure 5(a)). Greenery areas and open spaces
were not capable of solving the wind nuisance within the city have linked each other to
completely. The passive mitigations methods enhance the unobstructed wind air circulation
can be natural wind breaks such as trees or as shown in Figure 5(b). Closely-spaced similar
technically advanced smart, automatic control heights tall buildings are prohibited to build in
systems. Less effectiveness and uncertainties in waterfront sites and designers are encouraged
alleviating high wind speeds are the to design buildings with varying heights or
disadvantages of passive mitigation methods. ascending building heights in the direction
parallel to the prevailing wind.
5
85
existing wind conditions with the expectations
citizen are a question to be addressed carefully.
As the previous sections demonstrated, it is
anticipated to experience both high and low
wind speeds in future Colombo city, and thus it
is recommended to employ criteria to address
both types of wind nuisance, simultaneously.
For instance, the future wind ordinance of
Colombo may propose a minimum wind speed
to maintain outdoor thermal comfort, several
wind speeds for different areas such as CBD,
residential areas, and recreational and a
maximum threshold wind speed to ensure the
safety of the citizens.
AnUrban Map of Colombo City
An urban map should be developed indicating

sensitive locations for the urban climate of the
city. Such an urban map can be utilised to
evaluate the impact of future development on
the wind conditions of its neighbourhood. The
urban map can also define the areas dedicated
Figure 6- (a) Breezeways and air paths in a city for different types of development, for example,
with low wind speeds, (b) linkage of open for the CBD or for a residential development,
spaces: air path A–A linking open spaces with and then to regulate design parameters such as
a breezeway; air path B–B linking open spaces building height, and ground covering ratio. By
with low-rise buildings; and air path C–C regulating building parameters ,the city
linking open spaces with a linear park. [23] authority can maintain an acceptable level of
wind circulation near the ground. For instance,
4.3 Long-Term Policies low building heights and small ground
covering ratios recommended for the coastal
Since Colombo city is still in its early stage of
sites would be advantageous to facilitate the sea
rapid urbanization, it is important to follow a
breeze penetration in to inland areas.
set of well-planned policies to avoid occurring
any kind of wind nuisance discussed in this Building design guidelines for Colombo City
paper. The long-term policies can be extended
in a wide range from educating the general A set of design guidelines should be introduced
public to providing specific design related for designers to design buildings without
information for engineers .In this subsection, imposing adverse effects on the urban wind
only the latter will be presented in detailed environment. Such building design guidelines
while non-technological long-term policies will can recommend certain building forms for
be discussed in section 5. different areas, for instance, podium structures
for locations experience high wind speeds and
A Wind Ordinance for Colombo City
‗lift-up‘ buildings for locations where the
prevailing wind speeds are low. Another
As along-term policy, the city authority should
advantage of stipulating the building design
stipulate a wind ordinance to address
guidelines is its ability to control minor design
anticipated wind-related issues in future
concerns such as the required percentage of
Colombo city. Several researchers [24, 25], have
greenery area, minimum building separation
reported the basic contents of a wind ordinance
distance, minimum clearance distance from
that addresses the two types of wind nuisances.
streets, and other parameters related to the
In general, a wind ordinance contains
wind circulation. On the other hand, design
followings; how the city decides which building
guidelines can set a common platform for both
projects are to be tested, how the testing is to be
the city authority and developers can agree and
carried out, criteria for whether the projects
it will reduce the risk of being modified the
meet the city‘s requirement, and how the
design in a later stage of the project. The format
results are reported. Among the above-
and some of the contents of the sustainable
mentioned contents, the criteria to assess the
6
86
building design guidelines stipulated by the meteorological data including air temperature,
Building Department of Hong Kong [26] can humidity, and solar irradiation, and roadside
beset as an example for building design air pollution for the planning purpose.
guidelines of Colombo.
6. Concluding Remarks
5. Applying wind engineering to
Colombo city design This paper presents two types of anticipated
wind-related issues: high wind speeds and low
Non-technical, long-term policies can be in wind speeds in the future Colombo city and
three major fields; education, research, and design strategies, mitigation methods, and
meteorology as discussed in the following long-term policies to eliminate the probable
subsections. wind nuisances. Moreover, several
recommendations are proposed to increase the
Education awareness of wind engineering in the urban
planning and possibility of integrating wind
It is necessary to educate all the stakeholders data for the future city design. However, with
including planners, engineers, authorities and the absence of exact data, the facts presented in
general public involved in the development of this article are based on the similar cases
the Colombo city to increase the awareness of reported in the literature. Therefore this article
urban wind climate. Particularly the should be referred as a source of the knowledge
professional courses for engineers, planners, on the urban wind climate rather than as a
architects should accommodate wind design manual established for Colombo.
engineering, and thus recognise wind as a
design parameter. Moreover, wind and its References
interdependency on other climate factors
should be illustrated to increase the awareness 1. http://www.worldbank.org/en/news/press
of technical professionals on the urban climate. -release/2015/09/29/urbanization-offers-sri-
lanka-major-economic-
Research opportunity,accessedon 08th February 2017.
2. Perera D. ―The way forward Colombo

A considerable amount of research is necessary
Metropolitan Transport Master Plan and
to estimate important design parameters such Areas for International Cooperation‖, 2014.
as the minimum wind speed required to Available on
maintain outdoor thermal comfort and http://www.uncrd.or.jp/content/documents
favourable buildings forms for Colombo. It is /22938EST-P5_Sri-Lanka-MOT.pdf
also necessary to follow an integrated research
approach to study the urban wind climate of 3. Emmanuel, Rohinton, H. Rosenlund, and Erik
Colombo than following individual research Johansson. "Urban shading—a design option
interests. An integrated research approach for the tropics? A study in Colombo, Sri
Lanka." International Journal of Climatology
would prevent establishing conflicting research
27, no. 14 (2007): 1995-2004.
findings such as promoting deep street
canyons, which are favourable in achieving 4. https://www.meteoblue.com/en/weather/f
outdoor thermal comfort but have adverse orecast/modelclimate/colombo_sri-
effects on wind circulation within the canyon. lanka_1248991,accessed on 09th February 2017
The local researchers should be encourage to
collaborate with experts in other countries to 5. Clarke, A.G., Swane, R.A., Schneider, L.M.,
exchange the existing knowledge and learn Shaw, P.J.R., ―Technical Assistance to Sri
from their past experiences. Lanka on Cyclone resistant construction-
Project Summary‖, Commonwealth of
Australia, Department of Housing and
Meteorological data Construction for Australian Development
Assistance Bureau,Vol-1, Part 1-4,
The scarcity of meteorological data is one of the October1979.
major obstacles inintegratingwindinto the
planning of the future Colombo city. It is 6. Holmes, John D., and Richard Weller. Design
advisable to establish a network of wind speeds for the Asia-Pacific region.
anemometers and small-scale weather stations Standards Australia International, 2002.
to collect the data of wind speeds (near the 7. Nandalal, K.D.W., Abeyruwan, H., ―An
ground and high altitudes) and other attempt towards wind zone demarcation for
7
87
Sri Lanka‖, APECWWReport,Tokyo 18. Tse, K. T., A. U. Weerasuriya, X. Zhang, S. Li,
Polytechnic University, Japan, 2010 (available and K. C. S. Kwok. "Pedestrian-level wind
on http://www.wind.arch.t- environment around isolated buildings under
kougei.ac.jp/APECWW/Report/2010/Srilan the influence of twisted wind flows." Journal
kab.pdf) of Wind Engineering and Industrial
Aerodynamics 162 (2017): 12-23.
8. Johansson, Erik, and Rohinton Emmanuel.
"The influence of urban design on outdoor 19. Penwarden, Alan David, and Alan Frederick
thermal comfort in the hot, humid city of Edward Wise. Wind environment around
Colombo, Sri Lanka." International journal of buildings. HMSO, 1975.
biometeorology 51, no. 2 (2006): 119-133.
20. Yim, Steve HL, Jimmy Chi Hung Fung, Alexis
9. Wijeratne, M. D., and M. T. R. Jayasinghe. Kai-Hon Lau, and See Chun Kot. "Air
"Wind loads for high-rise buildings ventilation impacts of the ―wall effect‖
constructed in Sri Lanka." Transactions Part: resulting from the alignment of high-rise
58-69. buildings." Atmospheric Environment 43, no.
32 (2009): 4982-4994.
10. Premachandra, W. R. N. R. "Study of new
wind loading code to be adopting on Sri 21. Fung, W. Y., K. S. Lam, W. T. Hung, S. W.
Lanka." PhD diss., M. Sc Thesis, Graduate Pang, and Y. L. Lee. "Impact of urban
school, Kasetsart University, 2008. temperature on energy consumption of Hong
Kong." Energy 31, no. 14 (2006): 2623-2637.
11. Weerasuriya, A. U., and M. T. R. Jayasinghe.
"Wind Loads on High-Rise Buildings by 22. Tse, K. T., Xuelin Zhang, A. U. Weerasuriya,
Using Five Major International Wind Codes S. W. Li, K. C. S. Kwok, Cheuk Ming Mak,
and Standards." Engineer: Journal of the and Jianlei Niu. "Adopting ‗lift-up‘building
Institution of Engineers, Sri Lanka 47, no. 3 design to improve the surrounding
(2016). pedestrian-level wind environment." Building
and Environment 117 (2017): 154-165.
12. Weerasuriya, A. U., and M. T. R. Jayasinghe.
"Wind Loads on a Medium-Rise Building by 23. Ng, E. "An investigation into parameters
Using Five Major International Wind Codes affecting an optimum ventilation design of
and Standards." Engineer: Journal of the high density cities." International Journal of
Institution of Engineers, Sri Lanka 47, no. 1 Ventilation 6, no. 4 (2008): 349-357.
(2015).
24. Arens, Edward, D. Ballanti, C. Bennett, S.
13. Emmanuel, Rohinton. "Thermal comfort Guldman, and B. White. "Developing the San
implications of urbanization in a warm- Francisco wind ordinance and its guidelines
humid city: the Colombo Metropolitan for compliance." Building and Environment
Region (CMR), Sri Lanka." Building and 24, no. 4 (1989): 297-303.
Environment 40, no. 12 (2005): 1591-1601.
25. Willemsen, Eddy, and Jacob A. Wisse.
14. Emmanuel, R., and H. J. S. Fernando. "Urban "Design for wind comfort in The Netherlands:
heat islands in humid and arid climates: role Procedures, criteria and open research
of urban form and thermal properties in issues." Journal of Wind Engineering and
Colombo, Sri Lanka and Phoenix, USA." Industrial Aerodynamics 95, no. 9 (2007):
Climate Research 34, no. 3 (2007): 241-251. 1541-1550.
15. Perera, Bimalka, Rohinton Emmanuel, and 26. Technical Guide for Air Ventilation
Sumal Nandasena. "Urban air pollution and Assessment for Developments in Hong Kong,
mitigation options in Sri Lanka." Proceedings by Hong Kong Planning Department. Visited
of the Institution of Civil Engineers-Urban on 15th of February 2017
Design and Planning 163, no. 3 (2010): 127- (http://www.devb.gov.hk/filemanager/tech
138. nicalcirculars/en/upload/15/1/jtc-2006-01-
0-1.pdf)
16. Emmanuel, Rohinton, and Erik Johansson.
"Influence of urban morphology and sea
breeze on hot humid microclimate: the case of
Colombo, Sri Lanka." Climate research 30, no.
3 (2006): 189-200.
17. https://skyscrapercenter.com/city/colombo,
accessed on 09th February 2017.
8
88
Annual Sessions of IESL, pp. [89
[1 - -10],
96],2017
2017
Contributory Factors for Cost and Time Overruns of

Construction Projects
Leelanath Daluwatte and Malik Ranasinghe
Abstract: Most construction projects experience overruns in cost and time when delivering the
agreed scope. Construction projects in Sri Lanka are no exception. As these delays negatively impact
the stakeholders of the projects, there is room for research on why such overruns take place and how
to control them. It is important to optimize the use of scarce resources (by way of controlling the cost)
and opportunity costs for the stakeholders (by way of controlling the time) of the projects. There are
many reasons for the overruns of cost and time as identified by this study.
The objectives of this paper were to identify and analyse contributory factors for the cost and time
overruns of construction projects. The literature review identified 42 factors that contribute to the
overruns of cost and time in construction projects. Most of these were due to the inadequacy of
analysis at the preconstruction stage of the projects.
Methodology of the study relied mainly on the literature reviews. Data collected from the review was
analysed. The following seven factors, design effects, planning and controlling, material issues, scope
change, finance, client influenced changes, and contractor inexperience contributed to 51%
occurrences of the contributory factors while top 17 factors contributed to 80% of the occurrences of
the contributory factors.
Keywords: Construction projects, Cost, Time, Overrun,

__________________________________________________________________________________________
1. Introduction identified contributory factors and (iii) to select

the key contributory factors.
Construction projects consume significant
quantities of scares resources like building The paper describes the extensive literature
materials, fossil fuels, machinery built using review conducted in the research area.
resources, human resources, irrespective of the
magnitude of the construction project [1]. This paper is based on ongoing research
However, construction projects are meant to focused on the cost and time overruns of
utilize optimum amounts of resources at construction projects and will followed by
optimum project duration to deliver effective further research in the industry.
project output. In most construction projects,
utilization of optimum resources or delivery of 2. Literature Review
the project on agreed time does not happen [2].
These require additional costs and time for the A number of publications related to cost and
project scope to be achieved resulting in lost time overruns of construction projects were
opportunity costs to stakeholders of the reviewed and during the research it was
projects. Most construction projects in Sri revealed that the contributory factors were
Lanka contribute to the above scenario, thus similar, irrespective of the type, cost, duration,
creating cost and time overruns. locality etc. of the project.
In preliminary review it was observed that Eng. Leelanath Daluwatte, B.Sc.Eng.(Hons), FIE(Sri
most of the construction projects exceeded the Lanka), C.Eng., P.E., M.ASCE, DBM, Dip HRM (Merit),
time and cost to deliver the agreed scope. In Dip(Cleaner Production), Dip(Human Rights),
Distinguished Toastmaster, Consultant in Project
the end this appear to incur losses to the
Management / Construction Management / Process
concerned parties and hence the study was Improvements.
focused in the concerned area. The objectives
of this paper are (i) to identify the contributory Eng. (Prof.) Malik Ranasinghe, B.Sc.Eng.(Hons), MASc,
factors for cost and time overruns of Ph.D., FIE(Sri Lanka), C.Eng., Senior Professor, Dept. of
Civil Engineering, University of Moratuwa.
construction projects; (ii) to analyze the
1
89
Further the findings of the literature review The following aspects of design were found to
was categorized according to different stages have contributed to cost and time overruns;
of the construction project are described in this Design errors [5],[6],[9],[18],[19]; Poor design
section. [14],[16];Delay in design [16],[20];Design
changes [9],[18],[19];Delays in approval of
2.1 Findings related to Preconstruction phase design [5],[16],[21];Client driven design
Ineffective investigations during changes [17], [18];Design changes driven
preconstruction caused cost and time overruns change orders [18];Failing to identify project
in construction projects [3],and inadequate risks at design stages [23]; Erroneous design
allocation of funds to the preconstruction stage work/process and Attitudes of the designer
led to shortcomings such as ambiguities, [17], [18].
design (errors and omissions) and incurred
additional costs during the construction It was stated that clarity in design at the early
phase[4]. Also, risks taken during stage and clearly defined scope led to more
preconstruction stage caused errors in accurate budget estimates of the projects[17].
processes (bidding, estimating, finance and
design) in construction projects [5]. As the clients were the main party that impose
changes and client-driven design changes
However, provision of all information to caused critical risks that resulted in cost
stakeholders as appropriate led to stakeholder overruns in construction projects [17].
satisfaction and avoided doubts and conflicts
among them regarding matters related to the 2.3 Findings related to Planning phase
projects [6] while analysis of stakeholder Ineffective planning was most significant and
involvement provided anevaluation of the common factor that caused cost overruns [16].
needs and expectations of stakeholders in Also, poor planning and controlling
relation to the main objectives of the project [24],[25],[26]insufficient planning and
[7]. scheduling [27], improper planning[28]and
It was found that lack of funding for bidding deficiencies in planning and scheduling caused
caused cost overruns [2] while limited time cost and time overruns in construction projects
given for submission of bids caused time [5], [29].
overruns in construction projects [8].
Slowness in decision-making process caused
It was also highlighted that contract conditions time overruns in construction projects [30].
were the most occurring and impacting risk in Efficient planning was one of the best methods
construction cost estimating and caused cost to reduce cost overruns [31].
and time overruns in construction projects [5]
and contract constraint risk caused cost 2.4 Findings related to Implementation phase
overruns in construction projects [9]. Financial difficulties faced by the client was a
major factor [14], [15], [21], [28], [32], [33] and
Ineffective project communication [10] and financial resource management was among the
inefficient communication between major contributors to cost and time overrun of
stakeholders had caused cost and time construction projects. Also, deficiencies in
overruns in construction projects[11]. finance management [34] and poor change
Unsupportive government policies[25], management caused cost and time overruns
uncertainties in regulations [13] and [35].
bureaucratic indecision [14] caused cost and
time overruns. It was also noted thatpoor site Delays in payments for contractors to complete
conditions [10], [13],[25] and unforeseen the work [25], [36] and cash flow and financial
ground conditions caused cost and time difficulties faced by contractors caused cost
overruns in construction projects [15]. and time overruns of construction projects
[37]. Long response time from utility agencies
2.2 Findings related to Design phase had caused time overruns in construction
Design related issues were identified among projects [27].
the most problematic [3],[4]and significant
[16],[17], which caused cost and time overruns Shortage of workers[28],[37],[38],[39],
in construction projects. ineffective relationship between management
and labour, [16], low productivity level of
workers [25],[40],[41], and poor
2
90
communication and coordination between in the cost of construction material [26] and
parties caused cost and time overruns in material related issues [10]caused cost and
construction projects [40],[42]. time overruns in construction projects.
Inexperienced[16], [28], [37],and incompetent
Poor site management and supervision [12], contractors caused cost and time overruns in
[16],[43], faulty contract management, construction projects [35].
[6],[34],[36],delays in decisions [15],[24], and
delays in approval of major changes of scope Though process-related factors influenced and
of work caused cost and time overruns in contributed to cost and time overruns
construction projects [41]. [10],methods of construction [31] and cost-
effective construction techniques [45] avoided
Proper management of site and supervision of cost overruns in construction projects.
the project would be ideal to avoid cost Similarly, decreased project duration and
overruns in construction projects [31]. increased productivity was achieved by
effective use of construction technology [38].
Changes in the scope was a significant factor
which incurred cost and time overruns in The review highlighted that cost and time
construction projects [6], [27], [28], [37]. The overruns spanned across all the phases
main source was the change in plans by the Preconstruction [4], Planning [25], Design
client, due to change of mind, which had a [4]and Implementation [26]) of a construction
critical impact on project cost, time and quality project and was spread over all five continents
of the projects [18]. Asia [15], Africa [41], America [4], Europe [17]
and Australia [29].
As requirements of the client expanded during
construction, additional works by the client 3. Methodology and Analysis
[40], [22], and the change orders significantly
contributed towards cost overrun[26]. Too After the extensive review of literature in the
many owner-generated change orders cause areas of cost and time overruns of construction
cost and time overruns in construction projects projects, the methodology identified and
[22].Change orders issued during construction record the contributory factors for cost and
generated cost overruns [26]as they had a time overruns of construction projects as
critical impact on cost, time and quality of highlighted by previous studies. The review
construction and it was concluded change showed that most of the factors identified
orders increased the contract value by about appeared in many of the other studies too. The
10% to 15% [18]. research was mostly based on qualitative
analysis method.
Changes in material specifications [44],
changes in quantities of material and
substituting material [18], unavailability of
material during implementation [45], increase
Table 1 - Contributory Factors Identified from the Literature Review
Rank Factor Occurrences Rank Factor Occurrences
1. Design aspects 29 22. Consultant's Issues 2
2. Planning and Controlling 20 23. Contract Conditions 2
3. Material Issues 19 24. Pre-Construction Issues 2
4. Scope Change 18 25. Disputes within parties 2
5. Finance 15 26. Equipment & Plant Cost 2
Client Influences -
6. 10 27. Government Policies 2
Changes
7. Management - Site 10 28. Information- Availability 2
8. Contractor Inexperience 9 29. Investigation - early 2
Poor Contract
9. 8 30. Perception on Issues 2
Management
10. Bid - Time given for 7 31. Quantities 2
3
91
11. Contractor - Defects 7 32. Social Issues 2
12. Construction Methods 7 33. Worker Productivity -low 2
Construction Technology Construction - Issues
13. 7 34. 1
related (Implementation)
14. Approval - Delays 6 35. Equipment - Productivity 1
15. Communication- Project 6 36. Estimation Issues 1
16. Management - Strategies 6 37. Internal Conflicts 1
17. Payment to Contractors 5 38. Optimism biased 1
18. Decision Delay 4 39. Procurement - Delays 1
19. Ground - Site Conditions 4 40. Records - Lack of 1
20. Worker Shortage 4 41. Uncertainties- Risks 1
21. Change Order Issues 3 42. Worker (Risky) Behavior 1
Total occurrences 237
The identified contributory factors for cost and
Three Highest Contributory Factors
time overruns of construction projects were
listed and the number of occurrences in the Material
Design
issues
literature review they were recorded as shown aspects
28%
in Table 1 above. 43%
Forty two contributory factors identified had a

total of Two hundred and thirty seven (237) Planning
occurrences in the literature reviewed (Table- and
1). Controllin
g
29%
Percentage Contribution by a factor = P
Number of Occurrences by a factor = n
Figure 1 - Three highest contributory factors.
Sum of Occurrences by all factors = N
Figure 2 below highlights the composition of
% of Occurrences by a factor = (n / N) x 100%
occurrences of the top seven contributing
factors which account for ~51% of 237 total
Sum and % contribution by the top ranking
occurrences.
contributory factors are given in Table 2 below.
Seven Highest Contributory
Site
Table 2- Sum and % Contribution of Top Client Factors manage
Ranking Factors. influence ment
Factors Sum of % from Total d 8%
grouped Contributions Contributions changes Design
(237) 8% aspects
Top 3 68 ~30% Finanace 24%
Planning
factors 12% and
Top 7 121 51% Scope Material Controlli
factors change issues ng
15% 16% 17%
Top 17 189 80%
Factors
Figure 2 - Seven highest contributory factors
Figure 1 given below indicates the composition
of occurrences by the top three contributing 4. Discussion
factors which account for ~ 30% of the total 237
occurrences. While 42 contributory factors for cost and time
overruns were found, the total sum of the
occurrences of the contributory factors in the
reviewed literature was found to be 237. The
highest occurrence of a contributory factor was
identified as 29, for Design aspects, and the
4
92
lowest contributory factor 1 was shared by 9 (41) contributory factors in construction
contributory factors (e.g. Optimism biased). projects.
(Table 1).
However, if an effective analysis was carried
Design aspects, planning and controlling and out on the said 41 contributory factors,
material issues were found to be the three perhaps at a relevant stage of a project (eg.
main contributors to cost and time overruns of preconstruction stage) cost and time overruns
construction projects. (Table. 1). of the projects could have minimised /
eliminated.
When the factors at the top of the list were
grouped, it was observed that sum and the % Further, through the review carried out, it was
of the contributions of the selected groups found that the industry sensitive contributors
were 68, ~30%; 121, 51% and 189, 80% for the for cost and time overruns of projects such as
groups of top three factors, top seven factors shortage of workers (Table 1, Rank 20), low
and top 17 factors respectively (Table 2). productivity of workers (Table 1, Rank 33) did
not appear as significant contributory factors
Above analysis indicated that factors ranked for cost and time overruns of construction
high with respect to occurrences dominated projects.
the contributions for cost and time overruns of
construction projects as indicated in Figure 3 It was revealed that there exist considerable
below. number of common factors contributing to
overruns irrespective of the type, cost,
Design aspects, Planning and controlling and duration, locality etc. of the project (Table -1).
Material related issues were identified to be
the top three key contributors to cost and time Focus of the review was to reveal type of
overruns of a construction project (Figure 3). possible contributory factors for overruns in
cost and time of construction projects. Hence
Some of the aspects found were, Design (delay the factors listed consists of contributions from
in design, failing to identify project risks at both cost overruns and/or time overruns. This
design stage, design errors, poor design, may be considered as a limitation.
change orders due to design changes etc.),
Planning and Controlling (incomplete 5. Conclusions
planning at the project implementation,
insufficient planning, poor planning, The review revealed there are 42 factors that
inaccurate / incorrect planning, etc.) and contributed to the overruns of cost and time in
Material (change in material specification, construction projects. The analysis highlighted
material deficiency, delay in material delivery, that most of these were due to the inadequacy
escalation of cost of material, substituting of proper analysis at pre-construction stage of
material by change orders, etc.) the projects. The analysis highlighted that
seven factors contributed to 51% of the
Out of 42 contributory factors identified, only a occurrences of contributory factors while top
single contributory factor, i.e. changes 17 factors contributed to 80% of the
influenced by the client, was identified as the occurrences of the contributory factors.
factor outside the agreed project scope and
hence out of control of the project team. All The analysis further revealed, design aspects
other 41 factors, i.e. 98% (41 out of 42), were (12%), planning and controlling (8%) and
within the agreed project scope and within the material (8%) were found to be the top three
control of the project team. Nevertheless, the key contributors with the highest number of
review carried out identified that all these 41 occurrences (Table 1). These three factors
factors had contributed to causing cost and contributed to ~30% of the occurrences (see
time overruns in construction projects. Table 2).Further, top 7 factors, design aspects,
planning and controlling, material issues,
Hence it is observed that the analysis carried scope change, finance, client influenced
out on identified contributory factors were changes and contractor inexperience,
inadequate and ineffective causing cost and contributed for 51% of the occurrences and top
time overruns by the identified controllable 17 factors contributed to 80% of the
occurrences (see Table 2).
5
93
The above indicates that the contribution for 6. Buys, F (2015) Five causes of project delay and
cost and time overruns for the construction cost overrun, and their mitigation measures.
projects by the identified factors were not Paper, LinkedIn.com
equally distributed and the contributory
7. Olander, S. (2007) Stakeholder impact analysis
factors ranked high in order contributed
in construction project management,
heavily (Table 1) for cost and time overruns of Construction Management and Economics,
construction projects. 2007, Vol. 25, Issue 3, 277-287.
However, 98% (41 out of 42) of the 8. Patel, C. H., Chaturvedi, H., Rao, M. R., Katta,
contributory factors were within the control of P. and Prasad, K. N. N. (2015). An analysis of
the project team the only exception being the cost and time overruns in construction
changes made by the owner (see Table 1). Yet projects. NICMAR Journal of Construction
Management, Vol. XXX April - June 2015 No.
these 41 factors within the control of the
II
project team continue to contribute to cost and
time overruns of construction projects. 9. Vu H.A., Wang, J., Min, L., Mai, S.H. and
Nguyen, H. P. (2016) Research on cost overrun
It is concluded that industry sensitive risk of construction phase of Vietnam highway
contributory factors like worker shortage and international contracting project, Engineering,
low worker productivity ranked 2016, 8, 86-98, Published Online March 2016 in
comparatively low in the analysis (Table 1). SciRes. http://www.scirp.org/journal/eng
http://dx.doi.org/10.4236/eng.2016.83011
Contributory factors appear to have surfaced
as the symptoms of the causes inducing cost 10. Adam, A., Josephson, P. and Lindahl, G.
(2014) "Implications of cost overruns and time
and time overruns of construction projects.
delays on major public construction projects".
Further research can be carried out to find the Proceedings of the 19th International
root causes and possible corrective actions of Symposium on the Advancement of
such contributory factors. Construction Management and Real Estate, 7-
9 Nov 2014, Chongqing
References
11. Priyadharshini, N. S. and Kumar S. S. (2015).
1. EU Technical report 2014-2478, Luxembourg. Project communication: Is key to productive
construction and its research needs in the
2. Shrestha, P. P., Burns, L. A. and Shield, D. R. future of construction engineering and
(2013) Magnitude of construction cost and management. International Journal of Science,
schedule overruns in public work projects. Technology & Management www.ijstm.com
Journal of Construction Engineering Volume Volume No 04, Special Issue No. 01, March
2013 (2013), Article ID 935978, 9 pages 2015 ISSN (online): 2394-1537
http://dx.do
12. Azhar, A., Farooqui, R. U. and Ahmed, S.M.
3. Jatarona, N. A., Yusof, A. M., Ismail, S. and (2008) Cost overrun factors In construction
Saar, C. C. (2016). Public construction projects industry of Pakistan, First International
performance In Malaysia. Journal of Southeast Conference on Construction In Developing
Asian Research Countries (ICCIDC–I) August 4-5, 2008,
http://www.ibimapublishing.com/journals/J Karachi,Pakistan, Conference paper,
SAR/jsar.html Vol. 2016(2016), Article ID https://www.reserchgate.net/publication/27
940838, 7 pages DOI: 10.5171/2016.940838. 7987526
4. Craigie, E. K., (2015) A framework for 13. Allahaim and Liu (2013) An empirical
estimating preconstruction service costs at the classification of cost overrun in infrastructure
functional level for highway construction projects by using analysis, University of New
projects. Graduate essays and Dissertations. South Wales, Built Environment,
Paper 14752. www.be.unsw.edu.au,
5. Ojo, G. K. and Odediran, S. J. (2015). 14. Morris, S. (1990), Cost and time overruns in
Significance of construction cost estimating public sector projects. Economic and Political
risks in Nigeria. International Journal of Civil Weekly, Vol. 25, No. 47, pp.M154-M168,
Engineering and Construction Science. Vol. 2, November 24, 1990.
No. 1, 2015, pp. 1-8. 2015; 2(1): 1-8 published
online April 20, 2015 15. Memon, A. H. (2014) Contractor Perspective
(http://www.aascit.org/journal/ijcecs) on Time Overrun Factors in Malaysian
6
94
Construction Projects. International Journal of 24. Bhatia, D. and Apte, M. R. (2015). Schedule
Science, Environment ISSN 2278-3687 (O) and Overrun and Cost Overrun in the
Technology, Vol. 3, No 3, 2014, 1184 – 1192... 1- Construction of Private Residential
Construction Project: Case Study of Pune,
16. Memon, A. H., Rahman, I. A. and Azis, A. A. India. International Journal of Technical
A. (2011). Preliminary Study on Causative Research and Applications e-ISSN: 2320-8163,
Factors Leading to Construction Cost www.ijtra.com Volume 4, Issue 2 (March-
Overrun. International Journal of Sustainable April, 2016), PP. 174-177.
Construction Engineering & Technology Vol 2,
Issue 1, June 2011. 25. Mulla, S. S. and Waghmare, A. P. (2015) A
Study of Factors Caused for Time and Cost
17. Jackson (2000) Project Cost Overruns and Risk Overruns in Construction Project and their
Management, School of Construction Remedial Measures. International Journal of
Management and Engineering, The University Engineering Research and Applications,
of Reading, Whiteknights, PO Box 219, www.ijera.com ISSN: 2015 Jan; 5(1):48-53.
Reading, RG6 6AW, UK,
https://www.reading.ac.uk>web>files, 26. Patel, A (2014) Effects and Causes of Cost
Visited 2nd May 2017. Overrun on Building Construction Projects,
Academia.edu.
18. Desai, N. J., S., Pitroda, J. and Bhavsar, J. J.
(2015) A Review on Change Order and 27. Emam, H., Farrell, P. and Abdelaal, M. (2015)
Assessing Causes Affecting Change Order in Causes of Delay on Large Infrastructure
Construction, Journal for International Projects in Qatar, Conference: 31st Annual
Academic Research for Multidisciplinary ARCOM Conference, At Lincoln, UK. DOI:
Impact Factor 1.625, ISSN: 2320-5083, Volume 10.13140/RG.2.1.1732.0400.
2, Issue 12, January 2015.
28. Bekr., G. A. (2015) Identifying Factors Leading
19. Olawale, Y. A. and Sun, M. (2010) Cost and to Cost Overrun in Construction Projects in
Time Control of Construction Projects: Jordan, Journal of Construction engineering,
Inhibiting Factors and Mitigating Measures in Technology & Management, Vol 5, No 3
Practice. Construction Management and (2015).
Economics, 28(5), 509-526.
29. Doloi, H.(2013) Cost Overruns and Failure in
20. Akinsiku, O. E. and Akinsulire A. (2012) Project Management: Understanding the Roles
Stakeholders’ Perception of the Causes and of Key Stakeholders in Construction Projects,
Effects of Construction Delays on Project Journal of Construction Engineering and
Delivery KICEM Journal of Construction Management, March 2013, Vol. 139, No. 3 : pp.
Engineering and Project Management Online 267-279 doi: 10.1061/(ASCE)CO.1943-
ISSN 2233-9582 25 7862.0000621).
www.jcepm.orghttp://dx.doi.org/10.6106/
JCEPM.2012.2.4.025.
30. Mukuka, M., Aigbavboa, C. and Thwala, W.
(2015). Effects of Construction Projects
21. El-Kholy, A. M. (2015) Predicting Cost
Schedule Overruns: A Case of the Gauteng
Overrun in Construction Projects,
Province, South Africa,
International Journal of Construction
http://creativecommons.org/licenses/by-nc-
Engineering and Management p-ISSN: 2326-
nd/4.0/). Peer-review under responsibility of
1080 e-ISSN: 2326-1102 2015; 4(4): 95-105.
AHFE Conference.
22. Kaveen, S., Aleem, M. I. A. and Thaarrini, J. 31. Shibani, A. and Arumugam, K. (2015).
(2015) Application of Value Engineering in
Avoiding Cost Overruns in Construction
Construction Project to Predict Time and Cost
Projects in India. Management Studies,
Overrun – An Overview; International Journal
August 2015, Vol. 3, No. 7-8, 192-202 D doi:
of Scientific Engineering and Applied Science
10.17265/2328-2185/2015.0708.003.
(IJSEAS) - Volume-1, Issue-9, December
2015ISSN: 2395-3470 www.ijseas.com.
32. Larsen, J. K., Shen, G. Q., Lindhard, S. M. and
Brunoe, T. D. (2015) Cost overrun, Denmark,
23. Edwards, P., Bowen, P., Hardcastle, C., and quality level, schedule delay)
Stewart, P. (2009) Identifying and http://dx.doi.org/10.1061/(ASCE)ME.1943-
Communicating Project Stakeholder Risks. 5479.0000391 Publisher: American Society of
Building a Sustainable Future: pp. 776-785. Civil Engineers.
doi: 10.1061/41020(339)79 ASCE.
33. Ameh, O. J., Soyingbe, A. A. and Odusami, K.
T. (2010) Significant Factors Causing Cost
7
95
Overruns in Telecommunication Projects in 40. Shibani, A. (2015) Time and Cost Overrun in
Nigeria. Journal of Construction in Construction Projects in Egypt,
Developing Countries, 15 (2). pp. 49-67. https://www.researchgate.net/publication/2
82219318, Visited, 2nd May 2017.
34. Rahman, I. S., Memon, A. H., Nagapan, S. and
Latif, Q. B. A.I. (2012). Time and Cost 41. Mukuka, M., Aigbavboa, C., and Thwala, W.
Performance of Construction Projects in (2015) Understanding Construction Projects’
Southern and Central Regions of Peninsular Schedule Overruns in South Africa. ICCREM
Malaysia. 2012 IEEE Colloquium on 2015: pp. 591-602. doi:
Humanities, Science and Engineering 10.1061/9780784479377.068 ASCE.
(CHUSER).
42. Ismail, I., Rahman, I. A. and Memon, A. H.
35. Long, N. D., Ogunlana, S., Quang, T. and Lam, (2013) Study of Factors Causing Time and
K. C. (2004) Large Construction Projects in Cost Overrun throughout Life Cycle of
Developing Countries: A Case Study from Construction Projects, Proceedings of
Vietnam, International Journal of Project Malaysian Technical Universities Conference
Management, Vol.22: 553-561. on Engineering & Technology (MUCET) 3-4
December 2013 , Kuantan, Pahang.
36. Okpala, D. and Aniekwu, A. (1988). Causes of
High Costs of Construction in Nigeria. Journal 43. Rahman, I. A., Memon, A. H., Karim, A. T. A.
of Construction Engineering and and Tarmizi, A. (2013) Significant Factors
Management, June 1988, Vol. 114, No. 2: pp. Causing Cost Overruns in Large Construction
233-244. Projects in Malaysia, Journal of Applied
Sciences 13 (2): 286-293, 2013 ISSN 1812-5654 /
37. Memon, A. H., Abdullah, M. A. H. I. A. M. R. DOI: 10.3923ljas.2013.286.293 O 2013 Asian
and Azis, A. A. (2010). Factors Affecting Network for Scientific Information.
Construction Cost in Mara Large Construction
Project: Perspective of Project Management 44. Patil, Y. K. and Bhangale, P.P. (2013)
Consultant. International Journal of Investigation of Factors Influencing Cost
Construction Engineering & Technology Vol 1, Overrun in High-Rise Building Constructions.
No 2, December 2010. International Journal in Latest Trends in
Engineering Technology (IJLTET), Volume 6,
38. Sepasgozar, S. M. E., Razkenari, M. A. and Issue 3, January 2016, ISSN: 2278-621X.
Barati, K. (2015) The Importance of New
Technology for Delay Mitigation in 45. Mahadik, U. A. (2015). Cost Reduction in
Construction Projects in Iran, American Construction Projects. International Journal of
Journal of Civil Engineering and Architecture, Engineering Technology, Management and
2015 3 (1), pp 15-20. Doi: 10.12691/ajcea-3-1-3. Applied Sciences. September 2015, Volume 3,
Special Issue, ISSN 2349-4476.
39. KPMG in India – PMI study (2012) Study on
project schedule and cost overruns in
infrastructure projects in India.
8
96
Annual Sessions
IESL,pp.
pp.[97[1- -103], 2017
10], 2017
Sub-Contracting and Accident Risk in Construction

Industry : A Case Study in Singapore
W.A. Asanka and Malik Ranasinghe
Abstract: Sub-contracting is a wide spread concept in construction project management.
Convenience in mobilization, competitive cost, lower welfare and relaxed management efforts are the
main factors that make sub-contracting popular. However, sub-contracting can increase the risk of
accidents in construction projects. Accident risk is higher in the construction sector, and it is an
adversed is advantages to the industry. Proper investigation and deployment of adequate mitigation
measures are key requirements for risk management. Risk factors such as negligence, knowledge,
legislations, communication, are aligned with the management system and relate to the sub-
contracting process as well.
Subcontractor’s contributions to construction projects can be as high as 90% of the total project cost.
Main contractors are taking the overall project responsibility and authority of the commence projects
work. Then, Sub contractors are invited to take part in specific work packages at execution stage. The
interaction between both parties in terms of accident risk, therefore, is an essential and sensitive
matter. Quantification of the accident risk is the main objective of this study. This study focuses on
subcontractor’s experience, communication ability, contribution and maintaining responsibility for
risk mitigation, and continuous improvement. Practicing engineers and supervisors were questioned
and the data was analysed to draw conclusions in this study. Practical project experience related to the
study was collected from Singapore, as the Singaporean construction industry has been considerably
successful in implementing safety and health regulations with low dangerous incidents.
Keywords: Subcontracting, Accident risk, Bayesian inference
1 Introduction quality and productivity standards[5],
coordination and supervision difficulties
1.1 Subcontractor in construction industry should be expected to due to temporary and
General contractor, the prime contractor, or the insecure relationships among site management
main contractor is the main party that team [6].
undertakes the responsibility for projects.
Contract administration, financing, material 1.3 Consequences
Minor, major or fatal injuries to employees are
and equipment procuring, monitoring and
consequences in construction incidents.
managing the progress are some of the main
Equipment or machinery damages are
contractor’s obligations[1]. Sub-contractors are
secondary effects. However, in terms of cost, all
below the main contractor as they are only
accidents are accountable losses on
responsible for specific work items.
organizational financial achievements.
Main contractors are sometimes reluctant to
Reported fatality rates, around the world, are
maintain in-house resources (humans and
evidence of the crisis of accidents and its
equipment) for construction activities. Sub-
diversified spread[6], [7]. Fatality statistics of
contractors, on the other hand,,are small
Singapore from 2006 to 2013 are presented in
companies, with expertise skills, knowledge
Table 1 as an example of the necessity for
and machinery for specific work categories.
further investigations on accident risks.
Supplying materials, manpower, tools,
equipment and designs [2] are some of the Table 1 - Fatality in construction sector, Singapore
[http://www.mom.gov.sg/]
functions of sub-contractor. Competitiveness in
cost, time and quality are key factors for their
2006
2007
2008
2009
2010
2011
2012
2013
popularity. Year
1.2 Issues in the process

Fatality
24
24
25
31
32
22
26
33
Sub-contracting has been identified as one of

the causes of workplace accidents[3], [4]. Short
operational durations and lack of concern about
safety and health aspects in contractual Eng. W.A.Asanka, AMIE(SL), GMICE,Msc(HW,UK),
agreements are pushing sub-contractor towards Design Manager, Dredging International Asia Pacific,
Singapore.
unsatisfactory performances. Ignorance or
Eng. (Prof.) Malik Ranasinghe, Senior Professor,
negligence is high in the process. Declining Department of Civil Engineering, University of Moratuwa.
1 97 ENGINEER
1.4 Objectives However, studies of subcontractor and main
Objectives of this study are the quantification of contractor with safety and health aspects are
the accident risk by the application of Bayesian limited in the literature[16].
Belief Network (BBN) model and to analyse the Lack of supervision due to distance operation
predicted accident risk in construction sector of sub-contractor’s official staff makes the
due to the performance of subcontractors. working crew less decisive and irresponsible.
The inherent risks in sub-contractors in the
1.5 Scope
Singapore Construction Industry are due to low
The study is an indication of general accident management competency [4]; Subcontracting,
risk associated with the sub-contracting process Foreign Workers and Job Safety [17].
in Singapore. Sub-contractors daily interactions
However, the dependence of time, quality and
with site management staff are the main scope
cost on sub-contractors may lead overall
of this study.
performance of main contractor [8], [18].
2 Literature Review Improvements of sub-contracting process can
2.1 General be by registration[20], [21], selection [22], [23],
Subletting or sub-contracting is a management [24], performance rating at construction stage
process, which divides the workload among [8], [18], [15], [25], [26] , [27] , [28].
specialised contractors. The popularity, as a In terms of stabilising organizational financial
fraction of the project cost, is between 50% [8] risks at business fluctuations, multilayer
to 90% [2] in total. Subcontracting is cascading contracting process is popular [2], [11], [29],
down to different layers, called multilayer or [30], [31], [22], [32],[33].Multilayer contracting
latent sub-contracting, sharing work schedules makes subcontractor’s overall performance
or damages among layers[9] can be highlighted more stable at business fluctuations, flexibility
in multilayer sub-contracting. in workforce coordination, and reduction of
The sub-contracting is more prevalent in the management cost [29] andare highlighted as
building construction industry than that of civil positive trends in the industry.
engineering or industrial sectors[10]. A Hong 2.3 Synopsys of literature review
Kong based study classified the subcontractors
Sub-contracting process has certain constraints.
to, equipment-intensive and labour-intensive
However, it is a management process and one
according to their specialist resource
of necessity in the construction industry.
availability[11]. Labour only sub-contractors
Therefore proper administration, management,
are reducing the costs on mobilising or mark
and regulatory aspects are required to maintain
ups. However, temporary workforce provides
healthier and safer work places.
the best advantages to the large firms [12],
[13].In Turkey, a study with 35 general Accident risk is one of the key disadvantage
contractors and 56 sub-contractors, concluded inherent with the subcontracting process in the
that work overload and time pressure is leading construction industry. Difficulties in the
them towards subcontracting to secondary interaction between sub-contractor with the
levels [14]. main contractor enhance the health and safety
risks, which have less attention from the
2.2 Pros and cons of the process research community.
A misunderstanding between both parties 3 Methodology
(main and sub-contractor) may result in costly
litigation or dissatisfied customers [15], while The methodology of this study is, constructing
declining standards of quality, falling a Bayesian Belief Network (BBN) by gathering
productivity, coordination difficulties, both quantitative and qualitative information
supervising and monitoring problems due to through expert opinions and literature review.
temporary and insecure relationship [5], [19] Parent and child nodes of the BBN were
cause sub-contracting accidents. defined according to the expert opinions. Then
further data will be used to learn BBN to use in
A UK based studies[3], [4], reported that accident risk prediction of sub-contractor
restricting the layers of subcontractors, working performance.
with a regular chain of subcontractors,
implementing a safety and health reward A freely available academic version of Uninet
scheme for subcontractors and insisting on non- software is used to construct BBN.Bayes
working sub-contractor foremen who have inference concept as described in Eqn. 1 is
direct responsibility for the safety of workers in inherent with the software. Normal copulas
their trade, improved safety at sites[4].
ENGINEER 2 98
were used in the current implementation to to risk mitigation, T1.4:Holding the
model influences from parent nodes. responsibility, and T1.5:Continual
The BBN model is based on the following improvement. These factors were taken as Tier
Bayesian concepts as given in Eqn. 1., 1, the parents of the accident risk in the model.
( ⁄ ) ( ) The construction site is a complex network with
( ⁄ ) ∑ anenormous amount of activities. Data
( ⁄ ) ( ) ( ⁄̅) (̅)
collection requires more tangible divisions of
( ⁄ )
risk factors. Therefore, four major site activity
( ⁄ ) ratings (Tier 2) and causes affecting their
( ) ⁄ ratings were proposed in the current study for
( ⁄ ) ( ⁄̅̅̅̅̅̅̅) data collection, i.e.T2.1:Method Statement
(MS),T2.2:Risk Assessment (RA),T2.3:Site Work
( ) (̅̅̅̅̅̅̅)
(SW),and T2.4:Site Management (SM). Finally,
3.1 Model development data were collected for the marginal probability
Initially, 5 construction industry professionals distribution of each cause through expert
with over 10 years of experience in Singapore opinions.
construction industry were interviewed to Causes affecting the ratings (Tier 3) have been
categorise risk factors related to Sub- set up using daily site activities of the site
contractors. All of them started their career as management team. Repetitive occurrences of
site engineers and proceeded to project those causes are much familiar with experts
engineer (2 of them) followed by project thus expected accuracy could be acceptable.
managers (3 persons). Singapore national statistics were accessed for
Five major risk drivers could be identified in estimation of accident risk at present. The
the interviews T1.1:Experience, model is presented in Figure 1.
T1.2:Communication ability, T1.3:contributing
Figure 1 - BBN with dependencies on each effect with cause
4 Data collection Risk Assessment (RA) are two main items in

sub-contractors documentation scope.
4.1 Nominating the four ratings in Tier 2
A method statement is a document expressing
Main contractor’s engineers deal with sub- sub-contractor’s expertise about the work they
contractors on a daily basis when managing the are about to engage. Proper reviewing of work
construction sites. Method Statement (MS) and scope of which have to be done is a good
indication of the sub contractor’s performance.
3 99 ENGINEER
MS represents industry updates, especially distribution (Eqn. 3) was used with alpha and
about updated safe work methods and incident beta fitted with least square fitting method.
studies. are vital in preparation of the content ⁄
of the MS.
( )
Similarly, the Risk Assessment also caters to
( )
determining applicable work sequence for
( ) ( )
specific site conditions and require expert ( )
guidance in assessing and predicting upcoming ( )
risks. Table 2 - Fatalities in Singapore, "Work place
Site work is a major concern, and it is the key to safety and health report,2015 National
statistics."
control accidents. Responsibility, commitment,
and vigilance are key areas investigated in this Year Fatality rate Rolling average
research study. Following are the brief content 2006 24 24.00
2007 24 24.00
of each item that has been studied
2008 25 24.33
Ratings of site work and site management is the 2009 31 26.00
second step of sub-contractors involvement 2010 32 27.20
with main contractor’s site staff. Over 90% of 2011 22 26.33
interviewees agree to use of the four ratings as 2012 26 26.29
sub-contractors' performance evaluation 2013 33 27.13
indexes. 2014 27 27.11
When nominating the tier 2 ratings, 15 site 2015 26 26.50
engineers and 20 site supervisors were 4.4 Questionnaire based data collection for
interviewed. All of them had over 5years of Tier 3: marginal distributions of risk factors
experience in Singapore construction projects.
A questionnaire was used to collect data for
4.2 Conditional relationship between Tier 3 risk factors. Background information of
nodes& prior probabilities the respondents was questioned at the
Four ratings at Tier 2 and causes in Tier 3, Risk beginning to confirm respondent’s position,
drivers in Tier 1 with ratings are connected work scope and company type followed by risk
with parent child relationships in Uninet factors. Then, all tier 3 factors were asked to
software. Interrelations for more than one rate in 1~5 scale. There were 55 respondents
parent per child are factorised equally, participated in the study.
assuming the same influence from each parent.
Safety Manager 5
Interviews with 3 experts from health and
Safety supervisor 4
safety departments were convinced about these Project Manager 3
relationships and followed their guidance. Tier Site Manager 7
1 factor’s effect on a number of fatalities is Site Engineer 36
taken as negative. Prior probabilities of Tier 1 &
0 10 20 30 40
2 factors assumed as uniform distributions with
No of personal
mean value 0.5.
4.3 Status of fatalities Figure 2 - Positions and number of
No of fatalities has been modelled using poison respondents
distribution (Eqn. 2) as prior information which All the respondents are selected above 5 years
is to be updated with influence data. Rolling of experience. All of them possesses basic safety
average of cumulative months has been used to and health training which is mandatory in
enhance the accuracy of predictions. However, Singapore (www.mom.gov.sg).
as poison distribution is not integrated with Average ratings for each factor are summarized
Uninet software, therefore equivalent Beta inTable 3.
Table 3 - Tier 2 and 3 Factors& Average ratings
Rate
ID Causes and Effects (Tier 2 & 3)
1 2 3 4 5
T2.1 Method statement (MS) Rating
T2.1.1 Content relevant to work scope 2% 15% 51% 25% 7%
T2.1.2 Properly reviewed the Main contractors drawings 2% 9% 40% 38% 11%
T2.1.3 Work sequence is relevant 2% 5% 22% 51% 20%
T2.1.4 Up to date safest Work method proposed 2% 13% 39% 33% 13%
T2.1.5 Proper equipment is proposed 5% 33% 44% 16% 2%
ENGINEER 4 100
T2.1.6 Equipment maintenance is up to date 2% 18% 38% 29% 13%
T2.2 Risk Assessment (RA) Rating
T2.2.1 Risk identification 2% 13% 41% 35% 9%
T2.2.2 Risk estimation 5% 35% 40% 16% 4%
T2.2.3 Contingency Plan provided & adequate 2% 4% 37% 44% 13%
T2.2.4 Recovery Plan provided & adequate 2% 11% 36% 40% 11%
T2.3 Site work Rating
T2.3.1 Relationship with main contractors site staff 2% 13% 36% 42% 7%
T2.3.2 Sufficient supervisory staff is deployed 7% 35% 38% 16% 4%
T2.3.3 Effectiveness of supervision 2% 33% 45% 18% 2%
T2.3.4 Proper equipment provided 2% 7% 33% 44% 13%
T2.3.5 Adhered to proposed work method in MS 2% 11% 33% 40% 15%
T2.3.6 Efficiency of incident reporting 4% 35% 42% 18% 2%
T2.3.7 Approach on rectification of unsafe acts or conditions 2% 5% 31% 49% 13%
T2.4 Site Management rating
T2.4.1 Competency of site staff 2% 2% 27% 51% 18%
T2.4.2 Actions to rectify unsafe acts / conditions 2% 7% 25% 47% 18%
T2.4.3 Instruction to site staff 2% 27% 51% 18% 2%
T2.4.4 Approach on MS review 4% 38% 40% 16% 2%
T2.4.5 Approach on RA review 7% 36% 36% 16% 4%
T2.4.6 Extend of foreign workers deployment 0% 0% 5% 29% 65%
T2.4.7 Awareness of emergency procedures 2% 19% 35% 30% 13%
T1.3 T2.4 0.2308
5 Analysis and Results Part 5
The advantage in the Bayesian inference model T1.4 T2.1 -0.0034
is learning with available data and predicting T1.4 T2.2 0.0121
the future situations with given initial
T1.4 T2.3 0.4957 Highest
conditions. Sensitivity analysis option built in
with Uninet software evaluates most sensitive T1.4 T2.4 0.4294
factors for current accident occurring. Table 4 Part 6
summarises the sensitivity data. T1.5 T2.1 0.3084
Table 4 - Sensitivity analysis T1.5 T2.2 0.0094
Predicte Rank T1.5 T2.3 0.3332 Highest
Base
d Correlatio Remark
Variable T1.5 T2.4 0.2908
Variable n
Part 1 Part 7
T1.1 -0.6003 Highest T2.1 T2.1.1 0.1551
NO_OF_FATAL
T2.1 T2.1.2 0.1556 Highest

T1.2 -0.5987
ITIES
T1.3 -0.5862 T2.1 T2.1.3 0.1481

T1.4 -0.5232 T2.1 T2.1.4 0.1467
T1.5 -0.5505 T2.1 T2.1.5 0.1379
Part 2 T2.1 T2.1.6 0.1481
T1.1 T2.1 0.2925 Part 8
Highest T2.2 T2.2.1 0.2342 Highest
T1.1 T2.2 0.2947
T1.1 T2.3 0.2732 T2.2 T2.2.2 0.2267
T1.1 T2.4 0.0071 T2.2 T2.2.3 0.2234
Part 3 T2.2 T2.2.4 0.2083
T1.2 T2.1 0.233 Part 9
T1.2 T2.2 0.2494 T2.3 T2.3.1 0.1212
T1.2 T2.3 0.2577 Highest T2.3 T2.3.2 0.1257
T2.3 T2.3.3 0.1346 Highest
T1.2 T2.4 0.2165
Part 4 T2.3 T2.3.4 0.1308
T1.3 T2.1 0.2587 Highest T2.3 T2.3.5 0.1315
T1.3 T2.2 0.2381 T2.3 T2.3.6 0.1274
T1.3 T2.3 0.2320 T2.3 T2.3.7 0.1254
Part 10
5 101 ENGINEER
T2.4 T2.4.1 0.1311 Analysis & Recommendation Report on Work
Highest at Height” by MOM, Singapore.
T2.4 T2.4.2 0.1375
Communication, mitigation, responsibility and
T2.4 T2.4.3 0.1302
continual improvements are also equally
T2.4 T2.4.4 0.1289 important in the prevention of accidents.
T2.4 T2.4.5 0.1272 Developed model in the current study is the
T2.4 T2.4.6 0.0950 status of Singapore industry. This model with
T2.4 T2.4.7 0.1110 marginal probabilities of risk factors (Tier 3
parameters) can be used for another similar
Rank correlation of each predicted variable is
environment, Srilankan industry, to predict
highlighting the individual contribution to its
fatality risks. The authors would like to extend
base variable. With respect to correlations, T1.1
the study towards Sri Lankan construction
(Experience) is the highest and T1.2
industry to identify steps necessary to improve
(Communication ability), T1.3 (contributing to
safety and health performance of Sri Lanka.
risk mitigation), T1.4 (Holding the
responsibility), and T1.5 (Continual References
improvement) have proven their importance 1. Benjaoran, V., “A cost control system
through significant ranks. development: A collaborative approach for
Each part of the analysis in Table 14 has its own small and medium-sized contractors,” Int. J.
highest correlation to base variable. When Proj. Manag., vol. 27, no. 3, pp. 270–277, 2009.
considering the tier 3 variables T2.1.2 (Properly 2. Kumaraswamy, M. M., and Matthews, J. D.,
reviewed the Main contractors drawings), “Improved subcontractor selection
T2.2.1 (Risk identification), T2.3.3 (Effectiveness employing partnering principles,” J. Manag.
of supervision), and T2.4.2 (Actions to rectify Eng., vol. 16, no. 3, pp. 47–57, 2000.
unsafe acts / conditions) it is highly clear that 3. Sabet, P. G. P., Aadal, H., Jamshidi, M. H. M.,
effectiveness in day to day work activities or and Rad, K. G., “Application of Domino
how individuals review their surrounding Theory to Justify and Prevent Accident
leads the accidents, paying attention on work is Occurrence in Construction Sites,” IOSR J.
Mech. Civ. Eng., vol. 6, no. 2, pp. 72–76, 2013.
required for mitigation.
4. Manu, P., Ankrah, N., Proverbs, D., and
All of these factors are management related and
Suresh, S., “Mitigating the health and safety
there is higher chance to achieve good
influence of subcontracting in construction:
performance by implementing proper The approach of main contractors,” Int. J.
management system. Proj. Manag., vol. 31, pp. 1017–1026, 2013.
Singapore achieved good health and safety 5. Heery, E., and Salmon, J., “Introduction: The
performance through careful deployment of Insecure Workforce,” Manag. Res. NEWS, vol.
competent personal to work sites. Required 21, pp. 7–8, 1998.
training and inductions prior to work is 6. Ling, F. F. Y. Y. F. Y. Y. F. Y. Y., Liu, M., and
mandatory. Regulatory framework is smart and Woo, Y. C. Y. C. Y., “Construction fatalities
wider to capture unpredicted situations of in Singapore,” Int. J. Proj. Manag., vol. 27, pp.
work sites. It is highly recommended to deploy 717–726, 2009.
experienced personal for certain works in all 7. Hämäläinen, P., Takala, J., and Saarela, K. L.,
construction projects since the experience is “Global estimates of occupational accidents,”
essential to deliver quality and reliability tasks Saf. Sci., vol. 44, pp. 137–156, 2006.
successfully.
8. Albino, V., and Garavelli, A. C., “A neural
6 Conclusions network application to subcontractor rating
Objectives of this study were to quantify the in construction firms,” Int. J. Proj. Manag.,
vol. 16, no. 1, pp. 9–14, 1998.
accident risk by the application of Bayesian
Belief Network (BBN) model and to analyse the 9. Shimizu, J. Y., and Cardoso, F. F.,
predicted accident risk in construction sector “Subcontracting and cooperation network in
due to the performance of subcontractors. building construction: a literature review,”
Proc. IGLC-10, Gramado-RS, 2002.
The study showed experience is the main cause
10. Clough, A., hr; Sears, “Construction
of accidents and it has –0.6003 correlation with
Contracting,” Constr. Contract. Law Manag.,
fatalities. Work procedure, work environment, vol. 7, no. Jan, pp. 12–31, 2005.
supervision are key elements, which can be the
cause of experience, identified in the “Safety 11. Thomas Ng, S., Tang, Z., and Palaneeswaran,
E., “Factors contributing to the success of
equipment-intensive subcontractors in
ENGINEER 6 102
construction,” Int. J. Proj. Manag., vol. 27, no. selection of sub-contractors for
7, pp. 736–744, 2009. refurbishment projects,” Beijing, China, 1996.
12. Bresnen, M., Wray, K., Bryman, A., 25. Chiang, Y. H., “Subcontracting and its
Beardsworth, A. D., Ford, J. R., and Keil, E. ramifications: A survey of the building
T., “The Flexibility of Recruitment in the industry in Hong Kong,” Int. J. Proj. Manag.,
Construction Industry: Formalisation or Re- vol. 27, no. 1, pp. 80–88, 2009.
Casualisation?,” Sociology, vol. 19, no. 1, pp.
26. Ko, C. H., Cheng, M. Y., and Wu, T. K.,
108–124, 1985.
“Evaluating sub-contractors performance
13. Beardsworth, a D., Keil, E. T., Bresnen, M., using EFNIM,” Autom. Constr., vol. 16, no. 4,
and a Bryman, “Management, Transience pp. 525–530, 2007.
and Subcontracting: the Case of the
27. Maturana, S., Alarcón, L. F., Gazmuri, P., and
Construction Site,” J. Manag. Stud., vol. 25,
Vrsalovic, M., “On-Site Subcontractor
no. 6, pp. 603–625, 1988.
Evaluation Method Based on Lean Principles
14. Assaf, S. A., and Al-Hejji, S., “Causes of and Partnering Practices,” J. Manag. Eng.,
delay in large construction projects,” Int. J. vol. 23, no. 2, pp. 67–74, 2007.
Proj. Manag., vol. 24, no. 4, pp. 349–357, 2006.
28. Wang, W. C., and Liu, J. J., “Factor-based
15. Sturts Dossick, C., and Schunk, T. K. path analysis to support subcontractor
“Subcontractor Schedule Control Method,” J. management,” Int. J. Proj. Manag., vol. 23, no.
Constr. Eng. Manag., vol. 133, no. 3, pp. 262– 2, pp. 109–120, 2005.
265, 2007.
29. Yik, F. W. H., and Lai, J. H. K., “Multilayer
16. Daykin, N., “Workplace health promotion: subcontracting of specialist works in
Issues and strategies for the insecure buildings in Hong Kong,” Int. J. Proj. Manag.,
workforce,” 1997. vol. 26, no. 4, pp. 399–407, 2008.
17. Debrah, Y. A., and Ofori, G., 30. Arditi, D., and Chotibhongs, R., “Issues in
“Subcontracting, Foreign Workers and Job Subcontracting Practice,” J. Constr. Eng.
Safety in the Singapore Construction Manag., vol. 131, no. 8, pp. 866–876, 2005.
Industry,” Asia Pacific Bus. Rev., vol. 8, no. 1,
31. Hsieh, T.-Y., “The economic implications of
pp. 145–166, 2001.
subcontracting practice on building
18. Mbachu, J., “Conceptual framework for the prefabrication,” Autom. Constr., vol. 6, no. 3,
assessment of subcontractors’ eligibility and pp. 163–174, 1997.
performance in the construction industry,”
32. Olson, R., “Subcontract coordination in
Constr. Manag. Econ., vol. 26, no. July 2015,
construction,” Int. J. Prod. Econ., vol. 5273, no.
pp. 471–484, 2008.
97, pp. 503–509, 1998.
19. Costantino, N., Pietroforte, R., and Hamill,
33. Perng, Y. H., Chen, S. J., and Lu, H. J.,
P., “Subcontracting in commercial and
“Potential benefits for collaborating
residential construction: an empirical
formwork subcontractors based on co-
investigation,” Constr. Manag. Econ., vol. 19,
operative game theory,” Build. Environ., vol.
no. 4, pp. 439–447, 2001.
40, no. 2, pp. 239–244, 2005.
20. Ng, S. T., Luu, C. D. T., and Chu, A. W. K.,
“Delineating criteria for subcontractors
registration considering divergence in skill
base and scales,” Int. J. Proj. Manag., vol. 26,
no. 4, pp. 448–456, 2008.
21. Ng, S. t T., and Luu, C. D. T., “Modeling
subcontractor registration decisions through
case-based reasoning approach,” Autom.
Constr., vol. 17, no. 7, pp. 873–881, 2008.
22. Okoroh, M. I., and Torrance, V. B., “A model
for subcontractor selection in refurbishment
projects,” Constr. Manag. Econ., vol. 17, pp.
315–327, 1999.
23. Arslan, G., Kivrak, S., Birgonul, M. T., and
Dikmen, I., “Improving sub-contractor
selection process in construction projects:
Web-based sub-contractor evaluation system
(WEBSES),” Autom. Constr., vol. 17, no. 4, pp.
480–488, 2008.
24. Okoroh, M. I., “A knowledge based decision
support and risk management system for the
7 103 ENGINEER
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[105 113],2017
2017
© The
© The Institution
Lanka
Techno-Economic Analysis of Building Energy System

with Net Meter Solar PV in Sri Lanka
K.T.M.U. Hemapala and T.M.S.C. Jayasinghe
Abstract: Optimum net meter solar PV capacity to be installed is depend on various factors
such as tariff category, solar irradiation, building load profile, maximum demand of the building etc.
Thus it is essential to do a techno economic analysis for deciding the most feasible PV system capacity.
A methodology is proposed to develop an optimum energy solution tool using techno- economic
analysis of building energy system with net meter solar PV in Sri Lanka considering three key variable
parameters; load profile, tariff scheme and the maximum demand. The software HOMER was used to
model the energy system and the tool is developed in Microsoft Excel. HOMER simulation has done
for 114 cases by changing the above mentioned key parameters. Selected results from HOMER
simulation such as PV system cost, capital cost, replacement cost, O &M cost, cost of energy,
renewable fraction, simple payback period, discounted payback period etc., are extracted and stored
in an Excel based database. Developed tool has three inputs as key variables; load profile, tariff
scheme and the maximum demand. And Optimum solar PV capacity to be installed, related PV
system investment cost, cost of energy, payback period are the outputs. If the user cannot go for the
optimum PV capacity due to the limitations on cost or area, then this tool can be used to select
whatever the other feasible capacity and to compare them. This tool is more convenient for any user
even without much knowledge on net meter solar PV installations and can see the feasibility of it
without having consultancy from solar panel providers. The tool developed has been verified
considering two actual situations.
Keywords: Optimum PV Capacity, Load Profile, Tariff Scheme, Maximum Demand
1. Introduction Thus at the conceptual design stage of a

building it is tedious to do such analysis.
Green building concept is widely considered in Instead user friendly method is preferred to
Sri Lanka. Optimizing the building energy obtain optimum combination of net meter PV
system with renewable energy sources is also system with important factors such as cost of
play a major role in this concept. Different energy, installation cost, payback period & etc.
types of renewable energy technologies can be and it will be really helpful at the conceptual
used depend on their feasibility, and net meter design stage. During the detailed design stage
solar Photovoltaic (PV) is very popular in Sri of PV installation, the results of the conceptual
Lanka. design could be considered as input
information.
However solar PV installation should be a
feasible solution in both economically and 1.1 Net Meter PV System
technically. Before the installation, a feasibility In the concept of the net metering, consumers
analysis should be carried out to obtain are allowed to connect their on site renewable
maximum outcome of it. Solar PV optimum generators to the grid. Consumers have to pay
capacity to be installed is depend on various only for the difference of export and import
parameters such as tariff category, solar energy units. Import energy is the amount of
irradiation, PV system technical characteristic, energy units which a user consumed from the
building load profile, maximum demand of the utility grid per month whereas the export
building etc. Thus it is essential do a proper energy is the amount of energy supplied by
analysis to decide the PV system capacity and the on- site generation to the grid. In the
also important to know about the investment, concept of net metering input / export energy
cost of energy, payback period etc. meter is used instead of normal energy meter.
Figure 1shows an arrangement of net meter
There are many computer software tools solar system. [14]
capable of analyzing the integration of Eng. (Dr.) K. T. M. U. Hemapala, MIE(SL),
renewable energy systems. For all those B.Sc.Eng. (Moratuwa), Ph.D.(Italy), Senior Lecturer,
Department of Electrical Engineering, University of
software need various parameters as inputs
Moratuwa.
such as tariff data, solar irradiation levels, PV Eng. (Ms.) T.M.S.C. Jayasinghe B.Sc.Eng. (Peradeniya),
system cost, load variation pattern and etc. M.Sc.Eng. (Moratuwa. AMIE (SL) Electrical Engineering
Department of Buildings.
105
The billing period is one month. At the end of payback period. But they have limitations such
a billing period the bill is prepared according as considered only a single category (Ex:
to the difference between import and export Residential), no load variation pattern is
energy units. There may be an excess considered etc. Sri Lanka Sustainable Energy
electricity due to the high number of export Authority (SLSEA) has developed a tool which
units than import units. In such cases, that helps to select the most economical net meter
excess amount will be carried forward to use solar panel system based on the investment
in future. The energy balancing of export and and the average electricity bill [15]. This tool
inputs units is carried out within the same has no capability to decide the optimum
tariff tier and credit in one tire is not system capacities. And also it is considered
transferable to another tier. only residential buildings and not considered
the load variation patterns.
3 Modelling and Simulation

3.1 Model Proposal
Figure 2 shows the proposed model which
contains three selectable inputs, a database and
an output. The first input is the type of the
building (cluster) which is based on the load
variation pattern of the building as
Commercial, Office, Hotel etc (12 clusters with
load curve are considered).The second input is
the tariff category which is based on applicable
Figure 1 - Solar PV Net Metering tariff scheme on Commercial, Office, Hotel
Arrangement [14] tariff schemes (GV2, GP2 & H2 are considered;
as CEB tariff scheme [16]) and the third input
1.2 Electricity Load Profile in Buildings
is the maximum load demand range of the
Load profile gives the variation of the electrical
building. Optimization model contains a
load versus time. This will vary according to
database which contain optimized net meter
the customer type such as residential,
solar PV system parameters which are
commercial and industrial. Load profile
analyzed using a micro power system
represents the loading pattern of various
optimization software.
electrical loads belong to an electrical
installation. When considering the Sri Lankan
buildings, there are several load profile based
on the building usages such as office buildings,
super markets, fashion outlets, banks,
hospitals, hotels, prisons etc. Those variation
details are obtained from a previous research
[13] and shown in Table 1.
2 Literature review
Several researches are carried out to optimize
the energy systems and to analyze reliable Figure 2 - Proposed Model
3.2 HOMER Simulation
electricity with renewable energy using
various software [2]. Software HOMER is
widely used to analyze a stand-alone systems
[6], [7]. Many researches have been done to
find out optimal renewable energy system
parameters including solar PV for grid
connected systems [3], [4], [5] and [12]. But in
many of these, only a single load variation and
one tariff scheme are considered. They are like
case studies and no tool was developed to
have a general idea for different cases. [9], [10],
[11].
There are few tools developed nationally and Figure 3 – Developed model in HOMER
internationally to find out capital cost and
106
Table 1- Cluster Description with Sample Installations [13)
Possible tariff
Cluster Example installations
categories
Government banks, Government offices, University, Educational

Cluster 2 Institute type 1 , Embassy offices type 1, High commissioner‟s GV2 , GP2
offices, Municipal Councils
Cluster 4 Financial institute office, Fashion Outlets type 1 GP2
Cluster 5 Private banks, Private company offices type 1, Insurance offices GP2
Cluster 6 Show rooms(shoes), Apartments GP2

Hospitals type 1, Embassy office type 2, Institute type 2, Private
Cluster 7 GV2 , GP2
company office type 2
Cluster 8 Utility service officers, Hotel type 1 GP2 , H2
Cluster 9 Hotels type 2, Hospitals GV2, GP2 , H2
Cluster 12 Hotels type 3 H2
Cluster 13 Prisons GV2 , GP2
Super markets(consumer goods, electrical items), Food outlets,
Cluster 14 GP2
Fashion Outlets type 2
Cluster 16 Railway Stations, Hotel type 4 GP2 , H2
Cluster 17 Fashion Outlets type 3, Private company office type 3 GP2
As shown in Figure 3, HOMER simulation is

consisted with a Primary load, Generator, PV, ….(2)
Converter and Grid. Electrical demand of a
power system at a time is identified as the
primary load in HOMER. But it will add where Com;gen is the O&M (Operational and
randomness to the synthesized data in order to Maintenance) cost in dollars per hour, Crep;gen
have a unique pattern for everyday load. the replacement cost in dollars, Rgen is the
Randomness is specified by the user. generator lifetime in hours, F0is the fuel curve
intercept coefficient in quantity of fuel per
HOMER assumes that the diesel generator fuel hour per kilowatt, Ygenis the capacity of the
curve which is shown in Equation 1, is a linear generator (kW), and Cfuel;eff is the effective price
variation [1]. Fuel curve is obtained of fuel in dollars per quantity of fuel.
considering the fuel consumption per hour
and respective output power generation Marginal cost of energy of the generator
capability. which is the additional cost other than the
fixed cost in dollars per hour is calculated
.....(1) using equation 3 [1]in HOMER.

WhereF0 is the fuel curve intercept coefficient, ….(3)
F1 is the fuel curve slope, Ygen the rated capacity
of the generator (kW), and Pgen the electrical WhereF1 is the fuel curve slope in quantity of
output of the generator (kW).Unit of F is fuel per hour per kWh and Cfuel;eff is the
depend on the unit of fuel if fuel is in liters, F effective price of fuel in dollars per quantity of
is l/h. fuel.
Fixed cost of energy of the generator is PV array can produce DC electricity in

calculated using the equation 2 [1]. HOMER. Electricity generation of the PV array
is directly proportion to the global solar
107
irradiation on it. Power output of the PV array from the grid under the grid off tariff, as
is given by equation 4[1]. shown in Figure 4. This will also require a
modification of the grid feed-in tariff schedule
..…(4) to prevent the sale of electricity back to grid
that is not available. This can be achieved by
Where, fPV is the PV derating factor,YPVis the altering the HOMER grid inputs for grid feed
rated capacity of the PV array (kW), ITis the in tariff to be $0 when the “Grid off” is
global solar radiation incident on the surface of scheduled. So this will not affect the net
the PV array (kW/m2), and ISis the standard present cost. It affects the net purchases figure,
amount of radiation used to rate the capacity but this aspect has not been considered for the
of the PV array. ISis 1 kW/m2. development of the tool in the present study.
Solar resource data should be provided by the

user in order to model a system containing a
PV array. Monthly average solar irradiation
level and the latitude for the location of
interest can be given. Then HOMER generates
synthetic hourly solar irradiation data. For this
HOMER use an algorithm developed by
Graham and Hollands [8].
In HOMER, grid is also available. Consumers

can purchase AC electricity from grid and sell
back to grid. The tariff scheme can be modeled
with peak, off- peak, day, etc. To model the
Figure 4 - Simulation Results When Grid
purchasing cost from the grid,kVA demand
Outage
charge can be also modeled. HOMER
considers all the tariff charges and demand Total net percent cost is the main parameter
charges for a billing period. HOMER can also which represent the system life cycle cost. By
model net metering systems with the one considering the capital, maintenance, running
month billing period. (fuel), replacement cost and salvage values of
each and every component in the micro power
There is no direct method to model the grid system, net percent cost is calculated in
outage in HOMER. HOMER generator will not HOMER.
be running unless the cost of grid purchasing
power unit exceeds the cost of generator Salvage value for each component is calculated
power. Thus a generator to be run in the by Equation 5 [1]. This is the component value
system grid power unit cost should be higher at the end of the project life time.
than generator power unit cost. In order to
model the grid outage above mentioned …..(5)
feature in HOMER was used. By making the
generator to be forced ON and introducing where S is the salvage value, Crep the
higher rate for grid power as compared with replacement cost of the component, Rrem the
cost of generator power during that period, remaining life of the component, and Rcomp the
0grid outage has been modeled. lifetime of the component.
In order to figure out the grid outage HOMER uses the total Net Present Cost (NPC)
percentage and its occurrence two studies are as its primary economic figure of merit and
done for real situations; standby generator calculates using Equation 6 [1].
operation at Geological Survey &Mines Bureau
–Pitakotte and at “Suhurupaya”, ( ) …..(6)
Baththaramulla. Based on above real data grid
outage is assumed as 1%. Where Cann;totis the total annualized cost, i the
annual real interest rate (the discountrate), Rproj
Based on this assumption, generator and grid the project lifetime, and CRF() is the capital
inputs are adjusted. New tariff is introduced as recovery factor, given by Equation 7[1].
„grid off‟ and generator is forced „ON‟ during
that that period. No energy will be purchased …..(7)
108
Where i is the annual real interest rate and N is tariff category, load curve (cluster type) and
the number of years. HOMER uses Equation 8 maximum load demand.
[1] to calculate the levelized cost of energy:
For a single case there is an optimum energy
…..(8) combination. But this optimum point may not
be achievable due to the practical limitations
such as area limitation, investment cost etc.
where Cann;tot is the total annualized cost, Eprim
Therefore other feasible combinations are also
and Edef are the total amounts of
considered to develop this database tool.
primary and deferrable load, respectively, that
the system serves per year, and Egrid,sales is the This tool gives the optimum net meter solar PV
amount of energy sold to the grid per year. capacity with minimum cost of energy and
other feasible combinations with payback
The average cost per kilowatt-hour of electrical
period. Graphical output shows how the
energy produced by the energy system is
system parameters such as CoE, payback
called as levelized cost of energy (CoE).
period, area requirement, solar PV system cost
3.3 Comparison of the Simulation Results and renewable energy percentage varies with
For the simulation, some parameters such as feasible PV combinations. And also this can be
solar irradiance level, economic parameters, used to compare other feasible systems with
PV cost, Inverter cost, Generator cost are the optimum PV capacity.
considered as constant for all analysis. But the
Always the best energy combination is the one
tariff category, maximum load demand and
which gives minimum CoE. User can select
load curve is changing accordingly.
other feasible solar PV systems as well but
Figure 5 shows, the variation of the optimum with the net metering concept, solar PV
PV system capacity with maximum load installation beyond the optimum capacity is
demand and cluster type in same tariff economically less feasible.
category (GP2). It clearly shows that the
4.1 Tool Development
optimum PV capacity is highly consumer
Figure 6 illustrates the algorithm used for the
based. Thus every building should be analyzed
tool development. Selected results from
separately to decide the optimum PV capacity.
HOMER simulation such as PV system cost,
4 Analysis of the Tool capital cost, replacement cost, O &M cost, cost
of energy, renewable fraction, simple payback
Development period, discounted payback period etc., are
extracted and stored in an Excel based
HOMER simulation has done for 114 cases by database.
changing the parameters of base model such as
Figure 5 – Optimum PV capacity for different clusters with varying maximum demand
109
Microsoft visual basic in excel is used for
programming. HOMER results are compared
to select best suitable energy combination
based on system cost and area limitations.
Those comparison shows how the CoE and
renewable energy fraction varies with solar PV
capacity, how the PV system cost and the area
requirement varies with solar PV capacity and
how the simple and discounted payback
Figure 7 – Input Selection Window
periods are varies with solar PV capacity.
If the selectable inputs are possible then the
tool will display all possible feasible PV
systems and its techno economic details with
comparison graphs. Otherwise it give an error
message which says “NOT REAL. Check your
input parameters as the given table in front
page”.
5 Case Studies
To verify the developed tool, two actual
buildings considered. This tool gives techno
economical details on energy system which is
much similar to actual situation. This verifies
that the user can use this tool to get correct
idea on their building energy system and can
be used to decide on future net meter solar PV
system.
Figure 6 - Tool development algorithm 5.1 Study 1
4.2 Data Retrieval from the Tool As the Study 1, “Food City Nawala”, No 264,
Nawala Road, Rajagiriya Net meter PV system
Microsoft visual basic in excel is used for
is selected.60 kW net meter PV system has
programming to retrieve the relevant data
been installed in 2014 September at cost of 11.8
from the database. Front page of the tool
million rupees. 396 m2 area was utilized to
contains all necessary instruction to user for
install 60 kW solar system. The PV system was
data obtaining from this tool. There is a table
expected 6900 kWh average PV units monthly.
which describe the cluster with example
The applicable tariff category is GP-2. Since
installations and the applicable tariff category.
this is a supermarket (consumer goods) cluster
By pressing (Ctrl+D), a window will popup 14pattern is considered as load profile.
which contains 3 selectable inputs. User can
According to the Electricity bill (Acc:
select these inputs; tariff category, average
0103133008) at Cargills food city Nawala, their
maximum demand and load profile (cluster
maximum demand is around 75 kVA and they
type) based on their building with the details
are under GP2 tariff category.
in front page. By pressing the “Build” button,
the tool will give the relevant details to the According to the tool output for CoE which is
user required building. By pressing the shown in figure 8, optimum solar PV capacity
“Clear” button, user can delete his which minimize the CoE is 80 kW. But they
requirement and select parameters again and have installed 60 kW due to the area
By pressing the “Exit” button, user can exit limitation. It is also feasible.
from the window. Figure 7 shows the input
window with selectable inputs.
110
electricity under the tariff category GP2.
Maximum demand is considered as 175 kVA.
Figure 9 shows tool output for the CoE which

contain optimum PV capacity and other
feasible system capacities as well. Optimum
solar PV capacity which minimize the CoE is
180 kW. But they have installed 55 kW. Table 3
compares actual data and simulation results at
GSMB on PV system cost, CoE, renewable,
average area requirement and simple payback
period for PV system which again validate the
Figure 8 - Tool Output for CoE Period for
developed tool.
Study1
Table 2 - Comparison of Actual Data and

Tool Output for Study 1
Parameter Actual Tool Results

data
PV system 11,800,000 10,037,625
cost ( Rs.) (in 2014)
Cost of 24.15 27.45
energy (Rs.) Figure 9 - Tool Output for CoE Period for
Renewable 39 % 42% Study 2
Fraction
Table 3 - Comparison of Actual Data and
Area 396 408
Tool Output for Study 2
requirement
(m2) Parameter Actual Tool
Simple 6.53 5.03 data Results
payback (If project
PV system cost 9,350,000 9,224,465
period for PV cost is
(Rs.)
system (years) 10,037,625
Cost of energy 24.48 27.26
this
(Rs.)
become
Renewable 28 % 28 %
5.56)
Fraction
Average area 363 374
Table 2 compares actual data which are
requirement (m2)
calculated using actual system performance
data and simulation results on PV system cost, Simple payback 5.81 4.65
CoE, renewable, average area requirement and period for PV
simple payback period for PV system which system (years)
validate the tool developed.
5.2 Study 2 6 Conclusion and Discussion

Solar PV installation at Geological survey &
Mines bureau (GSMB) – Pitakotte is selected as This research was done to analyze the
study 2. 25 kW net meter PV system was feasibility of incorporating solar net metering
installed since 2014 investing 4.2 million for commercial building in Sri Lanka.
rupees and expected monthly solar PV
generation was 2,750 kW. Then they have Optimum solar PV capacity to be installed is
added another two 15 kW systems in 2016 at a depend on tariff category, solar irradiation,
cost of 5.15 million rupees with 3400 kWh technical performance of the system, building
units monthly production. Expected total PV load profile and maximum demand of the
generation is 6150 kWh per month. This is a building. But there was no user friendly
government office building which has the method to analyze the feasibility of PV
cluster 2 load pattern. They are paying for installation and its outcomes for commercial
buildings in Sri Lanka. This paper present a
111
tool to analyze solar PV incorporating at the retail type domestic, commercial and hotel
conceptual design stage of the project. electrical installations (electrical installations
Variation of technical performance of PV where its contract demand is below 42 kVA).
systems and solar irradiation is not considered
in this tool. References
 User can select optimum solar PV
1. Lambert, T., Gilman, P. and Lilienthal, P. (2005)
capacity to be installed and get details Micropower System Modeling with Homer, in
on investment cost, cost of energy, Integration of Alternative Sources of Energy
payback period etc. (eds F. A. Farret and M. G. Simões), John Wiley
 If the user cannot go for the optimum & Sons, Inc., Hoboken, NJ, USA.
PV capacity due to the limitations such doi: 10.1002/0471755621.ch15.
as cost or area, then he can select
whatever the possible capacity using 2. Sunanda Sinha, S.S. Chandel, Review
this tool. It can be used to decide ofsoftware tools for hybrid renewable energy
whether to install solar and if installed systems, Renewable and Sustainable Energy
Reviews, Volume 32, April 2014, pp. 192-205,
what the payback period is.
ISSN 1364-0321.
 This tool can be used even with the
conceptual design stage of the 3. Patrick Mark Murphy, SsennogaTwaha, Inês S.
building with estimated electrical Murphy, Analysis of the cost of reliable
demand. That will help to utilize the electricity: A new method for analyzing grid
area for solar panels and the project connected solar, diesel and hybrid distributed
cost. electricity systems considering an unreliable
 When consider the government sector electric grid, with examples in Uganda, Energy,
projects, this tool will help to do Volume 66, 1 March 2014, pp. 523-534, ISSN
0360-5442.
proper budgetary allocations for net
meter solar PV system at the initial
4. Makbul, A. M. Ramli, AyongHiendro, Khaled
stage. This will avoid price variations. Sedraoui, SsennogaTwaha, Optimal sizing of
 Any user without much knowledge on grid-connected photovoltaic energy system in
net meter solar PV installations, can Saudi Arabia, Renewable Energy, Volume 75,
see the feasibility of it without having March 2015, pp. 489-495, ISSN 0960-1481.
consultancy from solar panel
providers. 5. Kyoung-Ho Lee, Dong-Won Lee, Nam-
ChoonBaek, Hyeok-Min Kwon, Chang-Jun Lee,
In this research only the net meter solar PV Preliminary Determination of Optimal Size for
Renewable Energy Resources in Buildings
installation is considered. In 2016 government
using RETScreen, Energy, Volume 47, Issue 1,
introduced a concept called “Battle for solar November 2012, pp. 83-96, ISSN 0360-5442
energy” which has three options; net metering,
net accounting and net plus. In net plus, the 6. ShafiqurRehman, Luai M. Al-Hadhrami, Study
electricity production and the consumption are of a Solar PV–Diesel–Battery Hybrid Power
independent. So no need to optimize the System for a Remotely Located Population near
capacity. But this can be included net Rafha, Saudi Arabia, Energy, Volume 35, Issue
accounting concept using the same simulation 12, December 2010, pp. 4986-4995, ISSN 0360-
software HOMER. Further this tool can be 5442.
developed to decide the most feasible solar
installation method; net metering, net 7. Tamer Khatib, A. Mohamed, K. Sopian, M.
accounting and net plus. Mahmoud, Optimal Sizing of Building
Integrated Hybrid PV/Diesel Generator System
for Zero Load Rejection for Malaysia, Energy
Sensitivity analysis are not included in this
and Buildings, volume 43, issue 12, December
tool. But this tool can be further developed by 2011, pp. 3430-3435.
adding sensitivity analysis for solar
irradiation, tariff scheme changes, technical 8. Graham, V. A., and Hollands, K. G. T., A
performance of PV panels etc. Method to Generate Synthetic Hourly Solar
Radiation Globally, Solar Energy, Vol. 44, No.
The results of this research is only applicable 6, pp. 333–341, 1990.
for bulk installations. This is not applicable for
domestic sector electrical installations. Also 9. Mukhtaruddin, R. N. S. R.; Rahman, H. A.;
this is not applicable for multi category Hassan, M. Y., "Economic Analysis of Grid-
electrical installations which are consisted as Connected Hybrid Photovoltaic-Wind System
112
in Malaysia", Clean Electrical Power (ICCEP),
2013 International Conference on June 2013,
vol. 11, no.13, pp.577-583.
10. Moniruzzaman, M., Hasan, S., "Cost Analysis

of PV/Wind/diesel/grid connected hybrid
systems," Informatics,
Electronics & Vision (ICIEV), 2012 International
Conference on May 2012, vol.18, no.19, pp.727-
730.
11. Kolhe, M., Ranaweera, K. M. I. U.,

Gunawardana, A. G. B. S., "Techno-Economic
Optimum Sizing of Hybrid Renewable Energy
System," Industrial Electronics Society, IECON
2013 - 39th Annual Conference of the IEEE ,
vol.10, no.13, pp.1898-1903, Nov. 2013.
12. Rashayi, E., Chikuni, E., "The Potential of Grid

Connected Photovoltaic Arrays in
Zimbabwe," Electrotechnical Conference
(MELECON), 2012 16th IEEE Mediterranean,
vol.25, no.28, pp.285-288, March 2012.
13. www.opac.lib.mrt.ac.lk
Methodology to determine the maximum
demand of multi category bulk electrical
installations - [2015].
14. http://www.pucsl.gov.lk/english/wp-
content/uploads/2016/03/Net-Metering-
Report-Final1.pdfVisited, 20th June 2016.
15. www.energy.gov.lkVisited, 20th June 2016.
16. www.ceb.lkVisited, 20th June 2016.
113
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[115 120],2017
2017
© The
© The Institution
Lanka
Multi-Functional Floor Cleaning Robot for Domestic

Environment
H.G.T.Milinda and B.G.D.A. Madhusanka
Abstract: Floor cleaning is a major task in the household cleaning process. But nowadays people
are very busy with their works they do not have enough time to spend cleaning process. Multi-
functional floor cleaning robot was introduced to the world as a solution for above reason. The
proposed method is capable to identify and classify the dirt in the domestic environment. According
to dirt detected by the floor, the robot can decide and apply most suitable cleaning action among
sweeping, mopping and vacuuming. Dirt detection is based on vision system. The vision system is
introduced a novel method of mud detection and proposed vision algorithm is used to separate mud
and dirt elements. According to this vision base decision robot select cleaning method. The robot can
be identified more than 70% of dirt and mud elements in the domestic environment. Many Sensors
around the main body are used to avoid collision with the floor obstacles or the fall from the stairs and
sharp edges. The automatic floor cleaning action of dynamic mechanism and dust removal equipment
is controlled by both electric circuit and single-chip microprocessor. For the operation of functionality
and practicality, an Android-based application is an added to the robot with three models, including
manual operation, automatic operation, and cleaning operation. Our findings suggest that average
save more than two hours of people’s time per week, floors are cleaner leading to better homes and
lives, systematic cleaning is seen as an important feature, and modifications to the domestic
environment to support the navigation of the robot are largely accepted.
Keywords: Multi-functional cleaning, Dirt detection, Mud identification, floor cleaning robot
1. Introduction with more efficient cleaning using artificial

intelligent technology and image processing
A mobile robot is an automatic machine which which able to take its own decision.
is capable of locomotion and has the capability In today’s market, most of the robot having its
to move around in their environment. The own charging and control from any part of the
mobile robots are capable of navigating an world some robot can control using smart
uncontrolled environment without any physical devices. Some of them can be connected to
or electromechanical guidance. Mobile robots house automation system.
have become more commonplace in Most of the robot in the current market can only
commercial and industrial settings. This kind of perform one task at each time. Roomba [1],
robot is used in many fields and new Neato [2] and LG Hom bot [3] are the best-
researchers also going on several fields as well. selling brands in the robot market. These robots
Domestic robots are consumer products, can only performance sweeping and vacuum
including entertainment robots and those that functions only.Any robot in the market does
perform certain household tasks such as not identify what sort of the dirt it is. They are
vacuuming, cleaning or gardening. always cleaning the floor and it is the waste of
Floor cleaning robot is the domestic robot as energy of robot. No robot on the market neither
well as the mobile robot. So it has tries to clean the floor if it is necessary or all of
characteristics of both. People in modern them.
society very busy with their work and they do
not have time spent on household cleaning Eng. H.G.T. Milinda, BTech(Eng)(Hons)(OUSL),
purposes. Nowadays floor cleaning becomes Student Member of IESL, Department of Mechanical
waste of time by comparing his daily rooting. Engineering, The Open University of Sri Lanka.
Also, modern world manual work is taken over
Eng. B.G.D.A. Madhusanka,
the robot technology and many of the related BTech(Eng)(Hons)(OUSL), MSc(Moratuwa),
robot appliances are being used extensively AMIE(SL), MIEEE, Probational Lecturer, Department of
floor cleaning also takeover by the robot. Mechanincal Engineering, The Open University of Sri
Lanka.
In the cleaning purpose, robots were
introduced for making the efficient cleaning
pattern. The present floor cleaning robot comes
1
115
2. Proposed System
Floor cleaning robot available only can Flow cleaning Robot
identified dirt but dirt is not specifically
separate as mud spot or dirt element. . Reason
for that is no robot able to identify the dirt and
separately so it cannot perform correct cleaning Sweeping Mopping Vacuuming
action for floor. All the prevailing robot used
basically used three method to identified dirt
• Ultrasonic scan Sweeping Using micro
function fiber pad
• Laser beam scan
integrated
• Image processing techniques with vacuum
But perform the multi-tasking need the dirt Need
specifically identification using the Ultrasonic machanisum
to move pad
and Laser beam it is impossible to separate dirt There has two
sweeping up/down
specifically. Image processing is an only
brushes
solution for detect the dirt.
So the design the multi-tasking robot first robot
should read the collision sensor to check
whether robot can move forward and also take
the image in front of robot. Then identified Counter
Clockwise
Clockwise
objects in the floor. In this research mainly
focused on identified object in the floor whether
it is dirt or mud and clean it.For that purpose Figure 2 - Overview of Cleaning
prevailing three methods only image
The sweeping unit consists of two sweeping
processing techniques can used separate mud.
brushes one rotates clockwise and other rotate
So dirt detect process image processing
counterclockwise. Vacuum unit consists of
techniques are going to be used
dustbin and vacuuming fan this unit is
integrated with the sweeping unit as given in
figure 3(b). Mopping unit has consisted of
microfibre cloth and water tank as in figure
3(a). From the Moping unit and Sweeping unit
only one unit work at the time.
Figure 1- Proposed model diagram
Firstly objects are identified from image using (a) (b)

template matching [4]. Then identified objects Figure 3-(a) Proposed mopping unit.
once again sorted whether dirt or mud. For that (b)Proposed sweeping unit
purpose expected to use combination of colour
matching and pattern recognition techniques 3. Methodology
coming under the image processing techniques. Floor cleaning robot to process the above task
should contain the following units shown in
According to identification image, cleaning figure 2. This unit should be designed or
action is automatically decided by the robot selected according to requirement. Mopping
and clean specific floor area after that. Mainly and Suction units going to be design and
robot can perform few cleaning action. sensors and processing and the controlled unit
must be selected.
2.1 Overview of cleaning method
Cleaning System has consisted of three main Image Acquisition unit is used to capture the
unit mopping, vacuuming and sweeping. image of floor and collision avoiding sensors
Sweeping and vacuum unit are work together are used to detect the object in the working
mopping function is operated separately. Dirt environment. Image processing should be done
detection is done using image processing separately because one controller is not suitable
techniques as mention earlier done both image processing and real-time robot
2
116
controlling at the same time. If the same Then the area of the each MSE region is
controller is used for both purposes working calculated and checked with the threshold.
efficiency of controller going to low then it is Mean and standard deviation of each region
not satisfied. The motor driving unit is used also checked. If any object is under both
move robot around the floor. This driver is threshold identified as dirt. Furthermore
used control the motor according to the signal MSERs are detected at different scales. Figuar 5
given by the controller. Gyro and shows the MSER detection of floor image.
Accelerometer sensors are used to tracking the
navigation on the floor. Robot scan collision
sensor first then take an image of ground and
processed the image and return the decision to
the main controller according to cleaning action
is taken and run motors according to that.
Image Image Sweeping

Aquastion Processing and suction (a) (b)
Camera) unit Figure 5-(a) Original dirt and (b) MSER dirt
identified Feature
Control Unit
Collision
Avoiding Mopping
Sensors Unit
4.2 Mud Detection
Mud detection novel method is introduced [8].
The new concept is based on the colour and
Gyro and
Accelometer Motor area distribution of mud. Though research is
sensor Driving done to service robot for the outdoor
Unit
environment by A. Rankin and L. Matthies,
[6]. That method uses two stereo cameras and it
is cannot be implemented for the small robot.
Figure 4 - Block diagram proposed cleaning
MSER function cannot identify mud most of the
robot cases because mud normally having the
property of spread on the floor than the dirt
Read sensor for any kind of collision: - Sonar particles and difficult a stable sharp point in the
sensor is used to detect the collision. Though mud marks .in this novel method first spectral
collision can check using the image it not can residual filter[7] is used identified the object
sudden collision tracking because it takes time from background then apply hue value mask.
to process. So sonar sensor read the collision Resultant image frame apply the morphological
Read the image first and identified dirt using operation to make close regions.then calculated
template machine:-take the image of floor and area, mean and standard deviation finally
process the image template matching and checked with threshold values and detected as
identified dirt then identified dirt separate:-This mud.Figure 6 shows that calculation of
identified dirt especially separate as mud or varicence and identified mud areas.
non-mud dirt using object identification
method using colour matching and pattern
recognition. Then select most suitable action:-
According to dirt speciation activate the
mopping unit or sweeping and vacuum unit.
4. Image processing and dirt

classification
4.1 Dirt Detection
The concept of MSERs (Maximally Stable
Extremal Regions) was proposed by Baek [5].
MSERs denote a set of distinguished regions
that are detected in a grey scale image. All of (a) (b)
these regions are defined by an extremal Figure 6-(a) Variance Calculation each region
property of the intensity function in the region (b) Mud detection according to threshold
and on its outer boundary. First identified MSE
regions and apply several morphological filters.
3
117
5. Prototype implementation and 5.3 Mud Detection on floor experiment
testing The experiments are conducted same
A Real-time working module of the robot was environments where the dirt detection
implemented and tested proposed algorithm experiment are done. One square feet area of
can identify and dirt and mud separately and tile floor is selected. The footprint of the man
cleaning it according to dirt detected. Real time with mud is used as mud element. There 7cm
working module of robot is shown in Figure 7. ×10 cm area of mud sign is used. From this
mud sign, several areas are detected as the mud
5.1 Analysis of dirt detection experiments by the algorithm. Several kinds of mud sign is
Several experiment is recoded at domestic also used on the same floor and tested.figuare 9
environments with a variety of dirt particles of shows some experiment cases.
different sizes in different floor to check
whether proposed algorithm can identified
mud spot and dirt separately.
Figure 7 - Woriking module of floor cleaning

robot.
5.2 Dirt Detection on floor experiment

These experiments were done several
household floors like tile, red cement, wood
and carpet. In this experiment 1| ×1| (one
square feet) area of tile floor is selected. Figure 9 - Example images from house
Newspaper particles with the different colour environment surfaces. Left column: original
are used as dirt particles which are having the mud spot images I. Right column: detected
area less than 25cm2. From these dirt element mud spot indicated by yellow rectangles
number of pieces are detected as the dirt by the
algorithm. Figure 8 shows that several images 5.4 Floor sweeping and vacuuming
of experiment results . Experiment
Sweeping and vacuuming action is done
several types of floors check whether sweeping
is successful done in each floor. Area of the 1|
×1| (one square feet) is selected as tested area
figure 10 (a) showed that dusty area. After
cleaning by the robot floor shown as figure
10(b).
(a) (b)
Figure 10 - Experiental result of sweeping

Figure 8 - Example images from house fuction
environment surfaces. Left column: original
images , Right column: detected dirt indicated
by yellow rectangles
4
118
5.5 Floor mopping Experiment Detection of Dirt
Mopping function is checked on tile floor. Area 90%
of the one square feet is selected as tested area 85%
figure 5. Mopping function is checked on the 80%
tile floor. Area of the one square feet is selected
75%
as tested area figure 5.17 (a) showed that mud
70%
spot. After cleaning from the robot floor shown
Floor tiles Red Cement Wood Carpet floor
in figure 5.17(b).In there robot mopping one floor surface
time routing its mopping unit in this area. All Detection of Dirt
most all of the mud spot is cleaned by the robot.
Figure 11- Graphical representation of mud
detection result
6. Results & Discussion
In these experiments try to identify the 6.2 Analysis of mud detection experiments
foreground object of the floor many Several experiments are recorded at above
experiments have been done this experiment environments with a variety of mud particles of
values given most accurate values at the different sizes. Table 2 summarizes the sets of
present. This algorithm good enough to surfaces in home environments, provides
identified 70% of the solid Dirt in the floor. percentages of mud spot detected on them and
comments on typical problems with mud
6.1 Analysis of mud detection experiments detection. Please notice that the detection results
Table 1 summarizes the sets of surfaces in home come with a low false positive rate of a few
environments, provides percentages of dirt percent. The reported percentages are computed
detected on them and comments on typical as the area of found mud spots divided by the
problems with dirt detection. Please notice that total area of present mud spots. The results were
the detection results come with a low false counted manually since the labelling of the
positive rate of a few percent. The reported streams would not have been most accurate. A
percentages are computed as the number of mud spot was considered to be found by the
found dirt spots divided by the number of algorithm if it was detected in several frames.
present dirt spots. The results were counted
manually since the labelling of the streams Table 2 - Summary of the mud identification
would have been too tedious. A dirt spot was result
considered to be found by the algorithm if it Detection of
was detected in several frames. Surface Mud(Appro Remarks
ximately)
Table 1 - Summary of the Image processing Reflection and
result Floor tiles 80%
edges with mud
Red Dark colour
Cement 75% difficult to
Detection
Surface Remarks floor identified mud
of Dirt
Most similar to
Floor Most of the dirt Wood
82.4% 70% mud colour so
tiles partials are bigger surface
difficult detect
Red
very dark and small Design in the
Cement 85.2% Carpet
black objects 79% carpet made
floor floor
some errors
Wood
77.6% fine-grained dirt
surface
In there robot sweep two times routing its
Carpet some false positives
75.3% sweeping brushes and vacuuming unit in this
floor in reflection of light
area. More than 98% of the dust is swapped and
Red cement floor get the most accurate result in
vacuumed by the robot. Mopping function is
the dirt particles detection carpet floor get
checked using the robot floor In there robot
minimum accurate results because algorithm
mopping one time routing its mopping unit in
compare MSER values it is difficult done in
mud area. All most all of the mud spot is cleaned
carpet floor because. Many experiments are
by the robot.
done in these floors average time values is
included into report. Many points is detected as
sharp edges in the carpet.
5
119
3. Shanavan, "LG Hom-Bot robotic vacuum -
Detection of Mud(Approximately) ocean black (VR65704LVM) | LG US,"
2017.[Online]. Available:
85% http://www.lg.com/us/vacuum-cleaners/lg-
80% VR65704LVM. Accessed: Feb. 21, 2017.
75%
70% 4. V. S and S. A, "Template Matching Technique
65% for Searching Words in Document
Floor tiles Red Cement Wood Carpet floor Images", International Journal on Cybernetics
floor surface & Informatics, vol. 4, no. 6, pp. 25-35, 2015.
Detection of Mud(Approximately) 5. I. Baek, "A Method to Detect Object of

Interest from Satellite Imagery based on
MSER(Maximally Stable Extremal Regions)",
Figure 12 - Graphical representation of mud Journal of the Korea Institute of Military
detection result Science and Technology, vol. 18, no. 5, pp.
510-516, 2015.
6. A. Rankin and L. Matthies, "Passive sensor

7. Conclusion evaluation for unmanned ground vehicle mud
The implemented low cost solution consists detection", Journal of Field Robotics, vol. 27,
with Raspberry PI cluster, load balancer and no. 4, pp. 473-490, 2010.
intelligent automated management platform. It
replaces the current manual process of physical 7. X. Hou and L. Zhang, "Saliency Detection: A
node IaaS configuration and management task Spectral Residual Approach", 2007 IEEE
of network administrator. Conference on Computer Vision and Pattern
Intelligent management controls the roles of Recognition, 2007.
worker nodes and changes worker roles
8. H. Milinda and B. Madhusanka, "Mud and
according to the available demand. Therefore
dirt separation method for floor cleaning
cloud can adapt with request types dynamically robot", 2017 Moratuwa Engineering Research
without interaction of the user. Conference (MERCon), 2017.
Dynamically expandable nature provides more
flexibility to cloud administrator in expansions 9. [Online].Available:
and management by reducing server down https://en.wikipedia.org/wiki/Kalman_filter
time and service interruptions. The floor . Accessed: Feb. 21, 2017.
cleaning robot is mainly focused to dirt .
identification part only. The robot navigation
part does not cover in this research. In next
stage of this research is develop the vision base
auto navigation algorithm. Also the robot
should be done auto charging when its battery
needed to be recharged. Virtual fence for floor
cleaning robot is to be suggested when cleaning
area. Advanced dirt identification algorithm is
to be developed which is not only for the
domestic environment but also adaptive to any
environment scenarios with decision taking
capabilities.
References
1. C. W. 2009, "How do Roomba robot vacuum

cleaners work?," Explain that Stuff, 2016.
[Online]. Available:
http://www.explainthatstuff.com/how-
roomba-works.html. Accessed: Feb. 21, 2017.
2. N. Robotics, "Neato robotics," in Wikipedia,

Wikimedia Foundation, 2017. [Online].
Available:
https://en.wikipedia.org/wiki/Neato_Roboti
cs. Accessed: Feb. 21, 2017.
6
120
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[121 124],2017
2017
© The
© The Institution
Lanka
Development of CZTS thin Film from Spin Coating for

Solar Cell Application
H.A.D.I. Dilshan, W.D.D. Weerasinghe, N.L.W. Wijewantha, W.P.S. Wijesinghe
and D. Atygalle
Abstract: Copper Zinc Tin Sulphide (CZTS) layer fabrication was done using the spin coating
method and optimum conditions for CZTS layer fabrication were found to be 2000rpm – 6000rpm spin
coating speed with 30 seconds of spinning time, 38 °C – 42°C precursor solution temperature, 140°C –
160°C drying temperature, and 280°C annealing temperature (30 mins) that promotes the formation of
CZTS. The obtained CZTS layers have shown a bandgap in the range of 1.45eV to 1.55eV within which
the theoretical bandgap 1.50eV of CZTS lies.
Keywords: Bandgap, Copper Zinc Tin Sulphide (CZTS), Photovoltaic, Spin Coating
1. Introduction conversion efficiency for single junction CZTS

solar cells are more than 30% [4].
The main objective of manufacturing thin film
A vast variety of CZTS fabrication methods
solar cells is to reduce photovoltaic electricity
have being studied by different researchers in
generation costs while maintaining the
the past decade. Each of these methods has its
conversion efficiencies comparable to first
advantages and drawbacks. Some are in the
generation bulk material solar cells. Thin film
high technological set category, which are also
solar cells have anabsorber layer of thickness
costly. Pulsed laser techniques, co- evaporation,
less than few micrometres, which is several
sputtering are some of them. Other than them,
orders magnitude thinner compared to
it is possible to identify, low-cost methods like
hundreds of micrometre thick siliconwafer
spray pyrolysis, spin coating and electro
based solar cells. In other words, thin film solar
deposition, which also reportedly produced
cells use much less material and thereby saves
encouraging results [5]. In our research, we
energy and money.
used spin coating method as our deposition
At present, the efficiencies of thin film solar method. It has several advantages over other
cells are comparable to that of polycrystalline low-cost deposition methods. It is a simple and
silicon; for instance in terms of efficiency, 21.3% cost effective method relative to other methods.
for multi-crystalline Si and 22.3% for copper Thin and uniform coating can be achieved
indium (gallium) diselenide (CIGS) [1]. Though within few seconds and coating will dry very
the Si polycrystalline solar cells dominate the quickly after deposition due to high airflow
market, the market share of the thin films during high speed rotating [6].
photovoltaic module is increasing [2]. Thin film
solar cells based on polycrystalline cadmium
2. Experimental Procedure
2.1 Preparation of CZTS Precursor solution
telluride (CdTe), copper indium diselenide
(CIS), and copper indium (gallium) diselenide The CZTS precursor solutions were prepared
(CIGS) have reached a commercialization stage; by mixing Copper(II) Chloride Dihydrate
however, restrictions on heavy metal usage, (98%), Tin(II) Chloride Dihydrate (98%),
such as Cd and scarcity of supply for In and Te Zinc(II) Chloride (99%), Thiourea (98%), 2-
have raised concerns on sustainability of certain Methoxy Ethanol (99%) and de ionized water
PV technologies. In this circumstance, Copper [2]. CuCl2.2H2O [0.9 M], ZnCl2 [0.7 M],
Zinc Tin Sulphide (CZTS) quaternary SnCl2.2H2O [0.5 M], and Thiourea [4 M] were
semiconductor compound has identified and used for thin films. All the dry chemicals were
promoted as one of the promising candidates.
H.A.D.I Dilshan, B.Sc.Eng.(Moratuwa),
W.D.D. Weerasinghe, B.Sc.Eng.(Moratuwa),
CZTS is nontoxic, and elements that are Eng. N.L.W. Wijewantha, B.Sc.Eng.(Moratuwa),
contained in CZTS are abundant in the earth AMIE(SL).
crust. CZTS has an absorption coefficient of 10 4 Eng. W.P.S. Wijesinghe, B.Sc.Eng.(Moratuwa),
AMIE(SL).
cm-1, and the band gap is around 1.5 eV, which
Dr. D. Atygalle, B.Sc.(Sp) (Colombo), Ph.D.(Toledo),
makes it a very promising absorber layer for Senior Lecturer, Department of Material Science &
thin-film PV [3]. Theoretically, the predicted Engineering, University of Moratuwa.
1
121 ENGINEER
measured according to the required (Infinity1-MEIJI) to determine if the
concentrations and mixed with the de ionized microstructure is uniform and has a good gain
water and 20 ml of 2-Methoxy Ethanol for a growth. The film thickness of the samples was
100ml of the solution. The solution was stirred determine by microscopic observation of the
for 1 hour at 40-450C until the metal salts were cross section of the glass slide coated with
dissolved. Crystal clear yellow coloured CZTS layer.
solution was obtained Dual Beam Shimadzu UV-1800
Spectrophotometer was used to collect the
2.2 Thin film fabrication transmittance data of different CZTS thin films
fabricated on glass slides under predetermined
Thin films were fabricated by the spin-coating conditions. The wave length range used was
method, which involved two steps: first, spin- 300nm to 1, 100nm, and the collected data was
coating precursor solution on the glass analysed to obtain absorbance bandgap of the
substrate to forma film; and second, CZTS thin films. Since the accuracy of
annealing thin films in a proper atmosphere calculated band gap value heavily dependent
[7]. on the film thickness, a Carl ZeissEvo 150
The experiments were carried out under the scanning electron microscope (SEM) was used
coating conditions, spin coating speed ranges to verify the optical microscopy measurements.
from 2000rpm – 12000rpm and amount of CZTS The film composition was determined using
ranges from 0.5ml to 1.5ml, to determine Energy Dispersive X-Ray Spectroscopy (EDS).
optimum fabrication conditions because, Electrical measurements of the CZTS thin films
according to literature, spin coating speed was were obtained to evaluate if the layer possesses
a determinant factor when it comes to CZTS photoconductivity. Since the CZTS layer
thin film thickness. Then, the drying was done thickness was in the micron range, the
at 150oC to promote the formation of CZTS measurement of the current being generated
when exposed to light was done, and the dark
Table 1- Regulated conditions for CZTS thin current and light current was measured for
film fabrication and annealing varying voltages.
Speed Amount Spin 3 Results and Discussion

Annealing
Range of CZTS Coating
Temperature 3.1. Microscopic Observations of CZTS
(rpm) (ml) Time (s)
4000 - 8000 1.0 30 310 Thin Films
4000 - 8000 1.0 30 280 Microstructure observationsprove that the
4000 - 8000 1.0 30 250 CZTS thin films fabricated at 40oC show a
2000 - 6000 0.5 30 310 higher level of uniformity compared to other
2000 - 6000 0.5 30 280 temperatures. Figure 1shows that the coating
2000 - 6000 0.5 30 250 was applied with uniformly distributed grain
2000 - 8000 1.0 30 310 structure. Uniformity and the film thickness of
2000 - 8000 1.0 30 280 fabricated CZTS thin films were determined by
2000 - 8000 1.0 30 250 optical micrograph as shown in figure 2.
It was identified that the speed of the spin

coating plays a dominant role in determining
the uniformity and film thickness. Thereafter,
samples were annealed at elevated
temperatures to promote grain growth and
grain refining. Table 1 shows the conditions
that were used in making samples.
2.3 Analysis of Thin Films

d=237.39υm
Prior to getting electrical measurements, thin
films on glass slides were visually examined for
cracks, unevenness and pinholes to eliminate
poor coatings.
Figure 1- Optical Micrograph of CZTS layer
Microscopic observations of the thin films were fabricated on glass slides (2000-6000rpm with
performed using an optical microscope 3 Coatings)
ENGINEER 2
122
3.3. Electrical Measurements of CZTS
Thin Films
According to the Figure 4,a considerable

change of current was observed in dark and
light environments of both samples annealed at
280oC. In the illuminated environment, the
resistance of the film gets reduced with the
generation of electrons due to the interaction of
light photons. In dark environment, it is not
possible due to unavailability of light photons
to generate electrons and holes pairs. Photo
resistive nature is a good proof for the photo
voltaic effect of CZTS material to use in solar
Figure 2 - Micrograph of CZTS thin film cell applications.
fabricated at 40oC with 2000-8000rpm and
0.5ml of CZTS solution
3.2. UV-Vis Spectroscopic Analysis of

CZTS Thin Films
Data obtained from UV-Vis Spectrophotometer

were analysed and obtained the absorbance
curves of different CZTS coatings applied on
glass substrates. In Figure 3, the absorbance
curves derived for CZTS fabricated on glass
slides at 40°C solution temperature for different Figure 4 - Voltage vs. Percentage of
fabrication conditions were plotted in one Photoconductivity graph for CZTS thin film
graph. annealed at 250oC and 280oC
When the bandgap of CZTS thin films was 3.4. SEM and EDS Analysis of CZTS Thin
calculated using the obtained graph (αhυ)2vs Films
(hυ), the bandgap was found to be in the range
of 1.40eV to 1.60eV, and it proves that the From the results obtained, it was evident that
obtained CZTS layer has suitable bandgap for the CZTS composition was not with the
thin-film solar cells. expected atomic ratios. In all the samples, the
Zinc content was relatively higher, and Sn
content was relatively lower while Cu and S
content remained almost constant. This can be
attributed to the fact that the Zn content was
increased by the researchers to hinder the
formation of secondary phases in Cu as
described in the literature.
Figure 5 - EDS graph of CZTS thin film

Figure 3 - Bandgap calculations of CZTS thin annealed at 280oC
films fabricated at varying conditions
3
123 ENGINEER
Table 2 - Composition of CZTS thin film 11% efficiency: Characteristics of state-of-the-art
annealed at 280oC cu 2 ZnSn(S, se) 4 solar cells,” Advanced Energy
Element Weight % Atomic % Net Int. Materials, vol. 3, no. 1, pp. 34–38, Aug. 2012.
SK 26.43 45.65 148.97

4. Shockley, W., and Queisser, H. J., “Detailed
SnL 22.78 10.63 47.42 Balance Limit of Efficiency of p-n Junction Solar
CuK 28.42 24.77 40.75 Cells,” Journal of Applied Physics, vol. 32, no. 3, p.
510, 1961.
ZnK 22.37 18.95 26.31
7. Shi, Z., Attygalle, D., Jayatissa, A.H., "Kesterite-

Based Next Generation High Performance thin
4. Conclusions Film Solar Cell: Current Progress and Future
Prospects" Journal of Materials Science:
Spin coating is an appropriate method to Materials in Electronics, Vol. 2 (28), pp. 2290-
fabricate thin films from CZTS as an absorption 2306, 2016.
layer for thin film solar cells. The
controllablethickness of few micrometres can be 6. “Spin coating: A guide to theory and
achieved for CZTS layer by spin coating techniques.” [Online]. Available:
method. http://www.ossila.com/pages/spin-coating.
[Accessed: 03-Jan-2016].
Thickness can be conveniently controlled via
varying the speed at which the spin coating 7. Todorov, T., Gunawan, O., Chey, S. J., De
was done. Optimum conditions for CZTS layer Monsabert, T. G., Prabhakar, A., and Mitzi, D.
fabrication was 2000rpm-6000rpm spin costing B., “Progress Towards Marketable Earth
Abundant Chalcogenide Solar Cells”. Thin Solid
speed with 30 seconds of fabrication time, 38 °C-
Films, Vol. 519 (21), pp. 7378–7381, 2011.
42°C precursor solution temperature, 140°C-
160°C drying temperature, and 280°C annealing
temperature that promotes the formation of
CZTS.
The obtained CZTS layers have shown a

bandgap in the range of 1.45eV to 1.55eV within
which the theoretical bandgap 1.50eV of CZTS
lies. Annealing temperature control was found
to be critical because, with the higher
temperatures, CZTS thin film tends to be
changed in composition and turning into a
semi-metallic while deteriorating the
semiconductor properties.
CZTS thin film has shown photoconductive

properties and temperature dependent
resistance variation as expected for a
semiconductor material.
References
1. Green, M. A., Emery, K., Hishikawa, Y., Warta,
W. “Solar Cell Efficiency tables (Version 35)”.
Progress in Photovoltaics: Research and
applications, Vol. 18, pp.144–150, 2010.
2. Dr. Gerhard Angerer, Dr. Frank Marscheider,

Weideman: “RohstoffefürZukunftstechnologien”
[English Translation]. FraunhoferInstitutfür
System- und Innovationsforschung ISI. Report
published on 15, May, 2009.
3. Todorov, T. K., Tang, J., Bag, S., Gunawan, O.,

Gokmen, T., Zhu, Y., and Mitzi, D. B., “Beyond
ENGINEER 4
124
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[125 133],2017
2017
© The
© The Institution
Lanka
Characterization of Fuel Briquettes made for

Household Cooking from Sawdust and Rice Husk
M.Y.A.R. Ahmed, U.M.A. Devaraja and K.A.R. Mihiranga and
B.A.J.K. Premachandra
Abstract: Fossil fuels are non-renewable source of energy which is commonly used in many
applications. Liquid Petroleum Gas (LPG) is a common derivative of Fossil fuels which is widely used
for cooking purposes in many households around the world. Alternative sources of energy plays a
wider role in many rural households that do not have access to electricity or LPG distribution.
Biomass is one of the most commonly used alternative sources of energy in the world for household
cooking, and two of the most important reasons for this popularity are low cost and being a solution
for waste management. Sawdust and rice husk are abundantly available biomass types in Sri Lanka.
About 112,000 MT of sawdust is produced annually, around 3% is used for energy purposes. Annual
production of rice husk is estimated around 700,000 MT and around 54% is used for energy purposes.
Rest of these are dumped in to the environment or burnt. Since still most of the Sri Lankan houses use
traditional fire wood stoves, using these wastes as a fuel for household cooking in an efficient way is a
good option for waste disposal problem. Briquetting is the most commonly used technique to use
sawdust as a fuel in an efficient manner. The research was carried out to produce and characterize
sawdust and rice husk mixed fuel briquettes using a Mount press machine at a pressure of 5000 psi.
Physical and mechanical properties of these fuel briquettes such as gross calorific value, moisture
content, ash content, volatile matters, density, compressive strength, shatter index, fixed carbon etc.
were tested according to ASTM standards. We mainly focused on calorific value variations of the
samples and other mechanical properties. After checking all these parameters it is found that sawdust
and rice husk mixture is suitable to use as fuel briquettes even though there is a slight reduction in
calorific value when compared to briquettes made only with sawdust.
Keywords: Biomass, Sawdust, Rice husk, Briquettes, Fossil Fuel
1. Introduction softwood like Pine and Albizia. When

Fossil fuel is the most commonly used energy comparing the sawdust production, hard wood
source for industrial and domestic purposes. usually produces more sawdust than soft
Generally fossil fuel is categorized as oil, gas wood[2]. Since Moratuwa timber industry
and coal. With the growing population of the mostly use hardwood, sawdust production is
world, energy demand has increased very high. According to a previous research[3]
drastically. According to Ecotricity in UK, fossil done in 2005 there are about 4000 sawmills in
fuels are being depleted very fast and it is Sri Lanka. Since the industry is developing this
expected that fossil fuels will be remaining only number should be higher than this now.
until 2088[1]. Therefore sawdust is an abundantly available
solid waste in Sri Lanka. Among these sawmills
Since fossil fuel is depleting drastically, there are around 400 sawmills in Moratuwa. It
alternative energy has gained lot more interest is found in a research done in 2002, the annual
to replace fossil fuel. They are widely available production of sawdust in country is about
and environment friendly also they cause little 112,000 MT[4] and in Moratuwa around 1800
or almost no pollution. Solar, wind, geothermal, MT of sawdust is generated per month. Almost
hydroelectric, ocean and biomass can be listed only 2% of this sawdust production in
as most widely used alternative energy sources. Moratuwa is used for energy purposes and also
Among these biomass is one of the most since there is no proper way to dispose them in
important alternative energy sources. an environmental friendly way the remaining
98% is burnt or dumped in to the sea or
In Sri Lanka, the city Moratuwa is producing river[3]. Therefore Moratuwa is under serious
highest amount of furniture. Timber can be environmental threat and need a proper waste
categorized as hardwood and softwood. Most management system.
of the wood used in Moratuwa such as Teak, M.Y.A.R. Ahmed, B.Sc. Eng. (Hons)(Moratuwa),
U.M.A. Devaraja, B.Sc. Eng. (Hons)(Moratuwa),
Ebony, Halmilla, Jak, Kumbuk, Nedun,
K.A.R. Mihiranga, B.Sc. Eng. (Hons)(Moratuwa),
Satinwood, Suriya Mara and Mahogani are Prof. B.A.J.K. Premachandra, B.Sc. (Colombo), M.Sc. (Sri
hard wood and there few saw mills that use J’pura), Ph.D. ( Cincinnati, USA).
1
125
Several researches are conducted to find the 147.15kN to 588.16kN. Since the reduction in
availability of getting a commercial value from transportation cost, increment in durability
sawdust[2],[5],[6]. Most of these researches are transformation of primary biomass into solid
conducted to find the availability of using compact fuels, such as pellets and briquettes, is
sawdust as an efficient energy source. But there very common in Europe and around the world
are some researches done to find the possibility [7]. Several factors are there which affects
of using sawdust as bricks with some other compacting biomass, such as raw material type,
secondary raw material[2]. Almost all the moisture content, size of sawdust particles,
sawdust used for energy purposes are used for aspects related to manufacturing process
boilers in factories, but not in an efficient way. (pressure applied, temperature reaching during
Manufacturing fuel briquettes using sawdust the compaction). Hardness of briquettes
and rice husk can a viable option for waste increases moderately as the particle size
management and the rising energy demand in decreases. When the particle size is large,
Sri Lanka. briquettes tend to fracture easily[11].
2. Literature Review Rice husk is a waste of agricultural industry

and since agriculture is one of the major
Biomass is organic matter derived from living,
industries in Sri Lanka rice husk is abundantly
or recently living organisms and it can be used
available. According to [3]there were about
as a source of energy. Compared to fossil fuel,
7000 rice mills in Sri Lanka in 2005 and the
biomass is a sustainable energy feedstock and
annual production of saw dust is about 2 MT.
supports the reduction of greenhouse gas
This number might be nearly the same in 2016.
emission and low impact on climate[7]. Also
According to the same research, among these
biomass is highly available in the world and it
2MT of rice husk about 54% is used for energy
is cost effective. As well as biomass is
purposes, 7% of it is used for other purposes
considered as an ecofriendly energy source
and 39% is not used. With the increasing
because of its low sulfur content, low nitrogen
interest for sustainability these values might be
content and low heavy metal content compared
different now and the not used percentage
to fossil fuel. When it comes to Sri Lankan
might be less now. But this is a serious
context more than 80% of the biomass is
environmental threat and need a solution very
consumed for household cooking and heating
quickly. Several researches are done around the
purposes[3]. Fire wood is a type of biomass and
world to find the availability of using rice husk
it is an important backup fuel for countries like
as a fuel in an efficient way.
Sri Lanka as it is cheaper and abundantly
Density of rice husk in Sri Lanka varies
available. According to Sri Lanka Sustainable
between 90-110 kg/m3 and moisture content is
Energy Authority, 77% of the fire wood
about 8.5-12% lower heating value varies
consumption in Sri Lanka is for household
between 12.5-19 MJ/kg. One of the most
energy requirement[8].
important characteristic of rice husk is that it
has high volatile matters (68%). Rice husk has
When sawdust is going to be used as an energy nearly 17.4% fixed carbon and 14.1% ash
source it should be translated into a user content[3].
friendly, advance manner that can be easily
handled[9]. Making densified fuel briquettes is 2.1 Chemical properties
the promising technology to be applied to solid
The main components of the wood are the
fuel production from these wastes because it is
cellulose, hemi-cellulose and lignin. In addition,
the simply technology which can be adjusted
main types of natural binding agents of
for the local area. The bulk volume of the
biomass particles are lignin, protein and
material can be reduced by applying a
starch[12]. Protein acts as a binder at plasticized
compressive load for easy handling,
state which needs processing at high
transportation and storage [10]. There are
temperature. Sawdust contains 40-50% cellulose
several pressure values for briquetting being
and 30% lignin[13]. Therefore starch, lignin or
used in previous research papers. InTanakorn
water soluble fibbers are needed to be added
et al.[10]compression pressure varies between
externally in order to increase the bonding of
56-166MPa for 20 seconds for each pellets. In
biomass in briquettes. When applying pressure
Lela et al.[6]it is mentioned that compression
on the briquette, lignin will squeeze out of the
force should be at least 100kN to obtain
biomass cell and agglomerate the fuel briquette.
briquettes with satisfactory strength and in
Lignin is not softened at the ambient
their research applied pressure varies between
temperature and pressure [13]. To get to lignin
2
126
work as a binding agent, the glass transition
point has to be reached[14]. For dry native
lignin the temperature is about 200 and this ….. (1)
temperature depends on the moisture content
of the biomass. Softening temperature for  Shatter index
biomass is around 90-100 at 30% moisture Shatter index of a briquette is defined as the
and around 130 at 10% moisture[15]. After percentage of a briquette retained on a sieve of
cooling solid briquette becomes harder. stated aperture after being subjected to a free or
However, previous researches have suggested gravitational resistance to fall or drop[5]. This
that binding materials need to be supplied measures the resistance to mechanical action
externally for ambient temperature which affects its transportation and handling.
processes[15]. In biomass combustion fuel gases This test is done either in a rotating drum or by
burn and release heat and liquid and other repeatedly dropping samples from a specific
solids do not burn themselves. But they height. The briquette fraction retained after
consume heat in the drying and volatilization screening is used as an index of its friability.
processes needed for them to be chemically Shatter should attain a value of higher than 90%
converted into fuel gasses. The main fuel according to a previous research results. In a
intermediates are volatile hydrocarbons and previous research each briquette were subjected
energy rich organic molecules, Carbon to gravitational fall. The initial mass of the
Monoxide (CO) and Hydrogen (H2)[16]. The briquette was taken and then subjected to the
process of gas evaluation due to heat is called fall of gravitational fall from a constant 2m
pyrolysis, and progressively converts the height three times. Then samples are passed
biomass into gasses, volatile liquids, and a through a 2.36mm sieve while the mass of the
carbon rich solid residue called bio char[16]. briquette retained on the sieve was
Biochar isa carbon rich (65 to 90%), fine-grained recorded[18].
and porous substance obtained when biomass,
such as wood, manure andleaves, is heated or
natural burned under oxygen limited condition
and at relatively low temperatures (<700°C).
Biochar yield decreased with increasing
temperature and the ash and calorific value of … (2)
the biomass increases with the increment of the 2.3 Physical properties
heating temperature[17].  Moisture content
Moisture content of a briquette is one of the
2.2 Mechanical properties
most important properties to consider for a fuel
 Compressive strength briquette. Usually the average moisture content
The maximum compression strength of a of sawdust is about 27% at dry basis. But this
briquette is defined as the capacity of a number varies with the type of sawdust. Since
briquette to withstand axially directed pushing it’s difficult to collect different types of sawdust
forces. This is usually performed to identify the in pure form from sawmills, a mixture of
ability of the briquette to withstand the sawdust has to be collected and therefore an
crushing load during handling, transporting average value for the moisture content should
and storing. There is no standard method be taken. Moisture content affects briquettes
established for testing the compressive physical properties like durability, density and
strength[15].In a previous research the most importantly the calorific value and the
compressive strength is measured using a combustion efficiency[6]. After the densification
compressive testing machine. The briquettes process briquettes should be air dried to
were placed in between plates of the machine, remove the moisture. It is foundthat the
which ensured uniform stress distribution over optimum moisture content for a briquette is 12-
the cross sectional area of the biomass fuel. The 15%[19]. In most of the previous researches
briquettes were subjected to uniform loading briquettes are air dried to reduce the moisture
until it failed. The corresponding strengths content[10],[5],[15],[20]. In[11]found that if the
were recorded from the scale of the moisture content is over 50 %( wet basis) it’s
compressive strength machine at the time of impossible to form a bound aggregate from a
failure[5]. forestry or wood residues. Also when the
moisture content is less than 12% low
comparison strength can be observed[15]. But
water is also an important factor in the
3
127
compression process, as it can intervene gives an indication of the proportion of char
chemically in the particle bonding process, that remains after the depolarization phase.
facilitating binding between the hydrogen
bridges in the cell wall. Moreover, dry Fixed Carbon = 100 - (Volatile Content + Ash
lignocellulose material acts as thermal Content + Moisture Content) … (3)
insulation between which obstructs heat
transmission, a key factors in the compacting  Calorific value
process[11]. In[6]it is found that moisture Calorific value can be defined as the amount of
content decreases with the increasing drying energy produced by the complete combustion
temperature. of a material. This heating value of biomass is
depend on the carbon content, ash content and
 Density the moisture content and reduces with
Density is an indicator of strength and stiffness; increasing moisture content and it has a linear
it varies widely between wood species and relationship with ash content and moisture
types. There are several methods used in content[21]. Ash content and fixed carbon
previous researches to determine the density. content doesn’t change with compression
One method is measuring the height and the process and fixed carbon content depends on
diameter of the briquette by a venire calliper the cellulose content of the selected
and measuring the mass by an electronic biomass[15]. According to[6] ash content in
balance[15]. Another method is by measuring obtained after the combustion of a briquette is
the volume by water displacement method. To increases with the reduction of the compression
prevent water absorption into the briquette a load. Therefore according to[21]when the
wax coating is applied to the briquette. Used compression load reduces, the calorific value of
volume of the wax is measured before using it the briquette reduces. Thus calorific value can
and from the difference of volumes, the volume be used to evaluate the competitiveness of a
of the briquette measured[5]. Second method is processed biomass fuel in a given market
the most accurate method to measure the situation. But the production cost of a briquette
density among these methods since briquette is independent of their calorific value. In
may not be uniform. In[3] it is found that the previous researches calorific value is given by
density of the sawdust is 370 kg/m3. Lower heating value (LHV) or Higher heating
value (HHV). Because of the difficulty to collect
 Volatile matter pure sawdust types from sawmills, calorific
Volatile matter signifies the components of value of different types of sawdust is hard to
carbon, hydrogen and oxygen present in the measure. Therefore in most of the previous
biomass that when heated turn to vapour. researches an average value or rage for calorific
Usually it is a mixture of short and long chain value is given. According to[3]average calorific
hydrocarbons. value (HHV) of sawdust collected from
Volatile content has been shown to influence Moratuwa at 28% moisture content is 18.8
the thermal behaviour of solid fuels, but this is MJ/kg. But since the HHV is independent from
also influenced by the structure and bonding the moisture content, this value has no effect on
within the fuel. the moisture content that is mentioned in this
research paper. In[6]it’s found that the
 Ash content compression doesn’t have a significant impact
Ash has a major influence on the heat transfer on HHV. But in[13] it is mentioned that
to the surface of the fuel, as well as the briquetting enhances the calorific value. Since
diffusion of oxygen to the fuel surface during there is a contradiction in these statements and
char combustion. in[13] there is no proper argument on this,[13]
argument can be taken as the correct one.
 Fixed carbon
The conversion of inorganic carbon (carbon  Environmental impact
dioxide) to organic compounds by living When considering a fuel environmental impact
organisms is called carbon fixation. The fixed is one of the most important factor that should
carbon of a fuel not equal to the total amount of be considered. Fossil fuel are the most
carbon in the fuel because there is also a commonly used energy sources in the world
significant amount released as hydrocarbons in and Petroleum and Coal are the most impactful
the volatiles. It is the percentage of carbon energy sources among them. But these two
available for char combustion. Fixed carbon non-renewable energy sources are the most
important environmental pollutants in the
4
128
present time. These are the main reasons for the Some of these pollutants can lead to formation
global warming and greenhouse gas of acid rains and snowfalls[23]. Even though
emissions[22]. In a previous research paper[23] sawdust has these negative impacts on the
it is mentioned that due to the discharge of environment, when it comes to densified
harmful substances to the environment carbon sawdust briquettes these impacts are
cycle in the nature is strongly affected. comparatively lesser than wood[23].
Therefore renewable energy should be given
more attention. Since the world is paying more 3. Objectives
attention towards sustainable energy, several Objectives of this research are finding an
types of renewable energy are introduced and environmental friendly solution for sawdust
people interest toward biomass has increased. and rice husk disposal problem and also
When compared to coal burning wood is more introducing an alternative fuel for energy
beneficial to the environment point of view due demand with higher calorific value and easy
to the lowest total emission of greenhouse handling for domestic use.
gasses (GHG) [7]. Wood combustion is
considered as a carbon neutral process since the
4. Materials and Methodology
CO2 produced in the process of wood
combustion is considered as a part of the 4.1 Collection and pre-treatment of raw
carbon cycle in nature[23]. Briquetting biomass materials
is one of the most commonly used and efficient Saw dust and rice husk are collected from saw
way of using them as a fuel. In Poland the mills and rice mills around Moratuwa area and
government has banned installing coal fired air dried for 3 days. Both raw material were
boilers in new family houses. Moreover, boiler grinded and sieved using a 2 mm mesh.
users are becoming increasing interested in
renewable fuels, mainly briquettes and 4.2 Sample Preparation
pellets[7]. Lack of proper sawdust waste
management system in Moratuwa one of the Dry sawdust and rice husk were blended
major problems that the timber industry in together according to four mass ratios.
Moratuwa faces. Sawdust is one of the major
wastes resulting from wood processing, which Table 1 - Mixed mass ratios of sawdust and
stored in uncontrolled manner can be resulting rice husk
environmental pollution[23]. From about 400
sawmills in Moratuwa around 2000 MT of Sample Mass ratio of Mass ratio
sawdust has accumulated in Moratuwa over Name Sawdust of Rice
the years and the Not for Profit Organizations husk
(NFPO) currently helps to dispose about 350- Sample A 100% 0%
400 MT per month[24]. Therefore there are Sample B 75% 25%
several researches done to find a solution for
these two problems in Sri Lanka and also Sample C 50% 50%
several researches have conducted around the
world too.Most of these researches are about Sample D 25% 75%
manufacturing efficient fuel briquettes from
sawdust or developing them. But also there is a 4.3 Briquetting
research conducted in 2007 to find the
availability of manufacturing bricks from Briquettes from every samples were made
sawdust and limestone dust[2]and they have using the Mount Press Machine (Figure 1)
found that it’s a viable option. which is available in the Department of
As well as there are lot of beneficial effect in Material science and Engineering, University of
using sawdust briquettes as a fuel there are Moratuwa. Few series of samples were
some negative impacts on the environment too. prepared using the mount press. Mould was
These pollutants can be categorized as ash and heated to 130 °C and kept for 5 minutes after
gaseous pollutants proceed though the applying 5 Psi (34.5 KPa) pressure. Then
combustion gases. Carbon Monoxide (CO), briquettes were removed and kept for cooling
volatile organic compounds, oxides of nitrogen, for few minutes.
oxides of sulfur and particulate emissions.
5
129
 Fixed carbon
FC (%) = 100 – (A +B+C) … (5)

Where,
A = moisture content
B = volatile matter
C = ash content
 Shatter index
Procedure: Weight the samples ( . Released

the samples in to 1 m height at three times.
Samples are sieved in to 2 mm sieve and
Figure 1 - Mount Press weighted remaining part (
Machine
4.4 Testing Methods … (6)

 Compressive strength
Several tests are conducted to find
characteristics of fuel and some of these tests Apparatus: Universal Tensile Testing Machine
are conducted to find characteristics of sawdust Procedure: Samples placed between plates of
and rice husk mixed saw dust. the machine. The briquette was subjected to
uniform loading until it failed / ruptured.
 Density Corresponding compressive strengths were
Density was determined for each briquette as recorded from scale of the machine.
ratio of briquettes weight to volume. Compression strength (MPa) = Stress/Strain ... (7)
Apparatus: Venirecalliper, three decimal scale

Procedure:  Water permeability
Measured the weight, diameter and height of
samples. Water permeability of the briquette is as the
Diameter of sample = change in volume and weight when the
briquette is submerged in water for 2 minutes.
Height of the sample =
Procedure: Weighted the samples ).
Volume of briquette = Measured the dimension of briquettes
(diameter = di, height = hi). Samples are
submerged in to water for 2 minutes. After
Density = ... (4)
weighted samples ( ) and measured
dimensions (d2, h2)
 Moisture Content
Method: ASTM E871

... (8)
Moisture content of the briquette is found by
using ASTM E871 standard method.
 Volatile Matters .. (9)

Method: ASTM E872 Where,
Volatile matters of the briquette is found by Vi = initial volume
using ASTM E872 standard method. Vf = final volume
 Ash content
Method: ASTM E830

Ash content of the briquette is found by using
ASTM E830 standard method.
6
130
Figure 4 shows the shatter index of briquettes
Density (kg/m3) are decreased with rice husk percentage (Figure
4).
1100
Density (kg/m3)
Therefore when the rice husk percentage

1000 1079.28 1097.52
1046.51 increased bonding strength of briquettes is also
979.95 increased.
900
A B C D According to above results mechanical
properties of briquettes increase with
increasing rice husk percentage.
Figure 2 - Density Variation of briquettes
When considering physical properties such as
When comparing density variation of moisture content, volatile matter, fixed carbon
briquettes, rise husk helps to increase the and ash content (Figure 5) sample A is shown
density of briquettes. Highest density is found positive qualities for combustion. As household
from sample D (25% saw dust and 75% risk fuel briquettes ash content should be reduce.
husk mixture) (Figure 2). Higher volatile matter and fixed carbon implies
higher combustion properties.
Calorific value of saw dust is higher than rice

Compressive strength (MPa)
husk and it is shown in figure 6. Therefore
strength( MPa)
Compressive
550 when decreasing saw dust percentage, calorific

500 value also decrease.
524.72 537.6
450
469.26 475.19 Figure 7 shows result of water permeability
400 test. According to visual inspections (figure 7)
A B C D “sample A” has higher water permeability than
others. When increasing rice husk, water
Figure 3 - Compressive Strength Variation of permeability is decreased.
briquettes
The figure 3 shows compressive strength of Physiochemical properties are decreased with
briquettes vary with different compositions. rice husk percentage and mechanical properties
Sample D is shown highest compressive are increased with rice husk percentage.
strength and it implies more rice husk give to
high compressive strength.
Shatter index (%)

Shatter index (%)
100
99
98 99.515 99.135 99.048
98.503
97
A B C D
Figure 4 - Shatter Index Variation of

briquettes
7
131
A B C D
73.496
79.91
68.06
65.9
11.634
11.77
11.55
10.79
10.84
13.3
9.96
9.38
9.04
6.58
5.83
1.96
Moisture Cont (%) Volatile Mat (%) Fixed Carbon (%) Ash content
Figure 5 - Variation of briquettesPhysical Properties
Calorific value (MJ/kg)

18
17
16
15 17.28
16.44
14 15.60
14.76 75% sawdust 25% rice
100% sawdust briquette
13 after water husk briquette after
A B C D permeability test water permeability test
Figure 6 - Calorific Value Variation of
briquettes
6. Conclusions
Calorific value is the most important factor that
defines the energy available in a fuel. When
50% sawdust 50% rice 25% sawdust 75% rice
considering the results, calorific value of husk briquette after water husk briquette after water
sawdust is higher than that in rice husk. So permeability test permeability test
when it comes to the briquette made with
sawdust and rice husk mixture, calorific value
is less than that of sawdust. Therefore the Figure 7 - Water Permeability Variation of
obtainable energy from the briquette is slightly briquettes
less than sawdust briquettes.
But durability, ease of transportation and
storage are some important factors that has to Acknowledgement
be considered when manufacturing briquettes.
In the present research it is found that stability Authors wish to thanks the Department of
of the briquette is higher when briquettes are Chemical and Process Engineering and
made using sawdust and rice husk mixture Department of Material Science Engineering of
than sawdust briquettes. Also when both University of Moratuwa for providing the
sawdust and rice husk is mixed, it doesn’t laboratory facilities and also wish to
require any binding material when the right acknowledge the assistance given by the IESL
pressure and temperatures are applied at in preparation of this guide.
densification stage.
Therefore to use as a fuel and for a waste
management solution, sawdust and rice husk
mixed briquettes can be given as a good
solution.
8
132
References briquettes with sawdust,” International Journal of
Scientific and Research Publications, vol. 5, no. 3, pp.
1-8, 2015.
1. A. Gore, “The end of the fossil fuels - our green
enegry,” 2010. [Online]. Available: 16. R. P. Overend, “Direct Combustion of Biomass,”
https://www.ecotricity.co.uk/our-green- Renewable Energy, vol. 1, pp. 1-9, 2008.
energy/energy-independence/the-end-of-fossil- 17. G. d. S. A. B. A. Wan Azlina Wan Ab Karim Ghani,
fuels. [Accessed 09 June 2016]. “PHYSICO-CHEMICAL CHARACTERIZATIONS
2. H. M. A. Paki Turgut, “Limestone dust and wood OF SAWDUST-DERIVED BIOCHAR AS
sawdust as brick material,” Building and POTENTIAL SOLID FUELS,” The Malaysian Journal
Environment, vol. 42, no. 9, pp. 3399-3403, 2007. of Analytical Sciences, vol. 18, no. 3, pp. 724-729,
2014.
3. P. R. S. S. A. S. S. B. P. A. S. K.K.C.K. Perera,
“Assessment of sustainable energy potential of 18. G. Borowski, “The Possibility of Utilizing Coal
non-plantation Biomass resources in Sri Lanka,” Briquettes with a Biomass,” in Environment
Biomass and Bioenergy, no. 29, pp. 199-213, 2005. Protection Engineering, Lublin, Lublin University of
Technology, 2007.
4. P. R.M.Amarasekara, “Resource Potential of
sawdust and its spatial distribution in Kandy 19. A. S. Ayhan Demirbas, “Evaluation of biomass
district,” Integrated Development Association, residue: 1. Briquetting waste paper and wheat
Kandy, Sri Lanka, 2002. straw mixtures,” Fuel Processing Technology, vol.
Volume 55, no. Issue 2, 1998.
5. B. A. C. Antwi-Boasiako, “Strength properties and
calorific values of sawdust-briquettes as wood 20. C. H. A. S. B. N. Emre Altun, “Influence of coal
residue energy generation source from tropical briquette size on the combustion kinetics,” Fuel
hardwoods of different densities,” Biomass and Processing Technology, vol. 85, pp. 1345-1357, 2004.
Bioenergy, vol. 85, pp. 144-152, 2016. 21. P. Z. M. V. G. O. D. M. D. Bolortuya, Physics of
6. M. B. S. N. ˇ. e. B. Lela, “Cardboard/sawdust Particles and Nuclei Letters, vol. 10, pp. 723-726, 2013.
briquettes as biomass fuel: Physical–mechanical 22. Rinkesh, “Conserve Energy Future,” 2015. [Online].
and thermal characteristics,” Waste Management, Available: www.conserve-energy-
vol. 47, pp. 236-245, 2016. future.com/NonRenewableEnergySources.php.
7. M. K. z. W. Mariusz Jerzy Stolarski, “Energy, [Accessed 1 10 2016].
economic and environmental assessment of heating 23. L. F.-T. F. G. Teodora Deac, “Environmental impact
a family,” Energy and Buildings, vol. 66, no. of sawdust briquettes use - experimental
18/07/2013, pp. 395-404, 2013. approach,” Energy Procedia , vol. 85, pp. 178-183,
8. S. E. A. S. Lanka, “Sri Lanka Enegy Balance,” 2014. 2016.
[Online]. Available: 24. I. Nizam, “The Island,” 5 9 2010. [Online].
http://www.info.energy.gov.lk/. [Accessed 29 09 Available:
2016]. http://www.island.lk/index.php?page_cat=article-
9. Nalladuari Kaliyan, R. Vancey Morey , “Natural details&page=article-details&code_title=6011.
binders and solid bridge type binding mechanisms [Accessed 9 2016].
in briquettes and pellets made from corn stover and 25. M. S.Ericson, The briquetting of agricultaral wastes
switchgrass,” Biosource Technology , 2009. for fuel, Rome: FAO, 1990.
10. T. P. N. S. N. S. M. T. Tanakorn Unpinit, “Fuel
Properties of Bio-Pellets Produced from Selected
Materials under Various Compacting Pressure,”
Energy Procedia, vol. 79, pp. 657-662, 2015.
11. S. V. M. L. Y. A. I. Relova, “Optimisation of the
manufacturing variables of sawdust pellets from
the bark of Pinus caribaea Morelet: Particle size,
moisture and pressure,” Biomass and bioenergy , vol.
33, pp. 1351-1357, 2009.
12. R. V. M. Nalladurai Kaliyan, “Factors affecting
strength and durability of densified biomass
products,” Biomass and bioenergy , vol. 33, pp. 337-
359, 2009.
13. L. L. G. J. M. J. M. E. Granada, “Fuel lignocellulosic
briquettes, die design and products study,”
Renewable Energy , vol. 27, pp. 561-573, 2002.
14. M. Stahl, “Improving Wood Fuel Pellets for
Household Use,” Karlstand University Studeis,
2008.
15. A. A. N. S. Daham Shyamalee, “Evaluation of
different binding materials in forming biomass
9
133
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[135 142],2017
2017
© The
© The Institution
Lanka
Assessing the Performance of a Destructive

Measurement System: A Case of an Industrial Glove
Manufacturing Environment
H.A.D. Perera and P. Gamage
Abstract: Quality improvement is a key aspect to gain a competitive advantage by any

organisation. The main focus of quality improvement is to reduce the variation of any system which is
inherent with product/process variation and the Measurement System Variation (MSV). Thus, Gauge
Repeatability and Reproducibility (Gauge R&R) study provides a better solution in verifying the
precision of the Measurement System by assessing the MSV as a percentage of total variation in a
system. Gauge R&R study is challenging when it is used in assessing destructive measurements
where the product is affected by the measurement. Literature suggests that there is a poor focus on
destructive Measurement System Analysis (MSA). Hence, this study is focused on assessing the
performance of a destructive measurement system for measuring the abrasion resistance level of
industrial gloves using Gauge R&R and Analysis of Variance (ANOVA) method. The results show
that the percentage study variation of total Gauge R&R is 17.41%, which is acceptable based on the
American Automotive Industry Action Group (AIAG) guidelines.
Keywords: Gauge R&R, Destructive Measurement System, Industrial Gloves
1. Introduction the processes as the real process variation can

be concealed by the Measurement System
With the growth of the global economy Variation (MSV). Therefore, understanding the
manufacturing organizations strives to be MSV is essential as it corporates in identifying
competitive by satisfying customers [18]. the real process variation and preventing
Engineers engaged in design, development and wrong quality management decisions [19].
actual manufacturing of products, play a vital
role in achieving process efficiency and process MSV can be categorised into accuracy closeness
capability of their manufacturing processes. of the mean in measured values to the true
Another important aspect of process efficiency value and precision closeness of replicated
is minimising the variation of the process observations [9, 19]. Precision can be
output so that the process output or the quality categorised into repeatability measuring the
characteristic under consideration stays well same sample using the same instrument by the
within the specification limits [1, 3]. same appraiser and reproducibility measuring
that includes the same sample using the same
Quality management decisions are depending instrument by different appraisers as shown in
on the quality of the data and knowledge that Figure 1.
could be gain through analytical studies. It is
likely to be low or high if the quality of the data Gauge R&R study is a designed experiment to
is respectively low or high—which is used to assess the capability of the MS to determine the
make an actionable conclusion [17]. total variability due to gauge, to isolate the
Measurements are considered as the data that is
essential for understanding and improving the Mrs. H. A. D. Perera, BSc. Eng. Hons (Thailand), MSc.
quality of product/ process or any kind of Candidate, Department of Manufacturing and Industrial
Engineering, Faculty of Engineering, University of
service [5, 10]. A measurement system is a Peradeniya.
process used to obtain measurements which are
a collection of gauges, appraisers, methods, Dr. P. Gamage, BSc. Eng. Hons (Peradeniya), Ph.D. (New
Zealand), Lecturer, Department of Manufacturing and
instruments, etc. [19]. Establishing a correct Industrial Engineering, Faculty of Engineering, University
Measurement System is a key requirement of Peradeniya.
before checking the stability and capability of
1
135 ENGINEER
-sources of variability in the system. collecting any data to understand the process
measurement system to obtain reliable data for stability, capability or to perform any quality
quality improvement activities [6]. improvements activities of product/ process
because these results contain do get obscured
Measurement systems can be broken down into by Measurement System variation. Inadequate
replicable measurement systems where the measurement systems cause loss to the
reading can be replicated for each part and non- company as well as for the customer due to the
replicable measurement systems [6, 19] where wrong quality decision with unreliable data [10,
the part is being damaged/destroyed or apply 12, 17].
changes to the characteristics of the part by the
Gauge R&R study is challenging when it is
measurement process [19].
used in assessing a destructive measurement
system where products are damaged when
Luo et al. [21] depicted that the requirement of
taking the measurements [11, 17]. Measuring
further research on the effective analysis of data
reproducibility and repeatability can be
obtained from a destructive measurement
system is limited in the current literature. performed under a strong assumption on batch
Therefore, the aim of this paper is to assess the homogeneity [16]. De Mast and Trip [11] have
capability of the measurement System using explained that good estimate of the
ANOVA method and few other capability measurement system can be obtained if the
measures such as P/T ratio, Signal-to-Noise experimenter can arrange a Gauge R&R study
ratio (SNR), and discrimination ratio (DR). with the homogeneous conditions which is not
Thus, this research is focused on analysing a completely solved the confounding of
destructive measurement system with
measurement system variation with other
Homogeneity Sample method followed by a
sources of variation.
case example of an ABC industrial glove
manufacturing plant.
Lou et al. [17] have studied the application of
2. Literature Review homogeneity method to analyse destructive
measurement system with a case example of
Total variation of any process and/or product electronic industry, and further research is
can be categorized into two major components: recommended to analyse destructive
the process and/or product inherent variation measurement systems effectively. ANOVA
and Measurement System Variation (MSV) as method has the capability of handling any
shown in Figure 1. experimental setup, estimating variances more
accurately and finding the interaction between
parts and appraisers though its numerical
computation is complicated [19].
According to AIAG [11] guidelines, (a) < 1%

contribution from Gauge R&R is ideal and 1-9%
contribution is acceptable, (b) < 10% Study
variation is ideal, 10-20% is acceptable, 20-30%
is marginal but acceptable if it is not
economically feasible to improve the
measurement system [4, 19]. Apart from
ANOVA method, few other criteria (P/T ratio,
SNR, and DR ) are used to validate the results
obtained from a Gauge R&R study, for a better
conclusion of acceptability of a measurement
system [7, 8, 14] Following section describes
different capability measures which were used
Figure 1 – Total process variation (Source: [9, 19]) to assess the capability of Gauge R&R study in
this case example.
Many literatures [2, 6, 10, 12, 19] have been
emphasized that conducting a Gauge R&R
study to analyse the MSV is important before
ENGINEER 2
136
2.1 Capability Measures Due to the fact that P/T ratio, SNR, and DR
criterions are being subjective to each process
Precision to Tolerance ratio (P/T ratio) and do not explain the capability of the
measurement system adequately, Burdick et al.
The precision to tolerance ratio is one of the
[5] highlights the need of inducing confidence
quality measures that use to evaluate the
capability of a measurement. The P/T ratio can intervals to the assessment process.
be calculated using the formula shown below
[5, 19]: Table 1 describes the criteria and their
acceptability ranges—extracted from
literature—which are discussed in this case
….(1) research to assess the capability of the
measurement system.
Where MS is the variance of the measurement
system, USL and LSL are the upper and lower Table 1 – Acceptability of evaluation criteria
specification limits of the process respectively,
and K is the standard deviations between Source
―natural‖ tolerance limits of a normal process.
Moderately
Accepted
Accepted
Criterion
Rejected
According to AIAG [20] guidelines, the
measurement system is capable when P/T ratio
 10%, and incapable when P/T ratio  30.
R&R% < 10% 10%-30% > 30% [5, 14, 15, 19]
Montgomery and Runger [10] have explained P/T <8% [17, 15, 19]
that there is a risk in relying on P/T ratio which ratio
can be made smaller by increasing the ndc 5 2-4 4 [5, 14, 19]
specification limits. Therefore measurement SNR 5 <5 [17, 19]
system is measured as a function of the ratio of DR >5 4-5 <4 [5, 6, 13]
process variance to measurement variance .
Signal-to-Noise Ratio (SNR) Table 1 describes the criteria that are discussed
here in this research with their acceptability
SNR is another capability measure which is a ranges extracted from the literature.
function of that can be calculated by the
formula given below [5, 19]. 3. Case Study
XYZ Company is a glove manufacturing

..(2)
company that produces many types of gloves
such as industrial gloves, surgical gloves, etc.
Where;
Abrasion resistance is considered as the major

quality characteristic of ABC industrial gloves
according to the customers’ requirement. (XYZ
…(3)
and ABC are pseudonyms given by the
p :Variance of the process
2
2: Variance of the measurement system researcher to prevent the identification of the

MS
company and product respectively).
Researchers are planning to conduct an
Further, Wheeler and Lyday [4], and Woodall
experiment to improve the abrasion resistance
and Borror [13] have presented Discrimination
level of industrial gloves to satisfy the
ratio (DR) as another capability measure which
customers by being competitive with other
is also a function of to examine the
glove manufacturers. An abrasion testing
applicability of a measurement system. DR is
machine is used to measure the abrasion
calculated as follows:
resistance of ABC industrial gloves.
…(4)
Where, …(5)
3
137 ENGINEER
Figure 2 – Abrasion resistance testing process
The Martindale abrasion resistance tester

(LIAB007) can test 9 samples at a single run that
takes around 4 hours for each set. Figure 2
explains the testing procedures which are
following by the operators to determine the
abrasion resistance of ABC industrial gloves.
Each sample is damaged during abrasion test.
Therefore, this measuring system will fall into
the category of destructive tests. Gauge R&R
study can be conducted with the key
assumption that the homogeneity of samples
within a batch to analyse the destructive
measurement system [17, 19]. Figure 3 – Research Methodology
Thus this case study example is focused on 4.1 Design the Experiment
analysing the measuring system which used to
measure the abrasion resistance of ABC
industrial gloves as the first step in quality
improvement process to make sure that the
measurement system variability itself is
consistent and comparably small, relative to
process variability.
4. Methodology
The methodology of this research can be

represented schematically as shown in Figure 3
below.
Figure 4 – Selection of homogenous samples
Gauge R&R study is designed with two

appraisers (operators A and B) with a single
abrading machine using 48 samples of gloves.
Since the nature of the test is destructive, Gauge
R&R study was conducted with the assumption
ENGINEER 4
138
that ABC gloves at each dipping row (see Sample Method based on ANOVA method and
Figure 4) have homogeneous properties. This with other criteria (P/T ratio, SNR, and DR) for
study assumes that researchers used the a better conclusion of acceptability of a
homogenous batch of parts which can be measurement system. Figure 5 and Figure 6
assigned at least two parts from each batch to show the graphical plots and ANOVA results
each operator to conduct the destructive Gauge which were generated by Minitab 17 software
R&R study. Figure 5 shows how the batches package respectively. Figure 7 shows the
were selected with the homogeneous properties percentage study variation of total Gauge R&R
to conduct the destructive Gauge R&R study. is 17.41%. This result shows that the
Samples were allocated to each operator, and measurement system is acceptable according to
the measurements were taken in the the AIAG guidelines [19, 20]. ANOVA results
randomised pattern which generated by a show that causes of gauge variation are due to
software package (Minitab 17). repeatability. Repeatability is larger compared
to reproducibility, and this may cause due to
5. Results excessive within part variation (96. 97%) [19].
The number of distinct categories (7) is also
This research was analysed the measurement greater than 5 and shows the acceptance of the
system which used to measure abrasion level of measurement system
ABC industrial gloves as a destructive
measurement system using Homogeneity
Assumption : Parts within Lot i (Li) is homogenious. (i= 1,2,……,12)
MSA (48 Samples)
Operator : OPERATOR A OPERATOR B
Lot: 1 2 ……… 12 1 2 ……… 12
Sample: S1 S2 S1 S2 S1 S2 S1 S2 S1 S2 S1 S2
Figure 5 – Crossed model Gauge R&R design
Figure 6 – Graphical representation of Gauge R&R study based on ANOVA method
5
139 ENGINEER
In Figure 6, the component of the variation chart shows that the precision of the
graph shows that part-to-part variation is the measurement system where the variation due
dominant source of variability, which basically to repeatability is low relative to part-to-part
implies that given the degree of part-to-part variation. The box plot shows the consistency of
variation, the measurement gauge is able to the measurements and the variability among
produce measurements that are repeatable and the two operators. The Sample × Operator
reproducible. The R chart shows that there are interaction plot shows that lines do closely
no points outside the control limits. Therefore, follow one another, thus showing high
this is indicated that the measurement is reproducibility.
repeatable and reproducible. The chart X-bar
Figure 7 – ANOVA results of Gauge R&R study
ENGINEER 6
140
In addition to this, P/T ratio, SNR, and DR References
capability measures and their confidence
intervals were calculated as shown in Table 2 to 1. Rao, L. P. Carr, I. Dambolena, R. J. Kopp, J.
Martin, F. Rafii, et al., A Cross Functional
validate the results obtained from this case
Perspective. . NY: John Wiley & Sons, 1996.
Gauge R&R study.
2. Montgomery, D. C., Introduction to Statistical
Table 2 – Acceptability levels of evaluation criteria Quality Control, 6 ed. Hoboken, NJ: John Wiley &
Sons, 2009.
Criterion Value 95 % CI Decision
3. Deming, W. E., Out of the Crisis. Cambridge,
Gauge R&R % 17.41% (9.82, 46.67) Moderately MA: MIT Press, 2000.
accept
P/T ratio 17.28% (14.16,52.35) Moderately 4. Wheeler, D. C., and Lyday, R. W., Evaluating the
accept Measurement Process, 2 ed. Knoxville, TN: SPC
ndc 7 (2.68,14.32) Accept Press, Inc., 1989.
SNR 5.65 (2.81,9.58) Accept
5. Burdick, R. K., Borror, C. M., and Montgomenry,
DR 7.07 (4.08,13.55) Accept D. C., "A Review of Methods for Measurement
Systems Capability Analysis," Journal of Quality
Technology, vol. 45, October 16–17 2003.
6. Discussion and Conclusion 6. Hamada, M. S., and Borror, C. M., "Analyzing
Unreplicated Gauge R&R Studies," Quality
In this research, we considered verifying the Engineering, vol. 24, pp. 543-551, 2012.
appropriateness of the destructive
measurement system that measures the 7. Burdick, R. K., Borror, C. M., and Montgomery,
abrasion resistance of industrial gloves through D. C., "Design and Analysis of Gauge R&R
Studies: Making Decisions with Confidence
use of statistical measures obtained from the
Intervals in Random and Mixed ANOVA
Gauge R&R and ANOVA method. The results
Models," American Statistical Association and the
are presented in Figure 6, 7 and table 2. Society for Industrial and Applied Mathematics.,
2005.
As shown in Table 2, R& R% = 17.41 %, ndc = 7,
P /T% = 17.28%, SNR = 5.65 and DR = 7.07%. 8. Al-Refaie, A., and Bata, N., "Evaluating
ANOVA results and the capability measures Measurement and Process Capabilities by GR&R
which were considered in this destructive with Four Quality Measures," Measurement, vol.
measurement system analysis are at acceptable 43, pp. 842-851, 2010.
levels at a 95% confidence interval.
9. Hare, L. B., "Gage R&R Reminders: Running
Additionally, from the comparison of all the
Gage Repeatability and Reproducibility Studies
considered measures, it is evident that this case Properly," Quality Progress pp. 62-64, 2012.
study example can be shown as a good gauge
capability analysis. Thus, the conclusion can be 10. Montgomery, D. C., and Runger, G. C., "Gauge
made that the destructive measurement system Capability and Designed Experiments Part I:
which used to measure the abrasion level of Basic Methods," Quality Engineering, vol. 6, pp.
ABC industrial gloves is capable of obtaining 115-135, 1993.
reliable data for any quality improvement
11. De Mast, J., and Trip, A., "Gauge R&R studies for
activities. Further studies have to be conducted
Destructive Measurements," Journal of Quality
to improve the dimensional measurement
Technology, vol. 37, pp. 40-49, 2005.
process further using Design of Experiment
(DoE) methods. 12. Sarkar, A., Mukhopadhyay, A. R., and Ghosh, S.
K., "Measurement System Analysis for
Implementing Design for Six Sigma,"
Acknowledgement International Journal of Productivity and Quality
Management, vol. 14, pp. 373-386, 2014.
Authors would like to acknowledge the 13. Woodall, W. H., and Borror, C. M., "Some
assistance and support given by the employers Relationships between Gage R&R Criteria,"
at the case study organization in conducting Quality and Reliability Engineering International
this research work. pp. 99–106, 2008.
7
141 ENGINEER
14. Cagnazzo, L., Sibalija, T., and Majstorovic, V.,
"The Measurement System Analysis as a
Performance Improvement Catalyst:A Case
Study," pp. 269-292, 2010.
15. Gunawan, I., and Sirodj, D. A. N., "Validation of

the Grader's Ability using Measurement System
Analysis," ARPN Journal of Engineering and
Applied Sciences, vol. 11, August 2016 2016.
16. Ya-Juan, H., and Zhen, H., "An Applied Study of

Destructive Measurement System Analysis,"
presented at the Second IEEE Conference on
Industrial Electronics and Applications, 2007.
17. LOU, Z. B., Shi, L. X., He, Z., and He, Q. M., "The
Study of Destructive Measurement Systems,"
presented at the Industrial Engineering and
Engineering Management (IE&EM), 2011.
18. Longo, R. J., "An Assesment of Total Quality

Managmeent in the Finanace Services of United
KIingdom and Brazil," Doctor of Philosophy,
Department of Management Studies, University
of Newcastle, Newcastle upon Tyne, UK, 1997.
19. AIAG, "Measurement System Analysis," ed.

Detroit: AIAG, 2010.
20. AIAG, "Measurement System Analysis," Chrysler

Group LLC, Ford Motor Company, General
Motors Corporation USA 978-1-60-534211-5, 2010.
ENGINEER 8
142
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[143 150],2017
2017
© The
© The Institution
Lanka
Development of ANN Based IMC for Biomass Pack

Bed Gasification Plant with Enhanced Efficiency
G.V.C. Rasanga, K.T.M.U. Hemapala and A.G. Buddhika P. Jayasekara
Abstract: Energy crisis and emerging negative impacts on environment are the driving factors
of process industries to increase the share of sustainable resources in energy production. The
utilization of biomass has a growing attention as a replacement for fossil fuels due to its wide
availability and sustainability. However, biomass gasification has to face most of its drawbacks
because of unpredictable variability of biomass properties, process complexity, and controllability of
the process which creates additional barriers for a wider utilization of biomass gasification.
This study focusses on the development of biomass gasification plant controller to enhance the process
efficiency. The plant control was developed based on internal model controller (IMC) architecture
while gasification plant model was developed using neural network based nonlinear autoregressive
model with external output (NNARX). Plant inverse model which was developed based on the
NNARX was used as the controller. The plant model predicts the output for one time step of five
seconds and sends the error between prediction and real output to the controller.
Developed neural network based internal model controller (NNIMC) was tested using 15kW pack bed
imbert type downdraft biomass gasifire and controller algorithms development was then analysed for
72 minutes of the plant operation. The analysis showed that with the introduction of NNIMC control
solution process, efficiency can be improved up to 12%, with the increase of performance in terms of
stability.
Keywords: Biomass, Gasification, Internal Model Controller, Neural Network
1. Introduction technical requirement for an example, partially

composition of the produced gas. The
Sri Lankan industrial energy demand is being composition of CO and H2 are greater than 20%
increased with the population growth and and 10% respectively, while tar content should
urbanization [1]. Therefore, the cost of be lower than 100mmgNm-3.
electricity and fossil fuel, which are the primary
energy sources of current Sri Lanka, is also However, this technology is still not much
continuously going up. In this context, usage of acquainted for small scale domestic power
renewable energy sources is becoming more generation application. Because of complexity
popular, particularly, industrial sector is and operation difficulties of those processes
interested in biomass-based solutions to meet such as maintaining continuous process output
their thermal, electricity and other forms of due to process uncertainties and process
energy demands due to low cost [2]. disturbances like fuel properties.
Most of past researchers tried to overcome
“Gasification” is one of the process of those problems by improving reactor geometry
converting solid fuel, which was invented and fuel pre – processing. Although there is a
during World War II [3]. In this process, there is considerable practical development in past,
a thermochemical conversion of solid fuel into
Eng. G. V. C. Rasanga, AMIE (SL), B.Sc.Eng. (Ruhuna),
high calorific and useful gaseous fuel or
Mechanical Engineer, Sri Lanka Ports Authority, Galle, Sri
chemical products [4], which can directly use Lanka.
for internal combustion engine and gas burners.
Furthermore, researchers have tried to produce Eng. (Dr.) K. T. M. U. Hemapala, CEng, MIE (SL),
B.Sc.Eng. (Moratuwa), PhD(Italy), Senior Lecturer,
useful chemicals and gases using gasification
Department of Electrical Engineering, University of
produced gas and produced gas is also used for Moratuwa, Colombo, Sri Lanka.
fuel cell. The overall efficiency of electricity
generation by gasification process is higher Eng. (Dr.) A. G. Buddhika P. Jayasekara, AMIE (SL),
B.Sc.Eng. (Moratuwa), M,Sc.(Moratuwa), PhD(Saga),
than direct combustion of biomass [4]. For
Senior Lecturer, Department of Electrical Engineering,
power generation, biomass gasification University of Moratuwa, Colombo, Sri Lanka.
produced gas has to meet some special
1
143 ENGINEER
Gases
(CO, H2,
CH4, H2O)
Liquids CO, H2, CH4, H2O,

Gas – phase reactions
(tar, oil, CO2 cracking +
(cracking, reforming,
Biomass Drying Pyrolysis napthalin) 5%products
combustion, shift)
Oxygenated
Compounds
(phenols,
acid) CO, H2, CH4,
H2O, CO2
Solid Char gasification reactions unconverted
(Char) (gasification, combustion, carbon
shift)
Combustion Gasification
Figure 1 - Schematic Diagram of Gasification Process [3]
number of gasification projects has increased all uncertainty is related with properties of
over the world, especially in Europe. biomass [6] such as gasification agent, process
temperature, moisture content, and air flow
Typically, biomass gasification includes basic rate [3, 4, 7, 8 and 9].
steps as drying, pyrolysis, combustion and char
reaction (gasification) as illustrated in Figure 1 Modelling is very effective in terms of
[3]. optimizing the operation of existing gasifiers
and explore operational limits instead of sizing
Moisture content of fresh wood and dry the reactor. A model can identify the sensitivity
biomass are usually varied between 30% - 60% of gasifier performance according to the
and 10-20% respectively. Drying is the stage variations of for different operations and design
where bounded water of biomass irreversibly parameters [8]. Furthermore, there are different
removes at above 100℃ due to heat from high modelling approaches, such as thermodynamic
temperature zone (combustion zone)[3]. equilibrium, kinetic, computational fluid
dynamics and artificial neural network.
In the stage of pyrolysis, large hydrocarbon
molecules in biomass are broken down in to The maximum yield that can be accomplished a
small hydrocarbons and char, due to lack of desired product in a reacting system can be
oxygen at relatively low temperature as shown calculated using a thermodynamic equilibrium
in Figure 1. Gasification pyrolysis occurs model. Equilibrium of thermochemical system
between 300℃ - 650℃, which is called as is determined based on the equilibrium
pyrolysis temperature. Pyrolysis gas reacts with constant or minimization of Gibbs free energy
oxygen at combustion zone and results high [8, 10]. Furthermore, there are two equilibrium
temperature (1000℃-1350℃). Then combusted models such as stoichiometric and non-
gas (CO2, H2O) reacts with red hot char and stoichiometric depending on above two
produced syngas (CO, H2 and CH4). techniques.
According to gas – solid contacting mode and Thermal chemical equilibrium cannot be used
gasification medium, there are several type of for gasifire reactor [8], because gasification is a
gasifires has been developed [3, 4]. Among dynamic process. The main advantage of the
them, downdraft pack bed gasifire is very much equilibrium model is that independence from
appropriate for small scale power application gasification geometry. Different authors had
as it is simple, low cost construction and has developed gasification equilibrium models both
low tar production [3, 5]. in non-stoichiometric and stoichiometric
Biomass gasification influences on several models for downdraft gasifire [11, 12]. The
process parameters. The main variable of main purpose of equilibrium modelling is to
2
144 ENGINEER
predict the performance of gasifier in design operating points or with data derived due to
stage. lack of extensive measurements. Less number
of artificial intelligence systems based on
Development of kinetic models to evaluate and biomass gasification models have been reported
imitate the gasifier behaviour is caused by the because ANN model does not create intensive
inadequacy of equilibrium model to correlate measurement.
the reactor design parameter with the final
product gas composition, reaction rate, ANN is a universal function approximator that
residence time, reactor geometry. can approximate any continuous function to an
arbitrary precision even without a prior
Ozgun Yucel (2016)[13], has developed a kinetic knowledge on structure of the function that is
model for downdraft and predicted the reactor approximated [17]. Furthermore, ANN models
performance for different throat angles. have proven their potential of forecasting the
However, due to nonlinearity and complexity process parameters in energy related processes
of the model, gasifire kinetic model is also such as in biodiesel production process [18] and
difficult to use for online plant control. applied for different thermochemical reactor
applications such as flotation column, packed
Hybrid multilayer feedforward neural distillation column [19, 20]. For that, ANN
networks (HMFNN) were improved by Guo et models use a non-physical modelling approach
al. [14] to forecast the plant output of fluidised which correlates input and output data to form
bed gasifire and they had only considered a process prediction.
temperatures of gas and bed as model inputs.
Furthermore, Puig- Arnavat et al. [15] Macmurray and Himmelbealj (1995) [20],
developed an artificial neural network (ANN) developed ANN based controller for
model for fluidised bed gasifire. That model distillation column. After that, Mohanty (2009)
consisted of input output layer, one hidden [19], developed ANN based model predictive
layer, and composition of the biomass in terms controller for floatation column. Robert (2015)
of C, H, and O, equivalence ratio, gasification [21], later improved the gasification process
temperature, ash and moisture content as efficiency by 25% modifying the plant
model inputs. Predicted outputs were controller using ANN based control map.
percentage values of CO, CO2, H2 and CH4 in
produced gas as illustrated in Figure 2. According to the research outcomes of above
researches, ANN based predictive, model based
controller has proved its capability for
nonlinear process and plant control. Among
them, internal model controller (IMC) is one of
the simplest model based predictive controller
which can easily link with ANN [22].
2. Methodology
Methodology is comprised of experimental set

up of gasification plant and development of
internal model controller.
2.1 Biomass Gasification Plant

Figure 2 - ANN Model Structure to Predict Imbret type downdraft gasifier of 15-25kW
Produced Gas Composition of CFB Gasifier thermal output was used in the experimental
[15] setup as shown in Figure 3.
Equilibrium and kinetic models are useful in Plant consists of variable speed driver (VSD),
prediction of gasifier performance under which controls 1hp centrifugal blower that
number of different stationary operating pulls air through the gasifire and a swirl type
conditions therefore, they are often used for burner. The temperature of char bed, throat
preliminary design and optimisation purposes. and flue gas out were measured by k type
According to Ahmed [16], many equilibrium thermocouples as illustrated in Figure 4.
models have been verified just on several
3
145 ENGINEER
Figure 3 - Experimental Set-up of Downdraft Gasifire
The mass flow rate through the swirl burner 2.2 Artificial Neural Network (ANN)
was kept constantly using induced blower Based Internal Model Controller (IMC)
which is also displayed in Figure 4.
Architecture of the ANN based IMC is shown
in Figure 5.
K type
thermocouple
Figure 5 - Structure of Neural Network Based

Internal Model Controller
P, C and M in Figure 5, represent nonlinear

plant, nonlinear controller and nonlinear plant
model respectively. F is first order linear filter.
Both M and C are nonlinear autoregressive
network with exogenous inputs neural network
Produced gas
(NARXNN). Temperature at char bed, and
throat, speed control, signal of VSD (which
controls blower rpm) and previous predicted
plant output are the inputs of the plant model.
Gasification plant was operated for 340 minutes
at different blower rpms and temperature of
char bed, throat, and flue gas were then
Fresh air recorded for every 5 second. Plant model NN
was trained using recorded values using
Figure 4 - Swirl Burner Matlab time sires tool.
4
146 ENGINEER
Best fitted NN for plant model is shown in D is delay of inputs and outputs. Similarly, b_x
Figure 6. C is the gasification plant inverse is base value of layers
model. It was also trained by the same data set
while and are and weights of network
which was used in Matbab time sires tool. Best
respectively. The developed NNIMC was
fitted NN for plant inverse model is shown in
implemented on LabView graphical interface.
Figure 7.
Developed NNIMC is shown in Figure 8. There,

f1 denotes logtan function and f_2 represents
linear function.
Figure 6 - Neural Network Plant Model (M)
Figure 7 - Neural Network Inverse Model (C)
Figure 8 - Developed NNIMC for Gasifire Plant
5
147 ENGINEER
3. Results and Discussion According to the above result, the overall
improvement of energy output is 12.60%.
Two experiments were arranged to investigate
the performance of proposed IMC system for
biomass downdraft gasifire. In first experiment, 25
gasifire was operated with the proposed IMC
system using 12kg of coconut shells. Set value
Power (kW)
20
was fixed during the operation as the dash line
in the Figure 9. Step response was provided at
t = 306, and t = 465 to controller within the 15
range of set value from 960 K to 980K.
According to the Figure 9, flue gas temperature
10
follows the set value and it levels off all over
64
1
127
190
253
316
379
442
505
568
631
694
757
820
Time (in 5 seconds)
1070 Power (without controller)
1020
Power (with IMC)
Temperature (K)
970
920 Figure 10 - The graph of Power Output vs.
870 Time
820
770 Figure 11 and 12 show temperatures at different
720 locations; char and throat, when gasifier was
670 operated with and without proposed internal
620
729
105
157
209
261
313
365
417
469
521
573
625
677
781
833
1
53
1200
Time (in 5 seconds)
Temperature (K)
1150
Flue Gas temperature (Without Controler)
1100
Flue gas temperature (With IMC)
1050
Set valuve
1000
950
Figure 9 - The Graph of Flue Gas 900
Temperature vs. Time
827
119
178
237
296
355
414
473
532
591
650
709
768
1
60
the operation as indicated in orange line. Time (in 5 seconds)

Secondly, gasifire was operated at constant T throat no C T throat C
blower speed of 3260 rpm for 12kg of coconut
shells. Black colour continuous line represents
Figure 11 - Temperature Variations at Throat
the flue gas temperature at burner and it is not
stable in the second experiment. Also, ignition
was disturbed and flame did not appear model control. According to the temperature
particularly at the lowest temperature points.
variations, introduced IMC system operates the
Figure 10 displays the variation of power gasifire at higher process temperatures. Higher
output for both experiments. Flue gas flow rate gasification temperatures reflect a better
was remained at a constant value of 0.029kg/s, oxidation, low tar production and higher
therefore, variation of power output is calorific value of syngas [11].
proportionate to the flue gas temperature
changes. Higher gasification temperatures provide better
formation rate of CO, CH4 and H2 which lead to
As shown in Figure 10, total energy output of higher syngas heating value which is shown in
two experiments (With IMC and Without IMC) Figure 09. When introduced IMC system
were 83201.57kJ and 73878.74kJ respectively. operates gasifire, flue gas temperature of the
6
148 ENGINEER
burner is higher than without IMC. There are In this research, the neural network based IMC
high temperature impulses indicated by ash has been trained only for coconut shells.
Therefore, this plant control may not applicable
1200 for other biomass types. As future directions,
network should be developed for different
Temperature (K)
1150 biomass with different physical properties.

Moreover, gasification plant with internal
1100
combustion engine is one of very popular
1050 application. However, composition of
gasification produced gas is not constant. This
1000 issue disturbs for smooth running of internal
950
combustion engine because of the ignition
preparation of gas component is different, for
900 an example, ignition propagation of H2 is high
and ignition propagation of CO is very low
111
166
221
276
331
386
441
496
551
606
661
716
771
826
1
56
Time (in 5 seconds) than H2. Therefore, advanced predictive

controller will be an effective solution for
T chr no C T chr C
smooth governor control
Figure 12 - Temperature Variation at Char

References
colour line in Figure 11 at around t = 309 and 1. Sri Lanka Sustainable Energy Authority „‟Sri
450. Those are resulted by the stuck char Lanka Energy Balance 2007 and Analysis of
particles at throat. When the pressure Energy Sector Performance‟‟, Chapter 5, 2017.
difference increased by them across the throat,
it causes a rapid high air flow and consequently 2. Johansson, T. B.; Kelly, H.; Reddy, A. K. N.;
Williams, R.H. Renewables for fuels and
flame at combustion area goes to char burning electricity, 1992, UNCED.
zone and throat.
3. Basu, P., Biomass Gasification and Pyrolysis, 1st
4. Conclusions Ed.
Internal model based control architecture was 4. Milieuonderzoek, K. et al, „‟ Review of

used for developing plant controller because of applications of gases from biomass gasification-
high process nonlinearity of biomass ECN Biomass‟‟, 2006.
gasification process. Plant inverse model was
also developed by using neural network. And 5. Available: http://biofuelsacademy.org/web-
modules/process/gasification/cross-draft-
developed biomass gasification plant model
gasification/ accessed on 22/11/2016.
was then used as the internal model. The neural
network technics can be easily combined with 6. Fournel, S., Palacios, J. H., Morrissette, R.,
IMC control architecture. Furthermore, Villeneuve, J., Godbout, S., Heitz, M., and Savoie,
according to the results, neural network can be „‟ Influence of Biomass Properties on Technical
effectively used to model and control and Environmental Performance of a Multi-Fuel
nonlinearity system. Boiler during on-Farm Combustion of Energy
Crops‟‟, Applied energy, 141, 2015, pp. 247-259.
Figure 9,10show that the syngas production
7. Budhathoki, R., „‟Three Zone Modeling of
stability is improved by IMC all over the plant
Downdraft Biomass Gasification: Equilibrium
operation. At the same time, IMC increases
and Finite Kinetic Approach„‟, 2013.
power output of the gasification. However, the
plant output takes more time to reach to the set 8. Patra, T. K., and Sheth, P. N., ‟‟Biomass
value because gasification is time delaying Gasification Models for Downdraft Gasifier: A
process. State-of-the-Art Review‟‟, Renewable and
Sustainable Energy Reviews, 50, 2015, pp. 583–
During the development of plant model, it 593.
revealed the importance of grate shaking effect.
9. Ghassemi, H., and Shahsavan, R., - Markadeh,
However, the effect of grate shaking could not
„‟Effects of Various Operational Parameters on
be considered for implementing the control
Biomass Gasification Process; A Modified
because of a technical problem i.e., burnout of Equilibrium Model‟‟, Energy Conversion and
the grate shaking motor. Management, 79, 2014, pp. 18–24.
7
149 ENGINEER
10. Available: http://www.bios- 22. Hunt, K. J., and Sbarbaro, D., „‟Neural Networks
bioenergy.at/en/electricity-from for Nonlinear Internal Model Control‟‟, IEE Proc.-
biomass/biomass-gasification.html accessed on D, 138(5), 1991.
22/ 11/2016.
11. Mendiburu, A. Z. et al, Thermochemical

Equilibrium Modeling of a Biomass Downdraft
Gasifier: Constrained and Unconstrained Non-
Stoichiometric Models, Energy xxx, 2014, pp. 1-
14.
12. Mendiburu, A. Z., et al, „‟Thermochemical

Equilibrium Modeling of Biomass Downdraft
Gasifier: Stoichiometric Models‟‟, Energy 66,
2014, pp.189 – 201.
13. Yucel, O., and Hastaoglu, M.A., „‟Kinetic

Modeling and Simulation of Throated Downdraft
Gasifier‟‟ Fuel Processing Technology, 144, 2016,
pp.145–154.
14. Bing Guo et al, „‟Simulation of Biomass

Gasification with a Hybrid Neural Network
Model‟‟ Bio Resource Technology, 76, 2001, pp.
77-83.
15. Puig-Arnavat, M. et al, „‟Artificial Neural

Network Models for Biomass Gasification in
Fluidized Bed Gasifiers‟‟, Biomass and bioenergy,
49, 2013, pp. 279-289.
16. Ahmed, T. Y. et al, „‟Mathematical and

Computational Approaches for Design of
Biomass Gasification for Hydrogen Production: A
Review‟‟ Renewable and Sustainable Energy
Reviews, 16, 2012, pp. 2304– 2315.
17. Robert Mikulandric et al, „‟Artificial Neural

Network Modelling Approach for a Biomass
Gasification Process in Fixed Bed Gasifiers‟‟
Energy Conversion and Management, xxx, 2014,
pp. xxx–xxx.
18. Piloto-Rodríguez, R. et al, „‟Prediction of the

Cetane Number of Biodiesel using Artificial
Neural Networks and Multiple Linear
Regression‟‟ Energy Conversion and
Management, 65, 2013, pp. 255–261.
19. Mohanty, S., „‟Artificial Neural Network Based

System Identification and Model Predictive
Control of a Flotation Column‟‟, Journal of
Process Control, 19, 2009, pp. 991–999.
20. Macmurray, J. C., and Himmelbealj, D. M.,

„‟Modeling and Control of A Packed Distillation
Column Using Artificial Neural Networks‟‟,
Computers chem. Lllgng, 19(10) 1995, pp.1077-
1088.
21. Mikulandric et al, „‟Process Performance

Improvement in a Co-Current, Fixed Bed Biomass
Gasification Facility by Control System
Modifications‟‟ Energy Conversion and
Management, 104, 2015, pp. 135–146.
8
150 ENGINEER
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[151 159],2017
2017
© The
© The Institution
Lanka
Organic Rankine Cycle (ORC): Performance of

Working Fluids and Energy Recovery Potential in Sri
Lankan Thermal Power Plants
B.S.R. Fernando, R.A.C.P. Ranasinghe and H.K.G. Punchihewa
Abstract: ORC based power generation is becoming popular as a way of generating electricity
from low-grade heat sources such waste heat. Working fluid selection and system optimization based
on heat source temperature are two critical aspects of ORC design. In this work, eleven fluids
comprised of hydrocarbons and refrigerants were theoretically investigated to maximize the work
output for a range of source temperatures. Results show that Heptane, Pentane and Decane show
favourable results in terms of work outputs while, in terms of efficiency, Decane and Heptane are
better. Further, it was found that Pentane outperforms, when source temperatures are between 45 –
190 0C, while Heptane for 190 – 260 0C. Decane is more suitable for 260 – 340 0C range. Based on the
theoretical analysis, a new summarized graphical chart was developed for Pentane, Heptane and
Decane, where one point on the graph can denote approximate work output, efficiency, pressure,
temperature and other required data for the initial design process and fluid selection of an ORC plant.
Subsequently, the economic feasibility of ORC was assessed considering WH data of all the thermal
plants of in Sri Lanka. Possible electric power outputs were computed for each selected plant, for
selected fluids from the above theoretical analysis. Then, maximum work out of each case was
selected for further economic evaluation under seven different scenarios, which represents the future
economic situation in the country. Investment cost was estimated pertaining to the maximum work
output and payback time was estimated to evaluate the investment feasibility. Interestingly, results
show that some of the high volume power stations are very good candidate for ORC, which has very
short payback time even with the worst possible economic situations considered.
Keywords: ORC, Thermal power performance, Thermal Energy recovery
1. Introduction fluid [14-17]. Therefore, the present study aims

to investigate the technical and economic
Thermal energy is still the major source for potential for low-grade waste heat recovery
power generation, worldwide. Efficiencies of from thermal power plants in Sri Lanka, using
most of the thermal plants are less than 50% [1]. ORC with different working fluids.
Consequently, the total emission of CO2 due to
fossil fuel combustion was estimated to be 2. Organic Rankine Cycle
32,381 Mt in 2014 [4], and many environmental
issues related to fossil fuel combustion have
Expander
also been identified [2, 3]. Dumping of low- 1
grade waste heat to the environment is one of

the key issues of such power plants, yet Evaporator
inevitable. One of the solutions would be

2
harnessing back waste heat, but presently, [5,6]
the available technologies are less cost-effective 4
for investment [7,8] unless there is a significant Condenser
volume of heat for recovery. Pump

3
A number of low-grade heat recovery cycles [9]

Figure 1- Basic configuration of ORC system
have been developed: Organic Rankine Cycle
(ORC), Kalina Cycle [10], Goswami Cycle Eng. B. S. R. Fernando, B.Sc. Eng. (Hons), M.Eng.
(Energy), CEng (UK), AMIE (SL).
[11,12], and Trilateral Flash Cycle [13]. Out of
these, one of the promising technology for Dr. R. A. C. P. Ranasinghe, Department of Mechanical
conversion of low-grade heat is the Organic Engineering, University of Moratuwa.
Rankine Cycle “ORC”, where principle Dr. H. K. G. Punchihewa, Department of Mechanical
operation is similar to the conventional Rankine Engineering, University of Moratuwa.
Cycle, but, with a different organic working
1
151 ENGINEER
ORC has become a field of intense research and When the line (1-2) slope is positive (i.e. point 2
appears as a promising technology for lies outside the vapour region), droplet
conversion of low-grade heat into electricity formation in the expansion is avoided and
[18]. It is exactly like the steam Rankine cycle, hence called dry fluids. In negative slope, the
except that the working fluid in the system will droplet will be formed in expansion and called
be a refrigerant or a hydrocarbon [19]. wet. Normally, the formation of droplets during
the expansion is undesirable for turbine
operation. Further, the fluids having an infinite
slope (1-2 vertical) is considered as isentropic
fluid as the entropy remains unchanged during
the expansion.
2.2 Superheating and critical point of working

fluid
Superheating an ORC can increase the thermal

efficiency, but more significantly decreases the
second law efficiency [20]. Further, overheating
increases the cycle pressure which increases the
Figure 2 - Basic T-s diagram for ORC system
investment costs of the system. Hence, it is not
recommended [24] unless to gain more power
Figure 1 shows the major components and their
at the expense of losing efficiency [20, 25].
connections in a basic ORC configuration,
Therefore, all the fluids considered in the
consisting of an expander, condenser,
present work have been analysed on saturated
evaporator (heat recovery unit) and a working
state (prior to expansion) [23].
fluid pump [20]. The cycle consists of four
thermodynamic processes shown in the T-s
Closer to the critical condition of the working
diagram in Figure 2. In process 1 – 2, organic
fluid, sudden abnormal changes of mass flow
fluid passes through the expander to generate
rate can happen. These changes near critical
mechanical power. Ideally, this process should
point may cause instability in the system [26].
be an isentropic process, but in the actual case,
However, performance near critical is
isentropic efficiency (η_is) is below 100%.
favourable than supercritical [26], hence, during
Exhaust organic vapour from the expander is
the present analysis, the maximum possible
fed into the condenser where it is cooled to
pressure for a given fluid was considered to be
liquid in isobaric process 2 – 3. In process 3 – 4,
4 bar less than the fluid’s critical pressure.
condensed liquid is pressurized and sent to the
evaporator. In the evaporator, heat absorption
takes place from the heat source denoted as 3. Thermodynamic Modelling of
process 4 – 1. During the heat absorption, liquid ORC
fluid is transferred to saturated or superheated
vapour phase in the outlet [20]. In practical Based on the work of [21, 27, 28], the ORC is
ORCs heat required for the process, 4 – 1 can be assumed to operate in steady, stable conditions.
obtained from waste heat sources such as For the calculated figures to be more realistic,
exhaust flue gas, jacket or condenser cooling the isentropic efficiency for expansion process 1
water, lubrication oil cooler, blowdown of – 2 is assumed as 75 % (η_is = 0.75) [29]. Note
steam boilers or by direct burning of fuels. that expander efficiencies vary between 70-85%
in practice. In addition, it is assumed that the
2.1 Wet, Isentropic and Dry working fluids mechanical efficiencies of fluid pump, expander
and generator are 75% [29], 96% and 98%
The operating costs of the ORC system are respectively. It is further assumed that feed
strictly linked to the thermodynamic properties pump work is isentropic. For simplification,
of the working fluid. Selection of working fluid pressure drops across pipeline, heat exchanger
is the most critical factor of ORC for efficient and condenser are assumed negligible for all
and economical operation [21]. Further, operating conditions. For calculation and
working fluids can be divided as dry, wet and comparison purposes, mass flow rate of the
isentropic, based on the expansion curve organic fluid is taken from 1.0 kg/s for all
gradient (1-2) on T-s diagram [22]. fluids in all conditions. The modelling
equations used in the present work are
ENGINEER 2
152
summarized below. Further information and ̇ ̅ ( )
… (7)
derivation procedures can be found in [20, 29]. ̇ ̅ ( )
The expander work output W _exp for a Where is mean average temperature
mass flow rate m of the working fluid is
at evaporator outlet, is mean average
expressed as;
temperature at evaporator inlet, average
̇ ̇ ( ) source temperature, is evaporator inlet
…(1)
temperature, ̅ denotes the specific heat
capacity of the working fluid and ̇ denotes
Where, and are the enthalpies at the state the fluid mass flow rate.
points 1 and 2 respectively in the T-s diagram. The
isentropic efficiency of the expander is related to the
actual power output ̇ and the isentropic power 4. Fluid Analysis
output ̇ as given by equation 2.
The fluid analysis here was performed by
varying the evaporator temperature.
̇ ⁄ ̇ …(2) Evaporator temperature denotes the exhaust
temperature of the heat source. Hence,
Heat absorption rate of the working fluid at the evaluating the temperature variation will
evaporator ̇ can be calculated from represent the ORC performance under different
equation 3 given below; source temperatures [35]. This will easily
provide maximum effective ranges of each fluid
̇ ̇ ( ) …(3) with optimum evaporator temperatures. These
factors are critical when designing ORC for heat
recovery. Also, this analysis provides a
reference for selection of fluids based on heat
̇ denotes the power requirement of
source temperature.
the pump, which is less than 100% practically,
as part of the energy that goes to raise the 4.1 Fluid Selection
temperature of outlet fluid. Hence, pump actual In the fluid selection, 11 fluids which are
pump requirement can be defined as; commonly recommended in the literature for
ORCs were chosen [21], where 06 of them are
̇ ( ) organic fluids, and other 05 are refrigerants.
̇ .(4)
Table 1 & 2, show the chemical formulae,
physical properties and few other important
characteristics of selected fluids.
The overall cycle efficiency can be calculated
from total energy input (Heat energy absorbed
Table 1 - Fluids Molecular Formula and
by the evaporator & pump work) and work
Physical Data
output from the expander.
̇ ̇ ̇ Molecula Physical Data [30]

..(5)
N r NBP(o Tc(o Pc(ba
Fluid
o Formula C) C) r)
̇ 1 R 245fa C3H5F5 14.9 154.1 36.4
̇ 183.6
…(6) 2 R 123 C2HCl2F3 27.82 36.62
̇ 8
Toluen
̇ ̇ 3 C6H5-CH3 110.6 318.6 41.26
e
214.0
4 R 113 C2Cl3F3 47.58 33.92
Amount of heat recovered from the waste heat 6
available depends on the efficiency of the Benzen
5 C6H6 80.08 288.9 48.94
e
evaporator , which is defined as follows; R 174.4
6 C3H3F5 25.13 39.25
245ca 2
̇ 344.5
7 Decane C10H22 174.12 21.03
5
̇
8 R 134a CH2FCF3 -26.07 101.0 40.59
3
153 ENGINEER
6 of evaporator were maintained within
ClF2CCF2 145.6 saturated fluid range by controlling the heat
9 R 114 3.586 32.57 input and fluid flow rate. Further, expander
Cl 8
Pentan outlet also maintained at the saturated state and
10 C5H12 36.06 196.6 33.7
e then fed to the condenser for further cooling.
Hepta 266.9
11 C7H16 98.38 27.36
ne 8 4.3 Results & Discussion
Power output and efficiency plot for each and

Table 2 - Physical, Safety and Environmental every fluid was done with varying the
Data of selected fluids evaporator temperature. Figure 03, shows the
plotted graph for fluid Decane, by varying
Safety expander power against evaporator
Environmental Data
Data temperature. The plot further shows the power
[31,32,33] Expan
[36]
No Fluid sion in & out a variation with the temperature from
GWP
ASHRA Atm OD [30,34] 180 oC to 340 oC.
(100
E Life P
yrs)
1 R 245fa B1 7.7 0 1050 Dry
2 R 123 B1 1.3 0.01 77 Dry
3 Toluene - 2.4 - 2.7 Dry
4 R 113 A1 85 0.85 6130 Dry
5 Benzene - 12.5 - - Dry
6 R 245ca 6.5 0 726 Dry
7 Decane - - - - Dry
8 R 134a A1 14 0 1430 Dry
9 R 114 A1 190 0.58 9180 Dry
10 Pentane A3 0.009 0 20 Dry
Figure 3 - Variation of Expander work with
11 Heptane - - - - Dry evaporator temperature variation for Decane
(Note: Tcon = 180 oC, Pcon = 1.1739 bar, ṁ = 1 kg/s and
Note: Tc – Temperature at critical point other parameter fixed including ƞ pump = 0.75, ƞ exp. =
Pc – Pressure at critical point 0.96 & ƞ elec. = 0.98)
NBP – Temperature at normal boiling point (pressure
at 1 atm) Efficiency variation also plotted for each fluid
ODP – Ozone depletion potential
GWP – Global warming potential
accordingly, and Figure 4, shows the sample
plot for Decane. When evaporator temperature
4.2 Evaporator Temperature Variation close to 340 oC, cycle efficiency is around 40%.
The following were also some

criteria/conditions used in addition to the
assumptions stated previously. The condenser
temperature was assumed to maintain at 45 oC
which is normally 15 oC higher than the
atmospheric temperature in a tropical climate.
However, some of the fluids are in gaseous
phase at 45 oC, at condenser pressure. Hence, in
such cases, the condenser temperature was
suitably adjusted to meet the temperature
requirements. Further, upon the selection of
temperature at the evaporator, corresponding Figure 4 - Variation of Expander work &
pressure values were fixed. Condenser pressure efficiency with evaporator temperature
was maintained higher than the atmospheric variation for Decane
level to avoid air mixing in a possible leakage. (Note: Tcon = 180 oC, Pcon = 1.1739 bar, ṁ = 1 kg/s and
The output of the evaporator was maintained other parameter fixed including ƞ pump = 0.75, ƞ exp. =
0.96 & ƞ elec. = 0.98)
as saturated vapour and is the input to the
expander. The outlet temperature and pressure
ENGINEER 4
154
Combining individual plots for each fluid, the same plot against the output on the Y axis
complete graphs for power output and and, equal temperature lines were drawn.
efficiencies of the ORC fluids were developed Further, system efficiencies of each point were
and compared as shown in figure 5 & 6 obtained, and iso-efficiency lines were drawn.
respectively. In the calculations of above points, unit mass
flow rate (1 kg/s) of ORC fluid has been
considered. One point on the developed
summarized charts will denote pressure,
temperature, work output and efficiency at a
given evaporator and condenser temperature
for a given fluid.
5. Economic Analysis
Economic feasibility of potential waste heat

opportunities of thermal plants in the country
was analysed. Then, plant capacities were
decided according to available source, and
investment costs have been calculated. Total
investments with payback were compared
Figure 5 - Work output variation with considering seven possible scenarios governed
different evaporator temperatures by external factors of the economy for better
(Note: Tevap.min = 45 oC, Tevap.max = 340 oC, ƞ pump =
0.75, ƞ exp. = 0.96 & ƞ elec. = 0.98) prediction.
5.1 Investment cost of ORC
Based on some worldwide reputed

manufacturers of ORC systems, investment
costs were derived for each opportunity. Those
manufacturers have given some basic
guidelines for cost estimation with required
source temperatures and cost per kW. The cost
of the organic fluid is small compared to
component cost, hence working fluid does not
have considerable impact. Further,
manufacturers have done the cost estimation
including the fluid costs.
Based on the evaluation, recommended fluids
Figure 6 - Cycle efficiency variation with for individual plants and their economics of
different evaporator temperatures output were shown in Table 3.
(Note: Tevap.min = 45 oC, Tevap.max = 340 oC, ƞ pump =
0.75, ƞ exp. = 0.96 & ƞ elec. = 0.98) Table 3 - Estimated plant capacity &
investment
Results show that Heptane, Pentane and
Decane show favourable results in terms of Organic Expected Estima Invest Total
work outputs (figure 5) while, in terms of Fluid Electric ted ment Invest
efficiency, Decane and Heptane are better
Plant
Output Plant Cost ment

(kW) Capaci (USD) (mn
(figure 6). Further, it was found that Pentane
ty per USD)
outperforms, when waste heat source (kW) kW
temperature is between 45 – 190 oC, while for Decane 769.07 750 2200 1.65
SPS
Heptane the temperature should be 190 – 260
oC. Decane is more suitable for 260 – 340 oC LVJ Pentane 669.52 650 2200 1.43
range. JDP Decane 27.39 25 2500 .06
Decane 11667.9 11600 1800 20.88
In the development of summarized charts for KLP
7
those fluids, evaporator inlet temperatures were Decane 2335.20 2300 1900 4.37
KGT
varied to obtain theoretical work output at
different points. These points were marked on LBD Heptane 103.32 100 2500 .25
5
155 ENGINEER
Most of the country’s thermal power plants
Note: SPS - Sapugaskanda Power Station would generate power, more than 60% during
LVJ - Lakvijaya Coal Plant
the year. Especially, Lakvijaya coal plant is
JDP - Jaffna Diesel Plant
KLP - Kelevarapitiya Plant expected to run more than 80% during the year
KGT - Kelanithissa GT Plant to minimize power purchasing from private
LBD - Lakvijaya Boiler Blowdown owned plants. For calculations, annual running
hours were estimated as 60%, which means
Based on manufacturer’s criteria [31], following 5256 hrs per year. Further, it has been assumed
cost factors were defined and investment costs that operation pattern would continue
of each WH opportunity are summarized in unabated for the next 07 years.
table 3. Fluctuation of electricity prices and an inflation
Plants below 250kW - USD 2500 per kW rate of the country have made the financial
Plants above 250kW - USD 2200 per kW analysis part more critical and complex. The
Plants above 1MW - USD 1800 - 2000 per impact of different economic situations arising
kW out of unstable economic conditions referred to
as scenarios, on the investment has been
5.2 NPV Calculation considered in evaluating the project. Hence,
considering possible future occurrences, seven
Here, cash inflows are the cash generated from different scenarios are defined based on
the investment and cash outflows are the average selling price of electricity unit, future
expenditures including investment. Hence, net economic condition and bank interest.
cash flow is the difference of cash in and Investments are evaluated with reference to
outflow. NPV relationship between net cash those scenarios, to estimate the financial
flow and the investment can be determined viability of project implementation.
using the following equation;
Table 4 - Different scenarios on which NPV
∑ ( ) …(8) calculations were done for project feasibility
Where;
NPV - Net present value Practical
Average Selling
Bank Interest
B - Cash inflow or benefit Situations
Price of Unit
Rate %
(Rs. Per kWh)
C - Cash outflow or investment
Scenario 1 14.00 8%
A - Discount rate
Scenario 2 15.00 8%
Scenario 3 15.40 8%
The discount rate can be explained as the
Scenario 4 14.00 10%
interest rate used in discounted cash flow
Scenario 5 15.00 10%
(DCF) analysis to determine the present value Scenario 6 14.00 12%
of future cash flows. The discount rate in DCF Scenario 7 Refer table 5
analysis takes into account not just the time
value of money, but also the risk or uncertainty Table 5 - Plant running hours and interest rate
of future cash flows. Hence, greater the for scenario 7
uncertainty of future cash flows, the higher the
discount rate [32,33]. Averaged Exp. Average
Waste Heat
Annual Running Unit Interest
Recovery
The discount rate can be calculated as follows; Opportunity
Running Hours per Selling Rate
% Year Price Rs.
Sapugaskanda
…(9) Power Station 60 % 5256 15.00 10%
( )
Lakvijaya Coal
Where; Plant 70 % 6132 15.00 10%
i - Interest rate Jaffna Diesel
p - Period or years Plant 60 % 5256 15.00 10%
Keravalapitiya
Approximate capital investment for selected Plant 60 % 5256 15.00 10%
waste heat sources was estimated based on Kelanithissa GT
Plant 40 % 3504 15.00 10%
their potential output calculated previously. Lakvijaya Boiler
Further, suitable organic fluid also was selected Blow Down 70 % 6132 15.00 10%
by referring to the previous calculations. Scenario 7 in Table 5 represents the plants
average actual running hours for last few years
(till 2015). Further, average actual selling price
ENGINEER 6
156
of a unit is Rs. 15.40, but for scenario 7, unit along their respective regions and can be
price was taken as Rs. 15.00. Because, if the recommended for further study in WHR
evaluation is favourable even with low selling systems. Sample summarized graphs were
price, it will eventually ensure the profitability developed for above fluids as a guideline for
during implementation. Additionally, present interested parties. The developed summarized
bank interest rates are low as 6 – 8 % (in 2015), graphs provide approximate power output
which is not a normal situation in the country. against source temperature in a convenient
Hence, the average interest rate for next seven manner for the end user.
years has been taken as 10%. Meanwhile,
scenario 7 represent current situation of the Economic evaluation of the implementation of
country. Further, all overhead costs such as ORC system was done according to net present
operation and maintenance, spare parts, labour, value method. Expected plant capacities were
etc., have been estimated as 1% of the total decided upon maximum power output, and
investment for next 07 years. relevant investment costs were estimated
accordingly. Based on NPV method, recovery
5.3 NPV Results of investment was estimated under different
economic situation for 07 years. With reference
Calculations were done accordingly to to evaluation, favourable results are shown for
investigate the feasibility of ORC systems on ORC implementation. Detailed financial
selected heat sources, based on different analysis is further recommended.
scenarios. Net cash flows were estimated after
07 years and tabulated in Table 6, where
negative cash flows are shown as N and Acknowledgement
positive cash flows shown as P.
Author wishes to acknowledge the assistance
Table 6 - NPV results of WHR opportunities given by the Ceylon Electricity Board
authorities to provide required information.
Scenarios
WHR Opportunity
1 2 3 4 5 6 7
Sapugaskanda References
P P P P P P P
Power Station
P P P P P P P 1. Hung, T. C., Shai, T. Y., Wang, S. K., (1997, July).
Lakvijaya Coal Plant
A Review of Organic Rankine Cycle (ORC’s) for
Jaffna Diesel Plant P P P N P N P the Recovery of Low-Grade Waste Heat. Energy
Kelevarapitiya Plant P P P P P P P (issue 7) [Online] 22, Pp. 661-667.
Kelanithissa GT
P P P P P P N 2. Qiu, G., (2012, July). Selection of working fluids
Plant
for micro-CHP systems with ORC, Renewable
Lakvijaya
P P P N P N P Energy [Online] 48, pp. 565–570.
Blowdown
3. Steinfeld, J. I., (2001, June). Climate Change and
Energy Options: Decision Making in the Midst of
6. Conclusion Uncertainty. Fuel Processing Technology
[Online]. 71(issue 1-3), pp. 121-129.
This paper presents the performance of
different ORC working fluids under varying 4. International Energy Agency. Key World Energy
Statistics 2016.
evaporator temperature. The evaluation was
done in the subcritical region for all fluids 5. Smouse, S., Staats, G., Rao, S. N., Goldman, R.,
considering waste heat recovery applications. Hess, D., (1998, March). Promotion of Biomass
Evaporator temperature was varied from 45 - Cogeneration with Power Export in the Indian
340 oC and upon their output, the range was Sugar Industry. Fuel Processing Technology
divided into 03. The region between 45 – 190 (issue 1-3) [Online] 54, pp. 227-247.
oC, fluid Pentane has shown higher work
output than any other fluids while the region of 6. Tillman, D. A., (1982, January). Cogeneration
190 - 260 oC, fluid Heptane has shown the with Wood Fuels: A Review. Fuel Processing
Technology (issue 3-4) [Online] 5, pp. 169-181.
highest work. Further, temperature between
260 - 340 oC, fluid Decane has shown the 7. El Wakil, M. M., Power Plant Technology, 2nd
highest work. Significantly, these fluids have International Edition. New Delhi, India: TATA
shown predominantly higher work outputs McGraw-Hill, 1988.
7
157 ENGINEER
8. Sjoding, D., (2010, September). Overview of 20. Velez, F., Segovia, J.J., Martin, M.C., Antolin, G.,
Waste Heat Recovery for Power & Heat. U.S. F. Chejne, and Quijano, A., (2012, November)
Department of Energy. Chicago, U.S.A. Comparative Study of Working Fluids for a
Rankine Cycle Operation at Low Temperature.
9. Convert Low Grade Heat into Electric Power; Fuel Processing Technology. [Online]. 103. pp. 71-
Thermodynamic Cycles for the Conversion. 77.
10. Deepak, K., and Gupta, A.V.S.S.K.S., (2014, June) 21. Quoilin, S., Declaye, S., Tchanche, B. F., and
Thermal Performance of Geothermal Power Plant Lemort, V., (2011, October) Thermo-economic
with Kalina Cycle System. International Journal optimization of waste heat recovery Organic
of Thermal Technologies. 4(2). Rankine Cycles. Applied Thermal Engineering.
[Online]. 31 (issue 14-15), pp. 2885-2893.
11. Chen, H., Goswami, D. Y., Stefanakos, E. K.,(2010,
June). A Review of Thermodynamic Cycles and 22. Badr, O., Probert, S. D., and O'Callaghan, P.W.,
Working Fluids for the Conversion of Low-Grade (1985). Selecting a Working Fluid for a Rankine
Heat. Renewable and Sustainable Energy Cycle Engine. Applied Energy. [Online]. 21(issue
Reviews. [Online] 14 (issue 9), pp. 3059–3067. 1). pp. 1–42.
12. Vijayaraghavan, S., and Goswami, D. Y., (2006, 23. Bândean, D. C., Smolen, S., and Cieslinski, J. T.,
July). A Combined Power and Cooling Cycle “Working Fluid Selection for Organic Rankine
Modified to Improve Resource Utilization Cycle Applied to Heat Recovery Systems,” in
Efficiency using a Distillation Stage. Energy. World Renewable Congress, Sweden, 2011, pp.
[Online]. 31 (issue 8-9), pp. 1177-1196. 772-779.
13. Smith, I., Stosic, N., and Aldis, C., (1995). 24. Bombarda, P., Invernizzi, C. M., and Pietra, C.,
Trilateral Flash Cycle System: A high Efficiency (2010, February). Heat Recovery from Diesel
Power Plant for Liquid Resources, City Engines: A Thermodynamic Comparison
University, London – U.K., [Online]. pp. 2109- between Kalina and ORC Cycles. Applied
2114. Thermal Engineering. [Online]. 30 (issue 2-3), pp.
212–219.
14. Thorin, E., Power Cycles with Ammonia-Water
Mixtures as Working Fluid: Analysis of Different 25. Quoilin, S., “Sustainable Energy Conversion
Applications and the Influence of Thermo- Through the Use of Organic Rankine Cycles for
physical Properties, Ph.D. dissertation, Waste Heat Recovery and Solar Application,”
Department of Chemical Engineering and Ph.D. dissertation, Faculty of Applied Science,
Technology, Royal Institute of Technology, University of Liege, Belgium, 2011.
Stockholm, Sweden, 2000.
26. Pan, L., Wang, H., and Shi, W., (2012, January).
15. Matsuda, K., (2013). Low Heat Power Generation Performance Analysis in Near-Critical Conditions
System. Chemical Engineering Transactions. of Organic Rankine Cycle. Energy. [Online]. 37
[Online] 35, pp. 223-228 (issue 1), pp. 281-286.
16. Ferdinand, W. Preliminary Study of Binary 27. Tchanche, B. F., “Low Grade Heat Conversion
Power Plant Output Comparing ORC and Kalina into Power Using Small Scale Organic Rankine
for Low-Temperature Resources in Rusizi Valley, Cycle,” Ph.D. dissertation, Dept. Natural
Burundi, Ministry of Energy and Mines, United Resources and Agricultural Eng., University of
Nations University, Iceland, 2013. Athens, Greece, 2011.
17. Zamfirescu, C., and Dincer, I., (2008, October) 28. Jian, S., and Wenhua, L., (2011, August).
Thermodynamic Analysis of Novel Ammonia– Operation Optimization of An Organic Rankine
Water Trilateral Rankine cycle. Thermochimica Cycle (ORC) Heat Recovery Power Plant.
Acta. [Online]. 477(issue 1-2). pp. 7–15. Applied Thermal Engineering. [Online]. 31 (issue
11-12), pp. 2032-2041.
18. Peris, B., Esbri, J. N., Moles, F., and Babiloni,
A.M., “ORC Applications from Low Grade Heat 29. Khatita, M. A., Ahmed, T. S., Ashour, F. H., and
Sources,” in 3rd International Seminar on ORC Ismail, I. M., (2014, January). Power Generation
Power Systems, October 2015, pp. 12-14. using Waste Heat Recovery by Organic Rankine
Cycle in Oil and Gas Sector in Egypt: A Case
19. Saadatfar, B., Fakhrai, R., and Fransson, T., (2014) Study. Energy. [Online]. 64, pp. 462-472.
Thermodynamic Vapor Cycles for Converting
Low- to Medium-grade Heat to Power: A State- 30. National Institute of Standards and Technology,
of-the-art Review and Future Research Pathways. Chemistry Web Book: Thermodynamic Properties
Micro Journals. [Online]. 2 (issue 1). of Fluid Systems. U.S.A. Web
ENGINEER 8
158
31. IPCC, 2007: Climate Change 2007: The Physical
Science Basis. Contribution of Working Group I
to the Fourth Assessment Report of the
Intergovernmental Panel on Climate Change
[Solomon, S., D. Qin, M. Manning, Z. Chen, M.
Marquis, K.B. Averyt, M. Tignor and H.L. Miller
(eds.)]. Cambridge University Press, Cambridge,
United Kingdom and New York, NY, USA, 996
pp.
32. The Montreal Protocol on Substances that Deplete

the Ozone Layer, 9th Edition, UNEP, Printed and
bound in Kenya by UNON, 2012.
33. The Pollution Information Site; logged on

05.07.2016 at 15.30 hrs.
34. Rycroft, M., (2016). Organic Rankine Cycle and

Waste Heat Recovery. EE Publisher.
35. Korobistyne, M. A., “New & Advanced Energy

Conversion Technologies, Analysis of
Cogeneration, Combined and Integrated Cycles,”
Ph.D. dissertation, Energy Research Foundation
(ECN), Netherland, 1998.
9
159 ENGINEER
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[161 165],2017
2017
The Institution
© The InstitutionofofEngineers,
Lanka
Mechanized Fertilizer Applicator Operated by

Two-wheel Tractor for Coconut Plantation Industry in
Sri Lanka
S.A.P.S. Silva
Abstract: Fertilization of coconut trees is a time consuming and labour intensive process, which
has become a problem for coconut estate owners due to inadequate availability of human labour and
reliable technology for applying fertilizer. This innovative fertilizer applicator improves efficiency of
fertilization process, reduces the cost of fertilization process and increases the absorption of fertilizer
to plants. This applicator consists of a fertilizer-spreading mechanism, a feed hopper and a modified
rotavator. The modified rotavator facilitates loosening the soil surface and damages the feeding roots
of the palm tree. Average time taken for application of fertilizer for one palm tree is two minutes with
two labourers. According to the field investigation reports, it has been revived that overall saving of
cost for fertilization with applicator is about 35%. This paper describes design aspects of innovative
fertilizer applicator with performance details.
Keywords: Mechanized fertilizer applicator, Coconut, Fertilization, Two wheel tractor
1. Introduction At present, the fertility levels of the

most of the coconut growing soils have been
Coconut is one of the commercially declined; especially the coconut lands in the
important plantation crops in Sri Lanka. coconut triangle where the higher proportion of
Currently, the contribution of coconut sector to national yield come from. The inherent capacity
the Gross Domestic Product (GDP) is about to supply adequate nutrients to coconut palms
1.4%. It is also responsible for nearly 3.1% of the in degraded soils is very low or poor. Therefore
total foreign exchange earnings in Sri Lanka in nut production in such soils reported to be very
2009 [1]. Coconut is a tropical perennial low. Improper soil and nutrient management
plantation crop and the totals of 1 million acres with high exploitation leads to further soil
of coconut plantations are available in Sri Lanka. degradation, resulting poor nut production in
Generally in wet zone around 150 palms are the existing lands. Therefore proper soil and
harvested in one hectare due to its canopy nutrient management system should be adopted
structure requires a wide spacing between especially for the coconut, in long term as far as
palms. the nut yield and the sustainability is concerned
Currently production of coconut is [5].
going gradually decreased due to fragmentation Coconut palm being a perennial with a
of lands that has taken place extensively in the life span of 60 years or more requires a regular
recent past. Nevertheless, the potential per supply of plant nutrients to sustain its growth
capita availability of coconut has been reduced and yield in its production period. It is revealed
due to the fact that the production has not been that 45% of the variation in yield is influenced
increased sufficiently to meet the requirement of by nutrients. In coconut growing soils,
growing population [2]. As solutions to increase considerable amount of nitrogen (N)
the coconut yield, it is necessary to introduce phosphorus (P), potassium (K) and magnesium
higher yielding coconut varieties and also more (Mg) are depleted from the soil, as a result of
proper fertilization methods. Considering above continuous removal of Nut and other parts of
the proper fertilization contribute around 5% for the palm. [6].
increasing yield [3]. Nut yield was taken at two
month intervals and the total 120 nut production Eng. S.A.P.S.Silva, MEng.(Manufacturing Systems
per palm per year was considered for Engineering - Moratuwa), B.Sc.Eng (Mechanical ,First
Class Hons-Peradeniya), AMIE, Research Engineer, NERD
calculations [4]. From above it was found that
Centre.
yearly around 360Mn yield increase of coconut
nut production is possible in Sri Lanka.
1
161 ENGINEER
Traditionally the application of fertilizer wheel tractor is vital for evaluating the quantity
to coconut trees is a labour intensive and time- of the applied fertilizer to the palm. In two
consuming activity. In manual method, top soil wheel tractors, turning is done by actuating the
of the adjacent area of the plant is removed up clutches in each wheel. As an example,
to 6” depth (within 6’ feet diameter circle) and disengaging the left wheel clutch the tractor is
refilled with removed soil and other debris after turned to left [7].Maximum turning path is
application of fertilizer. The main disadvantages defined by the tractor manufacture and
of the manual method are, according to that this fertilizer applicator was
 Non-uniformity of fertilizer distribution designed. The PTO speed of a common two-
 Considerable wastages of fertilizer wheel tractor is around 80 rpm. The fertilizer is
 Low absorptivity due to cutting off of always filled with the hopper in operation.
feeding roots
 High cost involvement
2. Development of an innovative l1
fertilizer applicator
The innovative fertilizer applicator
shown in Figure 2.1 has been developed by
improving the efficiency of fertilization and
addressing the above-related issues. Basically
equipment is an attachment to the common two-
wheel tractor and it is powered by the power
transmission output (PTO) of the two-wheel
tractor. The rotavator in this attachment loosen
the surface soil with damaging the feeding roots
while spreading mechanism is applying
fertilizer. The above technique is assured the
d1
improvements of fertilizer absorption capacity.
Hopper Spreader
driving system
h1
Frame
Figure 2.2 - Schematic view of the fertilizer

PTO
spreader
Rotavator
driving system
Where h1 is the height of the flap, d1 is the
Rotavator
diameter of the cylinder, l1 is the length of the
Adjustable
rotor and γ is the speed of the rotor in rpm.
caster wheel
Volume rate of the fertilizer spread out of the
Figure 2.1 - Schematic conceptual design of a unit,
fertilizer applicator
{ }
2.1Analytical formulations in the design of
fertilizer spreader Weight of the fertilizer spread in unit time,
For analytical design formulations, it is
considered that the tractor speed and the PTO { }
(Power Transmission Output) speeds are
synchronized based on the tractor
manufacturer’s specifications while fertilizer is
applied. Further the turning radius of the two
ENGINEER 2
162
Palm tree
v a
Fertilizer
spread area
b Figure 3.1 - fertilizer applicator
Table 3.1 - Overall Features of the developed

Figure 2.3 - Schematic view of the tractor- machine and the results of the field testing
driving path
Where v is the tractor travelling speed, a is the Feature Value of the feature
inner radius of the fertilizer spreading area and Dimensions Width :550mm
b is the outer radius of the fertilizer spreading Length :700mm
Height :850mm
area.
Weight 80kg
Average radius drive around the palm, Width of the rotavator 450mm
Capacity of fertilizer hopper 40kg
[ ]
Fuel consumption depend on the tractor
Time taken for one palm, Application time max 2 min/tree
No of trees can apply inclusive of 200/day
travelling in estate
[ ]
Note: Above factors depend on the soil
Weight of fertilizer applied for one palm, condition, palm layout of the estate, Tractor
[ ] specifications etc.
{ }
4. Cost impact analysis
From above equation found the height of the Table 4.1 - Data for calculating the cost in year
flap (h1) and the diameter of the cylinder (d1). 2016
Skilled labour cost/day Rs.1500.00

Table 2.1 - values obtain from the design Un-skilled labour cost/day Rs.1250.00
calculations Labour Manual Un-skilled :03
Parameters Value requirements Developed Skilled :01
Inner radius of the fertilizer 0.6m machine Un-skilled :01
Diesel cost per liter Rs.95.00
spreading area (a)
Price of a Chinese brand new two Rs.350,000.00
Outer radius of the fertilizer 1.05m wheel tractor with attachment
spreading area (b)
Tractor travelling speed (v) 10m/min Note: Pre-mixed coconut fertilizer cost is used
Diameter of the cylinder (d1) 100mm and wear and tear is calculated for IMI 7 Two
Length of the rotor (l1) 450mm wheel tractor
Height of the flap (h1) 10mm
Density of the fertilizer ( 1725 kg/m3 Table 4.2 - Cost benefit analysis
Required speed of the rotor 3 rev/min Description Traditional Developed
in rpm (γ) method method
(Manual) (Fertilizer
Proposed travelling time 31s applicator)
around palm Fertilized trees per day 30 160
(8h shift)
Cost for fertilizer per 115 115
3. Fabrication of prototype tree (5kg) /LKR
Labour cost per 125 17
By considering the design concept andthe palm/LKR
formulation in section 2, and by considering the Diesel cost per - 12
palm/LKR
features mentioned in the below table, the
Wear and tear/LKR - 16
prototype of the mechanized fertilizer applicator Total cost per 240 160
as shown in figure 3.1 has been fabricated. palm/LKR
3163 ENGINEER
 Cost saving respect to the manual The lever system which connects and operates
the fertilizer ON/OFF chute was also more user
method = friend.
= 34%
6. Conclusions
 Labour saving respect to the manual The purpose of mechanization in coconut
cultivation is to produce more from existing
method =
land. Machinery is a complimentary input
= 86% required to achieve higher land productivity.
Additional benefits to the user may be
%
associated with a reduction in drudgery of farm
 Time saving respect to the manual work greater leisure, or reduction of risk.
method = Use of manpower for the application of fertilizer
is uneconomical due to high labor cost.
= 75% Therefore power tiller operated fertilizer
Normally fertilizer application is done in applicator for coconut cultivation was designed
twice per year and the palms per hectare are and constructed. This machine is not only useful
150 nos. to broadcast fertilizer on the soil but also to mix
So, it with the soil close to the palm on the weeded
 Minimum payback period for 15 hectare surface.
The designed machine consists of fertilizer
land = distribution unit and rotary unit. The Maximum
spreading width in 0.5 meter, Machine discharge
= 0.97 years
rate in 1.6 kg/min (3.3kg per palm) and
= 01year
Uniformity coefficient of spray distribution were
I.e. The equipment cost is cover within
considered as criteria for comparison of merits
one year for 15 hectare land with twice
and demerits. According to the field
application per year.
investigation reports, it has been revived that
overall saving of cost for fertilization with
applicator is about 35%.
5. Prototype’s field evaluation and
other advantages – The results
Acknowledgement
After completing the field trials in different soil
Express my warm thanks to Eng. D.D.A. Namal
conditions (different locations), each around 20
(Director General), Eng. G.G.K.A. De Silva
acre lands in Sri Lanka it was found that the
(Deputy Director General), Eng. K.Y.H.D.
following advantages,
Shantha (Director),Dr.A.Warahena (Principal
Research Engineer),Dr. (Mrs.) K.M.W. Rajawatta
 Applying fertilizer within two minutes
(Senior Scientist),Eng. (Mrs) P.M.Y.S. Pathiraja
time period.
(Senior Research Engineer) and,Mr. P.A.U.W.K.
 More than 35% of financial saving could Paranagampola (Engineer) for their support and
be achieved. guidance in National Engineering Research and
 Ability to fertilize at least 10 Nos. of Development Centre of Sri Lanka.
palms with one hopper load of fertilizer.
 Homogeneous spread of fertilizer around
the palm tree. References
 Ability of applying right quantity of
1. Annual report Central Bank of Sri Lanka, 2009
fertilizer. 2. http://www.statistics.gov.lk/agriculture/COP/
 Minimizing wastage of fertilizer. COP_Coconut.html
 As the machine damages the feeding roots 3. An Empirical Investigation on the Effect of Land
of the palm tree, absorption of fertilizer Size on its Productivity of Coconut Plantations in
becomes more efficient. Sri Lanka K. V. N. N. Jayalath1 , P. M. E. K.
 The center caster wheel which connects to Pathiraja1 , U. K. Jayasinghe-Mudalige2 and M.
T. N. Fernando3 1 Agricultural Economics&
the rotavator back side ergonomically
Agribusiness Division, Coconut Research
affects to the operator.
ENGINEER 4
164
Institute of Sri Lanka, Lunuwila, Sri Lanka, 61150
2 Dept. of Agribusiness Management, Faculty of
Agriculture & Plantation Management Wayamba
University of Sri Lanka, Makandura, Gonawila,
Sri Lanka (NWP) 3 United Nations Development
Project, Asia & Pacific Regional Centre, 23,
Independence Avenue, Colombo 7.
4. Abeywardana, V. (1971). Yield variations in

coconut. Coconut Research Institute of Sri Lanka.
Ceylon Coconut Quarterly, 22: 97-103. Bardhan,
P. K. (1973). Size, productivity and returns to
scale: An Analysis of farm-level data in Indian
agriculture. J. of Political Economy, 81(6): 1 370-1
386. Central Bank of Sri Lanka (2009).
5. http://www.cri.gov.lk/web/images/stories/statist
ics/satistics_on_coconut_oil_palm.pdf
6. Senarathne, S.H.S and Costa, M.J.I. (2009).

Comparison of several recommended cultural
practices for weed management and their effects
on yield of coconut in tropical coconut
plantations, Proc. 8th Workshop of the EWRS
Working Group: Physical and Cultural Weed
Control conf., Zaragoza, Spain, 17-23.
7. Peiris, K.&Adikarinayake,T.B. (2005).Design and

Analysis od a skidder Type Supporter for a
Combine Harvester.Engineer:Journal of the
Institution of Engineers,Sri lanka.38(p),pp7-12.
DOI:http://doi.org/10.4038/engineer.v38i2.72
10
5165 ENGINEER
Annual Sessions IESL,pp.
pp. [1 -- 10],
[167 171],2017
2017
© The
The Institution
Lanka
Water Loss Reduction by Understanding the

Reticulation A Case Study
S.G.G. Rajkumar
Abstract: Safe treated water is considered as an economic good, than a social good as scarcity of
water is felt in many parts of the world. The situation is made worse with increase in population and
climate change. Presently many parts of developing countries are provided with interrupted water
supply to share the resources amongst the entire citizen. Traditionally water for piped supplies was
obtained from perennial wells and water supply was limited to few households by the respective local
authorities. Presently the situation has changed and water authorities are supplying water for which
the costs have to be recovered. The consumers are used to paying for water and are conscious of the
amount they have to pay while they are aware of their consumption. Presently amount of water
consumed, wasted by leaks and unaccounted water are given priority. There are many steps taken to
address all those mentioned above from consumer’s level to operational level while illegal use is also
given due consideration.
Colombo City water supply is over hundred years old, the Non Revenue Water remained high for
many years. Many strategies were adopted; having understood the complexities of the reticulation,
water loss reduction through visual leak survey became a low cost approach and effective method to
make the change. The Non Revenue water reduced from 49% to 46% in Colombo City.
This paper describes the lessons learned to improve water utility efficiency by reduction of water
losses. Extensive work carried out in many planned distribution isolation within the city boundary
and on engineering judgments. While giving due consideration for holistic approach in dealing with
water losses, while considering every drop conserved would be useful for many who are deprived.
Keywords: Visual Leaks survey, Engineering Judgment, Low cost approach, lessons learned,
many steps
1. Introduction connections. Action was taken to identify the

causes of losses from part of the system, to
The National Water Supply and Drainage replicate the remedial measures to the whole
Board (NWSDB), is responsible for the supply distribution system.
of drinking water in Sri Lanka. Non Revenue
Water (NRW) has been a major issue in
2. Objective
Colombo City, where initial pipe lines were
 Detail study done in small area to
laid as far as in 1870’s resulting in aged pipes understand the longstanding NRW issue
in many parts of the city. The NRW was  Identify the causes in scientific manner
reported around 49% in Colombo City, which and
is higher than the nationwide average of 30% replicate the finding to whole of Colombo
set by NWSDB in 2010. In order to resolve this City.
issue, several steps such as leak detection/
repair, detection and regularization of illegal
connections, removal of public stand posts and
converting its users’ to be responsible for
individual connections and improvement to
billing was carried out. However, the NWSDB
still faces difficulties to reduce NRW to a
satisfactory level. Eng. S.G.G. Rajkumar, B.Sc.Eng(Hons), C.Eng., FIE(Sri
Lanka), M.Sc(Denmark), MBA(Sri.J), M.Eng (Moratuwa,
Presently, Deputy General Manager(Commercial), National
In view of resolving the NRW, a study area Water Supply & Drainage Board)
was selected. The area comprised of 5000
167
3. Methodology zones”[1] and “relatively high pressure”. Table 1
gives the details of the sub zone selection.
Colombo City has over one hundred thousand
connections. Non Revenue water has four Hydraulic isolation was planned based on
contributory factors, Leaks from the system, free available drawings. Some, valves were
water, administrative losses, unauthorized introduced to modify the existing distribution
consumption. Even though, many activities had system for hydraulic isolation with one or two
been done to arrest these four factors,NRW feedings [2]. as shown in figure 1.
remained high, for past 10 years. To identify the
main cause of losses indepth study was carried 3.1 Household survey
out by selecting a sample area in Colombo city.
The customers in the subzones were identified
with the available information and all the
The area was selected in such a way where the premises within the subzone were visited. All
pipe was over one hundred years old, and the the premises, the water bills was compared with
NRW was considered high. The selected area the water meter; the number of users were
had nearly 5000 connections, this was sub identified; the water meter accuracy was tested
divided into sub zones where each sub-zone has using a conical can; service leaks were checked
about 500 house connections, was carefully using acoustic principles[3], and bypass lines
selected paying attention to “less were checked.
interconnection with the outside of the sub-
Table 1 - Details on No of connections and pipe lengths
Sub-zone 1 2 3&4 5 6 7 9
No of
connections 397 426 1383 115 159 1545 200
Material PVC CI CI PVC CI CI PVC CI PVC CI DI PVC CI DI
3 61 963 319 872 148

4 407 993 3,168 63 38 1,536 262 2,153 1,589
5 300 414 119 314 165
6 692 737 171 925 114
7 724
8 173 212
9
10 1,449 1,034
12 771
Main Pipe Length (m)
16 298 290
Diameter (inch)
18 304 898
24 887
28
Sub Total (m) 692 707 1,468 173 7,160 867 336 2,307 262 4,617 304 114 3,669 887
Total (m) 1,399 1,468 7,333 867 2,643 5,183 4,670
168
In addition more field information was collected 4. Outcome
on unauthorized consumption, visible leaks
along roads and free water outlets etc. Free 4.1 Water loss reduction in study area
water outlets were metered and their
consumptions were measured. In sub zone 1 under the activity 1, action was
taken to regularize the 45 unauthorized use of
water; 19 defective meters were replaced; service
The initial NRW was established by measuring
leaks 52 nos and three main leaks were repaired
the inflow to the subzone. The composition of
14 standpost were metered. After completing
factors contributing to NRW was identified.
the activity 1, the status of NRW is given in
Remedial work was done and the water saved
Table 2.
was monitored. Key areas where large amount
of water saved was identified and that was
replicated to the Colombo City.
Since the percent NRW was still high, as activity
2 further search was made for possible leaks,
which resulted in repairs of 14 service leaks and
one main leak and the percent NRW reduced to
72.95. The other observation was the system
inflow dropped to 918m3/d, the Minimum
Night Flow (MNF) was 330 l/min and the
system pressure was 19.47m. Leakage search
activity was repeated by segmenting the
distribution into further small length and
locating leaks adopting step testing technique
and repair or replacement of pipes or replace
bundle pipes. Table 3 shows the outcome of the
four subzones after repeated water loss
reduction activity. It became clear one of the
main causes of losses is the leakage.
The search was initiated in Colombo city for the

visible leaks, during the night time when Figure 1- Sub-Zone Selection in Study Area
pressure was high visible leak survey was done
in a methodical manner [4], Table 4 shows the
number of leaks identified. During day time
culverts, road side drains etc. were searched for
losses Figure 2 shows the NRW reduction over
the years.
Table 2 - Status of reduction of losses in sub zone 1 after Activity 1&2
Component Water Balance Water Balance Water Balance

Before activity After Activity 1 After Activity 2
Total System Input (m3/d) 1295 1041 918
Billed Authorized consumption 190.88 248.29
(m3/d)
NRW 85.26 % 76.15% 72.95%
MNF (l/min) 690 480 330
169
Table 3 - Water saved by repeated water loss reduction activities
Sub-zone 1 2 3&4
Initial 1,295 1,245 4,240
System input volume (m3/d) After activities 571 933 3,989
Saving 724 312 251
Initial 85% 78% 73%
NRW (%)
After activities 56% 72% 71%
Table 4 - Visible leaks identified by Night leak survey in Colombo City
Number of
OIC Area Connections 2013 2014 2015
in the Area
Pamankada 15,700 130
81 257
Timbirigasyaye 13,911 169
Hulsdrof 14,555 150
32 304
Slave Island 10,910 58
Kotahena 11,338 94
122 524
Mattakkuliya 18,393 205
Maligawatta 18,690 113
57 277
Borella 11,180 126
Total 292 1362 1045
50
49
48
NRW %
47
46
45
44
2012 2013 2014 2015
Year
Figure 2 - NRW Reduction after Night Visible Survey Leak Repair
170
Jan -13
Nov 2015
10,000,000 Feb -13
Oct 2015 Mar -13
Sep 2015 Ap -13
Aug 2015 8,000,000 May- 13
Jul 2015 Jun 2013
6,000,000
Jun 2015 Jul 2013
4,000,000
May 2015 Aug 2013
2,000,000
April 2015 Sep 2013
Mar 2015 - Oct 2013
Feb 2015 Nov 2013
Jan 2015 Dec 2013
Dec 2014 Jan 2014
Nov 2014 Feb 2014
Supply Qty (m3)
Oct 2014 Mar 2014
Sep 2014 Apr 2014 Sold Qty (m3)
Aug 2014
Jul 2014 May 2014
Jun 2014
Figure 3 - No change to the inflow to the City, Sold volume has increased
5. Conclusion References
It became clear from the study area first round 1. Farley, M. and Liemberger, R. “Developing a
of activity that the NRW could be reduced non-revenue water reduction strategy planning
marginally. More emphasis needs to be given and implementing the strategy” Water Science
and Technology: Water Supply Vol 5 No 1 pp.
for leakage reduction, search for leaks and
41–50 © IWA Publishing 2005.
repairs need to be continued till satisfactory
level of NRW is reached. The process for leakage 2. International Water Association “District
search and repair needs to be routine work. It Metering Area” Guidance Notes, Feb 2007, pp. 1-
confirms that even in an older system, attention 10.
to leak reduction brings desired results. The
findings from small part of the distribution 3. Stuart Hamilton, Bambos Charalambous “Leak
system could be replicated to other areas so that Detection Technology and Implementation” 2013,
the whole distribution system could be covered pp. 7-9.
for reduction of percent NRW. The water saved
4. Rajkumar, S.G.G., “Managing Water Loss by
could be supplied to the existing and new
Understanding the Reticulation”, 2014 pp. 30-34
customers without bring in additional water as
shown in Fig 3. The estimated income from sale 5. Rajkumar, S.G.G., “Non Revenue Water
of additional water is estimated as Rs 500 Reduction in Colombo City By Adaptation of
million [5] during the period 2012 to 2015. Low Cost Strategies” Mar 2016 pp 3 of 5.
171
172
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[173 179],2017
2017
The Institution
Lanka
Implementation of Water Safety Plans for Rural Water

Supply Schemes Managed by CBOs in Sri Lanka –
Common Risks and Analysis
T. R. Samaraweera and Niranjanie Ratnayake
Abstract: Safe water is a major concern in the present global context. Even though water is
available and appears as safe, it is important to assure the extent that water can be safely used,
especially for drinking purposes. In most countries the urban communities receive drinking water
with adequate quality assurance, while the rural areas are receiving water without quality assurance,
most of the time. As far as Sri Lanka is concerned, the water quality in urban water supplies, which
are managed by the state institution National Water Supply and Drainage Board, is monitored and
regulated, and the implementation of Water Safety Plans (WSPs) has commenced. But a considerable
population is supplied water by the small community based organizations (CBO’s), to whom WSP is a
very new concept. However, these are the schemes that are mostly at risk of contamination, as the
monitoring and external surveillance of quality is lacking in these schemes. In this research several
case studies were conducted to identify and rank the risks according to the World Health
Organization (WHO) guidelines, in order to develop specific guidelines to help the CBO’s to focus on,
in providing safe drinking water to consumers. The study revealed that in water supplies with surface
water and spring water as sources, there were inherent typical risks with similar severities, such as
residual chlorine being not tested in water (uncertainty of presence of chlorine in water), head
variation of the chlorination tank, resulting in variations in the added chlorine amount, rusted GI
pipes in the distribution line and source pollution due to tea plantations surrounding the water
source. The water supplies that use ground water as the source encountered common risks such as
filtration or chlorination not being done, poor practices of water storage and usage, rusted GI pipes,
and source pollution due to houses being located close to the dug wells.
Keywords: Water Safety Plan, Rural Water Supply, Community Based Organizations (CBO), Risk
Identification and Assessment
1. Introduction  To treat the water to reduce or remove

contamination that could be present to the
Water is essential to sustain life, and a extent necessary to meet the water quality
satisfactory (adequate, safe and accessible) targets; and
supply must be available to all [3]. Safe  To prevent re-contamination during storage,
drinking water, as defined by the WHO distribution and handling of drinking water.
Guidelines, does not represent any significant [2]
risk to health over a lifetime of consumption,
including different sensitivities that may occur 1.1 Water supply and community based
between life stages. Safe drinking-water is organizations in Sri Lanka
suitable for all usual domestic purposes, Some of the water supply schemes extract water
including personal hygiene [3]. from a source with good or acceptable quality and
The most effective means of consistently distribute to the community while some extract,
ensuring the safety of a drinking-water supply treat and then distribute to the community
is through the use of a comprehensive risk (Statistics in NWDB). Since the community is
assessment and risk management approach that
encompasses all steps in water supply from
catchment to consumer called WSP. [1].
Eng. (Prof.) (Mrs.) N. Ratnayake, B.Sc.Eng.Hons (Cey),
M.Eng. (Wales), C.Eng., FIE(SL), Member IWA, Senior
The objectives of a water safety plan are to Professor in Civil Engineering, University of Moratuwa.
ensure safe drinking water through good water
supply practice, that is: Mr. T. R. Samaraweera, Department of Civil
Engineering. University of Moratuwa.
 To prevent contamination of source waters;
173
provided drinking water through such schemes it demographics and tariff data were also
must be assured they distribute safe water, so the gathered.
water safety plans (WSPs) are introduced recently
to the CBOs. As the concept of WSP is newly 2.3 Catchment to consumer process
introduced to the CBOs in Sri Lanka, apart from a identification
few studies being done on implementation of Catchment of each water supply scheme was
WSPs in rural scenario in some other countries, studied and mapped using Google maps with
there is a knowledge gap in adopting WSPs to the the aid of the GPS coordinates marked during
community based organization’s water supplies in the visits. The processes from source to
Sri Lankan context. Thus a proper study carried consumer were observed and all concerns were
out on water supply types and identifying the noted down.
common potential risks of contamination would
be worthwhile for improving the process of Finally the attention was given to study water
implementation of WSPs. storage and consumption methods at the
consumer levels.
1.2 Research and Objectives 2.4 Catchment to consumer risk identification
 To find out and analyse the variation of the Using the prepared catchment maps and
common risks with the common elements collected details the probable risks on
of the water supplies via a quantitative risk catchment contamination and pollution were
ranking so that a basis to categorize identified. Risks and practices relating to the
community based water supplies can be collection, treatment, storage, distribution and
developed for the design and storage at consumer levels were identified via
implementation of WSPs. the collected details by interview and informal
questionnaire survey at each supply.
 To provide an example risk analysis done
on CBO managed water supplies to aid
2.5 Ranking risks using semi-quantitative risk
WSP team.
matrix approach suggested by World Health
Organization
2. Methodology
As the WSPs are intended to treat and mitigate
2.1 Desk study and collecting details of
the risks of contamination, water supplies with
different water supply schemes in Kalutara
similar types of risks can be treated in a similar
district.
way in planning and designing the WSP by the
In selecting the CBO managed water supply
WSP team. All schemes studied were subjected
schemes, major concern was given to cover all
to the risk assessment using the semi-
main types of water sources such as ground
quantitative risk matrix approach suggested by
water, surface water and water springs. The
WHO [4].
collected details of water supplies were
analysed and it was found that within Kalutara Once the proper risk analysis and ranking was
District all types of sources can be covered. done, trends of the risks were studied and it
Thus, six CBO managed schemes within was attempted to find out whether the risks are
Kalutara Disrict with around 100 connections or varying in a common pattern or patterns
above were selected. Apart from Kalutara among the categories of schemes.
schemes the Kithulgala Liyanoya water supply
scheme which has been selected by UNICEF in
collaboration with NWSDB as the pilot CBO
managed water supply scheme for water safety
plan implementation was also selected.
2.2 Visiting the water supply schemes and

collection of data
For assuring the reliability of the collected data,
site visits were done to each water supply
scheme and visual observation of catchment to
consumer processes was done. During the visits
174
Table 1 - Detail of Water Supply Scheme
Name of the DS Grama- GND Type Address Num Source Chlori Treatm Quality
CBO Divis Niladhari No. of of the ber nation ent
ion Division Sche CBO of Process
me Conn
ectio
ns
Ridee Hora Koralaima 604A Pum Wijayagir 110 Shallo No None Good
Nimnaya na ping i w
CBO Viharaya, Ground
Gonapala
Junction
Suhada CBO Hora Kumbuka 607 Pum 158/A/1, 105 Shallo No None Fair
na East ping Kumbuka w
East, Ground
Gonapala
Junction
Janopakara Hora Kumbuka 607A Pum Saranathi 84 Shallo No None Good
CBO na West ping ssa Mw., w
Kumbuka Ground
West,
Gonapala
Junction
Randiya Ingiri Kurana 627B Gravi Siri 105 Shallo No None Fair
CBO ya North ty Nimal w
Road, Ground
Arakavila
,
Handapa
ngoda
Nagarika Bulat Bulathsinhal 824B Gravi Piriwena 1,323 Surface Yes Roughi Good
Sanwardana hsinh a ty Road, ng
Padanama ala Athura, Filter &
Bulathsin Chlorin
hala ation
Sathsewa Math Yatadolawatt 802D Gravi Yatadola 275 Spring No None Good
CBO ugam a East ty watta,
a Mathuga
ma
Randiya Math Welipenna 793B Pum Palliya 128 Shallo No None Fair
Dahara CBO ugam East - North ping Road, w
a Welipenn Ground
a,
Mathuga
ma
Kithulgala Kithu Malwatta 134 Gravi Malwath 515 Surface Yes Good
liyanoya lgala ty a Chlorin
developmen Rd,Kithul ation
t gala only
Source: NWSDB
Table 2 - Semi-Quantitative Risk Matrix
Risk ranking basis

Likelihood Severity Rating Risk ranking basis
Risk score Risk
rating
Almost Certain /Once a day Catastrophic public health impact 5 >15 very high
Likely/Once a week Major regulatory impact 4

Moderate/ Once a month Moderate aesthetic impact 3 15-10 High
Unlikely/Once a year Minor compliance impact 2 6-9 Medium
Rare/Once every 5 years insignificant 1 <6 Low
175
3. Results and Discussions
3.1 Kithulgala Liyanoya Development Water Supply
Table 3 - Example Risk Analysis Done Using Semi-Quantitative Risk Matrix Approach
Process Step Hazardous Hazard Type Likelihood Severity Score Risk Basis
Event (Source Rating(Before
Of Hazard) Consideration
Of Controls)
Source Tea plantation Chemical 3 4 12 High Potential toxic chemical addition
(surface surrounding above national standards ex;
water ) the water glyphosate(rating of 4)
source Can occur more than thrice a
year( likelihood rating 3)
Source People bathing Chemical & 4 2 8 Medium Potential introduction of
in the source Microbial detergents to water (severity
rating 2)
Can occur more than once a
week(likelihood rating 4)
Source New hotel Chemical & 3 4 12 High Potential for hazardous waste
being setup in Microbial specially when rainfall is
the low(severity rating 4)
upstream(addit
ion of waste) Can occur more than thrice a
year( likelihood rating 3)
The impact of Chemical 2 5 10 High Potential of electronic wasted
Source Yatiyanthota being added potential of mercury
repeater or lead addition(severity rating 5)
station(
addition of Uncertain likelihood (likelihood
electronic rating 2)
waste)
Treatment Residual Chemical 5 5 25 Very high Potential of pathogens intruding

chlorine is not the system without knowing
tested in (severity rating 5)
water(uncertain
ty of presence Daily occurrence(likelihood
of Cl in water) rating 5)
Treatment Head variation 5 4 20 Very high Not enough disinfectant dosage
of the present in water(severity rating
chlorination 4)
tank, hence the Daily occurrence( likelihood
added chlorine rating 5)
amount varies
Distribution Inadequacy of Physical 2 3 6 Medium Potential of people moving for
water supply in other water sources which may
dry season be polluted(severity rating 3)
Occurring twice a year(likelihood
rating 2)
Treatment The security Physical 2 5 10 High Potential of a deliberately caused
fence around microbial and risk (severity rating 5)
the water tanks chemical Uncertain likelihood( likelihood
rating 2)
Distribution Rusted GI Chemical and 3 5 15 High Potential for leakage and
pipes physical corrosion products entering the
system(severity rating 5)
Source Turbidity in Chemical 2 5 10 High Potential of shielding microbes
water from chlorine ( severity rating 5)
176
The main risks identified with the highest risk 3.3 Yatadolawatta sathsewa CBO’s water
score were residual chlorine is not tested in supply
water (uncertainty of presence of Cl in water),
Filtration or chlorination not being done, rusted
head variation of the chlorination tank, hence
GI pipes in the distribution line were the
the added chlorine amount varies and rusted
identified high risk events.
GI pipes in the distribution line. Tea plantation
Tea plantation surrounding the water source,
surrounding the water source new hotel being
people bathing in the source, leakages in the
setup in the upstream(addition of waste),
tank, high turbidity in water during wet season,
possible impact of Yatiyanthota repeater
lack of maintenance personnel and inadequacy
station( addition of electronic waste), broken
of water supply during dry season were the
security fence around the water tanks increased
identified moderate risks.
turbidity in water during wet season and
people bathing in the source were identified as
the risks with moderate impact.
Figure 3 - Yatadollawaththa Water Supply
3.4 Janopakara Community Based

Figure 1 - Kithulgala, Liyanoya Water Supply Organization- Kumbuka West
3.2 Urban Development Foundation water Filtration or chlorination not being done poor
supply - Bulathsinhala practices of water storage and usage and rusted
GI pipes were among the prevailing high risk
Residual chlorine is not tested in water
events.
(uncertainty of presence of Cl in water), head
The houses located close by the dug well,
variation of the chlorination tank, hence the
fertilizer used in paddy fields and inadequacy
added chlorine amount varies and rusted GI
of water during dry season were among the
pipes in transmission and distribution were the
moderate risk events.
identified high risk events.
Deforestation surrounding the intake tea

plantation surrounding the water source water
extraction from dug wells and surrounding
houses and possibility of intrusion of septic
tank water into the dug wells were the
identified moderate risk events. Inadequacy of
water supply in dry season and occasionally
cleaned consumer level storage tanks were also
among the identified risks.
Figure 4 - Dug Well, Kumbuka West
Figure 2 - Bulathsinhala Water Supply
177
3.5 Rideenimnaya Community Based
Organization- Koralaima
Filtration or chlorination not being done, poor
practices of water storage and usage and rusted
GI pipes were among the high risk events.
The houses located close by the dug well,
fertilizer used in paddy fields and inadequacy
of water during dry season were the moderate
risk events.
Figure 7 - Storage Tank, Welipenna
4. Discussion
The study showed that rural water supply

schemes with similar water sources show
typical risks in an almost similar increasing
Figure 5 - Dug Well, Koralaima order. Some of these risks, especially on source
protection are not within the capacity of the
WSP team to be mitigated.
3.6 Suhada Community Based Organization
Kumbuka East Most water supply schemes faced the issue of
inadequacy of water supply during the dry
A treatment method not being followed, rusted weather. Even though the risk assessment
GI pipes and the houses located very close to method used resulted in this being of lesser
the dug wells were the major risks identified. priority, the additional risk of the consumers
The unavailability of a parallel pump and the resorting to less safe sources during these
poor storage practices were among the mid period is not reflected in this analysis.
impact risks. Mitigation of this risk by proper catchment
restoration and aquifer recharge are important
aspects that need to be emphasized.
During water supply scheme design stage the

focus should be given on providing technical
assistance on problems on water treatment and
head variation problem in chlorination tanks.
Further study is encouraged on how the human

factor will affect the implementation of WSPs
and to investigate the legal background and
support on taking actions by WSP teams in
Figure 6 - Storage Tank, Kumbuka, East
issues such as agrochemical addition.
3.7 Randiya Dahara CBO -Welipenna,

Mathugama
5. Conclusions
Filtration or chlorination not being done, poor  The common issues identified in all CBO
practices of water storage and usage and rusted managed water supply schemes regardless of
GI pipes were the high risk events. the water source type were rusted GI pipes and
The houses located close by the dug well, the related leakages.
fertilizer used in paddy fields and inadequacy  The treatment process of surface water
of water during dry season were the moderate sources and spring supplies show typical risks
risk events. of water not being chlorinated, residual
chlorine not being tested and effect of head
variation of the chlorination tank which
178
prevents the chlorine dosage being applied at  As observed in the research the common risks
the correct rate during the course of the day. were shown with almost similar risk rankings
The mitigation measures suggested are in the water supplies with similar types of
introduction of chlorination if possible or else water source. During the WSP initiating stage
public awareness to boil the drinking water water supplies may be categorized and the
where water is not being chlorinated; try above issues can be commonly addressed
adopting ways for testing residual chlorine or depending on the type of the water source of
at least check presence of chlorine in water by the water supply.
DPD; chlorine addition to water chlorination
tank to be done twice a day so that during peak
hours enough water head will be remaining in References
the tank, respectively.
1. Bartram, J., Corrales, L., Davison, A., Deere, D.,
 In the catchment of surface water source Drury, D., Gordon, B., Howard, G., Rinehold, A.,
supplies, issues such as tea plantations in the & Stevens M. (2009). Water safety plan manual:
catchment directly discharging water into the step-by-step risk management for drinking-water
streams, people bathing in the upstream and suppliers. Geneva: World Health Organization
possible threats of future development along
the banks were identified. The mitigation 2. Davison A., Howard G., Stevens M., Callan P.,
Fewtrell L., Deere D., and Bartram J., Water
measures that can be followed are public
Safety Plans - Managing drinking-water quality
awareness & legal actions that can stop from catchment to consumer,
contamination by above issues. WHO/SDE/WSH/05.06 Water, Sanitation and
Health Protection and the Human Environment
 Ground water source supplies show typical World Health Organization Geneva (2005)
risks of proper treatment not being done,
houses located very close to the water wells and 3. World Health Organization, Guidelines For
poor safety practices of water storage at Drinking-Water Quality Third Edition
consumer levels, for which adopting treatment Incorporating The First And Second Addenda
methods or at least public awareness for boiling Volume 1 Recommendation (2008)
water before drinking, pumping over longer
4. World Health Organization, Water safety
periods at slower pumping rates, usage of
planning for small community water supplies;
screens on side walls of the well are suggested
step by step risk management for drinking
for mitigation of risks. water supplies in small communities.
179
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[181 187],2017
2017
The Institution
Lanka
A Study on Recontamination Risks in Water

Distribution Systems, using A Network Model
M.H.V. Lasantha and Niranjanie Ratnayake
Abstract: Water supply schemes aim to deliver safe and wholesome water to the consumers in
adequate quantities. However due to reasons like inadequate disinfectant residual, leakages,
stagnation, intermittent services, low pressures and cross connections, treated water can get re-
contaminated before reaching the consumers. In this study, selected contamination scenarios were
simulated using the open source network model EPANET developed by the USEPA in order to
demonstrate the application of this freely available software to identify safety risks in a distribution
system and to study the effects of proposed mitigatory measures.
Municipal water supply scheme of Raddolugama was modelled with EPANET 2.0 as a case
study. Presence and remediation of contamination risks such as inadequate disinfectant residual,
contamination by a conservative substance, stagnation of water and contaminant entering at a
pressure deficient location were studied. Simulation of chlorine level in the network revealed that
some sections have inadequate chlorine residual. The proposed chlorine booster provides an adequate
safeguard against this risk, showing that a network model can be satisfactorily used to identify and
develop mitigation actions for inadequate disinfectant residual. Simulation of transport of a
conservative substance helped identify various factors that affect the system’s behaviour. Time of
addition of the substance, with respect to water demand, is a critical factor that affects the peak
concentration. In an event of contamination, the rate of dispersion, especially within first few hours, is
a deciding factor on the severity of the outcomes and the mitigatory actions of the risk. This
simulation reveals that a contamination occurring at a peak time affects significantly more people. It
was also possible to identify the affected sections of the network due to a leaking pipe at a pressure
deficient location. The tracer study for determining percentage of water at any node originating from
each of the two tanks is useful for determining the affected area due to a contamination.
Keywords: Water supply, pipe networks, water quality modelling, contaminant transport
1. Introduction Common contaminants in water distribution

can be mainly categorized as chemical and
Water supply schemes aim to deliver safe and microbiological. Chemical contaminants in
wholesome water to the consumers in adequate piped water can be either the chemicals added
quantities. Due to inadequacies of treatment during water treatment and distribution
process and distribution infrastructure, and activities or entrants to the system during
also due to improper usage and lack of distribution. Thompson & WHO [2] identifies a
maintenance, contaminants can enter the large number of chemical contaminants,
distribution system and degrade the water categorised by the source, along with
quality. Changes of physio-chemical and information useful for risk management. A
radioactive characteristics of water and water treatment process may use various
presence of harmful microorganisms can make chemicals for purposes like coagulation,
water unsuitable for human consumption. softening and disinfection. Aluminium and
Since the effect of a substance on an organism Iron from coagulants and acrylamides from
changes with the dose and the exposure time coagulant aids can be present in the treated
[1] it is important to know not only the water. Treatment chemicals like chlorine,
toxicological properties of a contaminant but chloramine and ozone can also be present. Pipe
also the exposure limits for humans. Water materials like iron, lead, copper, zinc and vinyl
quality guidelines and standards establish safe chloride can be leached in to water.
exposure levels substances found in drinking Mr. M.H.V. Lasantha, B.Sc. Eng. (Hons) (Moratuwa).
water. Water treatment plants in Sri Lanka
Eng. (Prof.) Mrs. Niranjanie. Ratnayake is a Senior
produce drinking water following the Professor in the Department of Civil Engineering,
specifications of SLS 614:2013 standard for University of Moratuwa and President Elect, IESL for the
potable water. session 2016/2017.
1
181
Some of the above chemicals can give rise to chlorine as a disinfectant over ozone or UV
harmful by products. Toxic and carcinogenic radiation which however does not have a
disinfection by-products (DBPs) such as residual capacity. It is understood that health
Trihalomethanes (THM), haloacetic acids benefits of disinfection outweigh the harmful
(HAA) and bromate can be formed when the effects of chlorination by-products. There have
disinfectants react with natural organic matter been observed remarkable decreases of
or bromide ions. waterborne disease outbreaks in water supplies
that started chlorination. [5]
A significant proportion of reported piped Leakages
drinking water-related disease outbreaks are Leakages are the main constituent of
due to contaminations during distribution [3]. Unaccounted-for Water. Apart from loss of
Contaminations of water distribution networks quantity, leakages can cause loss of water
by pathogenic microorganisms are extremely quality as well. Leaks in pipes allow
dangerous in a public health point of view contaminants from outside to enter into the
because of the large number of people exposed system. When paired with low pressures and
to the infection risk. Common water borne intermittent supplies, their effect is highly
bacterial diseases include cholera, typhoid, significant. [6]
haemorrhagic colitis, paratyphoid, Intermittent service
salmonellosis, shigellosis, legionellosis, Due to increase of population beyond the
leptospirosis and dysentery. Hepatitis A and E design capacity, many water supply systems,
are caused by the respective viruses. Protozoa especially old ones, may not be able to provide
can cause amoebiasis and giardiasis. Many a continuous supply. In periods of drought,
types of parasitic worms can also spread treatment plants may face shortages of supply
through water contaminated with their eggs. and the service may be restricted to a limited
Effects of some of these pathogens like Vibrio period each day. Various studies have found
cholera, Escherichia coli O157:H7 and hepatitis evidence that, continuous supplies are less
E can be lethal especially when healthcare is prone to contaminations than intermittent
inadequate. Microorganism are found to cause supplies. [7] During the periods of restricted
chronic diseases as well. Diabetes, myocarditis, service, water is stagnant within parts of the
Guillain–Barré syndrome and gastric cancer system and the pressures are drastically
have been linked to waterborne pathogens. [3] reduced. This increases the chance of
contaminants in the surrounding being drawn
1.1 Contamination risks in water into the system. Increased occurrence of
distribution networks pressure fluctuations causes additional strains
on the pipes increasing the chance of fractures
Problems with treatment process, improper and leaks.
usage of distribution system and lack of Inadequate pressure
maintenance can lead to chemical or When the capacity of a water supply system
microbiological contaminations in the system. becomes insufficient, often the pressures will be
Some of the major contamination risks as inadequate at some regions. When negative
identified by WHO & UNICEF [4] are pressures occur, water will flow in reversed
inadequate disinfectant residual, low water direction creating a backflow. Backflow can
pressure, intermittent service, water leaks, cross draw in undesired contaminated water into the
connections and ageing of infrastructure. network. [5]
Inadequate disinfection residual Cross connections
Disinfection as the last step of water treatment Another source of contamination is cross
is intended for inactivating any microorganisms connections in the distribution system,
present in treated water. However, it is particularly where the consumers use dual
important for the water to have a residual systems or use hose pipes for delivering water
disinfection capacity to protect against from the tap to various collection vessels. In
contamination during distribution. Reasons for household level, cross contaminations between
absence of disinfection residual can be water supply pipes and wastewater pipes can
intentional. Chlorine which is the most occur due to poor plumbing practices. Backflow
commonly used disinfectant, has a residual is another hazard associated with distribution
capacity. Chlorination imparts a taste and networks. When there is a possibility of
odour to the water and there is also the harmful substances entering the pipe network
possibility of formation of by-products. These by flowing upstream, a backflow protection
factors may negatively influence the choice of device need to be installed. EPA [8] and WHO
2
182
[9] identifies many common cross connection 1.3 Description of the water distribution
hazards and presents different backflow network
prevention methods. Air gap, barometric loop,
atmospheric vacuum breaker, hose bib vacuum Raddolugama water supply scheme near
breaker, pressure vacuum breaker and check Seeduwa town in Western province, Sri Lanka
valve are some backflow prevention devices. was constructed in 1982 along with the housing
Opening or closing of valves, use of scheme. Using Dandugam-Oya as the source, a
unapproved material, third party access to conventional treatment plant produces water to
hydrants, open and unprotected service satisfy Sri Lanka Standards for potable water.
reservoirs and vandalism have also been [12] The distribution network covers an extent
identified as water quality hazards within of 3 km2 and serves 6,000 households. There are
distribution networks. [10] two water tanks in the network, each with 1800
m3 capacity.
For controlling growth of micro-organisms in
pipe networks, current practice is maintaining a 2. Aim and Objective
disinfectant residual. WHO [11] identifies
importance of rigorous application of good The aim of this study was to obtain an
operational and maintenance practices along understanding on transport of contaminants
with providing a disinfectant residual. and predict the behaviour of the system under
Strategies for controlling contaminations of possible contamination risk events.
distribution systems should start at the design The objectives of this study were,
stage of the water supply system. For example,  To study the behavior of residual
practices such as decentralised treatment and chlorine level in the network and
water zoning can facilitate better operational, identify residual chlorine deficient
maintenance and management routines. sections
 To propose chlorine boosters to remedy
Networks models can become useful any lack of disinfectant residual
throughout the life cycle of a water distribution  To simulate the transportation of a
network for different purposes like design, conservative contaminant in the
maintenance and troubleshooting. Hydraulic network
models which simulate the flow (quantity,  To study the transport of a contaminant
direction and pressure) in the system, steady- which enters the system due to a leak at
state water quality models which determine the a pressure deficient location
transport of contaminants including flow paths  To determine the percentage of water at
and travel times under steady-state conditions any node that originates from either of
and dynamic water quality models which the two tanks
simulate the movement and transformation of
substances in the water under changing
conditions are the three types of network
models which become useful under different
scenarios.
1.2 EPANET
EPANET software package for water

distribution network modelling was developed
by the United States Environmental Protection
Agency's (EPA) Water Supply and Water
Resources Division. It can perform extended-
period simulation of a network’s hydraulic and
water-quality behaviour. EPANET supports
simulation of spatially and temporally varying
water demand, variable speed pumps, and
minor head losses for bends and fittings.
EPANET is in public domain and its latest
version EPANET 2.0 was used for this study.
Figure 1 - Raddolugama distribution network
model in EPANET 2.0 showing locations of the
two tanks
3
183
3. Methodology
3.1 Data collection and hydraulic modelling
National Water Supply and Drainage Board

(NWSDB) operates the Raddolugama water
supply scheme. Data related to pipes and their
layout, water demands, demand pattern (figure
2) and tank details were obtained from
NWSDB. Using this data, a simplified network
having 1608 nodes and 1729 pipes (links) was
constructed. Hazen-Williams formula was
selected for calculating head loss. A global pipe
friction factor of 130 was applied as specified by
Clarke [13] for cast iron pipes.
Figure 3 - Graph of ln(Ct/C0) vs time
3.4 Application of the model
After the hydraulic model was developed,

water quality inputs were defined for the
selected scenarios. Hydraulic model was first
run using a 1-hour time step. Using the
information on flows and pressures in the
Figure 2 - Demand pattern applied for the system network from the hydraulic model, water
quality study scenarios were planned. Water
3.2 Field tests and sampling in
Raddolugama quality simulations were done using 5 minutes
for both hydraulic time step and water quality
Water samples from ten locations in the study time step.
area were tested to identify the prevailing
water quality characteristics of the network. Simulation of residual chlorine level in the
Chlorine level maintained in the clarified water network
sump of Raddolugama water treatment plant is For this simulation, water quality parameter
0.7 mg/l. was defined as a non-conservative substance
with first-order decay. Global bulk reaction rate
3.3 Determination of bulk decay rate for all pipes and tanks was defined -0.7 /day.
This simulation was run for progressively
In the laboratory, chlorine levels of water longer periods to identify steady state
samples from the network were tested over operating conditions.
time, to determine bulk decay rate of chlorine
in the studied network. Chlorine is known to be Determination of a suitable location for a
consumed in the bulk liquid phase and at the chlorine booster
pipe wall following a first order decay [14]. EPANET allows defining any node as a source
Results of the test, applied to the first order of the modelled water quality parameter.
integrated rate law (equation 1), are shown in Model was run until a suitable location was
Figure 3. found for booster dose to produce sufficient
chlorine levels at all deficient nodes. Source
... (1) type was selected as mass booster which adds
chlorine with a fixed concentration to the water
that enters the node. [15]
Ct is the chlorine concentration at time, t and C0
is the initial concentration. Bulk reaction Simulation of transportation of a conservative
coefficient is k. contaminant in the network
A conservative substance having zeroth order
Bulk reaction coefficient for chlorine in the growth and reaction rate of zero is introduced
network was found to be -0.7/day. to the system at the tank of the housing scheme
(Node T1). Method of addition is a mass booster
with a time pattern set to activate for one hour.
The Time of addition of the substance was
4
184
changed to observe the change in response of
the system. Peak demand periods of 0600h to
0700h, and 1800h to 1900h along with low
demand periods 1200h to 1300h and 2100h to
2200h were considered in the study.
Simulation of a conservative contaminant

entering at a pressure deficient location
Hydraulic simulation reveals that some nodes
experience negative pressures. There is an
increased risk of contamination as leaking pipes
can allow contaminants inside the network
when pressure becomes negative. This scenario
Figure 5 - Change of Chlorine level at node J-978
is simulated by assigning an emitter which
represents a fracture in a pipe, at a node which suddenly during midday (at around 1300 h).
experiences negative pressure. Pressure This can be correlated with the demand pattern
variation shown in Figure 4 is representative of applied to the network (figure 2). During the
the general pattern for most nodes that periods of low demand, chlorine levels tend to
experience negative pressures. drop.
Chlorine deficient nodes in the system were

identified in this study. Eleven nodes with zero
residual chlorine were identified. They are
located along two branches of the network close
to each other as shown in figure 6.
Simulation of Chlorine boosters

After several trial runs, node J-185 located
ahead of the chlorine deficient region was
selected as the location for chlorine boosting. A
chlorine source adding a 0.5 mg per litre of flow
through the node was set up. After this there
were no zero-chlorine nodes.
Figure 4 - Pressure variation at node J-1008

throughout 24 hours
Determination of the percentage of water at

any node that originates from each tank
By a tracer study, the percentage of water at
any node that originates from another node can
be determined. [15] In this study, water
originating from the two tanks were traced.
4. Results and discussion
(1) Simulation of residual chlorine

The system reached steady chlorine levels after
about 700 hours. Figure 5 shows change of
chlorine concentration at a selected node which
is representative of the whole network. Results
for 30th day were selected as the operating
conditions of the system for subsequent studies.
When variation of chlorine levels throughout a
24-hour period is examined, a general pattern
can be observed. Chlorine level shows a drop Figure 6 - Chlorine booster at J-185 and the zero-
during night time (2300 h – 0600 h) and chlorine nodes
5
185
Transport of a conservative substance
The dispersion of the substance within the
network, and the concentrations at selected
nodes were observed in the simulation.
Simulation showed that the peak concentration
at a selected node and the time to peak, changes
with the time of addition of the substance.
When a contamination event occurred during
peak time, the contaminant concentration at a
selected node was found to be lower than the
concentration resulting from a contamination
during low-demand period.
The number of nodes to which the contaminant
has spread at a certain time after addition, is
also affected by the time of addition. When Figure 9 - Contour plots showing the
percentages of water originating from the two
contaminant was added during 0600h and
tanks at 1200h.
0700h which is the peak hour, the number of
nodes that had got contaminated after one hour
5. Conclusion
i.e. at 0900h was 657. When the contaminant
was added during 2100h and 2200h the number
In this study a water quality modelling
of contaminated nodes at 2300h was 387.
approach was taken to assess contamination
risks in a water distribution system.
Contaminant entering at a pressure deficient
Simulation of chlorine level in the network
location
revealed that some sections have inadequate
Negative pressure occurs at the time of peak
chlorine residual. The proposed chlorine
demand (0600 h – 0700 h) as shown in Figure 3.
booster provides an adequate safeguard against
A contaminant that enters the system during
this risk.
this period disperses along the network. If can
When the network model was used for
be observed that, at end-nodes with smaller
predicting the transport of a conservative
demands, contaminants are retained for a
substance, various factors that affect the
longer period.
system’s behaviour could be identified. Time of
addition of the substance, with respect to water
demand, is found to be a critical factor that
determines the peak concentration that occurs
at a given node. In an event of contamination,
the rate of dispersion, especially within first
few hours, is a deciding factor on severity of the
outcomes. This simulation reveals that a
contamination occurring at a peak time affects
significantly more people.
In this study, the network model was
successfully used for identifying the affected
sections of the network due to a leaking pipe at
a pressure deficient location.
Figure 8 - Contaminated nodes at 0900 h due
Tracer study also presents its practical
to the leak at J-1008
usefulness especially in a system with multiple
Tracer study for determination of percentage sources. For instance, in an event where one of
of water that originated from a certain node the tanks is contaminated, this information
Figure 9 shows the contour plots for the becomes useful for determining the affected
percentage of water originating from each tank area.
at 1200 h. The two graphs complement each Outcomes of above studied scenarios suggest
other. that water quality models of water distribution
networks can be useful for designers and
operators of such systems by providing
information that supports activities like design
of pipe networks, troubleshooting, contingency
planning and emergency response.
6
186
References 14. Ozdemir, O. & Ger, A., “Realistic numerical
simulation of chlorine decay in pipes”,Water
Research, 32(11), 1998, pp 3307-3312.
1. Crump, KS; Hoel, DG; Langley, CH; Peto, R
(1976). "Fundamental carcinogenic processes
15. Rossman, L. A., EPANET 2 User’s Manual.
and their implications for low dose risk
Cincinnati, Ohio. US Environmental Protection
assessment"., Cancer Research, Volume 36, Issue 9
Agency, 2000, 79 p, 95 p.
Part 1, pp. 2973-2979
2. Thompson, T., & World Health Organization.,

(2007). Chemical safety of drinking-water. Geneva:
World Health Organization. PP 75-79
3. World Health Organization., Dufour, A., &

Organisation for Economic Co-operation and
Development, Assessing Microbial Safety of
Drinking Water: Improving Approaches and
Methods (WHO Drinking Water Quality Series),
IWA Publishing, 2003, 12p, 179 p.
4. World Health Organization & United Nations

Children's Fund, Global water supply and
sanitation assessment 2000 report, 2000. Retrieved
from World Health Organization,
http://www.who.int/water_sanitation_health/
monitoring/jmp2000.pdf
5. Lee, E. J. & Schwab, K.J., “Deficiencies in

drinking water distribution systems
indeveloping countries”, Journal of Water and
Health, vol 3, no. 2, 2005, pp. 109-127.
6. Fontanazza, C., Notaro, V., Puleo, V., Nicolosi,

P. and Freni, G., “Contaminant Intrusion
through Leaks in Water Distribution System:
Experimental Analysis”, Procedia Engineering,
119, 2015, pp.426-433.
7. Kumpel, E. and Nelson, K. “Comparing

microbial water quality in an intermittent and
continuous piped water supply”, Water Research,
47(14), 2013, pp.5176-5188.
8. EPA, Cross connection control manual, Office of

Water, 2003 (4606M)
9. WHO, Health aspects of plumbing, 2006, Geneva.
10. Bartram, J., Water safety plan manual. Geneva:

World Health Organization, 2009, Geneva.
11. World Health Organization, Safe Piped Water

Managing Microbial Water Quality in Piped
Distribution, IWA Publishing, 2004.
12. Rajkumar, S.G.J., Assessment and suggestions for

improvements to the treatment process for the
Raddolugama water treatment plant, 2005
13. Clark, R., Modeling water quality in distribution

systems. 1st ed. Denver, Colo. American Water
Works Association, 2011, 64 p.
7
187
188
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[189 196],2017
2017
The Institution
Lanka
Review of Irrigation Ordinance and Identifying the

Gaps to fulfil the Current Needs of the Society
Badra Kamaladasa, Lalith De Alwis and G.D.F.U. Perera
Abstract: During a recent intervention on amending Irrigation Ordinance, it was revealed that
most of the legislations related with water resources management in Sri Lanka need to be updated to
the current context of water usage. Irrigation Ordinance itself is decades old while the organisations
with vested power have undergone many structural changes which have not been updated in the
legislations. Further the water users and uses have been changed compared to the time formulating
the original legislations. Provisions made on dealing of those who violates the conditions are outdated
in terms of offences and also charges. This paper intends to discuss provisions in existing water
legislations in Sri Lanka, highlight drawback in current legislation related to irrigation management
and draw some experience gathered when amending the Irrigation Ordinance during 2016.
Keywords: Irrigation Ordinance, Water legislations, amendments to Acts
1. Introduction 2. Provisions in selected current

water Legislations
Water is considered a basic human need,
globally and nationally. Hence access to water Being an island country, Sri Lanka has its own
and regulations in usage, abstraction and Acts conferring at national level to make use of
pollution of this valuable resource should be a water resources, as no international treaties are
primary concern in legislations. Water plays required. Hence until 1987 all the enactments
different roles according to the usage, in approved by the parliament were valid island-
agriculture as a livelihood supporting wide. After the formation of Provincial
ingredient, in domestic use for drinking, Councils (PC) under the 13th amendment to the
sanitation and cooking purposes, as an essential constitution, PCs were given powers to enact
element for environment sustainability and an their own statutes subject to the conditions
indispensable component in hydropower explained in section 2.1.
generation. Hence when legislations are framed
water is considered to have stakes in different There are no dedicated national laws pertaining
dimensions in social welfare, environmental to water, but is covered with several Acts
sustainability and national economic introduced from the colonial period. [9] The
development. However, in Sri Lanka existing legal framework consists of many
legislations on water have been introduced enactments that directly or indirectly influence
sectorally addressing the issues arising from the water sector in areas such as flood
time to time. The first such ordinance is prevention, water resources management,
Irrigation and Paddy lands Ordinance that was drinking water supply, state lands, coast
enacted in 1856 by colonial rulers to promote conservation, inland fisheries, environment
traditions and customs in managing water in protection etc.
irrigated agriculture. [8] For other major service
or development sectors (such as domestic water Eng. (Mrs.) Badra Kamaladasa, B.Sc.(Eng.) UOMSL, M.Sc.
supply, power generation, Mahaweli river (Development Technologies) University of Melbourne,
basin development, inland fishing etc.), FIE(SL), Retired Director General of Irrigation.
legislations were introduced mainly to establish Eng. Lalith De Alwis, C.Eng., MIE(SL), M.Eng.
the organisations describing their functions and (EWRE&M) University of Moratuwa., Director (Water
Resources Management ) Ministry of Irrigation and Water
responsibilities.
Resources Management.
Eng. G.D.F.U. Perera, IESL (Civil), GCGI (UK) PGDM,
This paper describes the process followed in Dip.Com. Arb, FIIESL, MIE(SL), MIM(SL), Head of
amending the Irrigation Ordinance and also the Institute, Kothmale International Training
vital changes proposed in the amendments. Institute of Water Management.
1
189 ENGINEER
2.1 Enactments under Provincial Councils (PC) 2.2 Water related Law in Sri Lanka
In the 13th amendments to the constitution, There had been several legislations enacted
schedules of subject areas were given under the giving authority to different organizations in
three lists as described below on which PC can the areas such as irrigation, land, environment,
have direct control or shared responsibilities or
coast conservation, agriculture, inland fishing,
areas that PC do not have authority. [19]
forests and wildlife that have a stake in water.
i) The Provincial List specified the subjects in
respect of which Provinces could exercise Forest Ordinance No 16 of 1907, Fauna and
Legislative powers. Water related subjects Flora Protection Ordinance, No. 2 of 1937 (latest
listed here are Agriculture and agrarian revision in 2009), National Heritage and
services, rural development, irrigation, Wilderness Areas Act No. 3 of 1988, National
animal husbandry, mines and minerals. Environmental (Protection) Act, Soil
ii) The Reserved List specified the subjects in Conservation Act, Land Development
respect of which the Central government Ordinance, Crown Land Ordinance, Land
could exercise Legislative powers. Settlement Ordinance are among those).
iii)The Concurrent List specified the subjects in
respect of which either the Centre or the When multi-purpose water resources
Provinces could exercise Legislative powers. development in several individual river basins
In addition to the subject matters listed in
such as Gal Oya, Uda Walawe and Mahaweli
the Provincial List related to water,
were undertaken separate legal enactments
planning, protection of environment,
were adopted to manage the whole natural
tourism is included in the concurrence list
resources in the basin. Entirely new
that would have an impact on water.
institutional setups such as Gal Oya
PCs can have their own statutes over the Development Board, Mahaveli Development
subjects in the Provincial list and Concurrence Board, Mahaweli Authority and River Valley
list. However, before the PCs could pass a
Development Board were established with
statute on such a subject, it should consult the
Parliament for its opinions on the provisions special powers vested with.
contained in such a statute. Seven PCs have
already constituted their own statutes with 2.3 Integrated Water Resources Management
regards to Irrigation. The Provincial Councils
follow the legal provisions of Central It is a well accepted fact that management of
government where there are no statutes water to be done in a holistic manner, as it is a
adopted yet. [10] cross-sectoral subject.
The 13th amendment further specifies that Inter- The Water Resources Board Act (no 29 of 1964)
Provincial Irrigation and Land Development was introduced in view of formation an apex
projects will be the responsibility of the body for regulation of all the water related
Government of Sri Lanka. Such projects would activities in Sri Lanka. The Board was setup as
be; an advisory body to the Minister on all matters
(a) Within the Province initiated by the State concerning the control and utilization of the
and which utilize water from rivers following Water Resources in the country. The Board did
through more than one Province; a Provincial not function assuming the powers vested with
Council however, may also initiate irrigation it in the Act, but ended up as an
and land development schemes within its implementation organization related to ground
province utilizing water from such rivers; water management after when in 1978
(b) Within the Province which utilize water Groundwater Division of the Irrigation
through diversions from water systems from Department was transferred to the Water
outside the Province; and Resources Board. [18] The Water Resources
(c) All schemes where the command area falls Board Act was amended and passed by the
within two or more Provinces such as the Parliament in 1999 to enable the Water
Mahaweli Development Project. The Resources Board to pay more emphasis on
administration and management of such matters pertaining to Groundwater Resources
projects will be done by the Government of Sri in Sri Lanka.
Lanka.
ENGINEER 2
190
In year 2000, an attempt was made to introduce  Irrigation Ordinance (No 45 of 1917) [1]
Water Resources Law and introduce an  Irrigation Ordinance No 32 of 1946
institutional mechanism for regulation in the
water sector. However, misinterpretation of There were a few amendments thereafter to the
certain provisions paved way to the 1946 Ordinance by Acts 1 of 1951, 48 of 1968, 23
government to abandon the draft proposals. of 1983, 34 of 1990 and 13 of 1994 and Irrigation
Law no. 23 of 1973. [15]
3. Irrigation Ordinance
But those did not address fully the fast
changing scenario in the society. Irrigation
Irrigation Ordinance is the oldest prevailing
engineers face many difficulties in managing
legislation covering the water sector. Provisions
limited water resources available and there
that are based on the traditions and customs
were many requests to the policy makers to
inherited from ancient hydraulic civilization in
change the legislations to fulfil the current
Sri Lanka subjected to the regulation of British
needs. Recognizing these shortcomings, the
colonial administrative structure. Hence it has
Ministry of Irrigation and Water Resources
unique features having combined ancient Sri
Management took initiative in 2016 to identify
Lankan law and British law. In this chapter the
the current needs in order to amend the
process of developing the current Ordinance,
Ordinance. Since the irrigation infrastructure
institutional structure, provisions and gaps will
serves not only the irrigation sector, but many
be discussed. [11]
sectors such as domestic water supply, inland
fishing, environment, flood control, power
3.1 Provisions in current Irrigation
generation etc. comprehensive consultative
Ordinance
studies had to be conducted to cover all these
areas.
The British rulers initiated the establishment of
a legal framework and made financial
3.2 Institutional Set Up and
provisions for the restoration of ancient Responsibilities
irrigation infrastructure. [6] The first Ordinance
related to irrigation was the “Irrigation and Irrigation Development was initially
Cultivation of Paddy Lands No 9 of 1856”. The undertaken by the Government Agent in 1850’s
Government Agent was vested powers to and then the same was shared with the Public
organise meetings to revive the customs and Works Department (PWD) when the technical
traditions related with paddy cultivation in issues became complicated. Central Irrigation
each District and then adopt a similar legal Boards and Provincial Irrigation Boards were
framework in the relevant area. [4] The formed mainly to develop major irrigation
following legal enactments or amendments works, when British rulers found that the PWD
were introduced from time to time spanning to could not pay much attention to irrigation. [5]
nine decades until the currently valid In 1900 Irrigation Department was established
Ordinance was introduced in 1946, depending solely for this purpose. However, the role
on the changes of government policy or played by the Government Agent in facilitating
demand by the society. and organizing seasonal cultivations, land
 Paddy Lands Irrigation Ordinance (No. alienation, collection of irrigation rates etc. were
21 of 1861) [16, 17] continued in the same spirt in the Ordinance.
 Paddy Cultivation Ordinance (No 21 of Even though Irrigation Department was fully
1867) [12] responsible for planning and development of
 Amendment to above by 42/1884 [13] water resources, that fact has not been
 Irrigation and Paddy Cultivation acknowledged in the amendments or in new
Ordinance no 23 of 1887 [14] Irrigation laws that were introduced later.
 Irrigation and Paddy Cultivation
Ordinance (No 23 of 1889) The focus of the Irrigation Department was
 Irrigation and Paddy Cultivation mainly on restoration of ancient reservoirs until
Ordinance (No 6 of 1892) 1920’s. It was later shifted towards
 Irrigation Ordinance no 16 of 1906 development of new reservoirs. Government
3
191 ENGINEER
Agent was vested with legal powers to sell or panel to study the shortcomings in the current
alienate lands developed under the irrigation Ordinance and recommend a course of action to
systems, manage lands and water through the rectify them. The panel consisted of two retired
Director Generals of Irrigation (both engineers),
village councils and headman, fixing and
one of them appointed as the team leader, and
collection of irrigation rates, taking action two other retired officers of Sri Lanka
against irrigation offences etc. They played a Administrative Service (one officer a former
major role in the management of the cultivation Secretary to the Prime Minister and the other
while managing the revenue collection through officer, a former Secretary to the Ministry of
his assistants who were known as Chief Lands). The coordinating officer of the
Headmen and later or as District Revenue committee was a former Director in the
Irrigation Department. The panel conducted a
Officers.
literature survey and had extensive discussions
initially with key officials in the M/I&WRM,
When the current Irrigation Ordinance, that Irrigation Department and Agrarian
had been later revised several times, was Development Department (ADD) to identify
enacted in 1946, Government Agent remained a the gaps.
key player. [2] One other reason for continuing
the GA’s role in irrigation management would A special briefing session was conducted with
the Minister of I&WRM to make him aware and
have been, and especially the absence of a
get his feedback as the M/I & WM has overall
central coordinator in managing all the two key
responsibility to redress the shortcomings in
natural resources, land and water, which was irrigation legislations. The expert panel then
duly filled by the Government Agent. conducted a series of consultations at District
level covering 22 districts where interprovincial
One successful institutional structure major irrigation schemes are operated. For the
introduced under the Irrigation Ordinance of District level consultations, District Secretaries,
1946 was the District Agricultural Committee, Director of Irrigation, Chief engineers,
popularly known as DAC. In the dry zone of Sri Divisional Irrigation Engineers, Divisional
Lanka this is still a strong administrative entity Secretaries, Irrigation officials, Official in line
that takes decisions regarding irrigation and agencies and farmer organization leaders were
agriculture in the districts and also on water involved. After briefing the problem the
resources development. [15] participants were grouped in random and a
discussion paper was given for them to discuss
An Advisory Committee at district level was and express their views.
prescribed in the 1946 Ordinance. However
these committees do not exist in any district Their opinion over the shortcomings of the
now. current legal provisions was very vital since all
these groups confront problems associated with
With the 1968 amendment, the overall irrigation & water resources sector on a daily
supervisory power over the activities defined in basis.
the Ordinance was vested with the
Commissioner of Agrarian Development. A special workshop was conducted with all the
However, this has not been practiced so far and key officials of nine Provincial Councils.
was proven to be an unsuccessful Discussions were conducted with Ministry of
administrative proposal. Lands, Ministry of Agriculture and Mahaweli
Amendment introduced in 1994 empowered Authority of Sri Lanka. The veteran Irrigation
the farmer organizations and Project and water sector officers included the retired
Management Committees to play key roles in officials from the Ministry of Irrigation and
seasonal planning and management of Department of Irrigation; university academics
Irrigation schemes. and representatives from International Water
Management Institute were consulted to get
their feedback as well. A paper advertisement
3.3 Methodology Adopted for identifying was also published inviting the views of the
gaps in the legislation public over the shortcomings of the current
Irrigation Ordinance and suggestions for
Ministry of Irrigation and Water Resources improvements.
Management (M/I&WRM) appointed an expert
ENGINEER 4
192
After listing the feedback from all the  Ambiguity of the definition of
stakeholders, the expert panel reviewed the “Minister” – The overall power over
current (consolidated) Ordinance. The execution of provisions in the
amendments were considered in two formats; Ordinance is vested with the
[15] Commissioner General of Agrarian
Development. Hence logically
i) Changes to the existing sections in the regulations over the matters related in
Ordinance the Ordinance have to be originated
ii) Introduction of new sections to fill the from the Minister in charge of Agrarian
gaps. Development. As most of the matters
relevant to major irrigation works are
Representatives from the Department of Legal under purview of another Ministry, it is
Draftsmen (D/LD) were then invited and not practical to issue such direction
briefed the rationale of each and every proposal from a minister not responsible directly
of the amendment and obtained their feedback. to the subject.
This was aimed to shorten the time taken for  Absence of regulations – There are
D/LD to draft the amendments and produce provisions made in the Ordinance to
the final version of amendments for the introduce regulations in specific
approval from the parliament. matters. No regulations have been
passed in the parliament under this
3.4 Observed Gaps Ordinance. This has created a huge
vacuum in the legislations as many
important issues were not defined in
The expert panel summarized all the feedback
details there.
received during the consultations. The issues
raised by those who face practical difficulties in
 Non-availability of a complete list of
the field in managing existing irrigation
irrigation offences - specially some
schemes or by those who do planning and
construction of new water development pressing issues such as non-compliance
reflected the current gaps in the legislations. with the agreed decisions with regard
The farmer organization leaders were allowed to management of the seasonal
to express their views freely and hence the cultivation, conservation of water,
panel was able to grasp the issues from the protection of the infrastructure,
beneficiary’s point of view as well. The main
responsibilities of the stakeholders
shortcomings identified in the prevailing
irrigation laws were: [15] towards proper maintenance of the
infrastructure, encroachments on the
 Not introducing timely amendments- reservations, willful damages caused to
Since 1946 from when the current structures, unauthorized acquisition of
Irrigation Ordinance of was enacted, water, harmful activities within
many changes have occurred in the reservoir bed and reservations,
social, demographic and economic unauthorized constructions, failure to
spheres in the country. Demand for
comply with instructions for remedial
water for the increasing population and
new livelihood was increased action. Identification of the offences is
dramatically. Hence the monopoly over the first step which should be followed
water resources initially held by the up with un-delayed action to undo the
irrigation sector has been confronted by harmful effects of any such offence
the demand created by urban water committed.
supply, industry, tourism, wildlife,
inland fishery, aqua sports and other
 No clear definition for major and minor
activities. Amendments to the
Irrigation Ordinance made after 1946 irrigation works - classification of major
do not reflect such matters. and minor irrigation works requires to
Environmental flow that has to be be redefined in the Ordinance in view
released under the new environmental of the current practice. In practice
regulations has also made impacts over below 200 Acs. (80ha) irrigable area
the share of the agricultural water user.
appears to be the criterion for the
5
193 ENGINEER
classification of minor irrigation difficulties in taking legal action for the
schemes which are under the purview protection of such reservations.
of the Agrarian Services Department or
Provincial Councils while the larger  Difficulties in mobilizing FO to act
inter-provincial irrigation schemes are against the irrigation offenders -
under the purview of the Irrigation Farmer Leaders in the Project
Department as major irrigation works. Committees and Farmer Organizations
Both categories are governed by the show a reluctance to take action against
Irrigation Ordinance. their errant fellow-farmers. They
 Entrusting the Government Agent on request to be given recognition within
construction and maintenance of the Ordinance so that they could take
irrigation works for which GA has no action against their fellowmen in an
supporting institutional structure. official capacity rather than in a
According to the Ordinance, where it is personal capacity as at present.
proposed to construct any irrigation
work or any variation be made for the  Lack of provisions for management of
physical conditions of any irrigation irrigation systems have extended
work, a project proposal for that beyond district and administrative
purpose may be prepared by the boundaries - in the present Ordinance
Government Agent. The Commissioner decisions over the development and
of Agrarian Services approves the operation of irrigation scheme has been
irrigation scheme prepared by the GA. taken at the District level. It does not
Practically this procedure does not take identify the requirements of the
place as the Irrigation authorities complicated schemes spread beyond
attend the same. the district boundaries.
 Absence of provision for Water

 Time consuming process for the resources planning – Planning of water
prevention of offences - The Ordinance resources has not been addressed in the
mentions for the GA to serve a notice Ordinance. Irrigation development
for the person who committed an plans are not merely to serve the
offence (such as encroachment or irrigation requirements but more needs
obstruction of a water courses, channels nowadays such as urban water supply,
or tanks) to abate such encroachments inland fishing, environmental needs,
or obstructions and if the offender fails industrial demand, hydropower, flood
to do so cause it to be abated and mitigation and salt water exclusion.
recover the costs. It has not been much There is a need for holistic approach for
practiced. A quicker and effective comprehensive water resource
procedure is required planning taking into account the needs
. of all above sectors.
 Not sufficient empathy given for
 Absence of regulations in catchment
protection of reservations - The
management & riverine management-
responsibility of protection of
To ensure sustainable water resources
reservations of irrigation reservoirs,
of the country, regulation of catchment
structures, canals, drainage canals and
and riverine management is a must.
spill tail canals or other lands on which
This aspect has not been covered in any
buildings or other structures are
of the water legislations. Irrigation
constructed is not assigned to the
Ordinance is the most suitable
Irrigation Department, which is
legislation to incorporate these aspects.
responsible in managing these systems.
The ID is currently facing frequent
ENGINEER 6
194
4. Observations have become regulatory bodies in water
management. Hence it is high time for
Many observations were made by the panel establishment of an apex body to impose
during the process of consultations and regulatory measures, for each organisation
literature review. One notable observation is handling bulk water management to make
the initiative and active participation taken by use of water resources in the agricultural
the engineers in changing the legislations. Since sector to be more efficient and productive.
the irrigation engineers are in the forefront in
planning, designing and management of water
resources infrastructure in the country, they 5. Way forward
face immense difficulties in the field and also
are subjected to criticisms from the public due The panel submitted its recommendations to
to a lack of proper legal support. the Ministry of I&WRM in its final report, after
Representatives from Provincial Councils too 6 months of deliberations. Amendments
were mainly engineers and they highlighted the proposed to the existing Irrigation Ordinance
lack of legal provisions, training opportunities were submitted to the Department of Legal
and interaction with national organisations, Draftsmen for their perusal. After review by the
which are essential for quality enhancement of Legal Draftsmen, the draft will be referred to
their interventions in water sector. the Attorney General’s Department to get the
Following additional information were also views from all the stakeholders. Once this
taken into account when finalizing the process is cleared, the amendments will have to
proposed amendments to the Irrigation be passed through the Parliament which will
Ordinance. take nearly one year.
i. An extended responsibility has been 6. Conclusion

placed with Irrigation Department in 2014
by the National Policy on “Protection and Systematic approach in identifying the gaps in
conservation of water sources, their the current Irrigation Ordinance made a
catchments and reservations”. The pathway to make proposals for amending the
Irrigation Department will have to take current Irrigation Ordinance. These
care of riverine management, protection of amendments define the issues in broader
reservations, protection of watercourses context and it is necessary to frame regulations
and preservation of water quality etc. considering detailed requirements. However,
With this assigned responsibility, legal legislation for overall water resources
management is a need of the country, which
support is necessary to be in place.
cannot be fulfilled without the introduction of
the amendments to the Irrigation Ordinance, as
emphasized in many consultative discussions.
ii. Failure in publication of an irrigation
scheme in the Gazette as prescribed in the
Acknowledgement
Irrigation Ordinance has not been taken
place. The legal acceptability of an The authors wish to acknowledge with thanks
irrigation scheme rests upon GA making the approval given by the Secretary of Ministry
arrangements to gazette same with the of Irrigation and Water Resources Management
recommendation of Director of Irrigation in quoting the internal documents of the
and /or Survey General due to practical Ministry and also the input of the specialist
panel who conducted a comprehensive study in
limitations. This has not been attended to
framing a proposal for amending the Irrigation
in the past due to practical reasons. There Ordinance.
is no legal coverage for Irrigation
Department to claim boundaries of the References
infrastructure due to this shortcoming,
which has to be attended early. 1. Ceylon Government Gazette, Part II, Dec 21,
1917.
iii. Lack of regulatory mechanism is found to 2. Ceylon Government Press, Acts of Ceylon,
be a shortcoming in the whole process Chapter 453.
where implementing agencies themselves
7
195 ENGINEER
3. Ceylon Government Press, Supplement to 19. www.ep.gov.lk/Documents/13th_Amendmen
Ceylon Government Gazette Feb 16,1951. t.pdf.
4. Ceylon Government Publication, 1856.
5. Ceylon Sessional paper IXLV, 1905, Capt Sir

John Keane, Report on Irrigation in Ceylon.
6. House of Commons, Report of Lieutenant-

Colonel Colebrooke, 1832, Government of
Ceylon.
7. Irrigation Department, Proceedings of Seminar

on Irrigation, 2000.
8. Kamaladasa, Badra (1999), “Does the Irrigation

Ordinance provide the necessary legal
framework for Irrigation Development and
Management”, Transactions 1999 of annual
sessions of The Institution of Engineers of Sri
Lanka.
9. Kamaladasa, Badra, 2001, “Integrated Water

Resources Management-Need to think both
globally and locally”, International Seminar on
“Engineering management for Accelerated
Development in South and South east Asia”,
Institution of Engineers Sri Lanka, Colombo,
2001.
10. Kamaladasa, Badra, 2015, “Role of the

Irrigation Department in irrigation sector in the
context of devolution of power under 13 th
amendment”, National seminar on “Impact
made in the irrigation sector by devolution
power by 13th Amendment”, 2007, Seminar
Proceedings by HARTI, Colombo.
11. Kamaladasa, Badra, 2015, “What if the first

Irrigation Ordinance in Sri Lanka in modern
history was implemented?” INWEP
symposium, Marawila, Sri Lanka, October.
12. Legislative enactments of Ceylon, Vol II Ceylon

Government, Ordinances, 1867.
13. Legislative enactments of Ceylon, Vol III, (pg

454-457) Ordinances, 1884.
14. Legislative enactments of Ceylon, Vol III, (pg

88-92), Ordinances, 1887.
15. Ministry of I & WRM, 2016, Final report on

“Amending the Irrigation Ordinance”.
16. Ordinances of Ceylon, Ceylon Government

Gazette no 3264, 1861, pg 453-458).
17. Ordinances of Ceylon, Ceylon Government

Press, Ordinances 21/ 1861.
18. Water Resources Board, Annual report, 2011.
ENGINEER 8
196
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[197 204],2017
2017
The Institution
Lanka
Analysis of Long-term Trends of Climatological

Parameters in Kigali, Rwanda
D.P.C. Laknath, N.R. Josiah and T.A.J.G. Sirisena
Abstract: Kigali is the capital city of Rwanda in Africa. Rwanda is located astride two key
climate regions (i.e. East Africa and Central Africa). Even though Rwanda is located in the tropical
belt, it experiences a temperate climate as a result of its high elevation. To represent the climatic
conditions at a proposed dam construction site in the proximity of Kigali City, this study was carried
out to identify the long-term variation of climatological parameters in Kigali as it was the most
representative hydro-meteorological station within the area. Accordingly, Kigali Airport weather
station was selected considering the long length of data availability between 1971 – 2013 periods. For
the analysis, meteorological variables such as rainfall, temperature, relative humidity, evaporation,
sunshine duration, wind and atmospheric pressure were selected. To identify the trends of each
climatic parameter, annual anomalies were calculated. In this study, trend analysis was carried out by
using non-parametric Mann-Kendall test (1945). On the basis of Mann-Kendall test and Zc statistics of
each month and seasons, long term trend of each variable between 1971 – 2013 periods was identified.
As the outcome of the study, it has been identified that there is an increasing trend in temperature
during last 40 years at Kigali-Aero station. Also, rainfall has a decreasing trend during the same
period. Further, relative humidity has decreased during the same period. With the increasing
temperature and decreasing humidity, an increase in evaporation is expected. According to the
present analysis, a positive trend of evaporation was identified. Further, a positive trend in wind
speed and atmospheric pressure was identified for the past 40 years. In case of sunshine hours, a
significant trend was not identified.
Keywords: Design of dam, climatological data, annual anomalies, Mann-Kendall test, trend
analysis
1. Introduction
Rwanda (Figure 1). This dam site is situated
Generally, in many countries in the African approximately 30km north east of Kigali which
continent, the impact of climate change on the is the capital and largest city of Rwanda. To
continent„s scarceresources such as water represent the climatic conditions at the
resources seems to be completely ignored. Also, proposed dam construction site, this study was
as climate change issues are not of immediate carried out to identify the long-term variation
importance, there is a lack of enthusiasm to of climatological parameters for the most
understand about the climate change [1] in the representative hydro-meteorological station for
same region. Further, especially in Africa, a the proposed dam within the area. Accordingly,
knowledge gap has been identified in literature Kigali Airport weather station was selected
related to water resource management under considering the long length of data availability
climate change [2].However, in African between 1971 – 2013 periods. For the analysis,
countries, water resources are considered as an meteorological variables such as rainfall,
important and limiting constraint for economic temperature, relative humidity, evaporation,
development. Hence, development of hydraulic sunshine duration, wind and atmospheric
infrastructure has been identified as an pressure were selected. Thus, trends of each
important task for water resources climatic parameter were identified.
management and identifying long-term trends Eng. (Dr.) D.P.C. Laknath, B.Sc.Eng. (Moratuwa),
and variation of climatic parameters such as M. Eng. (Thailand) Dr. Eng. (Japan), Engineering
rainfall, temperature and atmospheric pressure Manager, Lanka Hydraulic Institute Ltd.
are crucial for the construction of hydraulic Eng. N.R. Josiah, B.Sc.Eng.(Peradeniya), Research
structures like dams. Engineer, Lanka Hydraulic Institute Ltd.
The present study was carried out focusing a Eng. T.A.J.G. Sirisena, B.Sc.Eng.(Peradeniya), M.Eng
(AIT), Ph.D. Candidate, UNESCO-IHE Institute for
potential dam site at a selected location in the Water Education, Delft, The Netherlands, University of
middle reaches of Nyabarongo River in Twente, Enschede, The Netherlands.
1
197 ENGINEER

Rwanda Land Boundary

Uganda
Tanzania
Nyabarongo River
Dam
DR
Congo
Kigali Airport
Akanyaru River
Figure 1 - Study Area: Proposed Dam Site and Kigali Airport Weather Station [3]
2. Study Area to the SCCE study [4], the average temperature

in Rwanda is around 200C. As a result of
Climatological study was carried out focusing a topography variation, the warmest annual
potential dam site at a suitable location across temperatures are found in the eastern plateau
the Nyabarongo River (Figure 1). This dam site (200C-210C) and Imbo and Bugarama Valleys
is situated approximately 30km north east of (230C-240C). The cooler temperatures are in high
Kigali which is the capital and largest city of elevations of the central plateau (17.5 -19 0C)
Rwanda. Rwanda has two principal basins and highlands (<17 0C). In the case of rainfall,
called eastern Nile basin (85% of entire basin) Rwanda experiences bimodal patterns of
and the western Congo basin (15% of entire rainfall, mainly driven by the progression of the
basin).These basins are divided by Congo-Nile Inter-Tropical Convergence Zone (ITCZ). The
alpine regions. The Nyabarongo and Akanyaru ITCZ is the area encircling the earth near the
rivers belong to the Nile basin. The proposed equator where the northeast and southeast
dam could aim to cater for municipal and trade winds come together. Based on
irrigation water and power needs of the country precipitation and climate, Rwanda can be
and stabilization of Nyabarongo River for flood divided into four climatic areas (i.e. low altitude,
control. Even though Rwanda is located in the high altitude, central plains and eastern savanna).
tropical belt, it experiences a temperate climate Rwanda experiences four seasons, including
as a result of its high elevation. The Albertine two rainy seasons and two dry seasons.
branch of the Rift Valley runs along the western Generally, long rains occur during rainy season
side of Rwanda. Much of the border with the 1 (March - May) due to the ITCZ moves to the
Democratic Republic of Congo (DRC) is north. Short rains occur during Rainy season 2
mountainous with elevation over 2000 m. (October - December) due to the return of ITCZ
Elevations decline toward the central plateau of to south. January – February and June to
Rwanda and again in the eastern plateau September periods are identified as Dry season
towards the border with Tanzania. According – 1 and 2, respectively.
ENGINEER 2
198
3. Methodology variables and record lengths at Kigali Aeroare
tabulated in Table 1.
In general, Fifth assessment report of the
Intergovernmental Panel on Climate Change 3.2 Trend Analysis
(IPCC) [9]set the technical guidelines for Before analysis of data, quality control tests
assessing climate change impacts and were carried out. For the identification of trends
adaptations. It sets the prerequisite of each climatic variable, data record of
characteristics of climate data sets necessary for sufficient length is required. Accordingly,
an adequate analysis of climate changes. For a period between 1971 and 2013 was considered
climatic data analysis, data availability, quality for the analysis. After selecting data between
and consistency of the data, observation suitable time periods, rough screening of the
practices and instrumentation are found to be data was done by computing and verifying the
the governing parameters [10].Further, totals of each parameter for the water years
Bocchiola [11] was presented different (hydrological years).By using water years, it
methodologies such as liner regression against removes any risk of the seasons being split over
time, liner regression against with NAO and two years. After performing a rough screening
global thermal anomalies, Mann-Kendall test, of the data, data was plotted for the selected
correlation against physiographic and climatic time step (year) to identify trends or
features and correlation against local climate discontinuities. Further, to identify trend of
variables to be useful for the prediction of each climatic parameter, annual anomalies
hydro-meteorological trend analysis. With this were calculated. Anomalies defined as
general idea, the climatic data and the suitable deviations from the average in climatology and
method of analysis were determined and the average was calculated over some time
discusses below. period of at least 20 or 30 years [8]. To detect
the significant trends in climatologic time series
3.1 Climatic Data parametric and non-parametric methods can be
In line with the project objective, climatic data implemented [6]. Independent and normally
analysis was carried out for the most distributed data is needed to use the parametric
representative hydro-meteorological station tests, while non-parametric tests require only
within or near the area. Accordingly, Kigali the independent data [6]. In this study, trend
Airport (Kigali Aero)weather station at (30.110 E analysis was carried out by using non-
-1.950 N) was selected considering the long parametric Mann-Kendall test [5]. The Mann-
length of data availability between 1971 – 2013 Kendall test statistic (S) can be calculated [7] as
periods (Table 1). For the analysis, following shown in 3.1.
meteorological variables were considered.
∑ ∑ ….(3.1)
 Rainfall
 Temperature where, n is the number of data, xi, xj are data
 Relative Humidity values in time series i and j (>i) respectively
 Evaporation and sgn(xj-xi) is the sign function as,
 Sunshine duration
 Wind
 Atmospheric pressure ( ) { ….(3.2)
Percentage availability of meteorological
Table 1 - Summary of Completeness of Meteorological Variables at Kigali Aero

Climate Parameters Interval % Availability Duration
Rainfall Daily 98.82 1971 - 2013
Mean Temperature Daily 98.66 1972 - 2013
Min/Max Temperature Monthly 98.64 1972 - 2013
Relative Humidity Daily 98.15 1972 - 2013
Evaporation Daily 91.63 1972 - 2012
Sunshine Duration Daily 97.34 1972 - 2001
Wind Speed/ Direction Daily 98.80 1974 - 2009
Atmospheric Pressure Daily 98.70 1972 - 2013
3
199 ENGINEER
The variance can be computed as,
…..(3.3)
where, n is number of data. For the sample size

n>10, Mann-Kendall statistic (ZC) can be
calculated as,
√
……(3.4)
Figure 3- Annual Rainfall Anomaly and 5 Year
{√ Running Mean Anomalies for 1971 - 2013
Period
Increasing and decreasing trends can be
indicated by the positive and negativeZC
There was a positive trend observed in January,
values. This statistical test is frequently used to
March and October while the other nine
analyse the significance of trends in
months had a negative trend. Thus, in overall, a
climatological time series [7] and it is being
negative rainfall trend for the considered 40
used for studying the spatial variation and
years has been identified. Table 2 shows the
temporal trends of hydro climatic series.
summary of the Mann-Kendall ZC values and
the long-term trend.
4.2 Temperature
4.1 Rainfall Mean, minimum and maximum temperatures
Annual average rainfall for water years within for the period 1972-2013 at the representative
1971– 2013 period at the representative station station were analysed. Average of mean
are illustrated in Figure 2. Further, anomalies temperature was calculated as 20.50C. The
for the same period are illustrated in Figure 3. mean values of the annual temperatures with
Anomaly shows the departure from the long- the 5 year running mean and anomalies are
term average of rainfall (983 mm). Thus, long shown in Figure 4 and 5 respectively.
term trend and inter-annual variability of
rainfall at the representative station were
identified.
Figure 4 - Average of Annual Mean

Temperature and 5 Year Running Mean for
1972 - 2013 Period
Figure 2 - Annual Average Rainfall and 5 Year
Running Mean for 1971- 2013 Period
Considering 5 year running mean rainfall and

anomalies, a slight decrease of rainfall over the
last 40 years was observed. This trend was
further analysed with Mann-Kendall test for
each month over considered time period. By
applying equations 3.1, 3.2, 3.3 and 3.4 static S,
sign, variance and ZC were estimated.
Figure 5 - Annual Average Temperature and 5
Year Running Mean Anomalies for 1972 - 2013
Period
ENGINEER 4
200
To understand the long term trend and inter- test was performed for each month
annual variability, Mann-Kendall test was individually. Table 2 presents the Mann-
performed for mean, minimum and maximum Kendall ZC values for Relative Humidity.
temperatures for each month over the
considered time period. It has been revealed that, except March, all
other months have negative trends. Thus, it was
Mann-Kendall ZC values are shown in Table 2. identified that there was a negative trend of
It has been noted that there is a positive trend relative humidity at the representative location.
of mean temperature of all months and a
gradual increase was identified since 1972. 4.4 Evaporation
Further, a large increase in minimum Annual total evaporation for the period of 1972-
temperature than maximum temperature was 2013 was considered for the analysis. Average
identified. This implies a reduction in diurnal annual total evaporation was calculated as
temperature range during the considered 1438.1 mm. Figure 8 and 9 illustrate the annual
period. total evaporation with 5 year running mean
and the anomalies respectively.
4.3 Relative Humidity
Annual average relative humidity for the years
1972-2013 at the representative station was
considered and the average relative humidity
was calculated as 73.9%. Figure 6 shows the
annual average humidity with 5 year running
mean while Figure 7 shows the anomalies.
Figure 8 - Annual Total Evaporation and 5

Year Running Mean for 1972-2013 Period
Figure 6 - Annual Average Relative Humidity

and 5 Year Running Mean for 1972-2013
Period
Figure 9 – Annual Evaporation and 5 Year

Running Mean Anomalies for 1972 – 2013
Period
There was an increase of annual total

evaporation as observed when considering the
5 year running mean and anomalies of total
evaporation. The trend was further analysed
with Mann-Kendall test for each month
Figure 7 - Annual Average Relative Humidity individually over the considered time period.
and 5 Year Running Mean Anomalies for 1972
– 2013 Period Positive trends were identified for all months
for the evaporation (Table 2). Thus, in overall
A decrease in relative humidity was observed there was a positive trend of evaporation at the
when considering 5 year running mean and the representative location.
anomalies. To ensure the trend, Mann-Kendall
5
201 ENGINEER
4.5 Sunshine duration respectively. The annual average wind speed
Annual average sunshine durations for the was calculated as 2.6 m/s.
period of 1972-2001 at the representative
location were used in the analysis. Figure 10 An increase in annual average wind speed was
shows the annual average sunshine duration observed by considering the 5 year running
with 5 year running mean. Further, average mean and anomalies. This trend was further
annual sunshine duration was estimated as 5.8 examined with Mann-Kendall test. Test was
hours. applied for each month individually over the
considered period.
Figure 10 - Annual Average Sunshine

Duration and 5 Year Running Mean for 1972-
2001Period Figure 12 - Annual Average Wind Speed and 5
Year Running Mean for 1974-2009 Period
Considering 5 year running mean and the
anomalies there were no significant trend
identified.
Figure 13 – Annual Average Wind Speed and

5 Year Running Mean Anomalies for 1974 –
Figure 11 - Annual Average Sunshine 2009 Period
Duration Anomaly and 5 Year Running Mean
Anomalies for 1972 – 2001 Period
The Mann-Kendall trend of annual average
wind speed showed positive statistics for all
Anomalies of annual average sunshine hours
months (Table 2). Thus it was recognized as a
are shown in Figure 11. Further, Mann-Kendall
positive trend of wind speed at the
test was performed for each month over the
representative location for considered period.
considered period to identify the possible
trends. Further, wind directions for the years 1974-2009
were also considered in the analysis. Winds are
Except January, March, October and December, dominant between 292.50-360.00 and 22.50-90.00
negative trends were identified (Table 2). In directions. A 73.74% of wind speeds belonged
overall, there were no significant trends to 4m/s – 8 m/s range and the percentage
identified for the sunshine at the representative occurrence of winds over the speed of 10 m/s
location. was very low during the analysis period.
4.7 Atmospheric pressure
4.6 Wind Annual average atmospheric pressure for the
Annual average wind speeds for the years 1974- years 1972-2013 at the representative station
2009 at the representative station were were considered for the analysis. Average
considered for the analysis. Figure 12 and 13 annual atmospheric pressure was 851.3 mbar.
show the annual average wind speed and 5 Figure 14 and 15 show the annual average
year running mean and the anomalies
ENGINEER 6
202
atmospheric pressure and 5 year running mean 4.8 Analysis summary
and anomalies respectively. Table 2 illustrates the summary of long term
variability of meteorological variables such as
rainfall, temperature, relative humidity,
evaporation, sunshine hours, wind speed and
atmospheric pressure at Kigali-Aero weather
station which is a representative location for the
study.
On the basis of Mann-Kendall test and

Zcstatistics of each month and seasons, the long
term trend of each variable between the years
1971-2013 was identified. Identified overall
Figure 14 - Annual Average Atmospheric trend of each variable is given under “long
Pressure and 5 Year Running Mean for 1972- term trend (1971-2013)” of Table 2.
2013 Period
5. Conclusions
This study was carried out to understand about

the long-term variation of climatological
parameters at a proposed dam site in the
proximity of Kigali City. Thus, as the
representative hydro-meteorological station
within the area, Kigali Airport weather station
was selected considering the long length of data
availability for 1971 – 2013period. For the
analysis, meteorological variables such as
Figure 15 – Annual Average Atmospheric rainfall, temperature, relative humidity,
Pressure and 5 Year Running Mean evaporation, sunshine duration, wind and
Anomalies for 1972 – 2013 Period atmospheric pressure were selected. As shown
in Figure 16, there is an increasing trend in
An increase in atmospheric pressure was temperature during last 40 years at Kigali-Aero
observed by considering the 5 year running station. Also, rainfall has a decreasing trend
mean and anomalies of atmospheric pressure. during the same period. Further, relative
Trend was further examined with Mann- humidity has decreased during the considered
Kendall test for each month over the considered period. With the increasing temperature and
period. decreasing humidity, an increase in
evaporation is expected and it was confirmed in
A positive trend of atmospheric pressure was the subsequent analyses (Figure 17).
identified for all months by Mann-Kendall test.
Table 2 - Summary of the Long-Term Variability of Meteorological Variable at Kigali Aero

Weather Station
Mann Kendal Zc Long-term
Parameter Units Average Dry -1 Rainy -1 Dry - 2 Rainy-2 Trend
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec ( 1971 – 2013)
Rainfall mm/year 983.5 1.35 -1.04 1.52 -1.28 -0.54 -0.75 -0.70 -0.18 -1.31 0.93 -0.18 -1.78 Decrease
0C
Temperature 20.5 3.90 3.74 2.40 4.43 4.70 4.58 4.20 4.24 3.10 2.27 5.41 3.88 Increase
Relative
% 73.9 -1.25 -1.78 0.47 -1.41 -0.76 -1.66 -2.16 -1.94 -1.42 -0.61 -1.11 -1.04 Decrease
Humidity
Evaporation mm/year 1438.1 1.80 2.02 0.16 0.36 2.00 3.10 2.57 3.06 1.41 0.67 1.08 0.25 Increase
Sunshine
hours 5.8 -1.64 0.34 -0.82 1.96 0.75 1.46 1.36 2.43 1.57 -1.18 0.71 -0.93 -
Hours
Wind Speed m/sec 2.6 2.49 2.83 1.86 3.58 1.68 2.61 1.26 1.83 1.85 1.59 2.06 1.68 Increase
Atmospheric
mbar 851.3 1.07 1.29 1.84 3.70 2.53 2.03 1.18 2.21 2.84 2.60 -0.22 0.11 Increase
Pressure
7
203 ENGINEER
Further, a positive trend in wind speed and 3. Henninger, S., “Does the global warming
atmospheric pressure is identified. In case of modify the local Rwandan climate?”,Natural
sunshine hours, a significant trend was not Science, Vol..5, 2013, pp. 124-129.
identified.
4. SCEE, “Rwanda‟s climate: Observations and
Projections”, Smith school of enterprise and the
environment, University of Oxford, UK, 2011.
5. Mann, H. B., “Nonparametric tests against

trend”, Econometrica, Vol.13, 1945, pp.245-259.
6. Gocic, M., and Tajkoviv, S., “Analysis of changes

in meteorological variables using Mann-Kendall
and Sen‟s slope estimator statistical tests in
Serbia”, Global and Planetary Change, Vol.100,
Figure 16 - Annual Trend of Mean Annual 2013, pp.172-182.
Temperature and Annual Rainfall for 40 Years
7. Douglas, E. M., Vogel, R. M. and Kroll, C. N.,
“Trends in floods and low flows in the United
States: Impact of spatial correlation”, Journal of
Hydrology, Vol.240, 2000, pp.90-105.
8. Jones, P. D and Hulme, M., “Calculating

regional climatic time series for temperature and
precipitation: methods and illustrations”,
International Journal of Climatology, Vol.16, 1996,
pp.361-377.
Figure 17 - Annual Trend of Annual Average
RH and Annual Evaporation for 40 Years 9. IPCC, “Climate Change: The physical science
basis, contribution of working group I to the
The methodology of this study can be used for fifth assessment report of the Intergovernmental
the developing countries in Asia, Africa and panel on climate change”, Cambridge University
press, Cambridge, UK, 2013.
other continents, where it has a lack in accurate
climatic data and very low technical adequacy.
10. Richard, R and Heim, Jr., “An overview of
Further, it is a need to improve the present weather and climate extremes – Products and
instrumentation techniques in-order to acquire trends”, Weather and Climate Extremes, Vol.10,
accurate data to do the future studies. It is 2015, pp.1-9.
recommended to develop the study further
with the available online climatic data to do the 11. Bocchiola, D., “Long term (1921-2011)
other suggested analysis methods and compare hydrological regime of Alpine catchments in
the results. Northern Italy, Advances in Water Resources,
Vol.70, 2014, pp.51-64.
Acknowledgement
The authors wish to acknowledge to Rwanda

Meteorology Agency for providing data for the
success of this research.
References
1. Nkhonjera, G. K., “Underatanding the impact of

climate change on the dwindling water
resources of South Africa, focusing mainly on
Olifants river basin: A review”, Journal of
Environmental science and Policy, Vol.71, 2017,
pp:19-29
2. Holman, I., “Climate change impacts on
groundwater recharge – uncertainty,
shortcomings and the way forward?”,
Hydrogeology journal, Vol.14, Issue.5, 2006,
pp:637-647.
ENGINEER 8
204
Development of a Flood Forecasting and Data

Dissemination System for Kalu River Basin in
Sri Lanka
S.Kokularamanan, A.W.M. Rasmy, Duminda Perera and Toshio Koike
Abstract: Flood hazard is one of the most frequent hazards in Sri Lanka which brings
devastating damages. At present flood forecasting in major river basins was based on upstream water
levels. This study focuses on the development of Flood Forecasting System (FFS) and data
dissemination system by incorporating improved spatial and temporal resolution of rainfall data and
numerical simulation to proactively predict the floods. Kalu River basin was chosen due to the reasons
that, high rainfall, more frequent flooding and still many people live in the flood-prone areas of the
basin. Since flood forecasting needs forecasted discharge and inundated area, Rainfall Run off
Inundation (RRI) model was used to develop FFS in this study. 2014 daily rainfall data was used for
calibration and 2013 daily data set was used for validation. Lead-time for this basin was found to be
nearly 19.5 hours In order improve spatial and temporal resolution of rainfall data, Global Satellite
Mapping of Precipitation (GSMaP) was used. In order to correct the bias of GSMaP, software
developed by Japan Aerospace Exploration Agency (JAXA) and International Center for Water
Hazard and Risk Management (ICHARM) was used. To determine the minimum number of ground
stations (and locations) needed for optimum bias correction, some combinations of ground daily
rainfall stations were analyzed. Rainfall stations located along the boundary of the basin gave better
results. To disseminate the forecasted data, a web based dissemination system was developed using
FORTRAN, Gnuplot, JavaScript, Openlayers and HTML. Combined automated system would be used
as emergency operation room tool during flood situations.
Keywords: RRI Model, Near-real time flood mapping, GSMaP, JavaScript, HTML
1. Introduction topographic, rainfall trend (highest annual

rainfall) and demographic reasons.
Flood occurs due to high intensity rainfall. Furthermore some studies indicate that
Since Sri Lanka has many rivers (103 major Rathnapura and Kalutara district flood
rivers) with very steep slope and flat lands damages show increasing trend [1] and around
around river mouth (and sometimes even in 132,000 people were living in flood prone area
upstream) flooding occurs in these flat lands. of this basin [2].
Kalu River (shown in Figure 1) is affected more At present, flood forecasts are made using
severely than any other river due to its upstream water levels obtained through
telephone and often criticized for inefficiency
and ineffectiveness. The objective of this study
is to develop a flood forecasting system for
Kalu River basin. Possibility of achieving this
with ground rainfall data and satellite rainfall
data would be considered. Even possibility of
dissemination of forecasted information
through modern information technology also
considered. Overall system would enable
Eng. S.Kokularamanan, B.Sc.Eng.(Peradeniya),C.Eng.
MIE(SL), M.Sc.(Grips-Tokyo), M.B.A.(Colombo), Chief
Engineer, Irrigation Department, Sri Lanka.
Eng.(Dr.) A.W.M.Rasmy, B.Sc.Eng.(Peradeniya), Senior
Researcher, ICHARM, Public Works Research Institute,
Japan.
Eng.(Dr.)Duminda Perera, B.Sc.Eng.(Moratuwa), Research
Specialist, ICHARM, Public Works Research Institute,
Japan.
Figure 1 - Kalu River basin map Eng.(Prof.)Toshio Koike, Director, ICHARM, Public
Works Research Institute, Japan.
205
Irrigation Department(ID) of Sri Lanka to use
this system as an operation room tool during
flood emergencies and might enable the
department to implement similar systems in
other flood prone basins in Sri Lanka.
For numerical modeling RRI was selected as it
could estimate inundation depth, extend and
duration as they were important for flood
forecasting. GSMaP rainfall estimates were
used as real-time rain gauges are limited.
However, it should be noted that if more real
time gauges are available, estimation of flood
forecasting could be improved and forecast
lead time also could be improved by 4.5 hrs.
2. Data and Methodology
Flood event of 1st, June,2014 is used for model Figure 2 - Lead time comparison based on 1st
development and calibration. May,2013 and June 2014 flood event
June data were used for validation. Lead time
Table 1 - Data used for calibration and validation
Daily Data Hourly Data

Period 2012 to 2014 01/05/2014 to
31/07/2014
Number of 18 2
rain gauges
Number of 4 4
river gauges
analysis showed that, discharge from most
upstream (Ratnapura) takes about 40 hrs. to
reach the most downstream area (Putupaula).
However, in the case of tributary (Milakanda)
Figure 3 - Edited DEM (Changed river path)
to Putupaula it takes only 24 hours. Hence
GSMaP data latency is around 4.5 hrs. brings
the overall lead time as 19.5 hrs. Rainfall and
discharge data sets used in this study is given
in Table 1. In order to validate the flood
inundation maps several Moderate Resolution
Imaging Spectroradiometer (MODIS) and ….(3)
Landsat data sets were tried and due to cloud Where equation 1 is discretized mass balance
coverage good results could not be obtained. equation and qxij,qyij are discharges in x,y
Finally and Advanced Land Observing directions. Furthermore, h is height of water
Satellite (ALOS) data set of 3rd, June, 2008 was from local surface and H is the height of water
used for partial validation. Several GSMaP from datum, r effective rainfall and f is the
data sets also used for bias correction infiltration rate. Equation 2 and 3 were based
verification. on diffusive wave approximation of Saint
Although there were several hydrological Venant equation for flood wave propagation
models available, in this study based on the and ka is the lateral saturated hydraulic
available data and purpose simple RRI model conductivity and da is the soil depth times
was adopted. RRI model is governed by both the effective porosity[4]. Using the RRI model,
continuity and momentum equations. FFS was developed by following the
methodology indicated in Figure 4. Initially 3
…(1) arcsec Digital Elevation Model (DEM) was
tried and due to noise and computational
power requirement 15 arcsec DEM was taken
for model development.
….(2)
2
206
DEM had to be edited (Figure 3) for river path bias correction JAXA/NTT/ICHARM
and after sensitivity analysis actual river developed software was used. Normal
dimensions also incorporated into the model. procedure for dissemination of flood alerts, ID
Simulation was done from 1st,April,2014 to had to convey the initial message by fax and
31st,July,2014. Sensitivity analysis and phone call to the DMC. Thereafter DMC had to
calibration were done using the ground pass the information to other agencies.
rainfall in this period and referred to as the However since web based dissemination
flood period. River width and depth system is available to other agencies also (like
parameters showed high sensitivity. Finally it local authorities), they can directly access the
was decided to incorporate the actual river information from the system.
dimensions in order to overcome following
problems. 3. FFS development details
1. Formulae (power formulae which Flood forecast and hazard information is so
dynamic and might be updated each hour
represents the river dimensions) assumes that based on the rainfall data. Therefore this type
river is expanding as it reaches the sea, of dynamic information might need dynamic
whereas actual river remains with fairly information dissemination system. Hence web
constant width in the upstream reach also. based forecast dissemination system might be
the most suitable. Most important advantage
2. Narrow point at Ellagawa cannot be of this system is interactive web based maps.
presented in the formulae method. Bias corrected GSMaP real-time data was used
After incorporation of actual river dimensions to run simulation of RRI as specified in
and adjustment of rainfall records, calibration methodology flow chart in Figure 5. In order to
make discussion easier system was broken into
was redone. In order to improve hourly data 5 components. FFS has 5 main components
(Figure 5).
Component 1: Pre-processing component of
the system details
Since system needs hourly data and hourly
rainfall data is sparse GSMaP data was used.
First part of GSMaP processing starts with
FORTRAN based executable. Based on the
input files from GSMaP it rewrite the last hour
and number of out files, in RRI control file
(RRI_input.txt). Beside this it also modifies
certain plot files which would be used as input
files for Gnuplot in the rain subfolder in order
discharge curves with starting time.
Figure 4 - Steps adopted in the study Furthermore system would prepare GSMaP
rainfall distribution for all 88 points with
resolution GSMaP was used. However GSMaP
temporal details.
has some systematic errors and this had to be Component 3 :Post processing
corrected. This is called bias Correction[5]. For Two important post processing functions were
done by FORTRAN based executable. They
are,
1. Producing predicted discharge and
water level information
2. Automated warning information
Figure 5 - Flood forecasting and Dissemination

system
Figure 6 - Inundation map produced at runtime

207
period of April 1st to July31st of 2014 using
ground rainfall data from Irrigation
Department. Initial calibration was not good.
However after incorporation of actual river
dimensions into RRI and a shift in the daily
Figure 7 - Warnings issued by the system
based on the threshold values given in the

location.txt Figure 9 - RRI simulation on 3 arcsec DEM
Gnuplot prepares the inundation map at
runtime with script as shown in Figure 6. User data was corrected recalibration results were
can set two levels of threshold for each station. good Nash Sutcliffe Efficiency (NSE) 0.91 and
Hence based on these threshold values, system Root Mean Square Error (RMSE) 56 m3/s at
would raise warning messages in the web site Putupaula .For validation 2013 data was used
as in Figure 7. (NSE 0.67 and RMSE 72 m3/s). Similarly 3
arcsec DEM model also was developed and
Component 4: Web technologies same calibration parameters were applied.
Openlayers, JavaScript, HTML and KML were
used. Inundation map is based on openlayers
and gives the user ability to dynamic zoom
and panning of the map. Demo is available on
the internet site http://kalu.cu.cc. DOS based
automation script was also developed.
Therefore if the system is able to retrieve the
DIAS data web site would be updated hourly
automatically.
Component 5: Directory structure and

automation
Directory structure of the system designed in Figure 10 - Inundation Map produced from
fixed manner so that automated script can 15arcsec simulation for the flood period of
execute each processes and transfer the 2014
relevant files to other processes. Therefore it is
a must to keep the sub folders name as it is.
4. Results
All these simulations were done for the
Figure 11 - Inundation area retrieved from

ALOS 2008 June3rd data set in blue colour
Figure 8 - RRI simulation on 15 arcsec DEM

4
208
basin boundary yielded good result (Figure 18).
Based on the basin area, minimum number
could be decided. For development of the web
based dissemination system, DOS based
scripting was used, since DOS based
computers are most common in irrigation
department. Automation script would run as
windows/DOS services for every hour and
web site would be updated automatically on
hourly interval.
Figure 12 - DMC Sri Lanka report on hazards

in 2012 inundation map (DMC, December
2012)
However, first model (15 arcsec- Figure 8) gave
good performance mainly Putupaula and other
station showed lesser NSE. Second model (3
arcsec –Figure 9) showed similar performance
for all four station and showed better average
NSE. In order validate the inundation MODIS,
Landsat ALOS (Three satellite image sources)
data were tried. MODIS and Lands at were
covered with clouds and could not get proper
inundation maps. However ALOS data for Figure 13 - Group 1
2008 June 3rd flood event gave partial
inundated area of Kalu River basin (Figure 11).
However Disaster Management Center (DMC)
published inundation Map (Figure 12) also
been compared.
GSMaP bias correction
As the hourly data for bias correction is from
automated gauges it is necessary to decide the
minimum number of stations and location of
the gauges to get the optimum bias correction.
For this five groups combinations were tried. It
was found that gauges located along the
boundary gave better bias correction (Figure
13 and 14).
5. Discussion Figure 14 - Group 2

Sensitivity analysis showed that river
dimensions plays important role in model
output. For flood inundation validation
satellite data was used. Since Landsat and
MODIS data during the flood period
concerned were mostly covered with cloud, as
a final resort 2008 June 3rd ALOS SAR image
and DMC [3] published map was used for
validation. During initial analysis of bias
correction parameter’s sensitivity analysis
grouping does not work well for this basin.
This may be due to the small catchment area.
To determine minimum number of the stations
required for optimum bias correction, 5
combinations (Figure 13 to 17) of 18 ground
stations were used. Basins located around the Figure 15 - Group3
209
Acknowledgement
Author wishes to acknowledge my heartiest

gratitude to the supervisors at ICHARM who
gave me the guidelines that enabled me to
complete this research work. Furthermore, he
wishes thank to JICA, which made financial
support and this opportunity to undertake this
research. He takes this opportunity to thank ID
for giving him duty leave for one year for this
research.
References
1. ADPC (2016, July 22), Urban Flood Risk

Figure 16 - Group 4 Mitigation in Kalutara City Case Study.
(Asian Disaster Preparedness Centre) Retrieved
July 22, 2016, from Asian Disaster Preparedness
Centre:
http://www.adpc.net/igo/category/ID222/do
c/2013-qvl8PC-ADPC-Safer_Cities_23.pdf
2. ADRC. (March, 2009). Comprehensive study on

disaster management in Sri Lanka. GOSL &
JICA.
3. DMC. (December 2012). Hazard Profiles of Sri

Lanka. Disaster Management Center of Sri
Lanka. Colombo: GOSL.
4. Sayama, T. (2015). Rainfall-Runoff-Inundation

Figure 17 - Group 5 Model ver. 1.4.2 user manual. Tsukuba, Japan:,
International Center for Water Hazard and Risk
Management (ICHARM) and Public Works
Research Institute (PWRI).
5. JAXA (2014). GSMaP Customization Interface

real time correction function Guide. JAXA.
Figure 18 - Comparison 5 group combination

Furthermore, when considering the DEM, due
to the noise in the 3 arcsec 15 arcsec was used
in this study. 3 arcsec model could be used for
forecasting if accurate DEM and enough
computational power is available.
Dissemination system could be further
improved by incorporating SMS and email
alerts. Development of simple web integrated
RRI model as described in this paper, for flood
prone basins might enable the stakeholders of
the basin to do better flood management.
6
210
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[211 217],2017
2017
The Institution
Lanka
A GIS Based Asset Management Approach to Upgrade

the Level of Service to Pipe Borne Water Consumers in
Western North Region
A.M.H.K. Abeykoon and B.U.J. Perera
Abstract: Geographic Information System (GIS) is a technology with a powerful set of tools that
many organizations are using to get a better grasp on their data. GIS is capable of organizing complex
water utility data into a clear visual representation that would allow water utility agencies such as
National Water Supply and Drainage Board (NWSDB) to make efficient operational decisions to
ensure a sustainable water supply. NWSDB implemented GIS and remote sensing based water utility
management project in the Western-North region to address the entire lifecycle and operations of
water infrastructure including, distribution and transmission pipeline mapping, design and asset
management, leak survey management, customer database management, water hydraulics, outage
management, operation &maintenance (O&M) management, field automation and operation
intelligence. Using the satellite images, we have prepared the low scale base maps. On top of the base
maps, all system components such as water mains, storage reservoirs, pump houses, pipelines, and
valves were digitized. Simultaneously, consumer locations and leak locations were marked using GPS
instruments. Water consumer related other attributes such as account number, address, contact details
were stored in the database. Water leakages related attributes such as diameter, leak type, repair
details were also stored there. Subsequently, organized data is utilized for O&M work by the Area
Engineer & OIC offices via more interactive interfaces such as Google earth, Google Map etc. Further,
we have started hydraulic modelling work. Hydraulic models are vital for the asset related decision
making. This whole process is incessant that we need to keep on enhancing & updating the current
knowledge base continuously with a trained O&M staff. Successful completion of this project will
enable water utility managers to make quick and effective judgments on routine operations that had
been extremely resource and time intensive during the past. Future focus areas include integrating
information from sensors, meters with the GIS maps, sharing interactive maps in the boardroom and
making functional maps available to field devices.
Keywords: GIS, Maps, Operations & Maintenance, Water Utility Management, Water
Distribution
1. Introduction system is yet to be implemented in order to

utilize them in the most efficient manner to
Drinking water supply to the Cities North of maintain a high level of service.
Colombo Region is mainly divided into six Water Asset Management (WAM) is meant to
water supply schemes namely; Biyagama, meet the required level of service, in the most
Mahara, Kelaniya, Welisara, Ragama and cost effective manner. Geographical
Ekala. The main reservoir is Church Hill, Information System (GIS) is one of the popular
which is fed by Kelani Right Bank water and worldwide accepted asset management set
treatment plant. The total region is fed by three of tools that can be used to achieve a better
main transmission pipelines coming down grasp of spatially located asset data. GIS is
from the Church Hill reservoir and connected capable of organizing complex water utility
to several water towers and ground reservoirs data into a clear visual representation that
which are located in high elevations. Main would allow water utilities such as National
transmission systems and distribution system Water Supply and Drainage Board (NWSDB)
consist of more than 3000 km of DI, CI and to make efficient operational decisions to
PVC pipelines, of which the diameters ranging ensure a sustainable water supply. GIS and
from 63mm to 1200mm. The system serves Eng. A.M.H.K. Abeykoon,, C.Eng., MIE(SL), B.Sc. Eng.
more than 175,000 domestic and commercial (Moratuwa), MSc, Civil Engineer, National Water Supply
type consumers. Though the NWSDB & Drainage Board.
maintains a massive quantity of assets such as Eng. B.U.J. Perera, C.Eng., MIE(SL), B.Sc. Eng.
(Moratuwa), M.Sc(Netherlands), Chief Engineer, National
pipelines, different types of valves, flow Water Supply & Drainage Board.
meters etc., a suitable asset management
1
211 ENGINEER
remote sensing based water utility platform becomes pivotal to ensure sustainable
management projects are implemented in the water utility operations. Water utilities can
Western-North region to address the entire reap following benefits by adopting GIS:
lifecycle and operations of water  Comprehensive and up to date asset
infrastructure, including distribution and information reduces time in the field
transmission pipeline mapping, design and spent locating assets; improves
asset management, leak survey management, planning for maintenance, renewal,
customer database management, water and growth.
hydraulics, outage management, O&M  GIS enables network analysis for
management, field automation and different functions such as system
operational intelligence. tracing and hydraulic modelling.
 Field GIS assists logistics through less
2. Literature Review work force, minimal travel time and
costs, integrates work flows, and
Water Asset Management (WAM) has been ensure real-time information flow.
identified as a responsible and proactive  Web GIS provides a single view of
approach for water infrastructure utility that integrates information from
management, efficient water utilisation and multiple sources into intuitive as a
minimisation of waste.WAM enables delivery map and dashboard. Web GIS
of institutional objectives in the optimal and applications can be integrated with
sustainable manner at the least lifecycle cost. many other systems used by the
Further more in today's environment of ageing organization.
infrastructure, an ageing work force,  Portal technology allows data
increasing regulations, and transparency accessible to everyone in the utility,
requirements of the communities served, allowing them to collaborate with
achieving sustainable water and wastewater contractors and other agencies. People
services require a geography-based platform. across the department can discover,
Asset management based on GIS transforms a use, make, and share maps on any
water utility into a more informed, efficient, device, anywhere, anytime.
fiscally responsible, and transparent
organization Water Loss Management
GIS facilitates water utilities to create, Age of the water system infrastructure is one
organize, and share geographic information of the main reasons for major breaks and
and tools within the organization on multiple chronic leaks. Using a properly updated GIS
devices at both office and field. It enables database, high leak prone areas due to the
everyone in the water utility sector to access aging pipes could be spatially identified, thus
the information they need and provides supporting Spatial prioritization of these
means to collaborate with other employees information to prioritize capital spending with
regarding the location and status of assets, objective evidence.
staff, and customers. ArcGIS is the most White House Utility District (WHUD) in the
commonly used software for water asset USA, adopted GIS to display water system
management aspects. ArcGIS can be used in asset information via interactive web maps.
multiple devices from desktops, the web, This information helped decision makers in
smart phones, and tablets and may run on any prioritizing projects based on data-driven
platform (i.e. local network or hosted in interpretations of the spatially enabled data.
clouds).It includes enterprise portal and Consequently, WHUD achieved significant
geospatial content management platform that operational cost savings in the first year.
supports online mapping and geographic Enterprise GIS enhanced asset management
analysis. This approach allows institutions to decision making that saved $32 million for
easily and securely manage content, while WHUD customers.
sharing the information with the key Moreover, undetected leaks cause major
stakeholders. resource loss, which can be crucial in water
Functions of water utility are unique per restricted regions. GIS platform offers a suite
location considered, such as distribution and of water-loss detection and response solutions.
collection networks, pressure zones, service By using advanced technology to isolate and
areas, customers, facilities, and crew locations. repair leaks, utilities avert costly acoustic
Hence, need for geography based data surveys and step testing as well as the even
ENGINEER 2
212
costlier consequences of long term undetected excavations were required to check the actual
leaks. Through the ArcGIS based leak- condition of pipes, valves and their
detection solutions enabled WHUD to save connectivity with each other, especially in
$37,000 per leak with total annual savings of junctions.
up to $219,000. Above data were centrally stored in a network
server in the geodatabase format. After
3 Methodology completing the databases of each scheme, it
was handed over to O&M staff, and they were
Some of the Satellite images of Gampaha trained to update the subsequent system
District were received from the UNDP, and the modifications. In addition, they were also
balance was downloaded from Google Earth. trained to map the locations of leaks within the
Those images were geo-referenced using distribution system, and a data format was
ArcGIS to be used as the background images given to collect details of every leak except
in the digitizing process. Road centre lines, service leaks. We commenced marking
road edges, railway lines, landmarks, consumer locations in a spatial database using
buildings etc. were digitized subsequently and GPS machines with the help of meter
those were used as the base maps to produce readers(Figure2). They were trained to
the water utility maps. On top of the base carryout this work parallel to their routine
maps, all water utilities such as water mains, duties. Building hydraulic models for the
storage reservoirs, pump houses, distribution existing schemes were started from the
pipelines, all kinds of valves were marked and Kelaniya Area by using Water GEMs software
relevant attributes of each type of asset were integrated with ArcGIS. Nodal elevations were
recorded (Figure 1). Main attributes collected extracted from Digital Elevation Models and
and recorded were the pipe diameter, pipe digital contours obtained from the Survey
material, type, constructed year, valve Department. Consumer pack boundaries were
position(open/close), flow meter type etc. marked in the model with the help of zone
Some of those data were collected from the officers. Nodal demands were estimated using
NWSDB mapping section. Data collected from the relevant pack demands and assigned to the
UDA and the Survey Department were also nodes by This sen Polygon method in
used in succeeding this work, especially for the WaterGEMs software. Peak demand factors
low scale base maps preparation work. were calculated by analysing logged flow data
Accuracy of the collected data was verified of critical points of the system for several days.
from the field sources. In certain locations, site Subsequently, the model was calibrated and
Figure 1 - Digitized Water Assets
3
213 ENGINEER
Figure 2 - Consumer Locations
verified using actual field data. future demands. Following parameter values
were used for the analysis.
Water quality models were also developed
using the relevant parameters to simulate the Bulk Reaction Rate -0.10(mg/L)^(1-n)/day
system water age and residual chlorine First Order Wall Reaction Rate -0.08 m/day
variation. Age analysis was performed for Diffusivity 1.2e-9m2/s
water in the system, assigning an initial age Initial chlorine concentration 0.2mg/l
as 0 for all nodes. There is an infinite supply of
source water (from the reservoir). The age of Water quality model was calibrated with the
water elsewhere in the system at a point in actual RCl values measured at site.
time represents the time from the start of the
run and when the water left the reservoir. The 4. Results and Discussion
analysis will be conducted run for 14 days (336
hours), in order to determine the equilibrium Using the updated database, we are basically
point of the system. Water age vs time graph capable of answering any type of query on
was used to monitor the time taken to stabilize those assets and make recommendations on
the system age. system rehabilitations or upgrading based on
Once a repeating pattern is reached, it can be available data analysis. The O&M staff could
observed that the age of the water fluctuates make use of these data for their related
between two fixed values (for a 24-hour activities such as pipe extensions, valve
period). After that, the water age of the system location identification in case of a leak, map
is supposed to be in equilibrium. Hence, the preparation work etc. Through leakage
age value could be estimated at any given time mapping, vulnerable areas for pipe bursts
period. The samples were collected in the could be identified, and remedial actions could
morning from 9 am to 12.30 pm. Therefore, the be proposed. In addition, cost of leakage
average age within that period was taken for repairs and loss of water could also be
the analysis. calculated.
The constituent analysis was carried out to NRW management through the establishment
monitor the chlorine residuals in the system of District Meter Areas (DMAs) is also another
over time under present demand as well as advantage that could be achieved by utilising
these data which we have planned to start
ENGINEER 4
214
Figure 3 - Displaying Valves on Google Earth
soon. Consumer complaint mapping and water Conclusions

quality mapping are also another two activities
we are planning to implement in the near Water Asset Management is a timely need for
future. Since all these assets could be viewed the sustainable development of the water
through Google Earth application, those data sector. The importance and the benefits should
has become more interactive and user- be convinced to the higher management of the
friend(Figure 3). Further, we expect to develop organization in order to maintain a continuous
a web based utility mapping system in order to progress and growth of this particular work.
benefit the mobile workers. Key advantages are the optimized usage of
After completion of consumer location assets, reduced wastage, water loss reduction,
marking work, we can contribute more to the asset renewal planning and communication
operational and design work, including with customers and enhance the quality of
supporting disconnection programs, service in respect of pressure, serving hours,
identification of defective meter locations, spot consumer response time, accurate and timely
estimations for new connections, identification billing etc. Further based on the risk and
of consumers affected by an interruption, valve growth assessments renewal, rehabilitation,
detection in case of a leak, demand estimations upgrade, and condition assessments could be
for hydraulic models etc. Hydraulic behaviour justified through a properly implemented asset
of the system could be identified through management process
hydraulic models and could propose remedial
actions for any underperforming situations.
Further, it will replicate the system under
various demand loading and operating
conditions, thus facilitate the decision making
processes. Hydraulic models and water quality
models could be used to monitor water quality
related parameters associated with RCl and
water age. Through that, we are able to trace
the areas where water stagnates or not having
residual protection (Figure 4,5 & 6).
5
215 ENGINEER
Figure 4 - Pressure Variation of the system at 10.00hrs
Figure 5 - Water Age Variation of the system at 10.00hrs
ENGINEER 6
216
Figure 6 - RCl Variation of the system at 10.00hrs
http://solutions.arcgis.com/water/documentati
References
on/,Visited , 05th May 2017
1. Booth, R. and Rogers, R., 2001. Using GIS

technology to manage infrastructure
capitalassets. Journal AWWA, 93(11), 62–68.
2. Ferreira, A. and Duarte, A., 2006. A GIS-based
integrated infrastructure managementsystem.
FIG Portugal. Proceedings of the Institution of
Civil Engineers. ISSN 0965-0903 159(2), 113-120
3. Garaci, M., Sutherland, J. and Mergelas, B.J.,
2002. Role of GIS and data managementsystems
for pipeline integrity. Paper presented at Pipeline
Division SpecialtyConference, Cleveland, OH.
4. Luettinger, J. and Clark, T., 2005. Geographic
Information System–Based Pipeline Route
selection Process. Journal of Water Resources
Planning and Management,131(3), 193–200.
5. City of Tampa, 2006. Tampa GIS strategic plan.
Tampa, FL: City of Tampa.
City of Tampa, 2008a. City of Tampa detailed

profile [online]. Available from:
http://www.city-data.com/city/Tampa
Florida.html, Visited5th September 2009].
\
City of Tampa, 2008b.City of Tampa GIS

strategic plan [online]. Available
from:http://www.tampagov.net, Visited,5th
April 2008].
6. http://www.esri.com/industries/water
http://www.esri.com/industries/water/water
7
217 ENGINEER
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[219 225],2017
2017
The Institution
Lanka
Application of 3D Physical Modelling for Verification

and Optimization of Hydraulic Designs for Salinity
Barriers in Sri Lanka
S.M.C.K. Subasinghe, K.P.S.T. Pathirana, N.L. Engiliyage and G.H.N. Rathnasiri
Abstract: Salinity Intrusion has become a major problem in several river basins of Sri Lanka
which are immensely used for potable water supplying. At present, the feasibility of non-structural
measures to get rid of this salinity intrusion problem is either impractical or economically non-viable
for many of these river basins. Hence, all preventive measures involve the construction of a full or
partial barrier to prevent the intrusion of saline wedge to main raw water intakes during low flow
period. However, the design of salinity barriers need to be done with minimum or zero impact to the
environment paying special attention to possible increase of flood levels, morphological changes to
river stretch with siltation and bank erosion etc., during flood events. Basic hydraulic design of
salinity barriers are carried out and checked by means of hydraulic models of both numerical and 3D
physical models. First, the conceptual model is developed and impacts are assessed with numerical
models. The physical model studies are carried out to verify or optimize the design parameters. The
objectives of the 3D physical modeling are the investigation of river hydraulics in the vicinity of
proposed salinity barrier location, estimation of head losses for different flows and suggest necessary
improvements to the barrier layout, assessment of possible water level rises during construction
stage, investigation of sediment regime in the river and investigation of impact of the barrier on
sedimentation and erosion pattern of river bed and bank, and finally the determination of the type of
erosion protective measure needed and its extent along the river reach. This paper discussed the
capabilities of 3D physical models applied to engineering studies, mainly for the construction of
salinity barriers for the purpose of locating water intakes. Further it elaborates the methodology on
verification of numerical model results and optimization of the hydraulic design of salinity barrier by
means of 3D Physical Model studies.
Keywords: Salinity Intrusion, Structural Solution, Flow Pattern, Siltation, Erosion, Fixed Bed
Model Testing, Movable Bed Model Testing
1. Introduction extent that some of the water intakes in rivers

are used occasionally. Hence, either non-
Salinity intrusion is the sea-water driving structural or structural measures are required
upstream of the river through the mouth by to be implemented to get rid of this problem
rising tides which resembles a form of a saline and provide safe drinking water.
wedge. Usually the river flow is large enough
to keep the salt water front at bay. But during Comprehensive studies have been carried out
very low flow periods due to drought or in recent past for water intakes that are badly
upstream water abstraction, salt water can
travel upstream. The extent or the length of Eng.(Mrs.) S.M.C.K. Subasinghe, C.Eng., MIE(SL),
salinity intrusion depends on the density B.Sc.Eng.(Peradeniya), M.Eng.(AIT), Engineering
Manager, Lanka Hydraulic Institute Ltd, Katubedda,
difference of fresh and seawater, fresh water Moratuwa.
flow rate, the depth of flow and the tidal range.
A higher ‘minimum flow’ is required to counter Eng.(Ms.) K.P.S.T. Pathirana, AMIE(SL), B.Sc.Eng.
act salinity intrusion into the river. Sand mining (Peradeniya), Research Engineer, Lanka Hydraulic
Institute Ltd, Katubedda, Moratuwa.
in rivers has a significant impact on saltwater
intrusion due to deepening of the river bed. In Eng.(Ms.) N.L. Engiliyage, AMIE(SL), B.Sc.Eng.
certain river basins of Sri Lanka, salinity (Moratuwa), Research Engineer, Lanka Hydraulic
Institute Ltd, Katubedda, Moratuwa, Sri Lanka.
intrusion has become a major problem and
restricts the water abstraction from the river Eng.G.H.N. Rathnasiri, AMIE(SL), B.Sc.Eng.
causing difficulties in supplying potable water (Moratuwa), NWS&DB, Water Supply and Sanitation
to public[2,3]. The frequency of occurrence of Improvement Project, Badulla District Support Unit,
Badulla.
salinity intrusion has been increased to the
1
219 ENGINEER
affected by salinity intrusion problem using This paper describes the application of 3D
possible non-structural measures such as physical model tests related to construction of
relocation of the intakes in further upstream, salinity barriers across Nilwala and Kalu
finding new alternative water sources etc [2]. Rivers. The numerical model studies were done
However, for some of the water intakes located using the MIKE 11 software and results were
in the river basin like Nilwala etc, the feasibility compared and verified with the physical model
of non-structural measures to get rid of this studies.
salinity intrusion problem is either impractical
or economically non-viable under the current 2. Objectives of 3D Physical Model
context. Therefore, as a solution, structural Studies for the Salinity Barriers
measures such as salinity barriers have been
across Rivers
proposed to get rid of this salinity intrusion
problem. In general, the salinity barrier type
 Investigation of river hydraulics in the
and location has to be decided with great care
vicinity of proposed salinity barrier
for stability of the structure while minimizing
locations.
the cost of the structure, cost for mitigating the
 Investigation of head losses for different
environmental impacts due to the introduction
flows and suggest necessary
of the structure and minimizing the flood
improvements to the barrier layout.
impacts too. When designing salinity barriers,
more attention should be paid for estimating  Assessment of possible water level rises
key design parameters and other factors such during construction stage.
as, locating the gated salinity barrier with small  Investigation of sediment regime in the
or zero impact on flooding, navigation and the river and investigation of impact of the
environment considering the 100 year (yr) sea barrier on sedimentation and erosion
level rise to prevent the salt water intrusion into pattern of river bed and bank.
water intakes. The conceptual design for  Determine type, size and length of the
salinity barriers are formulated to conform to upstream and downstream protection of
the hydraulic design criteria established the proposed barrier & effectiveness of
subsequent to the hydrologic and hydraulic bed protection.
studies. Design verifications prior to the
construction are important as millions of rupees 3. Theory Related to Physical
are generally spent on constructions and design Modelling
failure would be a loss and create harmful
impacts to the people. The theory of physical model is related to the
laws of similitude. There are three types of
3D physical models are generally proposed to similarities; Geometric similarity, Dynamic
verify the designs in addition to the application similarity and Kinematic Similarity. For a
of 1D or 2D numerical models. Application of proper physical model, all these three
physical model would produce a physical similarities need to be satisfied. For flows with
system of required study area which is scaled a free-surface, inertia and gravity forces are
down so that the major dominant forces acting usually dominant, and the Reynolds number
on the system are represented in the model in effects can be neglected if the flow is turbulent.
correct proportion to the prototype. Hydraulic Consequently, the requirement for
model studies with a 3D physical model to hydrodynamic similitude in a river system
verify the design parameters and observations between the model and the prototype is the
on flood levels, velocities, bank erosions etc., is Froude similarity:
more suitable as the appropriate equations
governing the processes will be considered (Fr) Model = (Fr) Prototype
without simplifications as in numerical models.
(V/√gd)m = (V/√gd)p
The results obtained from numerical modelling
system are verified by means of physical model
Where, d is the depth of water (m), g is the
for the effectiveness of the design. Further,
gravitational acceleration (ms-2), V is the mean
visual observation can be made during physical
velocity (ms-1) of flow and the subscripts m and
model testings and reproduction of complex
p, denote the model and the prototype,
boundary conditions can be done whereas non-
respectively.
linear effects, major difficulties are encountered
during numerical modelling.
The similitude ratios are obtained using the
above equations and the necessary model
ENGINEER 2
220
parameters are calculated. The Reynolds cohesive sediment range whose behaviour
numbers of the flow corresponding to each test would be completely different that of the real
condition is calculated to ensure that the flow is situation. Therefore, instead of scaling down
turbulent. the sediments according to the model ratio, the
available sand with mean diameter of 0.12 mm
The main scale ratios used for model was used for both these studies. As the
parameters are given in Table 1 where the sediment sizes are not scaled down according
selected geometrical scale ratio is 1: X,. to the similitude ratio, the model results such
as, erosion and deposition patterns were
Table 1 - Similitude ratios for main model interpreted only in qualitative terms.
parameters
Scale
Parameter
(model: prototype) 4.2 Model Approach
Length 1: X
Area 1: X2 The model tests are usually conducted in two
Time and Velocity 1: √X stages; for fixed bed and movable bed. From
Discharge 1: X5/2 the fixed bed model, velocity and water level
measurements and flow pattern investigations
4. Main Steps of Physical are obtained while the movable bed model is
used to get erosion patterns along the river
Modelling Process
banks and river beds. The fixed bed model tests
The main steps related to physical modelling
are carried out with and without the salinity
are;
barrier while the movable bed tests are carried
i) Scaling of the model
out only with the salinity barrier. In order to
ii) Model approach
check the river hydraulic changes during the
iii) Model construction
construction of the barrier, tests are carried out
iv) Model calibration
for coffer dam phases. The purposes of each
v) Model testing
testings are given in Table 2.
4.1 Scaling of the model
Table 2 - Purpose of Model Testing under
different Bed conditions
A model can be either distorted or undistorted.
Both Nilwala and Kalu river models were
selected to be undistorted models because these Condition Purpose
studies required observing the sedimentation
Fixed Bed  Calibration of the model:
patterns with movable bed model. Factors
Without  Observation of flow patterns
taken into consideration when selecting scale
Barrier and velocity distributions
ratios were;
under existing condition
 To minimize scale effects
Fixed Bed  Investigate head losses for
 Convenience for the observation of flow
With Barrier different flows and suggest
patterns and discharge characteristics
(Gates in necessary improvements to
 Available basin area
Open & Gates the barrier layout
 Available pumping capacities
in Closed  Assessment of the water
position) level rise during flood
The appropriate model scales for Nilwala and conditions and construction
Kalu rivers were 1:40 and 1:60, respectively. stage
Further, these selected model scales ensured
that comparatively large pooling area is Mobile Bed  Investigate impact of the
available at upstream and downstream of the With Barrier barrier on sedimentation and
model area to get a smooth flow and a (Gates in erosion pattern of river bed
sufficient depth for scour in the bed. Open, Higher and bank
Flows)  Effectiveness of bed
The extent of the flood plain to be modelled protection
and area to be bounded were calculated by
considering the 100 year return period flood.
Scale down of sediment sizes has been a typical 4.3 Model Construction
problem in physical model studies as the In order to represent the actual flow conditions
modelled sediments would eventually fall into in the model properly, the bed surface is
3
221 ENGINEER
finished in a manner, which would have the water levels and flow velocities are checked
same relative roughness as for the prototype with the generated water levels by numerical
condition during the fixed bed model tests. model studies. In case of any discrepancies
Adequate pooling areas are provided in incurred, the bed resistance of the river is
upstream and downstream of the river (Figure changed accordingly with appropriate
1). modifications such as applying paints to
smooth the bed and make is coarser with the
use of sand/silt for the model. Figure 3 and
Figure 4 show the calibration check against
water levels and velocities respectively.
Figure 1 - Fixed Bed Model with Proposed

Salinity Barrier (for Nilwala River)
For movable bed models, only the required

Figure 3 - Comparison of observed U/S water
river stretch from upstream and downstream of
levels under existing condition
the barrier are filled with appropriate sand. The
length to be filled is decided according to the
geometry of the river (Figure 2).
Figure 4 - Comparison of observed U/S

velocities under existing condition
Table 3 - Flow Conditions for Model Tests
Figure 2 - Movable Bed Model with Proposed Flow Condition Details

Salinity Barrier (for Nilwala River)
Fixed Bed Model
The water circulation system is planned in a 2 yr Return Period
way that it provides a smooth, continuous flow. 10 yr Return Period
The required model discharge (upstream Flood Flows 25 yr Return Period
inflow) was supplied by means of a regulating 50 yr Return Period
weir. A thin plate rectangular weir was used for 100yr Return Period
higher discharges while V –Notch was applied 5% Exceedance Flow of FDC
High Flows
for low discharges [01]. Downstream water 10% exceedance Flow of FDC
levels were also controlled by a regulating weir.
75 % exceedance Flow of FDC
Low and
4.4 Model Calibration and Testing Average Flows Required minimum Flow to
flush the salinity wedge
It is required to ensure that physical model Movable Bed Model
represents the actual hydraulic behaviour of the Flood Flow 50yrReturn Period
study area without been affected by the scale
effects etc. Model calibration is done for
existing condition with fixed bed model. Both
ENGINEER 4
222
Once the model is calibrated, testings are done The orientation of salinity barrier is proposed in
for the selected flow conditions. The different a ways that it does not obstruct the stream lines
flow conditions used for both Nilwala and Kalu which were observed under existing condition.
river models are illustrated in Table 3. According to the Figure 6, it can be seen that
the proposed orientation of the salinity barrier
High, low and average flows were defined is perpendicular to the streamlines. Therefore,
based on the Flow Duration Curve (FDC) at no changes had to be done for the location.
salinity barrier site. Steady state flow conditions
are generated in the model for the selected flow 4.2.2. Verification of Estimated Flood Water
conditions. Once the flow becomes steady, the Levels
water levels were measured at the selected
locations in the model area. During the model tests for the existing
condition with the fixed bed model, the
4.2.1 Observation of River Flow Pattern estimated flood levels from the numerical
model studies were verified. Thereafter, tests
Flow velocities were measured across some are carried out with the presence of salinity
selected profiles of the river and the flow paths barrier (Figure 7).
along the river were recorded by means of float
tracking. Finally float tracking was carried out
to check the streamline flow patterns. Figure 5
shows the observed flow pattern in Nilwala
River without and with the proposed salinity
barrier corresponding to 5yr return period
flood [2]. Streamlines are obtained with float
tracking (Figure 5).
Figure 7 - Simulation of Flood Condition with

gates opened position
Figure 5 - Float Tracking
With Proposed Salinity

Existing Condition
Barrier
Figure 6 - Streamline pattern for 5yr RP Flow

5
223 ENGINEER
Once the tests are completed for all scenarios,
the results are compared with the numerical
model results.
Table 4 - Comparison of Generated Extra

Head up with Presence of Salinity Barrier for
a Selected River
Water Level Rise (m) at

Upstream location of
Flow Observation
Condition
Numerical 3D Physical
Model Model Figure 8 - Simulation of Flood Condition with
partial gate opening condition
(2yr RP) 0.09 0.11
(10yr RP) 0.07 0.06
(25yr RP) 0.07 0.06
(50yr RP) 0.07 0.07
(100yr RP) 0.08 0.09
Referring to Table 4, it can be observed that the

water level rises due to the presence of the
Figure 9 - Model Simulations to observe
salinity barrier as simulated by the numerical impacts during the construction stage
model were validated with the 3D physical construction
model results [03]. The difference of anticipated
water level rises for different return periods are 4.2.3. Identification of Vulnerable Areas for
interrelated with the stage – area relationship of Deposition and Erosion
river cross sections at and in the vicinity of
proposed salinity barrier location. Further, once The objectives of the model runs with a
the magnitude of flood is increased, movable bed are to investigate the scouring/
considerable amount of water goes through a deposition patterns at upstream and
larger flood plain and there is a possibility that downstream of the proposed salinity barrier
the sill of the structure will become to fully structures and to decide on required lengths for
drowned condition. river bank and bed protections. Refer Figure 9.
Moreover, 3D model simulations facilitate to

study the several other scenarios such as, when
the gates are kept in partially or fully closed
position during a flood scenario (refer Figure
08). With this kind of observations, the
significance of having a proper gate operation
rule is clearly understandable. In addition, 3D
physical models also facilitate the observation
of water level rises during the construction
stage with coffer dam (Figure 8). This type of
model tests would lead to recommend proper
construction methodology with minimum
impacts.
Figure 10 - Identification of Vulnerable areas
for erosion and deposition
ENGINEER 6
224
5. Conclusions Acknowledgement
3D physical model studies facilitate the Authors would like to extend their gratitude for
engineers to observe and understand the river Lanka Hydraulic Institute Ltd for providing the
flow patterns before and after construction of resources needed to carry out this study and
salinity barriers across rivers and confirm the NWS&DB for the extended assistance.
hydraulic design of the structure with
minimum impacts to the environment.
Moreover, the visual observations and References
measurements of water levels and velocities
also provide a guideline for operation of the 1. Acker, P., White, W. R, Perkins, J. A., and
salinity barriers. Harrison, A. J. M., ‘Weirs and Flumes for Flow
Measurement’1978.
This paper discussed the capabilities of 3D
2. Lanka Hydraulic Institute Ltd. (2016),
physical models applied to engineering studies, Consultancy services for construction of Salinity
mainly for the construction of salinity barriers barrier across Nilwala River.
for the purpose of locating water intakes. In this
3. Lanka Hydraulic Institute Ltd. (2017),
regard, physical model studies already carried
Consultancy services for construction of Salinity
out for the proposed salinity barriers at Nilwala barrier across Kalu River.
and Kalu Rivers were taken as examples. With
the completion of both fixed and movable bed
model tests, following conclusions could be
made based on the test results.
 Appropriateness of the selected
location; the selected location of the
barrier is suitable since it does not
change the flow pattern during both
average and high river flows
significantly. The optimization of the
salinity barrier axis was done.
 The dimensions of the proposed
structure and sill levels are satisfactory
considering the observed water level
rises.
 Observed water level rises in the
physical model tests are in line with the
numerical model observations.
 The shape of the piers, with arcs at
upstream and downstream end caused
smooth flow. Consequently, the
hydraulic performance of the barrier is
acceptable.
 The boulder sizes selected for upstream
and downstream bed protections are
adequate.
 Adequacy of the extents of chosen
upstream and downstream river bed
protection is verified.
 Estimation of maximum head over the
gates during operational conditions i.e.
in the closed condition.
 The lengths of upstream & downstream
bank protections are assessed.
 Emergency plans are to be well
executed to keep the gates open during
flood conditions.
7
225 ENGINEER
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[227 237],2017
2017
The Institution
Lanka
Numerical Model Investigations on Dispersion of

Industrial Effluents in East Coast of Sri Lanka
S. Wickramaratne and K.P.S.T. Pathirana
Abstract: It is proposed to develop eastern Sri Lanka with an industrial zone in Eravur. This
planned development requires a proper environmental assessment of the implications arising from
discharging accrued industrial effluents to the sea. The discharge includes typical heavy metals
generating from multitudinous of industries proposed. The effluent discharge, in addition, is heated,
necessitating an investigation on its thermal effect to marine environment. Albeit the discharge is
significantly „neutralized‟ after being sent through a large treatment plant, one would still need to
account for the worst case scenario where a total blackout of the plant occurs, allowing untreated
discharge to gush out to the sea.
On this context, a numerical modelling study is formulated to observe probable dispersion patterns of
two key physical entities: concentration of solid matters and the effluent temperature. The study
aimed at spatially and temporally quantifying the dispersion of let discharge under a variety of tide,
wind, and wave conditions typically anticipated in the east coast. Landfalls of thermal or effluent
plumes were also monitored for their concentrations and the time durations at which the plumes are
in contact with shore are recorded. A state of the art numerical modelling platform MIKE 21 is utilised
for simulations and recommendations followed with the model derivations established therein. The
entire study provides sufficient guidance for any future studies particularly been the first ever
dispersion modelling completed for east coast of Sri Lanka.
Keywords: Dispersion of Effluents, Numerical Simulations, Eravur
1. Introduction The main issue that rises in an industrial zone

is environmental implications occurring due to
Adding to the number of industrial zones the discharge of harmful effluents. To address
operating in Sri Lanka, the newest is proposed this, dispersions have to be modelled and
export processing zone at Eravur (Batticaloa) necessary measures have to be taken
in the Eastern Province of Sri Lanka (Figure 1). depending on the level of concentrations
The zone includes heavy industrial factories present. As such, a bathymetric survey and a
and plants which are expected to discharge a numerical modelling were carried out for
considerable amount of effluents. After simulation of effluent dispersion of the sea
passing through a treatment plant, this effluent outfall related to the proposed export
discharge is significantly „neutralized‟ which is processing zone. A bathymetric survey of area
then discharged out to the sea via sea outfall. 2 km X 2 km was conducted to gather all
The outfall is approximately 1km from shore. important depth information for subsequent
planning, design, modelling and construction
work of proposed sea outfall (Figure 1).
2. Hydrodynamic Study
2.1 Introduction
Hydrodynamic (HD) study is a priori for
dispersion modelling for its distinct resolution
of water currents and fluxes which govern the
Eng. (Dr) S. Wickramaratne, C.Eng., MIE(SL), B.Sc.Eng.

(Moratuwa), Ph.D.(Calgary), Senior Engineering Manager,
Lanka Hydraulic Institute (LHI), Katubedda.
Eng. (Ms)K.P.S.T. Pathirana, B.Sc.Eng. (Peradeniya),

AMIE(SL), Research Engineer, Lanka Hydraulic Institute
Figure 1 - Location of the Project (LHI), Katubedda.
1
227 ENGINEER
dispersion. In principle, hydrodynamic study (m3/s/m) = (uh,vh); (u,v) =
is an investigation of the behaviour of water depth averaged velocities in x-
bodies subjected to a variety of natural force and y-directions
fields including tides, winds, and waves etc. C(x, y) Chezy resistance (m½/s)
This study launched an HD modelling for the G acceleration due to gravity
prime purpose of, (m/s2)
1. determining water currents, levels that f(V) wind friction factor
would attain in the study area under V, Vx, Vy (x, y, t) wind speed and components
probable, selected offshore wave attacks. in x- and y directions (m/s)
2. monitoring variation of the current field (x, y) Coriolis parameter, latitude
based on hydrodynamic parameters, and dependent (s-1)
3. deriving hydrodynamic input data (fluxes pa(x, y, t) atmospheric pressure
and levels) for dispersion modelling (kg/m/s2)
studies. ρw density of water (kg/m3)
x, y space coordinates (m)
2.2 MIKE 21 HD Module: Theoretical t time (s)
Formulation τxx, τxy, τyy components of effective shear
For a given water body with known boundary stress
information, currents and fluxes are resolved were resolved through the numerical
through two basic equations of conservation: process[1].
continuity and momentum (Eq 1 and Eq 2).
Required computational platform is created 2.3 MIKE 21 HD: Model Setup
with state of the art computer simulation tool: The approach used in formulating the HD
MIKE 21 HD [1]. This modelling package module is nested grid approach starting from a
provides hydrodynamic basis for larger “Regional model”, gradually reducing
computations performed in modules of to a smaller “Intermediate model” and finally
dispersion, sediment processes and to a much smaller “Local model” (Figure 2).
environmental hydraulics. Unsteady two Objective of sticking to the grid approach is to
dimensional flows in one layer fluids were use known tidal constituents to generate
simulated under the HD module. In particular, distinctive tides in the boundaries to drive the
three main relationships: Regional HD model, and to arrive at more
accurate and finer grid resolutions towards the
2D continuity equation area of interest (Local Model). The regional
model was set up to be with a resolution of
....(1) 500m x 500m covering an area of 425km x
260km. Stepping down from the Regional
model is to an intermediate model with a grid
2D momentum equation in y direction
resolution of 100m x 100m for an area of 105
√
km x 100 km. With a high grid resolution of
( ) ( ) 10m x 10m, the local model was set up which
( )] had a study area of 14km x 10km.
(
( ) ....(2) Regional Intermediate
Model Model
2D momentum equation in x direction
√
( ) ( )
( ( )]
( ) . ...(3)
Local Model
where
h(x, y, t) water depth (= ζ -d, m)
ζ(x, y, t) surface elevation (m)
d(x, y, t) time varying water depth (m)
p, q(x, y, t) flux densities in x- and y
directions Figure 2 - Nested Grid
ENGINEER 2
228
2.4 Model Calibration used for Local Model simulation. Wave and
The regional model was calibrated using wind have been chosen that they represent
constituent-generated tide at Trincomalee. The average (50%) and extreme (2%) exceedance
model predicted water levels were compared scenarios.
with that of Admiralty predicted (constituent- Table 1 - HD Local Model Scenarios
generated) and the best match is obtained via
varying Manning‟s coefficient of the model.
Hence calibrated model is utilized for
furtherance of HD study.
2.5 Model Simulation

Regional model was purely tide driven, and
evaluated overall hydrodynamic
characteristics of the region while the
intermediate model was tide and wind driven
for an accurate model formation. The regional
and the intermediate models were simulated
for a period of 15 days inclusive of 3hr warm In all simulations, wind was assumed to be
up period. Model scenarios simulated were constant in time and space, and applied to the
Southwest monsoon, Northeast monsoon and entire model domain. Further, a 3 hour warm
inter-monsoon. Necessary boundary up period was allowed for smooth and
conditions for the intermediate model were gradual model computations.
extracted from HD results of the regional
model simulations. 2.6 Hydrodynamic Model Results
The major aim of the water circulation study
Local HD model was simulated for 7 was to obtain water level and current
combinations of wind speeds/ directions and information to investigate whether any of
spring/neap periods. Selection of wind speed those parameters attain unacceptably
and direction was based on locally measured high/low values at any instance in the HD
wind by the Department of Meteorology (SL) simulation, except probably during the warm
and offshore wind provided by Wavewatch III up period. Hence, in the model time domain,
model. the instance where maximum currents occur in
the generic study area was plotted for each
Although the Regional model did not consider model scenario (Figure 4).
wave effect in its hydrodynamics, local model
being focused on nearshore was fed with An in depth assessment of current values in
appropriate wave information. In this context, each scenario indicates that currents present in
radiation stresses resulting from a series of the swash zone depict a significant level of
MIKE 21 PMS simulations provided required irregularities and perturbations due to the
wave information to MIKE 21 HD models. local wave breaking. However, the maximum
Table 1 presents the selected combinations water current in any modelled wave, wind,
and tide scenario does not exceed 0.5m/s.
(a)- NE Swell waves under Spring Tide (b)- SW Sea waves under Spring Tide
(Simulation No.3) (Simulation No.4)
Figure 3 - Maximum Water Currents Attained in the Study Area
3
229 ENGINEER
3. Effluent Dispersion Study Yet, only a handful of dispersion models have
ever been set for Sri Lankan perspective [7,8,9].
3.1 Introduction and Theoretic Given the distinctive two monsoonal cycle and
Formation related wind patterns prevailing in Sri Lanka,
In general, heat exchange is calculated on basis a high emphasis has been placed to compare
of the four physical processes. and correlate dispersion characteristics
i. sensible heat flux (or the heat flux due observed in Sampur thermal power project
to Convection ) which geographically is the closet available
ii. latent heat flux (or the heat loss due to study for Eravur [8].
Vaporization )
iii. net Short Wave Radiation 3.2 Technical details of the Model
iv. net Long Wave Radiation Contents of the discharged effluents can be
categorized into two main parts.
The heat dissipation follows a prescribed 1. Conservative matter (i.e. typical solid
function, where the decay factor F may take matters present in the discharge
the form: measured in mg/l)
2. Heat dissipation
F = 0.2388 / (r Cp H)(4.6-0.09(Tref +T)+4.06W)
exp(0.033(Tref +T)) ....(4) Government of Sri Lanka (GOSL) has
published maximum allowable concentrations
Tref is the reference temperature in degrees that can be safely discharged to the sea as
Celsius (same units as T), r is the density of wells as maximum allowable intake
water, Cp is the specific heat and H is the water concentrations of a treatment plant [10]. Since
depth [2]. W may be given constant or varying no predicted discharge concentrations of the
in time, following a prescribed function. In proposed outfall are available, the worst case
units consistent with former ones, r Cp is given scenario: a total blackout of the treatment plant
the value 106 (Cal/m3 ºC) [2]. forcing the maximum allowable intake
concentrations to be directly discharged to the
Conservative matters disperse as, sea was modelled. In other terms, maximum
allowable intake concentrations by the
( )
or ....(5) treatment plant have been taken as input
concentrations for the dispersion modelling,
where c is the concentration and t is the time.
and resulting maximum dispersed
concentrations at sea were checked with
The T90 value (the time it takes to reduce
maximum allowable marine concentrations
matter by 90%), is used to calculate c from,
adopted [10].
...(6)
Since there are no individual dispersion
Advection-Dispersion module (MIKE 21 AD) differences among the different Conservative
in the MIKE 21 Flow Model provides the basis Matters, the highest individual concentration
for computation. The AD model solves two- of 2100 mg/l at the intake (as published by
dimensional, depth-integrated transport GOSL) was selected as the representative
equation of the advection (dilution)-dispersion strength of Conservative Matter. The ratio of
type for heated effluents. The hydrodynamic observed dispersed concentration versus the
flow field of the study area is a pre-requisite intake concentration (2100 mg/l) is then used
for this dispersion modelling. Under the to scale up and derive the level of dispersion of
dispersion modelling, the dispersion all individual conservative matters. In terms of
characteristics of effluents, their extension of temperature, the maximum allowable intake
dispersion and probability of a landfall were temperature, 40ºC was assumed at the
checked. Given the very low discharge rate discharging point.
(625m3/hr) anticipated at the outfall, near field
dispersion is conservatively excluded in the Background pollution is not considered in this
modelling, paving way for more emphasis on study as the area concerned does not have any
critical far field dispersion. major polluting sources.
Modelling of effluent dispersion is not novel 3.3 Model Simulation

and the literature is rich with numerous The seven scenarios used in the Local HD
examples of international case studies [3,4,5,6]. model were simulated in parallel with the AD
ENGINEER 4
230
model. Model simulations were continued Maximum plume size in each of the model
until steady state dispersion was obtained. scenario was plotted in a bid to quantify the
Observed time duration for this was 2-3 days worst effect to the neighbourhood during the
for most cases. The simulation period included respective model scenario considered. Further,
3hr warm up time during which specified maximum length spans in which the plume is
model parameters attained their full values in contact with the beach (max. landfall
gradually. lengths) have been measured (Figures 5 and 6).
3.4 Dispersion of Effluents Following up with similar studies, given the

differences exist over the discharge rate of
3.4.1 Spatial Assessment cooling water in proposed Sampur thermal
Results depicted that the dispersion starts as a power plant (100,000m3/hr) versus the effluent
speck of the plume which then grows in size discharger rate of Eravur treatment plant
over the time. While the outer perimeter (625m3/hr), comparison of thermal plumes is
grows, conservative concentration or the plausible only for its dispersion pattern and
temperature rise reduces inside the plume, not the strength. It is rightfully observed that
weakening the propagation and eventually thermal plumes of both studies behave exactly
letting the plume collapse back to a smaller in accordance with the monsoonal wind and
size. It was also observed that patches of high wave patterns modelled [8]. Albeit the
accumulation of conservative matter are existence of obstructions for natural dispersion
present within the plume span creating in Colombo Port area, Mutwal dispersion
multiple foci temporary. After several cycles, a studies completed in 2004 also record similar
steady state variation, i.e. identical cycles in elongated thermal plume pattern[11,12,13].
expansion and retraction of plume is attained.
Conserv. Matter (mg/l)
Figure 4 - Maximum Thermal Plumes
Figure 5 - Maximum Effluent Plumes
5
231 ENGINEER
3.4.2. Temporal Assessment Extracted data related to temperature rise of
In order to gain an understanding of the the sea water and the variation of the
variation of seawater temperature over the conservative matter at the three locations and
time due to proposed discharge, three at outfall were plotted (Figures 8-9). In both
representative locations (Figure 7); one temperature and conservative matter scenarios
offshore of the outfall point, one at south and a very desirable dispersion cycle of approx. 12
the other at north of Punnaikudah were hrs was achieved for all model scenarios except
considered. in the inter-monsoonal condition where the
cycle time has increased to 3.0-3.5 days. No
accumulation of heated discharge or
conservative matter was present. For locations
north and south of Punnaikudah, the only
effect is from inter-monsoon condition, which
too does not record a higher value. Due to low
discharge rate of the outfall, the effluent
concentrations are too weak to inflict any harm
to the marine environment.
Figure 6 - Representative Locations for

Thermal Dispersion Study
(a)- Temperature Rise at Outfall (b)- Temperature Rise at Location A
(c)- Temperature Rise at Location B (d)- Temperature Rise at Location C

Figure 7 - Temporal Variation of Temperature
ENGINEER 6
232
(f)- Conservative Matter at Location A
(e)- Conservative Matter at Outfall
(g)- Conservative Matter at Location B (h)- Conservative Matter at Location C

Figure 8 - Temporal Variation of Conservative Matters
Although a landfall of the plume was evident landfall during any of the major monsoon
in the inter monsoon waves coupled with periods. Inter-monsoon period however
north westerly wind (simulation no.7), the creates a platform for effluent landfall along
average temperature that would be felt is only northern Punnaikudah beach for a length of
0.003°C higher than the ambient sea approximately 6km. Nonetheless, dispersed
temperature. In general, low outfall discharge concentrations of all effluents have been very
rate coupled with high winds and strong low and less than 1 mg/l. Simulated
monsoonal waves assist the quick dispersion. concentrations have been well within the
Events in which wind and wave directions are stipulated maxima by CEA and thus, no threat
distinctively apart as is the case in simulation exists over contamination of coastal
no.7, cause slow dispersion recording neighbourhood of Punnaikudah from effluents
relatively larger plumes. of proposed outfall discharge.
4. Conclusions Computer simulations of effluent dispersion

carry distinct advantages. The textbook
In this study, no accumulation of either heated benefits include being able to predict the final
discharge or effluents was observed, as such residue concentrations of harmful effluents
discharged materials/heat were effectively that may be present in the marine
dispersed within a maximum cycle time of 3.0- environment. In this instance, the modelling
3.5 days. In addition, no heated discharge ever results further indicate that the need of a
reached nearshore areas except for inter- treatment plant is not mandatory as the model
monsoon conditions where a landfall is scenario was the total blackout of the
expected at north Punnaikudah with a treatment plant. Further, this study will bear
maximum temperature rise in the sea water of ramifications for similar dispersion studies in
0.003°C. In practical terms, this level of future given the very low number of
temperature rise is hardly detectable and its modelling activities ever carried out for Sri
effect to the environment is negligible. Further Lanka.
the dispersion levels of Conservative Matters
have been excellent with no effluents making a
7
233 ENGINEER
Acknowledgement 7. LHI (2013). Dispersion Modeling Report,
Feasibility Study for Introducing LNG to Sri
Lanka.
Authors wish to thank BOI Sri Lanka and
Lanka Hydraulic Institute for their 8. LHI (2014). Final Report-Thermal Dispersion
authorization to publish the research findings. Study, Sampur Thermal Power Project.
Guidance and support rendered by Eng. IGI
Kumara for the numerical modelling 9. Samarawickrama, S. P., Hettiarachchi, S. S. L.,
component is also acknowledged. Jayaweera, M., O'Brian, D., Raveenthiran, K., &
Wickramaratne, S. (2008). Modelling the changes
in Water Quality due to the Colombo South
References Harbour, Proc. of the 7th International
Conference on Coastal and Port Engineering in
1. DHI (2009). User Manual, MIKE 21 Flow Developing Countries (PIANC-COPEDEC VII),
Module, Danish Hydraulic Institute. Dubai, United Arab Emirates. Feb 24-28.
2. DHI (2009). User Manual, MIKE 21 AD Module, 10. BOI (2011). National Environmental
Danish Hydraulic Institute. (Protection and Quality) Regulations, No. 1 of
2008, available from www.investsrilanka.com
3. Abbaspour, M., Javid, A. H., Moghimi, P., and
/images/publications/pdf/environmental_norms.
Kayhan, K. (2005). Modeling of Thermal
pdf, last accessed on Aug 20, 2017.
Pollution in Coastal Area and its Economical
and Environmental Assessment. Int. J. Environ.
11. LHI (2011). Development of Modern City at
Sci. Tech.,2(1):13–26.
Southern Border of the Colombo Port,
Mathematical Modelling Report.
4. Monteiro, A. J., Neves, R. J., Sousa, E. R., (1992).
Modelling Transport and Dispersion of Effluent
12. Martin, J. L., and McCutcheon, S. C., (1998).
Outfalls, Water Science and Technology, 25 (9) 143-
Hydrodynamics and Transport for Water
154;
Quality Modeling. Lewis Publishers, New York.
5. Baumgartner, D. J., Frick, W. E., and Roberts, P.
13. Scott Wilson (2005). Colombo Port Efficiency
J. W., (1994). Dilution models for effluent
and Expansion Project, Colombo South Harbour,
discharges. 3rd ed. USEPA, Pacific Ocean
Detailed Engineering Report.
Systems Branch, Newport, Oregon.
6. James, I. D., 2002. Modeling Pollution

Dispersion, the Ecosystem and Water Quality in
Coastal Waters: A Review. Environ.
Model.Softw.17(4): 363–385.
ENGINEER 8
234
Annex
(a)- NE Sea waves under Spring Tide (b)- NE Sea waves under Neap Tide
(Simulation No.1) (Simulation No. 2)
(c)- NE Swell waves under Spring Tide (d)- SW Sea waves under Spring Tide
(e)- SW Sea waves under Neap Tide (f)- SW Swell waves under Spring Tide
(g)- IM Swell waves under Spring Tide

(Simulation No.7)
A1 - Maximum Water Currents Attained in the Study Area
9
235 ENGINEER
(g)- IM waves under Spring Tide

(Simulation No.7)
A2 - Maximum Concentration of Thermal Plumes
ENGINEER 10
236
Conserv. Matter (mg/l)
(g)- IM waves under Spring Tide

(Simulation No.7)
A3 - Maximum Concentration of Effluent Plumes
11
237 ENGINEER
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[239 246],2017
2017
The Institution
Lanka
Numerical Modelling Study for RahfalhuHuraa Resort

Development in Maldives
M.S.L. Fernando, I.L. Abeygoonasekara and S.M.D.E. Suriyamudali
Abstract: Unavailability of sufficient land for the development has adversely affected booming
tourism industry of Maldives. Although the developers show interest to invest in resort development,
lack of land is a dire discouraging factor for them. As a solution, the Government of Maldives has
promoted formation of islands on shallow reefs through reclamation. Fulfilling the need,
RahfalhuHuraa Resort Development Project surfaced which once constructed will be the largest
reclamation undertaken in the Maldives for tourism development.
RahfalhuHuraa reef is located in North Male Atoll, around 50km northward to Male city. The
proposed development consists of six main islands which create 77.5ha land on top of the reef area
where average water depth is just 1m. In order to obtain design wave parameters, sea currents and
max/min water levels, a comprehensive numerical model study was carried out considering different
monsoon conditions and layouts. In this process, DHI’s MIKE 21 software package available with
flexible mesh and finite volume method was utilised for numerical modelling. Offshore wave and
wind data, tide elevation, bathymetry and proposed island layout were used as main input
parameters whilst wave breaking coefficients and sea bed roughness were used as calibration
parameters. Since Maldives is highly vulnerable to storm surge and sea level rise with its low level
lands, those factors were also taken into consideration. The study was formed to predict wave heights
in front of shore protection structures for different return periods. Further, current patterns with its
maximum values were obtained for existing and proposed layouts. In addition, influence for the water
level variation by proposed development was also discussed in the study.
Keywords: Wave Transformation, Hydrodynamic, Reclamation
1. Introduction The orientation of islands with their shapes

should be carefully designed to minimise the
The Maldives is a South Asian island country, effect of wave action and sea current. Further,
located in the Indian Ocean which consists of 26 coastal protection structures such as breakwater
ring-shaped atolls. More than 1000 scattered and revetment should be designed to withstand
coral islands are existed on top of atoll reefs severe waves. For this purpose, it is essential to
creating 298 km2 of total land area. estimate wave conditions affected on
structures. In this process, numerical models
There is a shortage of land area for the play vital role in order to transfer waves from
development and it has severely affected deep sea to project site. Hence, this paper
tourism industry. Most of the islands in atolls presents usage of numerical model in wave
surrounding Male city (nearly 90%) is being transformation. Additionally, influence of sea
utilized for various land uses and remaining currents and water level variation obtained
islands are too small for economic activities. through a hydrodynamic model were also
Hence, formation of new land on shallow reefs presented.
through reclamation is the only option
remaining. With this circumstance, the
Government of Maldives has promoted this
reclamation option to fulfil the requirement of
booming tourism industry. As a result, several Eng. M.S.L. Fernando, AMIE(Sri Lanka), B.Sc. Eng. Hons
resort island developments come up for the (Moratuwa), M.Sc. (Moratuwa), Engineering Manager,
Lanka Hydraulic Institute Ltd.
field. Among them RahfalhuHuraa Resort
Developments is a major tourist resort Eng. I.L. Abeygoonasekara, AMIE(Sri Lanka), B.Sc. Eng.
Hons (Ruhuna), Research Engineer, Lanka Hydraulic
development which was subjected to a Institute Ltd.
comprehensive coastal engineering study.
Eng. S.M.D.E. Suriyamudali, AMIE(Sri Lanka), B.Sc.
Eng. Hons (Peradeniya), Civil Engineer, National Water
Supply &Drainage Board.
1
239 ENGINEER
2. Project Background  Obtain and analyze of UK Met Office
(UKMO) hind-cast data (wind & wave).
The proposed resort island development site is
located in north of Kaafu Atoll (North Male  Carry out field investigations to obtain
Atoll), Maldives, and around 50km northward bathymetry, tide, current and wave data at
to Male city (Figure 1). A natural reef is proposed site and extract model required
available at site about 1m depth level with a data from reliable sources.
lagoon in the middle. The proposed island
formation would be placed on top of the reef.  Perform numerical modelling for wave
transformation from UKMO data location to
project site.
 Perform numerical modelling for water

circulation (hydrodynamic) for regional
model using known tidal data.
 Perform numerical modelling for water

circulation (hydrodynamic) for local model
using regional model and wave model
results.
 Extract the results at suitable locations,

analyse and discuss them
Figure 1 – Project Location
The RahfalhuHuraa Resort Development

4. Data Collection and Analysis
Project will be the largest reclamation
4.1 Field Investigation
undertaken in the Maldives for tourism
In order to find out sea bed levels, tidal
development with total area of 77.5ha. The aim
variation, wave and current condition at site a
of this project is to develop a multi-island resort
set of field investigation was carried out at the
and marina in North Male Atoll by reclaiming
beginning of study. Bathymetry survey was
the shallow water lagoon of RahfalhuHuraa.
done in July 2015 and the other measurements
The proponent aims to use the proximity to
were taken in December 2015. In addition to the
Male International Airport and continued
field investigation, regional bathymetry was
growth in tourism arrivals in the Maldives to
extracted from Admiralty Maps and offshore
establish a collection of successful and
wind/wave data was obtained from a Global
profitable resorts.
Spectral Wind/Wave Model of UK Met Office
(UKMO) [3].
The concept master plan encompasses
developing six main islands with four as high-
4.2 Offshore Wind/Wave Data
end resorts (5 – 12ha in size), one as a marina,
Wave data is very important for the planning
hotel and resort (34ha), and the other one as a
and designing of coastal development and it is
service island (11.5ha). The proposed project
necessary to measure wave data for a period of
includes land reclamation, shore protection,
5-10 years. Such a measured data are presently
and infrastructure development of marina and
not available for the project site. The visual
resorts [1].
data, based on ships observations are not
adequate for long-term wave characteristics of
3. Methodology an area. In order to obtain a uniform spatial-
temporal wave data base for estimating wave
Since measured wave data is not available at statistics, the study based on Global Wave Hind
the vicinity of site, Hind-Cast wind/wave data Casting Method is the only alternative.
obtained through a Global Spectral
Wind/Wave Model was used for the study. The The wind and wave data was obtained based
other data required to set up the numerical on the hind-cast data from the global wind and
models were also obtained through field wave model of the UK Met Office (UKMO) at
investigation and reliable sources. The basic offshore of the project area. The available
procedures of the methodology are given nearest suitable offshore data point (4.453N,
below. 78.180E; UTM 43 N) at 2500m water depth was
ENGINEER 2
240
selected with considering wave pattern as
shown in Figure 2. UKMO is regularly
forecasting sea state of global oceans using a
2DGlobal Spectral WaveModel (GSWAM). The
spectral energy changes due to wave
generation, propagation, decay, nonlinear
wave-wave interaction, etc. is catered in the
source term of the model. The input for this
spectral type wave model is analyzed global
wind fields generated from a Global Spectral
WindModel (GSWIM). The model assimilates
wave measurements from the radar altimeter
on the ERS-2 satellite and the spectral and
statistical outputs are available on 1.5 x 1.5 deg
in 3hr / 6hr basis for any region of interest [3].
Project Location
UKMO Data
Extracted Location
(2500m Depth)
Figure 4 – Annual Distribution of Offshore
Wave
Figure 2 – Offshore Wave Data Extracted
Location 5. Wave Climate Model
Six years data was obtained from UKMO at When implementing new coastal structures,
offshore of proposed site. The wave height, study on wave climate and wave modelling are
wave period and wave direction are available essential components. The wave data is not
with wind speed and wind direction during available for the desired location on the sea fit
January 2010 to December 2015. The data set for usage. Most of the time wave data
contents 8764 no of records with the interval of measured point is far away from the project
6 hours. Analyses were carried out to assess the location and it has to be transformed to a
distribution of wind and wave parameters and desired location by a model. In this case, MIKE
annual distribution pattern of wind and wave 21 SW numerical wave model developed by
are given in Figure 3 and 4 respectively. Danish Hydraulic Institute, Denmark was used
for wave transformation process. The objectives
of the wave climate modelling of this scheme
are,
 Establishing nearshore wave climate at the

proposed site.
 Evaluating the extreme wave events for

different returns periods at different
locations considering proposed island
Figure 3 – Annual Distribution of Offshore formation.
Wind
5.1 Wave Model Setup
The model bathymetry was established by
digitizing from the Admiralty charts no 1013 &
3323, and near shore measured bathymetry
3
241 ENGINEER
data. The preparation of model bathymetry was
completed using flexible mesh as a requirement
of SW model with reference to the UTM 43
coordinate system (for (x, y) coordinates) and
Mean Sea Level. Further, a rounded offshore
boundary was selected considering the wave
and wind approach directions. This rounded
boundary was almost fixed to the depth value
of UKMO wave point (2500m); thus the wave
Figure 6 – Calibration Curve for the Wave
condition is almost same along the boundary
Model
(Figure 5).
5.3 Wave Model Results
Accordingly all wave data records were
transferred from offshore to site location with
applying particular wind conditions. The
results were extracted at 8m depth in front of
Project Location
the project site and different locations near the
islands as shown in Figure 7.
Figure 5 – Wave Model Bathymetry - Regional
Additionally, a local model was also

incorporated to improve the performance and
accuracy of results. Then UKMO waves were
transferred to 1000m depth and applied to the
local model. Size of the elements in mesh was
reduced while moving to the area of interest
with consideration of keeping high resolution
near proposed site.
Figure 7 – WaveHeight Extracted Locations
5.2 Wave Model Calibration
Measured wave data at site location was used
Wave data analysis was done for the extracted
for the model calibration. Basically, sea bottom data at 8m depth for total year, South-West and
friction and wave breaking index were used as
North-East monsoon periods. South West
the calibration parameter [2]. Since this site is monsoon was considered as from May to
abundant with natural reef, sudden drops of
November while North East as from December
sea bed are available around the reef. Hence, it
to April [1].
is not like a normal beach, these steep slopes
directly affect to the wave breaking. As a result,
Wave statistic at 8m depth (Figure 8) shows
wave breaking index would be higher than that refracted waves have approached the near
general values, and then it would be significant
shore from east side with a narrow band. The
in calibration. most wave occurrence has happened between
900 and 1000, but high waves reach between
Finally, after several test runs, most suitable 600and 700. Further, occurrence of high wave
bottom friction and wave breaking index were
was observed during North-East monsoon.
set to Nikuradse Roughness value of 0.04m and
Gamma value of 1.2 respectively [3]. The
calibration curve is given in the Figure 6.
ENGINEER 4
242
Astronomical tide variation was estimated
using Admiralty Tide Tables [4]. Since tide data
at proposed site is not available in Admiralty
tide tables, the data at Girifushi (32 km
southward) was used for the estimation. The
relevant tide levels at Girifushi are given in
Table 1.
Table 1 - Tide Levels at Girifushi [4]
Level Level
relative to relative
Tidal Level
Figure 8 – Annual Wave Height Distribution Chart to MSL
at 8m Depth Datum (m) (m)
Highest Astronomical
1.2 0.62
5.4 Extreme Wave Analysis Tide (HAT)
In the design of any coastal structure, it is Mean Higher High
0.9 0.32
necessary to ensure the structural stability Water (MHHW)
against extreme wave events. In general, in the Mean Lower High
0.7 0.12
design of structures such as breakwater Water (MLHW)
revetments and quay wall, a significant wave Mean Sea Level (MSL) 0.58 -
height of 50-100 year return period is Mean Higher Low
0.4 - 0.18
considered as the design wave height. However Water (MHLW)
the selection of return period value depends on Mean Lower Low
0.3 - 0.28
design life of structure and appropriate Water (MLLW)
probability of extreme event occurrence.
Mean Higher High Water (MHHW) was
Accordingly the Extreme Wave Analysis was applied for all return periods as the tidal
done using the ACES software of the US Army elevation. A storm surge of 0.38m water level
Corps of Engineers (USACE). Analysis was increment was applied for the still water level
done by fitting the data to distributions of according to the UNDP research findings [5].
Weibull and Fisher-Tippet with varying K The sea level rise was taken as 3.6 mm per year
shape parameter. with considering regional sea level rise and
0.18m of sea level rise was estimated for 50
Mean Sea Level is not desirable to apply as still years design period for proposed structures [3].
water level for the modelling of designing Computation of design water level is expressed
purpose. The extreme condition should be in Table 2.
considered when designing coastal structures.
Therefore the still water level should be Table 2 - Design Still Water Level relative to
decided with considering following factors MSL
which are mainly influence the water level Water Level
Influence
variation in the sea environment. Variation (m)
Astronomical Tide 0.32
 Astronomical tide variation Storm Surge 0.38
 Storm surge Sea Level Rise 0.18
 Sea level rise Design Still Water Level
0.88
(m MSL)
Table 3 - Design Wave Height - Hs (m) at Different Locations

Return Location
Period
(yr) P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 8m Depth
2 1.24 1.26 1.07 1.09 1.24 1.06 1.31 1.15 1.13 1.01 1.12 2.06 2.75
5 1.33 1.29 1.09 1.12 1.27 1.08 1.34 1.17 1.16 1.03 1.14 2.18 2.98
10 1.4 1.31 1.11 1.13 1.28 1.09 1.36 1.19 1.18 1.05 1.16 2.27 3.15
25 1.49 1.34 1.13 1.15 1.31 1.11 1.39 1.21 1.2 1.07 1.18 2.39 3.38
50 1.56 1.36 1.14 1.17 1.33 1.13 1.41 1.23 1.22 1.08 1.2 2.48 3.55
100 1.63 1.38 1.16 1.18 1.34 1.14 1.43 1.24 1.24 1.1 1.22 2.57 3.72
5
243 ENGINEER
The results of extreme wave height analysis for element area of 100m2 was developed in local
different locations in front of islands are given bathymetry for marina area to observe the
in Table 3. The locations are according to the water circulation and flushing rates. The UTM
Figure 7. The results show a significant 43 coordinate system was used for (x, y)
reduction of wave height in front islands coordinates and Mean Sea Level (MSL) was
compared to 8m depth. Energy dissipation of used as the elevation datum. The local
low depth submerge reef is the main reason bathymetry with mesh is given in Figure 10.
for this wave height reduction. However the
location P12 has gained considerable wave
height with its higher depth and lower
protection of reef.
6. Hydrodynamic (HD) Model
In order to compute water circulation with

effect of tidal flow, wave and wind in and
around the study area, a hydrodynamic model
should be carried out. Hence, the numerical
model, MIKE 21 HD was used in this process.
6.1 HD Model Setup

Same as the wave model, Admiralty charts
and near shore measured bathymetry data
were used for the development of model
bathymetry. Since tidal variation would be
applied as main boundary condition for the Figure 10 – Local Bathymetry with Mesh
regional model, availability of tidal data was
considered when setting up the model. The regional model was run for a period of
Therefore the regional model extends from one year to cover both South-West and North-
Goidhoo (North Boundary) to Male (South East monsoon and the local model was run for
Boundary) where tidal boundaries were two set of periods with two weeks for each set
derived using the tidal constituents. The which covering spring and neap tide. One set
regional model covers the model area of 77 km was selected from South-West monsoon and
in north-south way and 162 km east-west way. the other one from North-East.
Figure 9 illustrates the bathymetry of regional
model and position of the local model on it. Water levels and discharges extracted from
Local model was selected considering the regional model at the boundaries of the local
bathymetry, and kept offshore boundary at model were used as boundary conditions for
1000m depth. the local model. Local model was run for
average and extreme wave/wind condition for
both monsoons. The average condition was
selected considering 50% occurrence of wave
Local Model
height of the wave data extracted at 1000m
depth from wave transformation model (MIKE
21 SW). In this process, most dominant
directional bands were selected for the
calculations. Same as wave, 50% occurrence of
Figure 9 – Bathymetry for Regional wind speed of UKMO wind data was selected
Hydrodynamic Model to obtain corresponding wind speed. Same
method was adopted to select the extreme
Triangular flexible mesh was generated by condition by considering 98% occurrence (2%
MIKE 21 Bathymetry Creator Tool while exceedance of wave height and wind speed).
considering finer mesh size for more The selected wave and wind conditions are
concerned area. Maximum mesh size in given in following tables.
regional model is 1x106 m2 and maximum
mesh sizes in local model, deep sea and
shallow areas are 500,000m2 and 1,000m2
respectively. Further very fine mesh with
ENGINEER 6
244
Table 4 - Applied Wave Conditions for the 6.3 HD Model Results
Hydrodynamic Model The main objective of the HD modelling is to
Average Extreme observe the flow pattern in the model area and
Dir
(50%) (98%) find out the effect for the current flow by the
Monsoon Hs Tp Hs Tp proposed island. Hence current speeds were
deg extracted in between islands as indicated in
(m) (s) (m) (s)
Figure 12 and the maximum current speed
South West 0.74 9.80 1.33 9.82 120
values are given in Table 6.
North East 0.8 9.35 2.18 9.42 75
Results show that an increment of maximum
current speed in between islands relative to
Table 5 - Applied Wind Conditions for the the existing condition. That increment
Hydrodynamic Model comparatively higher in North-East monsoon
Average Extreme than South-West.
Dir
(50%) (98%)
Monsoon Speed Speed Furthermore water levels are also extracted in
Deg front of proposed island as indicated in Figure
(m/s) (m/s)
12 and the results are given in Table 7.
South
4.96 8.58 270
West
North East 4.36 8.55 60
6.2 HD Model Calibration

Calibration was done considering the variation
surface elevations. Model predicted tide
elevation and measured water level at
proposed site were incorporated for the
calibration. The bed roughness was used as the
calibration parameter and Manning’s Number
of 32 was set as the suitable values. Figure 11
shows the calibration curve.
Current Extraction
Water Level Extraction
Figure 12 – Current Speed and Water Level

Extracted Points
Figure 11 – Calibration Curve for the HD

Model
Table 6 - Maximum Current Speed in m/s at Extracted Point
Average Extreme
Location South-West North-East South-West North-East
Existing Proposed Existing Proposed Existing Proposed Existing Proposed

CH 1 1.04 0.87 1.17 0.94 1.04 0.87 1.12 0.93
CH 2 1.20 1.03 1.35 1.07 1.20 1.03 1.30 1.07
C1 0.37 0.26 0.48 0.53 0.42 0.32 0.65 0.54
C2 0.32 0.61 0.39 1.09 0.39 0.76 0.57 1.10
C3 0.32 0.57 0.38 1.02 0.39 0.72 0.58 1.03
C4 0.32 0.43 0.40 0.86 0.38 0.54 0.59 0.86
C5 0.23 0.32 0.33 0.64 0.32 0.39 0.48 0.65
EC 0.23 0.19 0.28 0.32 0.27 0.23 0.44 0.31
M 0.09 0.03 0.11 0.03 0.09 0.05 0.13 0.05
L1 0.09 0.04 0.09 0.07 0.10 0.05 0.13 0.07
L2 0.07 0.03 0.09 0.02 0.07 0.05 0.07 0.03
7
245 ENGINEER
Table 7 - Maximum and Minimum Water Levels (m MSL) in front of Islands
Average Extreme
South-West North-East South-West North-East
Location
Max Min Max Min Max Min Max Min
Ext. Pro. Ext. Pro. Ext. Pro. Ext. Pro. Ext. Pro. Ext. Pro. Ext. Pro. Ext. Pro.
W1 0.45 0.46 -0.51 -0.52 0.48 0.63 -0.47 -0.14 0.46 0.51 -0.5 -0.52 0.54 0.63 -0.42 -0.13
W2 0.45 0.47 -0.51 -0.51 0.48 0.62 -0.47 -0.18 0.46 0.52 -0.5 -0.51 0.53 0.63 -0.42 -0.18
W3 0.45 0.46 -0.51 -0.51 0.48 0.62 -0.47 -0.21 0.46 0.51 -0.5 -0.51 0.52 0.62 -0.42 -0.21
W4 0.44 0.44 -0.51 -0.52 0.48 0.49 -0.47 -0.45 0.45 0.45 -0.51 -0.51 0.51 0.49 -0.43 -0.45
7. Conclusions beach crest level should be designed

accordance with these water level variations.
A significant reduction of wave height and a
directional change were observed at near shore Acknowledgement
waves compared to offshore waves. Presence
of low depth reef has been affected for wave Authors would like to express their gratitude
height reduction. Especially considerable to Dr. Ahmed Shaig (Director – CDE
energy dissipation happens through depth Consulting, Maldives) for all support given
induce wave braking when waves enter the throughout the project duration and Lanka
low depth area. Existence of hard bottom reef Hydraulic Institute for providing there sources
bed dissipates more energy through bottom to carry out the study. Also we thank to all
friction as well. The effect of low depth reef for friends for their support given in many ways.
the wave height reduction is clearly illustrated
by the results given in Table 3. The reef edge References
where the 8m depth point located had high
waves, but after energy dissipation, low waves 1. CDE Consultant, “Environment Impact
appeared on top of reef near islands. Hence Assessment for the Proposed RafalhuHuraa
this natural reef has provided a good Resort Development, at Mai Falhu, Kaafu Atoll”,
protection to islands by acting as a submerged April, 2016.
breakwater.
2. Danish Hydraulic Institute, MIKE 21 Scientific
Document; Spectral Wave Module, 2014.
Diffraction and refraction processes have
influenced to narrow the wide range of 3. Lanka Hydraulic Institute, Development on
offshore wave directions. Islands and reefs RahfalhuHurra, Maldives, Numerical
located in south of atoll cut off southern waves Modelling, Final Report, September 2016.
causing diffraction process. Furthermore
waves refract with depth reduction limited 4. United Kingdom Hydrographic Office,
wave directions to a narrow band. Admiralty Tide Tables, Indian Ocean and South
China Sea, Vol 03, 2015.
Current speed of channels either side of reef
5. United Nation Development Programme,
has been reduced by proposed island
Developing a Disaster Risk Profile for Maldives,
formation, but it has been increased in Vol 1 & 2, 2006.
between islands relative to the existing
condition. This current speed increment in
between island is more significant in North-
East monsoon season than South-West.
Interruption of the water flow on top of reef
would be the reason for this increment of
current speed in between islands.
Significant change in maximum water level

cannot be observed for the South-West
monsoon, but considerable increment has been
indicated for the North-East monsoon. Hence
ENGINEER 8
246
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[247 254],2017
2017
The Institution
Lanka
Wave Tranquillity Study for a Fishery Harbour using

Boussinesq Type Wave Model
D.P.C. Laknath, D.E.N. Senarathne and K.K.P.P. Ranaweera
Abstract: The knowledge about the behaviour of waves inside a harbour basin and entrance is
important for fishing boats mooring and handling, navigational purposes as well as handling
downtime. During the last 4 decades, Boussinesq-type models became well-known simulation
technique among the coastal engineering community. This is because of their ability to represent all
main physical phenomena of water waves in shallow seas and harbours. This study was carried out
for a fishery harbour planned to be implemented in the North-western coastal stretch of Sri Lanka.
The main objective of this study is to carry out a wave tranquillity study for the proposed harbour to
ensure the mooring arrangements, safe navigational through the harbour entrance in relation to
predefined criteria for acceptable wave disturbance. Boussinesq Wave (BW) module of MIKE 21
modelling system was used to simulate the wave disturbances inside harbour basins and harbour
entrance under different characteristics of wave incidences. Boussinesq type equations incorporate
frequency dispersion as opposed to the shallow water equations. With the hindcast, nearshore wave
climate in the study area, simulations were carried out for representative nearshore wave climate
conditions during the southwest and northeast monsoon seasons (for Hs: 0.7m – 1.7m & 2650N - 3050N
directional ranges). As the outcome of the simulation, it was identified that maximum significant
wave heights are less than 0.4 m for most of scenarios at the harbour entrance, inside and quay wall
areas indicating the safe navigation across the harbour entrance and convenient operations. Hence, as
per the PIANC Guidelines and simulation results, it was concluded that navigational and mooring
operations could be done in the basin with minimum disturbance for all seasons. Thus, suitability of
the harbour location and layout in terms of wave tranquillity was justified with the numerically
simulated results.
Keywords: Wave disturbances, fishery harbour, numerical simulation, Boussinesq-type model
1. Introduction
In addition to the existing 20 fishery harbours, optimization through wave tranquillity studies.
Sri Lankan government continues to develop This is because of their ability to represent all
new fishery harbours and landing facilities in main physical phenomena of water waves in
the coastal zone of the Sri Lanka [1]. Hence, shallow seas and harbours [3]. Generally,
improvement of harbour design methodologies Boussinesq Wave Model includes non-linearity
is an important factor to be considered in terms and frequency dispersion. Also, it has a
of enhancing the accuracy of the design and capability of simulating refraction, shoaling,
saving cost and time with an optimized reflection, diffraction, wave interaction and
harbour layout. During the harbour design other important non-linear phenomenon [4]. In
phase, mathematical modelling is considered as this context, MIKE21 Boussinesq Wave (BW)
an efficient and cost effective technique for the model is a powerful tool for determination and
reproduction of existing coastal phenomenon as assessment of wave dynamics in ports and
well as to select the best harbour configurations harbours in coastal areas. In this study, wave
out of several alternatives designs through a penetrations through a proposed Fishery Port
systematic harbour layout optimization process
Eng. (Dr.) D.P.C. Laknath, AMIE(SL), B.Sc.Eng.
[2]. Further, understanding about the wave (Moratuwa), M.Eng. (AIT), Dr. Eng. (Yokohama),
disturbances inside a harbour basin is Engineering Manager, Lanka Hydraulic Institute Ltd.
important for fishing boats mooring and
Eng. D.E.N. Senarathne, AMIE(SL), B.Sc.Eng.
handling during loading/unloading operations (Moratuwa), Research Engineer, Lanka Hydraulic Institute
as well as navigational purposes. Over the last Ltd.
40 years, Boussinesq-type models have become
Eng. K.K.P.P. Ranaweera, C.Eng., MIE(SL), B.Sc.Eng.
well known and are favoured by the coastal (Moratuwa), Project Director, Ministry of Fisheries and
engineering community for harbour layout Aquatic Resources Development.
1
247 ENGINEER
(a) (b).Waves in Study Area (Wennapuwa)
(c). Proposed Harbour

Study Area
(Wennapuwa) Sri Lanka
Colombo
SW Monsoon Waves
Figure 1 - (a). Study Area (Wennappuwa); (b). Wave Condition in Study Area (c). Proposed Harbour
(Sources: Field visit in July 2015, http://maps.google.com)
entrance and wave disturbance (i.e. agitation) the west coast of Sri Lanka is mainly governed
inside harbour basin (especially, quay wall area by the southwest monsoon. Comparatively,
and harbour entrance) have been studied by northeast monsoon is very much milder. The
MIKE21 BW module. This study was carried tides occurring in Sri Lanka is the semi-diurnal
out in connection with a fishery harbour type with an average tidal range of less than
planned to be implemented in Wennappuwa in 1m. Due to that, magnitudes of tidal currents
the North-western coastal stretch of Sri Lanka along the coast are not significant. In case of
(Figure 1(a)). Accordingly, the main objective of wave induced currents, the long shore coastal
this study is to carry out a wave tranquillity currents are strongest during the respective
study for the proposed harbour to ensure the months. Thus, during the southwest monsoon
mooring arrangements, safe navigational period, wave induced currents are in a
through the harbour entrance in relation to northerly direction while currents in a southerly
predefined criteria (i.e. Permanent International direction could be experienced during the
Association of Navigation Congresses - PIANC) for northeast monsoon time [5]. In terms of
acceptable wave disturbances. Hence, it was an sediment process, this study area was
attempt to assess the suitability of the harbour recognized as a highly sensitive area for coastal
location and layout by wave calmness study erosion before implementing the Coastal
with the numerically simulated results. Resource Management Project (CRMP) in Sri
Lanka between 2000 - 2006 periods. Though this
2. Proposed Harbour and Study area has been stabilized with beach
Area nourishment and hard structures under the
CRMP, still this coast can be considered as a
In this study, wave tranquillity study was highly sensitive and critical coastal stretch for
carried out for a proposed Fishery Harbour in sediment processes.
Wennappuwa (Figure 1(c)). Considering the
general wave and sediment processes in the Out of several significant coastal processes, we
study area, conceptual harbour layout was have studied each process with the state-of-the-
initially designed. As per the requirement of the art technologies to design a high-performance
fishermen and other stakeholders, the harbour layout with a minimal adverse impact
conceptual layout was further optimized. Thus, on the existing coastal processes. In this paper,
proposed harbour layout consists of two to assess the navigational and mooring capacity
breakwaters, quay wall, slipway facilities and of the proposed harbour, we only presented
onshore facilities. As most important areas, and discussed about the wave penetration
wave disturbances in harbour entrance, basin through the harbour entrance and resultant
and quay wall areas are mainly considered for wave disturbances inside the harbour basin
the assessment. Generally, the wave climate in areas for the identified dominant wave
conditions.
ENGINEER 2
248
3. Methodology addition, the selection of time step and grid
spacing must attest the Courant Number which
3.1 MIKE 21 BW Model is a stability indicator, to be less than or equal
MIKE 21 BW module is developed by Danish to unity.
Hydraulic Institute (DHI), offering the required
modelling platform to assess the tranquillity of 3.2 Wave Climate in the Study Area
the harbour basin. Boussinesq Wave model is a To identify dominant wave conditions for wave
state-of-the-art numerical tool for calculation penetration study, the establishment of the
and analysis of short and long period wave wave climate in the study area is a prerequisite.
disturbance in ports and harbours. Madsen and Detailed description about the hindcasting of
Sørensen [6] explained the theoretical wave climate for the study area can be found
formulation of enhanced Boussinesq equations. from the study of Laknath et al. [5]. As per this
The model has been extended into the surf zone study, the applied methodology to hindcast
by inclusion of wave breaking and moving wave climate is based on the application of
shoreline as described in Madsen et al. [7, 8] MIKE 21 SW (Spectral Waves) model to
and Sørensen [9]. MIKE 21 BW is capable of simulate wave propagation off the western
reproducing the combined effects of most wave coast (offshore) and using a Wave
phenomena of interest in ports, harbours and Transformation Matrix (WTM) approach to
related structures. These include: shoaling, transform wave recordings based on
refraction, diffraction, partial reflection and reproduced model results. Wave propagation
transmission, non-linear wave-wave modelling was carried out to transform wave
interaction, frequency spreading and recordings from Colombo to representative
directional spreading. BW simulation requires a locations in the study area. Thus, representative
series of input parameters (e.g. bathymetry, nearshore wave climate was established at 8 m
boundary conditions, simulation period, water depths in the study area (Wennappuwa).
internal wave generation, eddy viscosity, All wave measurements in Colombo during
bottom friction, porosity, and sponge). As the 1998 - 2015 periods were transformed to the
main output, wave parameters (i.e. wave offshore boundary of the model and thereafter
heights, time period and directions) and wave transformed back to the desired destination
disturbance coefficients (WDC) can be obtained locations (Wennappuwa). Thus, established
at the interested locations. During the model wave data at 8 m depth was analysed to use as
setup stage, selection of suitable time step and the boundary conditions of the MIKE 21 BW
grid resolution is an important factor to ensure model simulations.
efficient model simulation and the accurate
results. Time step size determines the frequency 3.3 Model Scenarios
at which Boussinesq equations are resolved. Wave tranquillity study was performed for the
The time step should be sufficiently small to proposed harbour layout for number of wave
resolve the shortest individual wave periods. In conditions. Thus, simulations were carried out
Table 1 - Simulation Scenarios for MIKE 21 BW Model

% Exceedance Wave Parameters
Case No Season Condition
50 2 Hs(m) Tp (s) Dir
1 SW Average √ 1.10 7.5 265
2 SW High √ 1.70 6.5 265
3 IM-1 Average √ 0.70 9.2 245
4 IM-1 High √ 1.50 6.2 255
5 NE Average √ 0.50 9.9 255
6 NE** Average √ 0.70 5.8 295
7 NE High √ 0.90 8.9 265
8 NE** High √ 1.10 4.8 305
9 IM-2 Average √ 0.50 9.8 245
10 IM-2 High √ 1.10 5.7 255
** NB: Additional wave conditions were selected for NE monsoon to represent North – West quadrant.
3
249 ENGINEER
for different wave boundary conditions at 8 m
Case 1 - SW Monsoon: Average Wave Condition
sea depth. Considering the percentage of
exceedance of established wave climate at 8 m
sea depth, wave parameters relevant to average
and high wave conditions were estimated for
each monsoon. Thus, 50 % and 2 % exceedance
of wave parameters for considered periods are
taken as ―average‖ and ―high‖ wave
conditions. Accordingly, wave tranquillity
study for proposed harbour layout was carried
out for 10 scenarios (Table 1).
4.1 Variation of Significant Wave Heights

For all scenarios (see Case No. 1 – Case No. 10
in Table 1), wave height variation in and
around the harbour was assessed. Thus, wave Figure 2 - Significant Wave Heights - Case 1
agitation and tranquillity areas are identified
and illustrated by using 2D plots (see Figure 2, Case 2 - SW Monsoon: High Wave Condition
3, 4 and 5 for Case No. 1,2, 6 and 8). Comparing
the prevailing duration of each season, SW
monsoons season is the longest season. Also,
wave effect is dominant during the same season
as a significant amount of waves comprise
generally high values. Hence, while analysing
2D plots of wave heights in the harbour area,
the outcome of SW monsoon was specially
considered. Basically, for the most
representative case (Case No. 1: SW – average),
significant wave heights (Hs) in basin area
varies between 0.1 m – 0.25 m. In harbour
entrance area, significant wave heights vary
between 0.25 m – 0.75 m. For south west high
case (Case No. 2) also, significant wave heights
are lesser than 0.25 m, indicating the calmness
inside the basin. At the entrance, significant Figure 3 - Significant Wave Heights - Case 2
wave height has increased up to 1 m compared
to SW – average condition (i.e. Case No. 1). For Case 6 - NE Monsoon - Average Wave Condition – 2
Case No. 3, 4, 5, 7, 9 and 10, relatively low wave
disturbances prevail at harbour basin and
entrance. In case of particular NE average and
high wave conditions (Case No. 6 and 8 in
Figure 4 and 5), slightly elevated significant
wave heights up to 0.25 m are observed inside
the basin. In front of quay wall area also, same
observation can be seen. This is due to the
direct penetration of wave during the NE
monsoon as wave propagation directions
belong to north – west quadrant (i.e. 295 0 N and
3050 N). However, considering the duration of
NE monsoon period and percentage of
occurrence of Case No. 6 and 8, expected
harbour operations are safe for these conditions
as wave heights are within the allowable range.
Figure 4 - Significant Wave Heights - Case 6
In general, the proposed harbour layout
provides sufficient wave tranquillity for
harbour operations.
ENGINEER 4
250
Case 8 - NE Monsoon - High Wave Condition – 2
the sizes of the vessels that will be operated
from proposed Wennappuwa Fishery Harbour.
The PIANC criteria have been specified for a
large fishing craft having a length overall
(LOA) 25 – 60 m.
Extracted significant wave heights for each

scenario are presented in Table 3. In front of
quay wall area, the maximum wave height is
0.25 m for Case No: 8 (i.e. NE High (Hs =
1.10m, Tp = 4.80 sec, Dir. = 305 0 N)). This could
mainly result as quay wall area is directly
exposed to 305 0N wave direction. However, as
per the allowable values presented in Table 2,
loading and unloading operations are safe at
quay wall as extracted significant values are
lesser than allowable value of 0.4 m. Further, in
Figure 5 - Significant Wave Heights - Case 8 turning circle area and spending beach area,
maximum significant wave heights are 0.18 m
4.2 Wave Heights at Selected Locations of and 0.13 m respectively. Hence, mooring
Harbour operations could be done in the basin with
For each scenario (see Case No. 1 – Case No. 10 minimum disturbance for all seasons. At
in Table 1), significant wave heights were harbour entrance, maximum significant wave
extracted at selected locations in the proposed heights over 0.4 m have been recorded only for
harbour (see Figure 6 (a)).To evaluate whether Case No: 2 (i.e. SW High (Hs = 1.70 m, Tp =
the simulated wave conditions inside the 6.50 sec, Dir. = 265 0N)) and Case No: 8 only (i.e.
harbour basin are within the acceptable limits, maximum values of 0.46 m and 0.51 m
the criteria tabulated in Table 2 is used. These respectively). Thus, these results indicate the
criteria were suitably adopted from the PIANC safe navigation across harbour entrance.
Guidelines (Criteria for Movement of Moored
Ships in Harbours), giving due consideration to
Table 2 - Allowable Significant Wave Heights for Safe Mooring and Operations of Vessels within
the Harbour Basin
Type of Boat Operation Allowable Significant Wave Height (Hs)
Loading/unloading operation at a Quay Hs should not exceed 0.4 m more than 1 week/year
Mooring in the basin Hs should not exceed 0.6 m more than 1 week/year
Mooring - survival Hs should not exceed 0.4 m during extreme condition
Figure 6 - Wave Extracted Locations and XY, AB Longitudinal Profiles for Wave Analysis
5
251 ENGINEER
Table 3 - Significant Wave Heights at Selected Points- BW Simulations
Incoming Wave Extracted Wave Heights (m)
Case No Season
Height (m) E1 E2 TC Q1 Q2 Q3 B
1 SW 1.10 0.37 0.23 0.13 0.14 0.15 0.13 0.12
2 SW 1.70 0.46 0.27 0.16 0.18 0.19 0.15 0.13
3 IM-1 0.70 0.19 0.12 0.06 0.07 0.08 0.07 0.06
4 IM-1 1.50 0.33 0.19 0.12 0.13 0.13 0.11 0.09
5 NE 0.50 0.17 0.10 0.06 0.06 0.08 0.07 0.05
6 NE** 0.70 0.37 0.23 0.12 0.13 0.13 0.09 0.06
7 NE 0.90 0.33 0.23 0.11 0.13 0.13 0.12 0.11
8 NE** 1.10 0.51 0.36 0.18 0.25 0.18 0.12 0.10
9 IM-2 0.50 0.15 0.09 0.05 0.05 0.06 0.05 0.04
10 IM-2 1.10 0.28 0.16 0.09 0.10 0.10 0.09 0.07
Further analysis of wave disturbances was resulting due to the lack of shelter for 3050 N
carried out for the proposed harbour at direction.
entrance and quay wall area (see Figure 6 (b)).
The analysis was carried out on the basis of Figure 8 illustrates wave height variation along
extracted wave heights profiles. Accordingly, E - Y longitudinal profile. It further clarifies the
extracted wave heights along X - Y (i.e. across observations which have been identified from
the entrance) and A – B (i.e. in front of quay Figure 7. It is clear that all wave heights inside
wall) longitudinal profiles for Case No. 1 – Case the harbour are less than 0.4 m.
No. 10 were compared (see Figure 7).
Figure 9 illustrates wave heights variation in
Wave height variation along the X - Y front of quay wall area along E - Y longitudinal
longitudinal profile for all wave conditions (i.e. axis. Generally, wave heights are below 0.4 m
Case No. 1 – Case No. 10) are shown in Figure for all cases ensuring safe loading and
7. It is clear that there is a remarkable reduction unloading operations at the proposed quay
of wave heights across the harbour entrance. wall areas. Highest waves are observed for both
Outside the harbour, the highest wave is Case No 2 (SW – Average) and Case No. 8 (NE
observed for Case No. 2 (SW – Average). Inside – High - 3050 N direction).
the harbour, the highest wave is observed for
Case No. 8 (NE – High - 3050 N direction),
Wave Heights along X-Y (Entrance)

1.80
1.60 Case 1 Case 2 Harbour Entrance
1.40 Case 3 Case 4
1.20 Case 5 Case 6
Case 7 Case 8
1.00
Hs (m)
Case 9 Case 10
0.80
0.60 E
0.40
0.20
0.00
0 50 100 150 200 250
X E Y
X - Y (m)
Figure 7 - Wave Heights Variation along Longitudinal Profile X-Y

(Entrance – Inside and Outside of the Harbour)
ENGINEER 6
252
Wave Heights along E-Y (Entrance)
0.50
0.45 Harbour Entrance
Case 1 Case 2
0.40 Case 3 Case 4
0.35 Case 5 Case 6
0.30 Case 7 Case 8
Hs (m)
0.25 Case 9 Case 10

0.20 E
0.15
0.10
0.05
0.00
160 180 200 220 240 260
E Y
X - Y (m)
Figure 8 - Wave Heights Variation along Longitudinal Profile E-Y (Entrance)

(Entrance – Inside the Harbour)
Wave Heights along A-B (Quay Wall)
0.50
0.45 Case 1 Case 2
0.40 Case 3 Case 4
0.35 Case 5 Case 6
0.30
Case 7 Case 8
Hs (m)
Case 9 Case 10
0.25 E
0.20
0.15
0.10
0.05
0.00 Quay Wall
0 50 100 150 200 250
B
A
A A - B (m)
Figure 9 - Wave Heights Variation along Longitudinal Profile A - B (Quay Wall)
5. Conclusions
Wave tranquillity study was carried out for the phenomenon. On the basis of already
development of proposed Fishery Harbour in established wave climate in the study area, BW
Wennappuwa. The main objective of this study simulation was carried out for 10 representative
was to carry out a wave calmness study for the wave conditions which cover southwest (SW),
proposed harbour to ensure the mooring inter-monsoon-1(IM1), northeast (NE) and
arrangements and safe navigational through inter-monsoon-2(IM2) seasons. As the outcome
the harbour entrance in relation to predefined of the simulation, 2-dimensional wave plots,
criteria for acceptable wave disturbance. For the significant wave heights at selected locations
simulation of wave disturbances inside harbour and longitudinal wave profiles were extracted
basins and harbour entrance, Boussinesq Wave and analyzed for all scenarios. From the
(BW) module of MIKE 21 modelling system analysis, it was identified that maximum
was used. Comparing with other wave significant wave heights are less than 0.4 m for
simulation models, Boussinesq type wave most of scenarios at the harbour entrance,
model is a powerful tool for determination and inside and quay wall areas indicating the safe
assessment of wave dynamics in ports and navigation across the harbour entrance. Hence,
harbours in coastal areas. Especially, it includes as per the PIANC Guidelines and simulation
non-linearity and frequency dispersion and has results, it was concluded that navigational and
capability of capture and simulate non-linear mooring operations could be done in the basin
7
253 ENGINEER
with minimum disturbance for all seasons. 4. Jitendra K. Panigrahi, Padhy, C. P., Murty, A. S.
Thus, suitability of the harbour location and N., "Inner Harbour Wave Agitation using
layout in terms of wave tranquillity was Boussinesq Wave Model". International Journal of
justified with the numerically simulated results. Naval Architecture and Ocean Engineering, Vol. 7,
Issue 1, January, 2015, pp. 70-86.
Acknowledgement
5. Laknath, D. P. C., Caldera, H. P. G. M. and
Ranasinghe, D. P. L., ―Wave Hindcasting and
The authors wish to acknowledge the Extreme Value Analysis for North-western Coast
―Construction of Fishery Harbours and of Sri Lanka‖, Proc., 27th International Offshore and
Anchorages Project‖ of MFARD (Ministry of Polar Engineering Conference, ISOPE, San Francisco,
Fisheries and Aquatic Resources Development) USA, 2017
in Sri Lanka for various supports for the success
of this study. 6. Madsen, P. A., and Sorensen, O. R., "A New Form
of the Boussinesq Equations with Improved Linear
Dispersion Characteristics". Coastal Engineering,
References
Vol. 18, 1992, pp. 183-204.
1. Laknath, D. P. C., and Weesakul, S. "Natural
7. Madsen, P. A., Sørensen, O. R., and Schäffer, H. A.,
Harmonize Design Improvements of Hikkaduwa
"Surf Zone Dynamics Simulated by a Boussinesq
Fishery Harbor". Proc., Conference on Coastal and
Type Model. Part I. Model Description and Cross-
Port Engineering in Developing Countries (COPEDEC
Shore Motion of Regular Waves". Coastal
IX), Rio de Janeiro, Brazil,2016
Engineering, Vol. 32, 1997a, pp. 255-287.
2. Santosh K. Kori , Amol S. Borkar, Vaidya, A. M.,
8. Madsen, P. A., SØrensen, O. R. and Schäffer, H. A.,
Kudale, M. D., " Design of Fishing Harbour Layout
"Surf Zone Dynamics Simulated by a Boussinesq
in High Littoral Drift Zone". Procedia Engineering,
Type Model. Part II: Surf Beat and Swash
Vol.115, 2015, pp.320-325
Oscillations for Wave Groups and Irregular
Waves". Coastal Engineering, Vol. 32, 1997b, pp.
3. Barve, K. H., Ranganath, L. R., M. Karthikeyana,
289-319.
Kori, S., and Kudale, M. D., ―Wave Tranquility
and Littoral Studies for Development of a Mini
Fishing Harbour". Aquatic Procedia, Vol. 4, 2015, 9. SØrensen, O. R., Schäffer, H. A. and SØrensen, L.
S., "Boussincsq—Type Modelling using An
pp. 72-78
Unstructured Finite Element Technique". Coastal
Engineering, Vol. 50, 2004, pp. 181-198.
ENGINEER 8
254
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[255 262],2017
2017
The Institution
Lanka
Verification of Wave Transformation to Establish the

Nearshore Wave Climate
D.P.L. Ranasinghe, N.L. Engiliyage, H.P.G.M. Caldera and I.G.I. Kumara
Abstract: MIKE 21 SW model simulation and wave transformation matrixes have been used to
transform the reordered wave data to desired locations. Both back and forward transformation were
carried out since the direct transformations were not possible due to the geophysical features exist
between wave sources and the destination. As the first step, all the measured data (Galle and
Hambantota) were transferred to a 100 m depth directly located off Gandara coast and then
transferred to Gandara and Kudawella (15m depth) which is the nearest location with recorded wave
data to Gandara. The comparison depicted that the South-west and Inter-Monsoon 1 monsoonal
waves transformed from Galle shows a statistical mismatch with on-site recordings while all the
transformed data from Hambantota is in lined with recorded data. Furthermore, it has been observed
that local features like headlands decline the southwest sea wave effect in Gandara which is prevailing
at southern coast. The transformed wave data which could not be statistically verified were not used
to develop wave data base at Gandara. Upon completion of building up of the wave data base for
Gandara, the extreme wave occurrences were estimated by extrapolating the wave heights within the
built-up data base. Furthermore, relevant peak wave period for extreme wave heights was also
obtained through a measured data in the vicinity. The obtained extreme values for 50 years return
period (Hs=3.89 m, Tp=14.9 s) at Gandara 20 m depth were compatible with past relevant studies
done on the southern coast. Thus, the importance of verification and screening process of transformed
wave data in the establishment of nearshore wave climate was evident.
Keywords: Galle offshore, Gandara, extreme peak wave period
1. Introduction numerical modelling platform in this study.

Even though a large number of data is given
1.1 General the higher accuracy on generated wave climate,
As the behaviour of a coastal zone is completely higher computational time is required to handle
different from one to another, site specific the data. The situation will be further
parameters are required prior to the design of complicated and time-consuming when the
coastal structures. Furthermore, the designs measured data is available at different
need to withstand against the extreme events locations. In such a situation usage of wave,
that could be occurred during its design life. transformation matrix method would be an
Therefore, extreme wave analysis is needed advantage.
through a long term wave data in order to have
a long lasting and cost effective design of 1.2 Study Area
coastal structures. Even though the relevant The proposed site Gandara is situated in the
details on the bathymetry and topography Southern coastal belt in Devinuwara Divisional
could be obtained through a single survey Secretariat division of Matara District (Figure
exercise, it is unable to obtain a considerable 1). Being a partially sheltered Bay, the proposed
measured wave data series at each and every location is currently acted like a fish landing
location that needs a coastal development. centre. However, the navigation difficulties
Therefore, numerical models are played a vital exist near landing site due to the rocky
role to transform the recorded wave data to the outcrops in the nearshore area.
desired location for the purpose of obtaining Eng. (Dr.) D.P.L. Ranasinghe, C.Eng., MIE(SL), B.Sc.
the design wave parameters which are locally Eng.(Ruhuna), M.Sc. (Moratuwa), Ph.D.(Tohoku),
Engineering Manager, Lanka Hydraulic Institute.
unavailable.
Currently, there are several commercial and Eng. N.L. Engiliyage, AMIE(SL), B.Sc.Eng.(Moratuwa),
non-commercial software that can be used for Research Engineer, Lanka Hydraulic Institute.
numerical model simulations to assess the wave
Eng. H.P.G.M. Caldera, AMIE(SL), B.Sc.Eng.
transformation process due to shoaling, (Peradeniya), Research Engineer, Lanka Hydraulic Institute.
refraction, wave breaking etc. Among them,
MIKE 21 SW module has been used as the Eng. I.G.I. Kumara, AMIE(SL), B.Sc.Eng.(Moratuwa),
Research Engineer, Lanka Hydraulic Institute.
1
255 ENGINEER
Figure 1 - Location Map and Local Bathymetry [1] (Source: http://maps.google.com)
Therefore, the necessity of the maximum effect of refraction causes the wave direction to
utilization of the Bay via breakwater is evident. change towards a shore normal direction.
Both south and north part of the bay is covered Therefore, local variations in swell wave
by the cliffs while small beach area is situated directions are observed in shallow nearshore
in the southwest of the bay. Therefore, waters. The sea waves are those waves being
sediment intrusion due to alongshore transport generated by the local wind fields. They have
seems to be a negligible and seasonal variation shorter periods compared to swell waves and
of the shoreline could be visible from the sandy are much steeper representing a chaotic sea
beach area [1]. Since Gandara on-site wave state with component waves widely scattered
recordings were not available, the in approach direction. The southern coast of Sri
transformation of wave data from other Lanka is primarily subjected to the influence of
available locations off the southern coast was sea waves generated by south-westerly
found to be necessary to establish the data base. monsoonal winds. However, moving towards
the south-east, progressively increased the
1.3 Objectives influence of north-east monsoonal sea waves is
The purpose of this study to establish nearshore observed.
wave climate at Gandara (20m and 10m depths) A summary of available wave data in Southern
using measured wave data at Galle offshore (70 Sri Lanka that has been considered in this study
m depth) and Hambantota near shore (17m and is given in Table 1, and the wave recording
10m depths) in Sri Lanka through a detailed locations are indicated in Figure 2.
verification using measured wave data at
Kudawella (15 m depth) and hence, to estimate Table 1 - Available Wave Data at Southern
the extreme waves. Coast
Duration
2. Wave Data Base and Existing Location/Depth
Wave Climate at Southern From To
Coast of Sri Lanka Galle (70m)

Feb 1989 Sep 1992
The southern coast of Sri Lanka is subjected to May 1994 Sep 1995
the dual influence of two distinct wave systems, Kudawella (15m) Jan 1996 Feb 1997
namely swell waves and sea waves. Swell
waves are those generated by wind fields in the Hambantota (17m) Dec 2002 Mar 2003
vast open sea areas of the Southern Indian
Ocean and have propagated beyond the Hambantota (17m) May 2006 May 2009
generating fetch to reach the coast. These waves
approach from a southerly direction in the deep Hambantota (10m) Apr 2013 Mar 2014
sea. However, as they approach the shore, the
ENGINEER 2
256
Figure 2 - Wave Recording Locations and Wave Transformation Sequences
(Source: http://maps.google.com)
The most comprehensive wave data base 3. Numerical Model Simulations

available for the southern coast is that the
directional wave recordings conducted around Since accurate assessment of wave conditions is
70 m water depth, about 8 km offshore of Galle very important for designing structures in
under the CCD-GTZ directional wave climate coastal regions or designing harbour structures,
study. Considering the deep sea location used wave modelling is essential. As wave induced
for recording waves, CCD-GTZ wave data can currents have a significant role in transporting
be regarded as representative of swell waves at sediment in coastal areas, knowledge of wave
that depth for the entire southern coast. climate is necessary in order to carry out coastal
Furthermore, these data can also be readily design structures with the minimal impact to
used for transformation of monsoons' induced the existing littoral movements.
sea waves to other locations [2]. Directional The spectral wave module, DHI MIKE 21 SW
wave recordings at 15 m water depth in developed by Danish Hydraulic Institute,
Kudawella were carried out in 1996-1997 about Denmark, is a state-of-the-art spectral wind-
13 km northeast of Gandara. Apart from that, wave model. The module enables to simulate
two sources of oceanographic data available growth, decay and transformation of wind-
from Hambantota near shore at 17 m and 10 m generated waves and swells in offshore and
water depths. coastal areas. The model includes the main
Due to close proximity to Gandara, physical phenomena, for example, the
Hambantota wave recordings can be processes of wave generation by wind,
considered to constitute a reliable data base for quadruplet wave-wave interactions, white-
transformation in the first instance. However, capping and bottom friction [3]. Further, this
the possible sheltering effect on sea waves by model handles processes that only affect waves
local coastline features like headlands could in coastal waters such as depth induced wave
restrict the usage of whole recorded wave data breaking and triad wave-wave interactions. The
at Hambantota for deriving wave conditions at formulations for the wave generation processes
Gandara. In some monsoonal sea waves are based on the wave action balance equation
approaching to Gandara exhibit a very weak (i.e. in MIKE 21 SW the waves are represented
influence on the Galle offshore wave recording by the wave action density spectrum). MIKE 21
location and vice versa. However, it has been SW uses an unstructured mesh for the
identified that establishment of nearshore data numerical solution. An unstructured mesh has
base using overall waves is more appropriate to the advantage that it is highly flexible and
estimate the design wave parameters. therefore suitable for modelling projects where
Therefore, waves transformed from Galle high resolution is needed in the area of interest,
offshore and Hambantota near shore to and coarse resolution is sufficient in the
Gandara should be used with caution and surroundings. The necessary model Input data
requires prior verification comparing with on of MIKE 21 SW Model can be divided into
site wave recordings from Kudawella which is groups such as domain and time parameters,
the nearest location with recorded wave data to equations, discretization and solution
Gandara. technique, forcing parameters, source function
3
257 ENGINEER
parameters, initial conditions and boundary waves are highly influenced by wind field, well
conditions. Model Output such as significant established Beaufort Scale standard was used to
wave height (Hs), peak wave period (Tp), mean incorporate the wind speed to the simulation.
wave direction (MWD) and radiation stress Accordingly, the wind speed was correlated
tensor can be obtained from MIKE 21 SW at with the offshore significant wave height based
each mesh point for each time step. on the Beaufort Scale [4] description of the
marine water (Table 3). It is assumed that wind
3.1 Wave Transformation directions are similar to the offshore wave
The applied methodology to hind cast wave direction over the entire model domain.
climate of this study is based on the application
of MIKE 21 SW model to simulate wave Table 3 - Wind Speed for Offshore Significant
propagation off the southern coast (offshore) Wave Height [4]
and using a wave transformation matrix Significant Wave Height (m) Speed (m/s)
approach to transform wave recordings based
0.5 2.9
on reproduced model results. Wave
propagation modelling was carried out to 1.0 4.8
transform wave recordings from Galle and 1.5 6.6
Hambantota to representative locations in the 2.0 8.3
study area. Thus, representative nearshore
2.5 9.8
wave climate was established at 20 m and 10 m
3.0 11.2
water depths in the study area (Gandara). Both
back and forward transformation were carried 3.5 12.5
out since the direct transformations were not 4.0 13.7
possible due to the geophysical features exist 4.5 14.8
between wave sources and the destination. As
5.0 15.9
the first step, all the measured data were
5.5 16.9
transferred to a 100m depth directly located off
Gandara coast and then transformed back to The southern coastal belt is influenced by two
the desired destination locations (Gandara and major monsoons of South West (SW: May –
Kudawella). For minimizing a large number of September) and North East (NE: December -
data transformation of individual wave February) as well as the two inter monsoons
recordings, a transformation matrix approach (IM1 and IM2) created due to changing of wind
which relies on the transformation of a selected patterns end of the major monsoons. Therefore,
set of wave conditions at the offshore boundary all the four seasons were considered for the
was adopted. A set of wave conditions simulations and the analysis in order get the
comprising combinations of following wave insight of the seasonal variation in wave
parameters were specified at the offshore climate.
boundary. Thus, MIKE 21 SW model
simulation was carried out to transfer 3.2 Assessment and Verification of Wave
combination of wave parameters as presented Transformation
in Table 2. The verification of wave data sets obtained by
transformation from different locations and on-
Table 2 - MIKE 21 SW Model Setup site recordings in the time domain is not
possible as there is no physical overlap between
Hs (m) Tp (s) MWD(˚N) different data sources. Therefore, the
comparison between different wave data
0.5 1,2,...........27 40,50,...........260 sources was made only in a statistical sense.
Such comparisons for measured and
1.0 1,2,...........27 40,50,...........260 transformed data sets at Kudawella are
illustrated in Figure 3 for overall seasonal
waves. In these illustrations, the mean value of
.....
.....
.....
a particular wave direction class interval

5.0 1,2,...........27 40,50,...........260 considered (e.g. 155° in the direction class 150° -
160°) represents that particular direction class
interval.
The specific significant wave heights of 0.5 m,
It is not realistic to expect similar distributions
1.5 m......and 4.5 m were assumed to represent
in wave heights in two data sources covering
offshore wave height ranges 0 m - 1.0 m, 1.0 m -
different durations. A longer duration wave
2.0 m.......and 4.0 m - 5.0 m respectively. Since
ENGINEER 4
258
data source is more likely contain high and constitute a representative directional wave
extreme waves. However, at a given location a data base.
particular wave system is known to approach
from a characteristic direction range over Table 4 - Comparison of Transformed Wave
different seasons. Therefore, in the statistical Data
comparison, these trends in directional Season Measured
distributions were looked at for different data
MWD(˚N) Direction Band(˚N)
sources. In this exercise, the two main
considerations for satisfactory comparison SW 165 160-180
between two data sets adopted were: similarity IM1 175 160-180
in dominant wave approach direction; and
NE 155 130-180
similarity in range of wave directions. The
magnitude of the percentage of waves IM2 165 150-180
representing the dominant wave direction is Transformed from Hambantota
only of secondary importance, as that will vary
with the time duration covered by the wave SW 165 160-180
data source. IM1 165 150-190
NE 165 120-180
IM2 165 160-190
Transformed from Galle
SW 205 190-220
IM1 195 170-220
NE 125 100-160
IM2 175 160-220
4. Nearshore Wave Climate at

Gandara
As concluded in Section 3.2, the following wave

data sets were combined to constitute the
directional wave data bases for overall waves at
Gandara 20 m and 10 m water depths (Table 5).
Therefore, transformed data from Hambantota
Figure 3 - Wave Roses for Measured and for SW and IM1 seasons and transformed data
Transformed Data at Kudawella 15 m Depth from Hambantota and Galle for NE and IM2
As shown in Table 4, the comparisons between were used to develop the nearshore wave data
different data sets clearly indicate that wave base at Gandara.
data transformed from Hambantota showed
similar directional trends throughout the year Table 5- Selection of Transformed Wave Data
as obtained from Kudawella site recordings. Transformed From
Galle offshore transformed wave data, although Season Hambantota Hambantota Galle
compared poorly with site recordings during 17m 10m 70m
SW and IM1 seasons. In contrast, Galle
SW √ √ X
transformed data indicates the very similar
range of approach directions in NE season IM1 √ √ X
while IM2 season shows similar dominant
NE √ √ √
direction with wider direction band. These
findings indicate that in the SW and IM1 IM2 √ √ √
seasons wave data transformed from
Hambantota can be combined with on-site The established wave data series at Gandara
recordings, while in the other two seasons of nearshore 20 m depth indicates that
the year additionally, wave data transformed comparatively high waves are approaching
from Galle offshore could also be combined to from the southern direction in SW monsoon
while during NE monsoon lower magnitude's
5
259 ENGINEER
wave is dominating with a wider direction 5. Extreme Wave Analysis
band in south-east quadrant (Figure 4). In other
words, swell component is higher than the sea In the design of breakwater structures for the
component in SW overall waves and vice versa proposed harbour, it is necessary to ensure the
in NE overall waves at 20 m depth of Gandara. structural stability against extreme wave
Table 6 indicates the characteristics of high and events. Since it is expected to design the
average waves in all four seasons. structures for 50 years of design life, a
significant wave height of 100 year return
Table 6 - Average and High Wave Conditions period (YRP) is considered as the design wave
at Gandara 20 m Depth height. However, very high waves are likely to
Average Extreme be attenuated due to breaking before reaching
(50% Exceedance ) (2% Exceedance ) the structure, whereas a wave of somewhat
20 m
Depth lower return period may approach the structure
Hs Tp MWD Hs Tp MWD
without significant reduction in height.
(m) (s) (˚N) (m) (s) (˚N)
Therefore, Extreme wave analysis was
SW 1.7 13.5 187 2.3 14.2 187 performed for different return period as well as
for both 20 m and 10 m depths. Results
IM1 1.5 11.8 187 2.5 15.5 187 obtained at 20 m depth will be used for the
NE 1.3 5.6 152 1.9 14.4 184 harbour tranquillity modelling while 10 m
depths data is needed to design the proposed
IM2 1.5 9.6 189 1.9 11.4 187 marine structures.
Due to the effect of local features like headlands 5.1 Significant Wave Heights
and the nearshore bathymetry, all the waves at The extreme significant wave heights for
10 m of Gandara Bay is approaching from annual overall waves for Gandara (20 m and 10
Southeast direction despite monsoonal change m) were determined through extrapolation of
(Figure 4). time series established a database. In the
method adopted, the maximum wave heights
occurring during 3 hours storms were selected
for extrapolation and data series was controlled
for the outliers [5]. This selection of wave
heights yields a statistically independent set of
data appropriate for making predictions. The
inclusion of comparatively low wave heights in
an extreme wave analysis would often lead to
underprediction. In order to exclude such
storms, the Peak Over Threshold (POT) method
was applied [5]. Selected storm events were
fitted to the standard statistical Extreme Value
Probability distributions and evaluated the
goodness of fit of each distribution. It was
observed that the selected wave data was well
fitted with three parameters (Shape parameter-
α, Scale Parameter-β, Location Parameter-γ)
Weibull Distribution (3P WD) and yielded most
consistent results compared with other
distributions. Therefore, 3P WD (with α=1.0)
was adopted to compute the extreme wave
heights for different return periods using
regression analysis [6].
Figure 4 - Wave Statistics at Gandara 20 m 5.2 Peak Wave Periods

and 10 m Depths The estimation of peak wave period that can be
associated with extreme significant wave height
is still a challenging task for the coastal
engineers. Since the estimated extreme wave
heights are not included in the primary data
series, a direct correlation is unable to obtain.
ENGINEER 6
260
On the other hand, breaking the effect of higher overall waves during each monsoon. Extreme
waves could be reduced when the extrapolation waves for the same return period wave
of peak period is used in the analysis. illustrated in Table 8 for better comparison. It is
Therefore, special care need to be taken while observed that the extreme waves for Gandara
choosing the appropriate extreme wave periods near shore are in line with the previous studies
since overtopping behaviour is completely done for the adjacent locations.
different in short (sea) and long (swell) periods
waves. Hence, available data records at the Table 8 - Summarized Results for 50 YRP
vicinity were carefully analysed in order to Extreme Waves in Southern Coast
obtain a correlation between wave heights and Designed
the wave periods. The design wave period for Wave
D Wave
particular extreme significant wave height was Location conditions Ref.
(m) Type
estimated using regression analysis [5]. This Hs Tp
regression line was generated for the measured (m) (s)
data at Hambantota 10 m depth as it contains a 15 4.37 16.5 AS
robust set of data. Since overall wave heights Suduwella [7]
were considered in this study, mean wave 15 3.85 8.5 ASea
periods (Tz) were considered instead of peak 15 3.41 16.9 AS
wave period (Tp) to develop the relationship. Kudawella [8]
Because the scatteredness of the peak wave 15 2.52 5.4 ASea
periods are higher than that of the mean wave 2.74 13.0 AS
periods for the same wave heights. In order to
Kudawella 15 3.33 5.0 SWS [9]
develop the better relationship, wave heights
were split into the classes of 0.2 m intervals, 2.86 4.0 NES
and the average of each class was plotted
3.58 16.8 AS
against particular mean wave periods, and the
regression line was obtained. Computed mean Hambantota 17 4.92 4.5 SWS [10]
wave periods for extreme wave heights were 2.60 4.5 NES
then transferred to peak wave period using
Tp=1.4Tz relationship which was estimated 3.86 13.5 AS
using recorded data in the vicinity. Hambantota 17 4.09 4.5 SWS [9]
The computed extreme wave heights and the
relevant peak wave periods are given in Table 2.59 5.0 NES
7. Gandara 20 3.89 14.9 AO
Present
Table 7 - Extreme Waves for Different Return Study 10 2.97 14.0 AO
Periods at Gandara 20 m Depth Note:D-Depth, AO- Annual Overall, AS- Annual Swell,
SWS-SW Sea, NES-NE Sea, ASea-Annual Sea
Return Period Hs/(m) Tp/(s)
2 3.28 14.4 7. Conclusions

10 3.58 14.7
Model verification using Kudawella wave
25 3.76 14.8 recordings indicated that the transformed wave
50 3.89 14.9 data from Hambantota for all seasons and Galle
transformed data only for NE and IM2 seasons
100 4.02 15.0
could be utilized to establish the wave data
base at Gandara Nearshore. The wave statistics
6. Comparison of Results at Gandara depicted that swell component is
higher than the sea component in SW overall
Since the structural stability and hence its waves and vice versa in NE overall waves at 20
durability depend on the accuracy of the design m depth of Gandara. Furthermore, all the
wave parameters, compatibility of obtained waves at 10 m of Gandara Bay are approaching
results has been compared with previous from Southeast direction despite the monsoonal
studies that have been done for the southern change. Estimation of peak wave period which
coastal belt. The studies done for the overall associates with extreme wave heights
waves could be directly compared with the concluded that swell component of extreme
present study while other studies provide the overall waves would be higher than the sea
indication of Sea and Swell component of the component in Gandara Nearshore.
7
261 ENGINEER
Acknowledgement
The authors wish to acknowledge the

“Construction of Fishery Harbours and
Anchorages Project” of MFARD (Ministry of
Fisheries and Aquatic Resources Development)
in Sri Lanka and the Project Director, Eng.
K.K.P.P. Ranaweera as well as to Dr. K.
Raveenthiran for the various supports given
during this research. Further, the authors'
gratitude also extended to Late Dr. P. P.
Gunaratne for his involvement and guidance
given for wave climate studies in Sri Lanka.
References
1. Lanka Hydraulic Institute, Interim Report 1 on
Detailed Design on Gandara Fishery Harbour,
2017.
2. Scheffer, H. J., Fernando, K. R. M. D. and

Fittschen, T.Directional Wave Climate
Southwest Coast of Sri Lanka. Colombo. CCD-
GTZ Coast Conservation Project, 1994.
3. DHI,MIKE21 Spectral Wave Module. Scientific

documentation, Danish Hydraulic Institute
(DHI), 2009.
4. https://en.wikipedia.org/wiki/Beaufort_scale,
Visited, 10th March 2017.
5. Caires, S., Extreme Value Analysis: Wave Data,

JCOMM Technical Report No. 57, World
Meteorological Organization, 2011.
6. Kamphuis J. W., Introduction to Coastal

Engineering and Management 2nd Edition:,
Advanced Series on Ocean Engineering-Volume
30, World Scientific, 2000
7. Uni Consultancy Services: University of

Moratuwa, Final Report on Feasibility Study for
the Development of a Fishery Harbour at
Suduwella, Kottagoda in Matara District, 2009.
8. Lanka Hydraulic Institute, Final Report on

Kudawella Fishery Anchorage Development,
1997.
9. Ranasinghe, D. P. L. and Gunaratne, P. P.

Assessment of Nearshore Wave Climate In
Southern Coast Of Sri Lanka, Annual Research
Journal of SLSAJ, Vol. 11, Japan, 2011.
10. Lanka Hydraulic Institute, Final Report on

Mathematical Model Test on Wave Field:
Hambantota Seaport Development – Phase I,
2008.
ENGINEER 8
262
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[263 269],2017
2017
The Institution
Lanka
Fabrication and Mechanical Property Analysis of

Microcrystalline Cellulose Based Polymer Composite
for Engineering Applications
H.A.D. Saumyadi, L.D. Rajapaksha, A.M.P.B. Samarasekara
D.A.S. Amarasinghe and L. Karunanayake
Abstract: In the last several years, polymer composites have been used heavily in aerospace,
automotive and other engineering applications. Natural fibers are recently getting attention from
researchers and academia to utilize in polymer composites due to their eco-friendly nature and
sustainability. In this research, Polypropylene matrix with microcrystalline cellulose (MCC) reinforced
composite was investigated for their mechanical properties.
MCC was subjected to surface modification technique to improve compatibility with hydrophobic
Polypropylene (PP) using silane treatment. Polypropylene was mixed with surface treated MCC by
varying MCC concentration (1% wt. to 5% wt.) in a laboratory type internal mixer. Impact, tensile,
hardness and water absorption tests were performed to evaluate the mechanical properties of the
developed composites. Experimental results showed that gradual increase of tensile strength,
hardness and impact strength with increase of MCC concentration. Polypropylene with 4 wt. %
sample showed the maximum impact strength and it was 18.2 KJ/m 2. Maximum water absorption
(0.02%) was observed in 5 wt% MCC containing sample. 5wt% MCC containing sample showed the
maximum hardness (74.4 Shore D). Developed composite showed the gradual reduction of density
from 1 wt% MCC (0.882 g/cm3) to 5 wt% MCC (0.826 g/cm3). Therefore, Polypropylene with MCC
polymer composite can be used for different engineering applications providing attractive benefits.
Keywords: Micro crystalline cellulose, composite, polypropylene
1. Introduction as reduced petroleum use, life-cycle emissions

and part weight while increasing recyclability
Cellulose is a naturally occurring of polymers. Natural fiber based composites
polysaccharide and it is the most abundant bio materials are widely used in different
polymer. Microcrystalline cellulose (MCC) is a industries due to their extraordinary properties.
member of cellulose family. Compared to The development and the application of
natural cellulose fibres, MCC possesses several polymeric composite materials filled with micro
advantages, such as micro scale dimension, sized rigid particles has been attracted both
high strength per mass and higher modulus, scientific and industrial interest [11-13]. The
higher specific surface area, unique optical objective of this research was to develop a MCC
properties, etc. MCC has high strength due to and polypropylene based composite with
their dense and orderly crystalline structure [1- improved mechanical properties.
7]. Ms. H.A.D. Saumyadi, B.Sc. Eng. (Moratuwa), Dept. of
Other benefits of cellulose crystals include not Materials Science & Engineering, University of Moratuwa.
costly and they are light weight with micro Mr. L.D. Rajapaksha, B.Sc. Eng. (Moratuwa), Dept. of
scale dimensions and unique morphologies. Materials Science & Engineering, University of Moratuwa.
MCC can be used as high-quality reinforcing
filler for polymers and biodegradable materials. Eng. A.M.P.B. Samarasekara, B.Sc.Eng.(Moratuwa),
M.Phil. (Moratuwa), AMIE(Sri Lanka), Senior Lecturer,
There are researches about glass fibre Dept. of Materials Science & Engineering, University of
/polypropylene composites to improving Moratuwa.
mechanical properties [8-10]. Increasing Dr. D.A.S. Amarasinghe, B.Sc.(Kelaniya), MSc.
mechanical properties such as shock (Moratuwa), M.Phil. (CUNY, USA), Ph.D.(CUNY, USA),
absorbance and strength is the aim of this Senior Lecturer, Dept. of Materials Science & Engineering,
University of Moratuwa.
research by using MCC/polypropylene
composite than other fibers with polypropylene Prof. L. Karunanayake, B.Sc. (USJP), Ph.D. (North
based composite. Furthermore, bio-based London), Professor, Department of Chemistry University of
plastics may provide so many advantages such Sri Jayewardenepura.
1
263 ENGINEER
2. Experimental Procedure 2.3.2 Silane treatment (Si -69 treatment)
20 g of MCC was added into 85 wt% ethanol,
2.1 Material dispersed by ultrasonic waves for 20 min, then
Ethanol (H₃CCH₂OH, Min. 99.85 %, ACS, VWR mixed with mixed solution which was 20 wt%
Chemicals), Stearic acid (Sigma-Aldrich), & 30 wt% Si-69(6 g) and 85 wt% ethanol
Toluene (C6H5CH3, ≥99.7%, ACS, Sigma- medium for 1 h. The mixture was dispersed by
Aldrich) were used for surface modification of ultrasonic waves for 5min, magnetic stirring at
MCC. 70 0C for 3 h. The modified material was
washed with ethanol and extracted. The
mixture was dried at room temperature to
2.2 Characterization of MCC
constant weight.
MCC was confirmed as pure MCC with correct
structure and size by using SEM image, FTIR
2.4 Composite preparation
spectrum (KBr) and TGA techniques.
Modified MCC and PP were mixed by varying
composition 1-5 wt.% to obtain good composite
2.2.1 FTIR Spectroscopy
properties. Maximum 5 wt% MCC was given
The MCC sample was mixed with KBr and
the optimum mixing homogeneity. Internal
grinded well to prepare fine powder and
mixing method was used as composite
compressed to make disc. FTIR spectrum was
preparation and compression moulding was
recorded on FTIR Bruker (α) alpha
used for sample preparation.
spectrophotometer in frequency range of 4000-
600 cm-1, with a resolution of 4 cm-1, in
2.4.1Composite preparation- Batch preparation
transmission mode. A total of 30 cumulative
Modified MCC (varying 1-5% wt.) and PP were
scans were taken.
well mixed by using internal mixer for 5 min.
(Brabandary type, Fill factor-0.8, temperature-
2.2.2Morphological investigation
150 0C, rotor speed – 65 rev/min).
The surface morphology of the MCC and
surface modified MCC particles were
investigated by Scanning Electron Microscope 3. Testing procedure
(CARLZIESS EBO – 18). A sample was coated
with gold layer until 90 seconds under high 3.1 Testing sample preparation
vacuum and then observed with an accelerating Prepared composite batch samples (by
voltage of 20 kV. considering MCC percentage variation and
modified method variation) were compression
2.2.3Thermogravimetric analysis (TGA) moulded under 165 0C for 8 min, with 0.75 psi
Thermo gravimetric analysis was performed pressure. Then samples were cooled under
with a TAS100 instrument to study the room temperature in 15 min.
degradation characteristics of the obtained
MCC. The samples were dried in oven at 80 0C 3.1.1 Impact test
for more than 8 h before test. The samples were 80 mm length, 10 mm width and 2 mm
heated from 30 to 800 0C at a heating rate of 10 thickness samples were cut from prepared
0C min-1 under N atmosphere.
2 testing samples and cut the notch according to
ISO 180 methods. Five specimens were
2.3 Surface modification methods measured for each sample by using Izod impact
The surface energy of polymer must compatible testing machine.
with MCC. Therefore, modification of MCC is a
critical process to develop composite material. 3.1.2 Tensile Test
Sample dimensions were 2-3 mm thickness, 3-5
2.3.1 Stearic acid treatment mm width and 30 mm length. The strain rate
Stearic acid 1.7 g was mixed with toluene (200 was 30 mm/min, extension range was 400 and
ml). MCC 20 g was added to stearic and toluene applied load was 1 kN at 25 0C. The five
solution and subjected to ultrasonic disruption samples were measured for each MCC
(50 kHz, 10 min) then stirred for one hour at percentage. Tensile forces and elongation at
room temperature. Then solution was dried at break were measured by using universal tensile
room temperature to constant weight to remove testing machine. This test was conducted
toluene and residual solvent. according to the ASTM D638 standard.
ENGINEER 2
264
3.1.3 Hardness
Hardness of samples was measured using a
shore durometer according to the ASTM D2240
standard. Several points of each sample were
measured to obtain Hardness values.
3.1.4 Water absorption test
Five samples were selected for each MCC
percentage. Samples were weighted and then
soaked in distilled water at room temperature.
The samples were removed at specific time
intervals (24 hrs), and removed the excess
water on the surface and immediately weighed.
This test was conducted according to the ASTM
D570 standard.
3.1.5 Density test Figure 2 - SEM image of Unmodified MCC
Densities of the samples were measured by
Specific gravity (SG) meter. Five samples were absorbed water [15]. By analysing FTIR
tested for each MCC weight percentage up to spectrum, we could confirm that obtained MCC
5%. This test was conducted according to the was pure.
ASTM D792 standard.
4.1.2 Morphological Investigation
4. Results and Discussion According to the SEM images in Figure 2,
4.1 Characterization of MCC needle type structure could be observed.
Length and diameter of the microcrystals were
4.1.1 FTIR analysis differed and had the sizes from 1 to 100
As per Figure 1, analysing a FTIR spectrum of microns [15].
the microcrystalline cellulose, it was possible to
tell that existence of a wide band in the field of 4.1.3 Thermogravimetric Analysis (TGA)
3348 cm-1. It showed that there are large Figure 3 shows the TGA and DTG curves for
number of various types of hydrogen bonds, MCC. MCC sample undergoes mass loss from
formed by –OH groups. [14] There was a peak 30 0C to 700 0C. There was an endothermic peak
at 2901 cm-1 due to the band of symmetric and at 75 0C due to the evaporation of water in this
asymmetric vibrations of CH2 groups. As well sample. So there was a 5.7% of weight loss due
as deformation vibrations of CH2 and CH to the evaporation of vapour. Additionally to
groups and angular deformation vibrations of the above degradation stage, there was another
C-O-H were located in the field 1637- 1372 cm-1. degradation stage. The 2nd degradation was
If there are hemicellulose and lignin, there believed to be due to the pyrolysis of MCC.
should be peaks at 1736 cm-1 and 1512 cm-1. There was 88.5 % of weight loss at the 2nd
However, that peaks were not appeared in this degradation stage. The total weight loss found
spectrum. That means, hemicellulose and lignin at 7000C was 94.2%.
have been removed.
The band at 1637 cm-1 was assigned to the
Figure 1 - FTIR Spectrum of Unmodified MCC Figure 3 – TGA and DTG Curves
3
265 ENGINEER
stretching vibration of the ester linkages (-C=O)
expected between 1725 and 1750 cm-1.
This confirmed that the formation of
carboxylate linkages between stearic acid and
MCC. The reduction of peak intensity was not
noticed at 3000-3700 cm-1, corresponding to the
-OH group in cellulose structure. According to
the Figure 5, stearic acid has been reacted with
some amount of -OH group in cellulose. So,
there were untreated –OH groups in MCC
structure [16].
There should be a peak at 450-500 cm-1 assigned
to Si-O-C bond in the structure. Spectral of Si-69
treated MCC (Figure 5) showed peak at 468 cm-
1 for this type of bond. Two absorption band
Figure 4 - SEM micrograph of surface
peaks at 2925 cm-1 and 1460 cm-1 were
modified MCC by (Si-69)
methylene, and on FTIR of surface modification
MCC have been significantly enhanced. As well
4.2.Characterization of Surface modified MCC
as, reduction of peak intensity was noticed at
4.2.1 Morphological Investigation 3000-3700 cm-1, corresponding to the -OH
According to the above SEM images, there were group in cellulose structure. According to the
bonded parts due to the silane (si-69) on the mixing percentage of Si-69, reduction of peak
surface of the MCC. So MCC had been surface intensity corresponding to the –OH group in
modified by Silane 69. the cellulose was varied. The amount of
4.2.2 FTIR Analysis reduction of peak intensity corresponding to
the –OH group in the cellulose had been greater
FTIR spectra of MCC (Figure 5) exhibited the in 30% Si-69 treated MCC than 20% Si-69
typical bands of cellulose. The -OH stretching treated MCC (Figure 5). So 30% Si-69 coupling
of intramolecular hydrogen bonds at 3,700– agent has been properly reacted with MCC. So
3,000 cm-1, the C-H stretching at 2,901 cm-1, the MCC was properly surface modified by 30% Si-
C-H bending at 1,370 cm-1, the C-O stretching at 69 [17].
1,059 and 1,033 cm-1 . The spectral data of pure
stearic acid showed (Figure 5) a sharp 4.3 Preparation of Composite.
absorption at (1700-1710) cm-1 wave number for
carbonyl (C=O). The spectral data of stearic 4.3.1. Casting evaporation
A proper film could not be obtained due to
acid treated MCC also showed absorption at
some reasons. Because thickening of the
1703 cm-1. This indicates that unreacted stearic
solution could not be kept constant.
acid is remained on the surface of the MCC.
The stearic acid treated MCC also showed a
4.3.2. Mechanical mixing
small peak at 1732 cm-1, corresponding to the
Laboratory type internal mixer was used to mix
polypropylene and MCC. Better mixed and
distributed composite batches were obtained by
this method. That means, polypropylene and
MCC had been properly bonded and mixed by
this method. So this method was used to
further composite mixing.
4.4 Physical properties of Composite.

The notched izod impact strength was shown
in Figure 6 for PP based silane modified MCC
composite. The impact strength was
significantly higher in the composites than pure
PP. The impact strength of the composites
increased when increased modified MCC
Figure 5 - FTIR spectra of (a) untreated MCC percentage up to 4% and then decreased
(b) 20%Si-69-treated MCC (c) 30% Si-69- furthermore.
treated MCC (d) stearic acid treated MCC
ENGINEER 4
266
Figure 6 – Impact strength of Developed Figure 10 – Water absorption of Developed
Composite composite
Figure 7 – Tensile Strength of Developed Figure 11 – Density of Developed Composite

Composite
Composites with modified MCC by silane
coupling agent, the impact strength seemed to
be dependent on modified MCC content. The
results strongly suggest that silane modified
MCC, acts as a compatibilizer, improves the
compatibility between hydrophilic MCC and
the hydrophobic PP matrix. That results strong
connections between MCC and PP due to
removing OH groups on MCC by silane
groups. But beyond 4% MCC was not working
as a filler material. Because MCC is a high
crystalline material. Therefore, high percentage
of MCC is not giving good properties as a filler
Figure 8 – Elongation at break of Developed
material to impact force and high crystallinity
composite
gives low impact strength.
The tensile strength at break of the composite
was shown in Figure 7. The tensile strength at
break exhibited sharp increases up to 5 wt% of
modified MCC.
The higher tensile strength shows in modified
MCC containing samples due to improved
compatibility between hydrophilic MCC and
the hydrophobic PP matrix. Therefore, a diffuse
interlayer can be formed between the PP and
MCC exhibiting excellent adhesion to each
other. So MCC has been act as a filler material
for this composite.
Figure 9 – Hardness of Developed Composite
5
267 ENGINEER
Elongation at break of pure PP was higher than can be used for light weight engineering
composite. (Figure 8) Elongation at break applications due to above tested properties.
slightly decreased with an increase of modified
MCC percentage. These results were possibly References
due to the interaction between silane groups
onto PP and OH groups of MCC that improve 1. Samarasekara, A. M. P. B., and Jayasuriya, E. A.
the interfacial adhesion of MCC with the PP P. C. D., "Synthesis of Biodegradable Polyolefins
matrix. Based Polymer Composites Using Degradable
Due to above reason tensile strength increased Natural Materials." Proceedings of International
and elongation at break slightly decreased with Forestry and Environment Symposium. Vol. 18.
2014.
an increase of modified MCC content. MCC
acts as a short fibers and distribution is 2. Umadaran, S., Somasuntharam, P., and
discontinue. These are other major reasons for Samarasekara, A. M. P. B., "Preparation and
this slight decrease of elongation at break. Characterization of Cellulose and Hemi-Cellulose
Hardness variation of composite was shown in Based Degradable Composite Material using
Figure 9 for Pure PP hardness value was lower Sugarcane Waste." Moratuwa Engineering Research
than composite and hardness value increased Conference (MERCon), 2016. IEEE, 2016.
with an increase of modified MCC content. The
results imply that resistance to indentation 3. Kahawita, K. D. H. N., and Samarasekara, A. M.
P. B., "Extraction and Characterization of
increased with modified MCC content. It can be
Cellulosic Fibers from Sawmill Waste." Moratuwa
described as stiffness of composite increases Engineering Research Conference (MERCon), 2016.
due to MCC gives high crystallinity region for IEEE, 2016.
composite. Therefore when MCC content is
increased, stiffness of composite slightly 4. Samarasekara, A. M. P. B., Somasuntharam, P.,
increases with MCC content. Therefore results and Umadaran, S., "Development of
strongly agreed with above hardness variation. Environmentally Friendly Cellulose Containing
It was important to analyse the water- Packaging Products from Waste Materials."
absorption (Figure 10) qualities of the (2015).
developed composites, because the dimensional
5. Samarasekara, A. M. P. B., Senavirathne, H. V. H.
stability of the final products decline as a result H., and Sandaruwan, A. H. W. O., "Preparation of
of water absorption. Also water absorption rate Biodegradable Polymer Materials using
increased with an increasing of modified MCC Agricultural Waste." Proceedings of International
content due to increase of slightly presence of Forestry and Environment Symposium. Vol. 17.
hydrophilic OH groups in modified MCC. 2012.
Density variation of composite and pure PP
was shown in Figure 11. Composite density 6. Medve, József, et al. "Hydrolysis of
was always lower than pure PP density and it Microcrystalline Cellulose by Cellobiohydrolase I
and Endoglucanase II from Trichoderma Reesei:
decreased with MCC content. MCC gave the
Adsorption, Sugar Production Pattern, and
low density for composite than pure PP. These Synergism of the Enzymes." Biotechnology and
results imply that MCC based composite can be bioengineering 59.5 ,1998, 621-634.
used for light weight product with high specific
properties. 7. Lee, H. V., Hamid, S. B. A., and Zain, S. K.,
"Conversion of Lignocellulosic Biomass to
5. Conclusions Nanocellulose: Structure and Chemical
Process." The Scientific World Journal 2014 (2014).
Experimental results showed that gradual
increase of tensile strength, hardness and
8. Kiziltas, Alper, et al. "Mechanical Properties of
impact strength with increase of MCC Microcrystalline Cellulose (MCC) filled
concentration. Polypropylene with 4 wt.% MCC Engineering Thermoplastic Composites." Journal
containing sample showed the maximum of Polymers and the Environment 22.3, 2014, 365-
impact strength and it was 18.2 KJ/m2. 372.
Maximum water absorption (0.02%) was
observed in 5 wt.% MCC containing sample. 5 9. El-Sakhawy, Mohamed, and Mohammad L.
wt.% MCC containing sample showed the Hassan. "Physical and Mechanical Properties of
Microcrystalline Cellulose Prepared from
maximum hardness (74.4 Shore D). Developed
Agricultural Residues." Carbohydrate
composite showed the gradual reduction of
polymers 67.1, 2007, 1-10.
density from 1 wt.% MCC (0.882 g/cm3) to 5
wt.% MCC (0.826 g/cm3). Therefore, 9. Amarasinghe, D. Shantha, Dehipalawage Sunil,
Polypropylene with MCC polymer composite and Harry D. Gafney. "Effect of CO Pressure
ENGINEER 6
268
Gradients on the Photochemistry and Thermal
Chemistry of Fe (CO) 5 Physisorbed into Porous
Vycor Glass." The Journal of Physical Chemistry
C 115.43 ,2011, 20824-20831.
10. Huang, Jun, and Denis Rodrigue. "Comparison of

the Mechanical Properties between Carbon
Nanotube and Nanocrystalline Cellulose
Polypropylene Based Nano-
Composites." Materials & Design (1980-2015) 65 ,
2015, 974-982.6.
11. Lokshina, I., et al. "Microcrystalline Cellulose:

Extraction and Analysis." Procedings of the
fourteenth Israeli–Russian Bi-National workshop,
Ariel, Israel. 2015.
12. Mathew, Aji P., Kristiina Oksman, and Mohini

Sain. "Mechanical Properties of Biodegradable
Composites from Poly Lactic Acid (PLA) and
Microcrystalline Cellulose (MCC)." Journal of
applied polymer science 97.5, 2005, 2014-2025.
13. Senevirathna, Sandhya Rani, et al. "The Effect of

Change of ionomer/polyol Molar Ratio on
Dispersion Stability and Crystalline Structure of
films Produced from Hydrophilic
Polyurethanes." Journal of Applied Polymer
Science 134.7, 2017.
14. Chauhan, Yuvraj P., et al. "Microcrystalline

Cellulose from Cotton Rags (waste from garment
and Hosiery Industries)." International Journal of
Chemical Sciences 7.2, 2009.
15. Spoljaric, Steven, Antonietta Genovese, and

Robert A. Shanks. "Polypropylene–
Microcrystalline Cellulose Composites with
Enhanced Compatibility and
Properties." Composites Part A: Applied Science and
Manufacturing 40.6, 2009, 791-799.
16. Gao, Tian Ming, et al. "Preparation and

Characterization Nano-Cellulose and its Surface
Modification by Silane Coupling Agent." Applied
Mechanics and Materials. Vol. 217. Trans Tech
Publications, 2012.
7
269 ENGINEER
AnnualSessions
Annual Sessionsof of IESL,
IESL, pp.pp.
[271[1- -276],
10], 2017
TheInstitution
© The Institution
of of Engineers,
Engineers, Sri Lanka
Sri Lanka
Synthesis and Characterization of Nanoparticle-

Incorporated Antimicrobial Coating for Food
Packaging Applications
S.P.A. Madushani, D.P. Egodage, A.M.P.B. Samarasekara, D.A.S. Amarasinghe,
H.T.S. Jayalath and S.M.N.S. Senerath
Abstract: The rising demand to increase fresh food shelf life, as well as the need of protection
against food-borne diseases, urged the development of antimicrobial food packaging. Nanomaterials
have increasingly been used in food packaging applications in recent years due to their extraordinary
properties when compared to bulk materials. Nanoparticles provide significant antimicrobial
properties in different environmental conditions. Nowadays, there is heightened attention in
designing nanoparticles incorporated food packaging with the introduction of nanotechnology. Silver
nanoparticles (AgNPs) based antimicrobial packaging is an innovative form of food packaging that
used to extend shelf-life of food and reduce the risk of pathogens. AgNPs are one of the most
powerful antimicrobial agents which can be used for increasing the shelf life of foods due to its
capacity to eliminate infectious micro-organisms. The present research work describes the preparation
of silver nanoparticles incorporated coating for polymer-based packaging components. Nanosilver
impregnated cross-linked polyvinyl alcohol coating was synthesized and applied to the polymer
surface. In this study, the wettability of the polymer surface was enhanced by a UV treatment. The
contact angle was measured to confirmation of the wettability. The adhesion of the coating to polymer
was obtained as load required to peel off the coating. Optimum UV treatment time was selected by
considering the contact angle measurements and mechanical properties of the polymer. Thermal
degradation of cross-linked nanosilver coating was determined using thermogravimetric analysis.
Presence of silver nanoparticles in the coating was confirmed by Surface Plasmon Resonance (SPR)
and Scanning Electron Microscopy (SEM). The nanosilver incorporated polymer was tested for its
biocidal action against model bacteria Escherichia coli using zone inhibition and applying foods in
nanosilver coated containers. Developed nanosilver incorporated polymer-based food packaging
products showed antimicrobial properties. This developed product can be used to improve the quality
of the food and extend shelf life, especially in food packaging applications.
Keywords: AgNPs, food packaging, antimicrobial coating, wettability
1. Introduction antimicrobial usefulness while complying with

Food packaging plays a vital role in preserving the current regulations [2].
food throughout the distribution. Food There are various nanomaterials that exhibit
containers with antimicrobial properties can antimicrobial activity.
withstand longer life than usual food
containers. Currently, available polymer food Ms. S.P.A. Madushani, B.Sc. Eng. (Moratuwa), Dept. of
containers can be attacked during the food Materials Science & Engineering, University of Moratuwa.
storage. Thus foods may be toxic or expired. So Ms. D.P Egodage, B.Sc. Eng. (Moratuwa), Dept. of
Materials Science & Engineering, University of Moratuwa.
food containers with antimicrobial properties
can be used to improve the quality of the food. Eng. A.M.P.B. Samarasekara, B.Sc. Eng. (Moratuwa),
The most effective way to improve the Dept. of Materials Science & Engineering, University of
antimicrobial properties in food container is Moratuwa.
applying a surface coating[1]. Antimicrobial Dr. D.A.S. Amarasinghe, B.Sc. (Kelaniya), MSc.
agents in food packaging is a full-grown (Moratuwa), M.Phil. (CUNY, USA), Ph.D.(CUNY, USA),
technology, concerns about the risks associated Senior Lecturer, Dept. of Materials Science & Engineering,
with the potential ingestion of the Ag ions
migrated into food still exist. The general Mr. H.T.S. Jayalath, B.Sc. Eng. (Moratuwa), Dept. of
requirement is not to exceed 0.05 mg/kg in
food. This gives that assessment of silver Mr. S.M.N.S. Senerath, B.Sc.Eng.(Moratuwa), Dept. of
migration profiles is necessary to assure
1
271 ENGINEER
Inorganic materials such as silver and gold approximately 102.10, for water. Therefore, PP
have been found as antimicrobial materials is a hydrophobic polymer. [6].
having nano range. Silver nanoparticles have a
high antimicrobial activity because they can The main feature of application of a coating is
insert within the cell membrane due to their that coatings adhere to the substrate, the
small size and release silver ions locally from surface energy of the substrate must be higher,
the silver particle [3]. or surface energy of the coating must be lower.
By increasing the surface energy of the
Nanosilver particles are prepared by chemical substrate, the contact angle will be reduced. To
and physical methods. A simple and easy increase the surface energy of polymer surfaces,
method is the chemical reduction of silver ions surface pre-treatments will be done such as
to nanosilver particles, in the presence of chemical treatments, plasma treatments, laser
reducing agent and capping agent. Polyvinyl irradiation and ultraviolet (UV) irradiation
alcohol (PVA) used to prepare silver treatments without affecting bulk properties.
nanoparticles. Thus it acts as a size reducing The increase in surface energy of plastics occurs
agent and as a stabilizer to stabilize the growth through the surface oxidation of polymer
of the particles and water-soluble synthetic chains by UV irradiation. [7,8]
polymer with excellent film-forming, non-toxic The objective of this paper is intended to
and adhesive properties and biodegradable describe the preparation of silver nanoparticles
polymer [3]. Nanoparticles of silver can be incorporated coating for polymer-based
prepared in an aqueous medium easily using packaging components. Nanosilver solution
PVA, thus making the material synthesis non- was synthesized by reducing silver nitrate by
toxic and environmentally friendly because of PVA and crosslinked the coating to avoid
the zero solid by-product [4]. solubility. The applicability of UV treatment to
increase the surface energy was investigated,
However, due to its solubility in water, it is and then nanosilver particles were coated on
often combined with other polymers in the substrate. The fabricated nanosilver based
multilayer structures, PVA being located in the coating was tested for adhesion. SEM and UV
core layer. An alternative route to avoid the use spectrometers were used to investigate the
of several polymers would be the crosslinking presence of nanosilver on the coating and
of PVA. For PVA, most commonly used analyse the thermal degradation of the
crosslinkers are dialdehydes and crosslinks.
polycarboxylic acids such as citric acid, succinic
acid and tartaric acid [4,5]). The advantage of Antibacterial properties against Escherichia coli
using Polycarboxylic acid is that it can be used were investigated for the developed products.
as food additives. Citric acid is a weak organic As Escherichia coli are a widespread, diverse
tricarboxylic acid. It occurs naturally in citrus group of bacteria, it can be found in
fruits. Citric acid used as a food additive environment and foods. It is an anaerobic
(European code E330). The advantage of using bacteria. Thus it can grow food containers even
citric acid is completely harmless natural though they are sealed.[9]
crosslinking [5] .
Many plastics have non-porous structure and 2. Materials and Methods

low surface free energy hence most of them are
hydrophobic and are not naturally wettable. 2.1 Materials
Wettability studies usually involve the Silver nitrate (AgNO3, 99.9% purity) was
measurement of contact angles as the primary purchased from Research-Lab Fine Chem
data, which indicates the degree of wetting industries, Mumbai and Polyvinyl alcohol
(PVA, 9000–10000 molecular weight, 88%
when a solid and liquid interact. Small contact
hydrolysis, and Industrial grade) were
angles that lesser than 900 correspond to high
purchased from Organic Trading, Sri Lanka.
wettability, while large contact angles that
Citric acid was obtained from Glorechem, Sri
greater than 900 correspond to low wettability Lanka and Polypropylene cups were purchased
[6]. Polypropylene is the second widest used from Starpack plastic product distribution
polymeric material for packaging applications agent, Borupana, Sri Lanka and other chemicals
due to stability, chemical and moisture and laboratory equipment were acquired from
resistance, lightness, flexibility, strength, and Department of Materials Science and
easiness of process, and suitability of recycling Engineering and Department of Chemical and
and reuse. For polypropylene has contact angle Process Engineering University of Moratuwa.
ENGINEER 2
272
2.2 Preparation of Silver Nanoparticles taken by 13 Mega Pixel camera, and 0.05ml of
Silver nanoparticles were prepared using de-ionized water drop using and the images
AgNO3 as precursor and PVA as reducing and were analysed by software.
stabilizing agent. 0.1wt% silver was prepared
by dissolving silver nitrate in 5ml of de-ionized 2.8 Mechanical Property Testing of UV
water. PVA solution was made by adding Treated Polypropylene
0.06gml-1 of PVA to 50ml of de-ionized water
and stirred magnetically maintaining a
Mechanical properties of the UV treated and
temperature between 60-700C. After PVA
non-treated polypropylene samples were tested
completely dissolved in de-ionized water
by Universal tensile testing machine. The
(transparent) the silver nitrate solution was
added drop by drop (drop per second) while polypropylene samples were cut 500×200mm2
maintaining the temperature between 60-700C size, and the width and thickness of the
with continuous stirring Magnetic stirring samples were measured by using Venire
continued until the solution became a faint calliper. The machine parameters were set as
brownish viscous liquid.[4] 500N load, 200mm extension. Tensile strength
was calculated and compared.
2.3 UV–Vis Spectrophotometry
The formation of the nanosilver particles in 2.9 Measuring UV penetration depth
aqueous solution was verified by recording UV Absorbance and transmittance spectrum of the
absorbance spectrum between 350-700nm wave 1.7mm thin polypropylene film was taken by
range by UV-1800 SHIMADZU spectrometer. using UV-1800 SHIMADZU spectrometer for
200 to 650 nm because mercury lamp spectrum
2.4 Preparation of Cross-linked Nano was distributed through this range which used
Silver Coating on Polypropylene Substrate for UV treatment.
Citric acid was used as a crosslinking agent to Absorbance was calculated by using Beer-
crosslink PVA. 0.0066g/ml of Citric acid were Lambert law for each wavelength of the
added to 50ml of de-ionized water and mixed transmission spectrum of polypropylene.
with pre-prepared PVA/Ag 50ml solution Percentage of transmittance was calculated for
while maintaining the temperature between 65- each mercury peak wavelengths for variable
700C while stirring magnetically. Then sample penetration depths.
polypropylene samples were coated with the
solution and were kept in the oven for 100- 2.10 Measuring Adhesion Load of Nano
1300C for 6 hours. Silver Coating
The adhesion load on coating was measured for
2.5 Thermo Gravimetric Analysis (TGA) nanosilver impregnated coated 48 hours UV
Thermal degradation of cross-linked polymer treated samples and non-treated polypropylene
coating was determined by using TGA. substrates by using the universal tensile testing
Thermal decomposition of the coating was machine.
obtained by TGA STD Q600 apparatus for
coated samples of different crosslinked The load required to peel off the coating was
temperatures (90,100, 120, 1300C). This analysis measured by setting the load cell parameters as
was done with 800°C ambient temperature with 500N load, 500mm extension, speed 20
a heating rate of 10°C/min in open ceramic pan mm/min and dimensions of the sample 4.5 x
under nitrogen environment. 4cm2.The maximum load to peel off the coating
was recorded.
2.6 UV Treatment
For a better adhesion of the coating on a 2.11 SEM Analysis
polymer substrate, UV treatment was done. The existence and shape of nanosilver particles
This was done by UV rays having intensity 35 in the prepared coating were determined by
W/m2. Samples were UV treated for 24 hours, ZEISS EVO 18 RESEARCH apparatus. The SEM
48 hours, and 72 hours time periods. analysis and the energy dispersive X-ray
spectroscopy were done by using prepared
2.7 Contact Angle Analysis coating under 10.00kV voltage with the
The wettability of the polymer after the UV magnification of 3.58K .
treatment was analysed by measuring the
contact angle of the polymer substrate. The 2.12 Antimicrobial Activities
contact angles for de-ionized water were Nanosilver coated, and non-coated
measured by taking photographs which are polypropylene samples were investigated for
3
273 ENGINEER
antibacterial activity by using Agar diffusion
method against Escherichia coli. This method
was carried out initially by pouring the Muller
Hinton Medium solid agar onto sterilized Petri
dishes and 106 CFU/mL concentration
Escherichia coli bacteria solution was spread
uniformly on the plate. The nanosilver coated
and non-coated polypropylene samples were
cut into a disc shape with 6mm diameter and
was placed on plates. The agar plate was
incubated for 24 hours at 37 °C, and inhibition Figure 3 - Graph of contact angle vs UV
zone diameter was recorded. treated time
The antibacterial activity of the nanosilver
coating observed visually by including foods in 3.2 Surface Modification of
the food packaging container. Polypropylene samples
Figure 2 is shown the images of the contact
3. Results and Discussion angle of the UV treated polypropylene samples
with de-ionized water and the Figure, 3 is
3.1 Characterization of Nanosilver
shown the contact angle values reduces with
Particles
the increase of the UV treated time. When solid
UV visible adsorption analysis was performed
and liquid interacts, the contact angle indicates
to verify the presence of silver nanoparticles in
the degree of wetting. The angles less than 900
the prepared solution. Silver nanoparticles have
results to high wettability and vice versa.
a surface plasmon adsorption between 400 nm
Therefore, this result shows that increasing of
and 450nm. Figure 1 shows the adsorption peak
the wettability of the polypropylene samples
at 415.43nm that clarified the formation of
because of reduction of the contact angle with
nanosilver particles and the presence of
increasing UV treated time of the
spherical or roughly spherical silver
polypropylene samples. That was a positive
nanoparticles.
solution for the adhesion difficulty of the
coating to the polypropylene surface. To
measure the contact angle, de-ionized water
was used since the nano silver solution is water
based.
3.4 Mechanical Properties of the UV

Treated Polypropylene Samples
The average tensile strength and elongation of
the polypropylene samples are very important
when using UV treatment as surface treatment.
According to the results shown in Figure 4 and
Figure 5 after 48 hours UV treated time, the
Figure 1 - Adsorption spectrum of the silver
tensile strength reduces 3%, but the elongation
nanoparticles colloidal solution.
reduces 47%. It is important to retain the
properties of the polypropylene sample up to
90% after the UV treatment. Therefore, 48 hours
was selected as best UV treatment time for the
polypropylene samples, since after 48 UV
treated time, the tensile strength doesn't show
significant change but, elongation value
reduces drastically.
3.5 Measuring UV penetration depth
Figure 6 shows the resulted transmittance
values for each penetration depth at peak
values of mercury. That shows the mercury
Figure 2 - Contact angle images peak values were chosen because 254, 300, 312,
334, 365, 405, 436, 546, 579nm values have the
highest tendency to penetrate due to the high-
ENGINEER 4
274
intensity UV rays will fully penetrate to the
sample as the thickness of the polypropylene
sample is 1.7mm. When 5% of UV transmitted,
95% UV was absorbed by the sample.
Therefore, 48 hours of UV treatment was
chosen since this effect will moderate at 48
hours.
3.6 Measuring Adhesion of Nanosilver

Coating
The adhesion load of the polypropylene Figure 7 - Temperature curves of TGA

samples (10.6 N) which 48 hours UV treated
showed a higher value than the non-UV treated
samples (4.7N). That interprets the 48 hours UV 3.7 Thermo Gravimetric Analysis (TGA)
treated time polypropylene samples gives the
best results of adhesion of the nanosilver Figure 7 shows the retain weight percentage of
coating.
the crosslinked polyvinyl alcohol using citric
acid with different crosslinking temperatures.
The thermal stability was analysed
qualitatively. High thermal stability can be
observed with increasing crosslinks. The non-
cross-linked films show weight loss at about
70°C, whereas the weight loss of the crosslinked
films start at about 100 °C. The first major
weight loss of both films initiated at about
120°C. But the non-cross-linked films shows
slightly more weight loss than the cross-linked
Figure 4 - Graph of Average Tensile Strength films. It can be seen that low weight loss of the
Vs. UV Treated Time cross-linked films between 120 and 220 °C. This
can be due to the higher thermal stability of
the films. According to the Figure 7, at 1300C
temperature crosslinked nano silver coating is
most stable due to a higher degree of the
crosslinking. Therefore, that is the most
favourable temperature for crosslinking.
3.8 SEM Analysis
The SEM microgram confirms that the surface

coating contains nanoparticles with around
400nm diameter with a spherical shape.
Figure 5 - Graph of Average Elongation Break
The energy dispersive X-ray spectroscopy
% Vs. UV Treated Time
confirms that the particles observed in the
scanning electron microgram are silver
particles. The energy dispersive X-ray
spectroscopy confirms that the particles
observed in the scanning electron microgram
are silver particles.
3.9 Antimicrobial Activity

Nanosilver coated polypropylene sample
shows 1mm clear inhibition zone around the
substrate and the non-coated polypropylene
sample doesn't show any inhibition zone
(Figure 10). That interprets that the E-coli
bacteria was inhibited because of the nanosilver
Figure 6 - Graph of Transmittance vs
particles, the clear zone appeared.
Penetration Depth of Polypropylene
5
275 ENGINEER
nanosilver incorporated polymer-based food
packaging product showed antimicrobial
properties.
4. Conclusions
The developed nanosilver particle incorporated
antimicrobial coated polymer showed
significant surface adhesion properties and
antimicrobial properties. This product can be
used in food packaging applications to extend
shelf life.
References
Figure 8 - Scanning Electron Microscopy
image of the coating 1. Metak and T. Ajaal, "Investigation on Polymer
Based Nano-Silver as Food Packaging
Materials", International Journal of Biological,
Biomolecular, Agricultural, Food and
Biotechnological Engineering, vol. 7, no. 12, 2013,
pp. 1103-1109.
2. Marilena Carbone, Domenica Tommas Donia,

Gianfranco Sabbatella, Riccarda Antiochia," Silver
nanoparticles in polymeric matrices for fresh food
packaging", Journal of King Saud University -
Science, Vol. 28, Issue 4, , 2016 pp. 273-279.
3. Yolanda Echegoyen, Cristina Nerín."

Figure 9 - Energy Dispersive X-ray Spectrum Nanoparticle Release from Nano-Silver
Antimicrobial Food Containers". Food and
Chemical Toxicology, vol.62, 2013, pp.16–18.
4. Abdul Kareem, T., Anu Kaliani, A., “Synthesis

and Thermal Study of Octahedral Silver Nano-
Plates in Polyvinyl Alcohol (PVA)”, Arabian
Journal of Chemistry, vol.4, 2011, pp. 325-331.
Figure 10 - Zone Inhibition of Non-coated 5. Birck, S. Degoutin, Tabary, N., Miri, V., Bacquet,
M., "New Crosslinked Cast Films Based on
Polypropylene and Nanosilver Coated
Poly(vinyl alcohol): Preparation and physico-
Polypropylene
chemical properties", eXPRESS Polymer Letters,
vol.8, no.12, 2014, pp.941–952.
6. Dehnavi, A. Aroujalian, A. Raisi and Fazel, S.,

"Preparation and Characterization of
Polyethylene/Silver Nanocomposite Films with
Antibacterial Activity", Journal of Applied
Polymer Science, vol. 127, no. 2, 2012, pp. 1180-
1190.
7. Ebnesajjad, S., Ebnesajjad C., Surface Treatment

of Materials for Adhesion Bonding, 1st ed.,
Norwich, NY, U.S.A,2006, pp. 3-18 .
Figure 11 - Microorganism Growth of Non- 8. Bracco, G., Holst, B., Surface Science Techniques,
coated Polypropylene and Nanosilver Coated 1st ed., Berlin, 2013, pp. 3-16.
Polypropylene
9. Michael Donnenberg, Escherichia coli Virulence
Figure 11 shows the storage of food in Mechanisms of a Versatile Pathogen, 1st ed.,
nanosilver coated and control containers.The California, U.S.A., 2002, pp. 214 .
control sample showed the clear
microorganism growth, but developed
ENGINEER 6
276
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[277 284],2017
2017
The Institution
Lanka
Assessing the Impact on Environment, Health and

Safety due to Releases from Chemical Process Routes
H.B.B. Anuradha, Manisha Gunasekera and Olga Gunapala
Abstract: The chemical process route selection is one of the important decisions that needs to be
taken during initial stages of plant design and development. Although conventionally the economic
factor has been considered in this selection process, presently the environmental, health and safety
(EHS) issues have also become main concerns, since the operational costs related to EHS can be largely
reduced by avoiding EHS hazards during initial stages of plant development. Most available
methodologies for chemical process routes assessment mainly consider environmental or health or
safety hazards individually or a combination of two of them. In this work inherent environmental,
health and safety hazard index (IEHSHI) is developed to assess chemical process routes based on
integrated EHS hazards due to daily operational activities of the plant as well as accidental releases.
The IEHSHI includes information of nine parameters available at the conceptual design stage of
process plant development. The environmental hazards and health hazards are assessed considering
both accidental and continuous releases from the manufacturing process. The safety assessment is
done considering chemical safety and process safety associated with the route. The environmental,
health and safety parameter values are combined individually using weights. The routes can be
ranked and compared according to IEHSHI values. The IEHSHI was applied in three routes to
manufacture methyl methacrylate.
Keywords: Accidental releases, Chemical process, Environmental hazards, Inherent Safety
1. Introduction Conceptual Design: Product specification, synthesis

route selection, process materials and overall process
Environmental impact, employee health and
safety are more concerned than before since
Process Selection and Design: PFD, recycling, product
they have become important in environmental
yield, quality and operating conditions
conservation and human life protection.
Chemical process route is considered as the
way to achieve a particular product using raw Detailed Engineering Design and Commissioning: P
materials and energy [3]. Literature reveals that & ID, utilities, process control, electrical and insulation
the operating cost of chemical manufacturing
plants can be considerably reduced by selecting
environmentally friendlier, healthier and safer Operation and Modification: Installing, maintaining
chemical routes during the design stage of the and operations
chemical process plant [4].
Decommissioning: Material, equipment abandon and
In inherently safer chemical process plants, waste treatment
hazards are avoided or removed rather than
controlling them with add-on protective Figure 1 - Process design stages.
equipment [3]. The inherent safety should be
considered or practised as early as possible in Chemical process route selection is one of the
the design stages because the impact on design key design decisions taken during process
decision is large at this stage.
The chemical process plant development Eng. H.B.B. Anuradha, B.Sc. Eng. (Moratuwa), Research
Student, Department of Chemical and Process Engineering,
process can be categorized into varies stages University of Moratuwa.
according to its function involved as they are
Eng. (Dr.) M. Y. Gunasekera, B.Sc. Eng.(Hons)
considered as key design decision-making (Moratuwa), M.Eng. (Moratuwa), Ph.D.(UK), MIE(Sri
points. The process design stages are defined Lanka), Senior Lecturer, Department of Chemical & Process
considering design stages [12] and process Engineering, University of Moratuwa.
design life cycle [5]. These stages are shown in Dr. Olga Gunapala, M.Sc. Eng.(Moscow), Ph.D.
figure 1. (Moscow), Senior Lecturer, Department of Chemical &
Process Engineering, University of Moratuwa.
1
277 ENGINEER
development, and it is taken at the conceptual and process hazards index (ICPHI), Material
design stage since it is the earliest stage. The harm index (MHI), Health hazards index (HHI)
conceptual stage is defined as that contains and the ratio between worker exposure
basic process information such as synthesis concentration (WEC) and occupational
process routes, process materials and product exposure limit (OEL) are used as parameters for
specifications. determining the PRHI. Small leaks and fugitive
emission in the workplace have been
An index called Inherent environmental (E) considered in the assessment. The chemical
health (H), and safety (S) hazards index process route with least PRHI value is
(IEHSHI) is developed considering information considered as the best process.
available at the conceptual stage. The IEHSH
index is developed combining environmental, The HQI [8] is another method which uses
health and safety parameters. Environmental airborne chemical concentrations in the
hazard assessment is carried out by considering working area to assess the impact on health.
catastrophic and continuous releases of the The fugitive emission rate and air volumetric
plant. Health hazard assessment is carried out flow rate are used to estimate airborne chemical
assessing acute and chronic hazards associated concentration. The health risk posed by a
with the chemical plant. In safety assessment, mixture of chemicals is given by HQI. The least
chemical and process safety are considered. value of HQI is considered in selecting the
chemical route with least health effect.
2. Methodologies available for
Hazards Assessment 2.3 The Safety Assessment
Methodologies
Inherent hazards assessment of chemical The I2SI proposed by Khan et al. [1] is
process routes has been done in various ways composed of two main sub-indices: a hazard
such as environmental hazards assessment or index (HI) and an inherent safety potential
health hazards assessment or safety assessment index (ISPI). The HI is intended to be a measure
and combining two of the above or all of tree of the damage potential of the process after
(EHS assessment). Noteworthy among them are taking into account the process and hazards
the Environmental hazards index (EHI) [3], control measures. The ISPI accounts for
Safety weighted hazard index (SWeHI) [6], applicability of the inherent safety principles to
Fuzzy logic based inherent safety index [10], the process.
Integrated inherent safety index (I2SI) [1],
Process route healthiness index (PRHI) [7], The RISI [5] is an extension of I2SI earlier
Inherent benign- ness indicator (IBI) [4], Health discussed. The risk involved in a base design
quotient index (HQI) [8], Sustainable (RiskBD) and the inherent safety risk for
assessment of novel chemical process and Risk- alternative designs (ISRisk) are integrated to
based inherent safety index (RISI) [5]. develop the RISI.
2.1 The Environmental Hazards 2.4 The Environmental, Health and Safety
Assessment Methodologies Hazards Combined Assessment
The environmental hazards index (EHI) is Methodologies
developed by Cave and Edwards [3] to estimate The IBI [4] combines 15 parameters related to
the environmental impact of a total release of safety, health and environmental impacts of a
chemical inventory. The EHI of a route is route. The chemical safety aspects of the routes
calculated by evaluating the effects and have been assessed by three parameters and
exposure of each chemical in the route. four parameters have been considered for
Chemical distribution and toxicity are main assessing the processing aspects. The health
parameters of EHI. Predicted environmental and environmental impact of each route is
concentration is determined by Mackay’s assessed by seven parameters. The toxicity is
fugacity model, and LC50 and LD50 data are also a property of a substance that is used as a
used for toxicity evaluation. parameter for the IBI. The parameter is scaled
to the range of 0 and 1, where 0 means not
2.2 The Health Hazards Assessment hazardous, and 1 represents the most
Methodologies hazardous level. The IBI is taken by integrating
The PRHI is developed by Hassim and the 15 normalized values of parameters for each
Edwards [7] for assessing inherent occupational route.
health hazards. In this work, Inherent chemical
ENGINEER 2
278
The sustainable assessment [2] evaluate the contribute to the final index value. The
chemical process based on economic feasibility, environmental and health hazards are
environmental impact, human health and risks. determined based on catastrophic and
Greenhouse gases and cumulative energy continuous released from manufacturing plant
demand are considered to assess the but in safety assessment, chemical and process
environmental impact. EHS index is proposed safety are considered.
by Sugiyama based on environmental, health
and safety [2]. The individual categories are 3.4.1 Environmental Hazards Assessment
aggregated using weight factors: 3.4.1.1 Accidental Release Assessment
environmental- 0.4, health- 0.2 and safety- 0.4. The Environmental Hazards Potential (EHP) is
proposed in this work which represents the
3. Inherent Environmental Health environmental impacts of raw materials, end
and Safety Hazards Index products and chemicals in the reactors and
separation stages. The parameter expects to
In this work, the proposed inherent assess the catastrophic releases from the
environmental health and safety hazards index chemical processing plant, which is the worst
(IEHSHI) aims to assess the chemical process case that would happen in the life time of the
routes in the early stages of process design. The plant.
method evaluates chemical process based on
continuous releases and catastrophic releases The EHP is based on following assumptions:
from the process plant. The process area  The total inventory stocks would be released
includes raw materials storage, plant operation to the environment during the accident and
area and end product storage. The framework that chemical released would be most
for estimating IEHSHI consists of mainly three concentrated, and consequently, effects
categories: environmental hazards, health would be most severe in the 1km2 area
hazards, and process and chemical safety. surrounding the plant.
 The released chemicals are accumulated in
3.1 Conceptual Design Stage Information the air, water and soil compartment within
The EHSHI assessment is carried out at the of 0.6 km radius circle around the plant.
conceptual design stage which is the earliest  The damage to the environment due to a
stage of process plant design (Figure 1). The chemical is directly proportional to the
chemical route synthesis and their conversion amount of chemical released.
factors, process operating conditions, enthalpy  The damage due to the release of a mixture
values and process block diagram are known of chemicals is additive.
information at this stage.
The EHP is a function of predicted
3.2 Development of Block Diagrams for environmental concentration (PEC) and
All Possible Chemical Routes permissible exposure limit (PEL). The ratio
The block diagram for each chemical route is between PEC and PEL values of a mixture of
developed considering main operating units chemicals are added to obtain the EHP value
such as reactors and separators. The mass and is shown in equation (1).
balance is compiled for the route using block
diagram, with considering main reactions, side
reactions and conversion factors.
3.3 Select a Chemical Route

The PEC is calculated using Mackay fugacity
The EHSHI assessment is used to compare the
approach [13] and this model has also been-
chemical routes which give an expected
used in inherent environmental hazard
desirable product. The calculation is begun
assessment by Cave and Edwards [3]. The mass
with considering one chemical route and
flow of chemical substances in the process are
determining all nine parameter values related
calculated for the production of a unit mass of
to that route. Once calculation for the first
product using stoichiometric equations. The
chemical route is completed, the other routes
resident time of chemical in storage is often
are analysed.
over 300 times of reaction and separation stage
[3]. Furthermore, the material stock is
3.4 Hazards Assessment
approximately 300 times of feeding rate of raw
In this analysis, the parameters are categorized
materials.
into environmental, health, and safety which
3
279 ENGINEER
In the calculation of this work, the product parameter, short-term exposure limit (STEL)
manufacturing flow rate is considered as one values are used as a reference for the
tone per hour and the raw materials in the stock concentration of airborne materials.
is 300 times of feeding rates, correspondingly,
the end products stock is also 300 times of The Dow chemical equation [11] for gaseous
manufacturing flow rate. releases and liquid releases are used for
calculation of AQ values. The hole diameter of
The chemical concentration in the air gases or liquid release is considered as at its
compartment is calculate considering sudden maximum value of 6.35 mm. this is used in
releases of available chemicals in the plant other research works as well such as the
premises according to the above calculation. inherent occupational health hazards index
developed by Hassim and Edwards [7].
The PEL data are collected for all chemical in
the plant and the ratios between PEC and PEL The volumetric flow rate (VF) is calculated as in
calculated. Moreover, the ratio values are Equation 5 considering a typical wind speed of
simply summed as shown in equation (1) for 4m/s inside the plant. The same has been used
the mixture of chemicals associated with the in Inherent occupational health assessment
considered chemical route. during preliminary design stage by Hassim and
The above procedure is repeated for all the Hurme [8]. Their work refers to a survey of
chemical process routes and the highest EHP wind speeds in indoor workplaces by Baldwin
value, is used to normalize the EHP values. The in arriving at this value [8]. Furthermore, the
normalized value is defined as and is average height of main unit operations’ leak
shown in equation (2). source is taken as less than (h) 7m. The same is
also applied in Hassim and Hurmes’
development where the reference is made to
the Process plant layout by Mecklenburgh [8].
The floor area of the process plant is calculated
3.4.1.2 Continuous Release Assessment by considering average floor area required for
The environmental hazards involved with the main unit operations of the plant and are given
continuous operation are assessed by in Hassim and Hurmes’ work [8].
Greenhouse Gas (GHG) emission. The
calculation of GHG emission rate depends on
the carbon dioxide production rate (mi) of the
manufacturing process route (i). The The concentrations due to airborne materials
stoichiometric equation for all chemical routes ( ) are given by largest AQ value dividing
is used to calculate the GHG emission rate. This volumetric flow rate as shown in Equation 6.
analysis is also carried out for one tone of the The largest AQ value is selected from possible
main product manufacturing process. chemical releases in the chemical process plant.
The values for all the chemicals in the
mixture are calculated.
The same procedure is followed for other

chemical process routes to calculate the ….(6)
respective GHG emission values. The
maximum value is used in the normalization The HAQ values are determined by getting the
step. Each GHG emission value is divided by ratio between and STEL value for relevant
the maximum GHG value, and the result is chemical as in following equation 7.
defined as .
The procedure for assessment of is

3.4.2 Health Hazards Assessment continued for all chemical routes and the
3.4.2.1 Acute Toxicity Assessment maximum value, which is defined as a Max
The Hazards in Airborne Quantity (HAQ) is ( ) is determined. The normalized value of
assessed by determining airborne quantity is determined for each value by
(AQ) resulting from small leaks, due to dividing the Max ( ) and the result is
accidental releases. The acute toxicity defined as (equation 8).
ENGINEER 4
280
3.4.3.1 Chemical Safety Assessment
The assessment of chemical safety aspects of
routes is carried out using two parameters,
3.4.2.2 Chronic Toxicity Assessment explosiveness and flammability.
Chronic toxicity parameter is used for long
term exposure to the chemical substance. Explosiveness (E)
Health hazards in continuous releases from The explosiveness measures the tendency of
chemical processing plant are assessed under chemicals to form an explosive mixture in the
chronic toxicity assessment. air. The explosive nature of chemicals is
represented by the difference between the low
The fugitive emission is used as a parameter to explosiveness limit (LEL) and the upper
assess the chronic toxicity hazard. The Hazards explosiveness limit (UEL) of each chemical.
in Fugitive Emissions (HFE) is taken as a
function of fugitive emission rate (FE), The summed value for explosiveness, Er of all
volumetric flow rate and time weight average chemicals (j) involved in the chemical route is
(TWA). The FE rate is determined by following considered for assessment of explosiveness of
the procedure applied in inherent occupational the chemical route. The Equation 11 shows the
health assessment by Hassim and Hurme [8]. determination of .
In this assessment, the main unit operation such

as reactors, separators and compressor are
decided considering reactions associated with
the chemical process route. The number of
piping components is calculated according to The explosiveness value, is determined for
the Hassim’s work [8]. all chemical routes and the maximum value is
selected for normalization. The normalization is
The pre-calculated emissions data are given in carried out as in Equation 12, and normalized
the U.S. Environmental Protection Agency values are defined by the notation, .
(EPA) for different piping components [14]. The
total emission rate is determined by a number
of piping component and its emission factor.
The individual emission rate of a chemical is
determined considering the molar ratio of the Flammability (F)
reactors and separators. The flammability measures the tendency of a
chemical to burn. The chemical safety from
The VF is assessed following the same accidental fires and explosions of flammable
procedure as in HAQ assessment, and TWA mixture could be measured by flammability
data for all chemical substance are collected values.
from the literature.
The flammability ratings of each chemical
involved in the route are taken from NFPA fire
rating [15]. The NFPA ratings have values
ranging from 0 (minimum impact) to 4
The procedure for assessment of is (extreme impact) as shown in table 1, which is
continued for all chemical routes and the developed by flash point (FP) and boiling point
maximum value, which is defined as a Max (BP) values.
( ) is determined. The normalized value of
is determined for each value by The added value for flammability of all
dividing the Max ( ) and the result is chemicals involved in the chemical route are
defined as is shown in Equation (10). considered for assessment of flammability of
the chemical process route and is defined in
Equation 13 as .
3.4.3 Safety Assessment

For safety assessment, chemical safety and
process safety are considered separately.
5
281 ENGINEER
The flammability value, determine for all The summed value of operating temperature
chemical routes and select the maximum value difference from ambient temperature (25OC) of
for normalization of flammability. The all chemicals reactions (n) in the chemical route
normalization is carried out as in Equation 14 are considered for hazards assessment and is
and normalized values are defined as a . shown in Equation 17 as .
Table 1 - NFPA fire ratings

The maximum value of is selected for
Flammability
Non- combustible 0 normalization of . The normalization is
FP > 60 OC 1 carried out as in Equation 18, and the
37.78 OC < FP < 60 OC 2 normalized values are denoted by .
37.78 OC > FP & 37.78 OC < BP 3
37.78 OC > FP & 37.78 OC > BP 4
3.4.3.2 Process safety assessment

Heat of Reaction (ΔH)
For process safety assessment, operating
Reaction enthalpy represents the heat changes
pressure, operating temperature and heat of
whether it is heat released or absorbed, and
reaction are considered.
both are potentially hazardous.
Operating pressure (OP)
The summed value of reaction enthalpy for all
The processes that operate under high pressure
chemical reactions (n) involved with the
or sub-atmospheric pressure are generally
chemical route is considered for the hazards
hazardous. High pressure can cause the leakage
assessment due to the heat of reaction for a
of chemicals. The difference of the reaction’s
chemical route and is defined in Equation 19 as
pressure range from atmospheric pressure is
.
chosen to represent the hazards due to
pressure.
The summed value for gauge pressure of all

chemical reactions (i) in the chemical route is The normalization is carried out as in Equation
considered for hazards assessment due to 20, and the normalized values are defined as
operating pressure, and it is defined in .
Equation 15 as .
3.5 Development of Hazard Scores

Hazard scores for the environment, health and
safety are developed by combining normalised
The pressure hazard values (Pi) are summed,
values by using equal weights. The equal
and Pr values are determined for all chemical
weight factors have been used by other
routes. The maximum value is selected for
research works as well such as in the
normalization of . The normalization is
sustainable assessment of novel chemical
carried out as shown in Equation 16 and
processes at an early stage by Patel et al. [2].
normalized values are defined as .
3.5.1 Environmental Hazards Score (EHS)
The environmental hazards level that is
resulting from combining both hazards of
catastrophic and continuous releases is
Operating Temperature (OT)
represented in EHS. The equal weights of 0.5
The process operating temperature is also a
are used for both normalized parameters,
direct measurement of hazards. This represents
heat energy available for release. A reaction and for chemical route r. The EHS for
condition higher or lower than ambient chemical route r is determined as in Equation
temperature is considered as unsafe. 21.
ENGINEER 6
282
3.5.2 Health Hazards Score (HHS) developed considering the Reactors and
The HHS is the health hazards level of chemical associated separator units. The environmental,
process route r, which is a result of combining health and safety hazard values are tabulated
both hazards chronic toxicity and acute toxicity. as shown in the following table 2. The model
The equal weights of 0.5 are used for both environment for PEC calculation included the
normalized parameters, and for air and water compartments. The normalised
chemical route r. The HHS for chemical route r values of each parameter are shown in Table 3.
is determined as shown in Equation 22. The environmental, health and safety score
values are given in Table 4.
Table 2 - Hazards parameter values

3.5.3 Chemical and Process Safety Score Parameter Route A Route B Route C
(CPSS) EHPr 9.83 63.93 69.69
The CPSS is determined by combining five GHGr 0.50 0.79 0.79
parameters using equal weights each of 0.2 HAQr 3.74 10.55 10.55
considering that all parameters are equally
HFEr 3962.39 1.25 1.25
important. The normalized values of a chemical
Er 109.30 108.3 102.7
route r, , , , and are combined as
Fr 23 20 17
shown in Equation 23.
Pr 6 0 0
Tr 1875 645 645
ΔHr 729.65 595.59 649.09
3.6 Determine the IEHSH Index Table 3 - Normalised values

Inherently safe processes allow for the Parameter Route A Route B Route C
reduction of hazards control costs. The 0.14 0.92 1
environmental, health and safety aspects of 0.63 1 1
chemical route r, are aggregated to arrive at one 0.36 1 1
score called the IEHSH index. The weight for 1 2.52×10-4 2.52×10-4
environmental, health and safety categories are
1 0.99 0.94
taken as 0.4, 0.2 and 0.4, respectively, which are
the weight factors used in work done by 1 0.87 0.74
Sugiyama [9]. 1 0 0
1 0.34 0.34
The IEHSH Index for route r is calculated as 1 0.82 0.89
shown in Equation 24.
Table 4 - Environmental, health and safety
score
Score Route A Route B Route C
EHSr 0.39 0.96 1
4. Case Study SHSr 0.68 0.5 0.5
CPSSr 1 0.6 0.58
The IEHSH index is demonstrated by applying
in three potential chemical process routes to The IEHSH index is presented for chemical
manufacture methyl methacrylate (MMA). The routes A, B and C in Table 5:
definition of each chemical route is considered
as the same as that is given by Adu et al. in Table 5 - The IEHSH Index values
their work on the comparison of methods for Route A Route B Route C
assessing environmental, health and safety
IEHSHr 0.69 0.72 0.73
hazards [9]. The MMA routes discussed are as
follows:
According to the individual EHS scores (Table
4) route A is the most environmentally friendly,
(A) Acetone cyanohydrin based route (ACH)
route B and C are the most healthy and route C
(B) Tertiary butyl alcohol based route (TBA)
is the one that is most safe. Considering the
(C) Isobutylene based route (i-C4)
final IEHSH index values (Table 5), the route A
is the most environmentally friendly, healthy
The IEHSH is calculated by using the simple
and safe. However, the rank could be different
Process Flow Diagram (PFD) that was
7
283 ENGINEER
if the weights are changed based on its 6. Khan, F. I., Husain, T., & Abbasi, S. A., “Safety
importance of each EHS. Weighted Hazard Index (SWeHI)”, J. Institute of
Chemical Engineering, Vol. 79, March, 2001, pp. 65-
80.
5. Conclusions
7. Hassim, M. H., & Edwards, D. W., “Development
The proposed method in this work is capable of of Methodology for Assessing Inherent
assessing environmental hazards, health Occupational Health Hazards”, J. Institute of
hazards due to both continuous and Chemical Engineering, Vol. 84, September, 2006,
catastrophic releases and safety risk assessment pp. 378-390.
for both chemical and process safety.
8. Hassim, M. H., & Hurme, M., “Inherent
Occupational Health Assessment during
Furthermore, the chemical routes can be ranked
Preliminary Design Stage”, J. Loss Prevention in
considering their environmental hazard level or the Process Industry, Vol. 23, December, 2009, pp.
health hazards level or safety separately. If the 476-482.
process route shows a higher hazard either on
the environment, health or safety, the 9. Adu, I. K., Sugiyama, H., Fischer, U., &
methodology is able to show the total hazard Hungerbuhler, K., “Comparison of Methods for
by all three aspects in a single score by using Assessing Environmental Health Safety Hazards
weights on EHSr, HHSr and CPSSr. in Early Phases of Chemical Process Design”, J.
Process Safety and Environmental Protection, Vol.
86, August, 2007, pp. 77-93.
Acknowledgement
10. Gentile, M., Rogers, W. J., & Mannan, M. S.,
This research project was supported by the “Development of a Fuzzy Logic-Based Inherent
University of Moratuwa Senate Research Grant Safety Index”, J. Institute of Chemical Engineering,
Number SRC/LT/2015/01. Vol. 81, November, 2003, pp. 444-456.
11. American Institute of Chemical Engineering,

References “Dow’s Chemical Exposure Index”, 1st ed., New
York, 1988, pp. 8-17.
1. Khan, F. I., & Amyotte, P. R., “Integrated inherent
safety index (I2SI): A Tool for Inherent Safety 12. Mannan, S., “Lee’s Loss Prevention in the Process
Evaluation”, Faculty of Eng. & Applied Science, Industry”, 4th ed., Texas, USA, 2012.
Memorial University of Newfoundland, Vol. 23, No.
02, June, 2004, pp. 136-148. 13. Mackay, D., “Multimedia Environmental Models”,
2nd ed., USA, 2001.
2. Patel, A. D., Meesters, K., Uil, H. D., Jong, E. D.,
Blok, K., & Patel, M. K., “Sustainable Assessment 14. EPA, “Protocol for equipment leak emission
of Novel Chemical Processes at Early Stage: estimates”, publish no. EPA- 453/R-95-017, 1995
Application to Biobased Processes”, J. Energy &
Environmental Science, June, 2012, pp. 8430-8444. 15. Edwards, D. W., & Lawrence, D., “Assessing the
Inherent Safety of Chemical Process Routes: Is
3. Cave, S., R., & Edwards, D., W., “Chemical there a Relation between Plant Costs and Inherent
Process Route Selection Based on Assessment of Safety?”, J. Institute of Chemical Engineering, Vol.
Inherent Environmental Hazard”, Department of 71, November, 1993, pp. 252-258.
chemical engineering, Loughborough, UK, Vol. 21,
1997, pp. S965-S970.
4. Srinivasan, R., & Nhan, N. T., “A Statistical

Approach for Evaluating Inherent Benign-ness of
Chemical Process Routes in Early design stages”,
J. Process Safety and Environmental Protection,
August, 2007, pp. 163-174.
5. Rathnayaka, S., Khan, F., & Amyotte, P., “Risk-

Based Process Plant Design Considering Inherent
Safety”, J. Elsevier, Safety Science, June, 2014, pp.
438-464.
ENGINEER 8
284
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[285 304],2017
2017
The Institution
Lanka
Antecedents of Work Engagement : Case Study of

Outsourced Employees in Telecommunication Service
Sector of Sri Lanka
K.A. Chathura S. Kumarasinghe
Abstract: This study evaluates the work engagement level of outsourced employees involved in
engineering service delivery in the telecommunication sector of Sri Lanka. It attempts to identify the
significant antecedents that have high influence on the work engagement. The conceptual framework
of this study was built on the job-demands resources model of work engagement proposed by Bakker
and Demerouti (2008a). This study adopts a mixed research approach. The quantitative study
involved analysing responses to a questionnaire by 115 outsourced employees to estimate their work
engagement level and to identify the significant antecedents. The relationship between work
engagement and nine antecedents were investigated through multiple linear regression analysis.
Qualitative analysis explored data collected from four interviews to identify the perceived influential
antecedents.
This study discovered that the work engagement level of outsourced employees in the
telecommunication sector of Sri Lanka was lower than desirable. Both the quantitative and qualitative
studies revealed that autonomy, managerial, supervisory coaching and opportunities for professional
development were significant influences on work engagement. Though there is no consensus between
qualitative and quantitative results, social support, self-efficacy, self-esteem and optimism have also
been found to be influential.
There is an increasing trend towards outsourcing engineering services in the telecommunication
sector of Sri Lanka. Telecommunication operators depend heavily on these outsourced employees to
deliver quality service to their subscribers. Therefore, there is much interest in the performance of
these outsourced employees. Though there have been ample academic studies conducted on the
financial and contractual aspects of service outsourcing, there are very few academic undertakings
that have focused on the human aspects of outsourcing. The study fills this void by investigating the
work engagement of outsourced employees and identifying influential antecedents. By knowing the
present level of work engagement and the factors that should be focused on, telecommunication
operators and outsource service providers should be able to devise an effective strategy to improve
the work engagement.
Keywords: Work Engagement, Outsourcing, Telecommunication, Service
1. Introduction having the lowest telecommunication tariffs in

the world (Fitch Rating, 2015).
The telecommunication industry is a capital-
intensive undertaking, where the operators In this context, operators have adopted
have to constantly keep investing in coverage measures to reduce the cost of production. One
expansion, infrastructure development and the of the key initiatives was to shift from the in-
introduction of rapidly changing new house operation model to outsource model.
technologies. Further, a large proportion of Different operators have adopted out-sourcing
operational costs to an operator, such as energy to different degrees. All operators have adopted
costs, site rentals, spectrum fees paid to outsourcing services at some level, and so there
government and network maintenance costs, is considerable involvement of outsourcing
are fixed costs that bear little relationship to the service providers in operating
number of customers served by the operator. telecommunication networks at present.
Hence, operators are constantly looking for Outsource service provider may be generally
avenues for increasing usage of the deployed responsible for the day-to-day maintenance of
network, as unused capacity at any given active network elements, maintenance of the
moment is wasteful. Therefore, passive infrastructure such as
telecommunication operators have opted to
slash their prices in order to win new Eng. K.A.C.S. Kumarasinghe, B.Sc.(Electronic &
Telecommunication Engineering – Moratuwa), MBA
customers. In fact, a severe price war among the (Leicester), C.Eng(SL & UK), MIE(SL), MIET (UK).
telecommunication operators has resulted with Email:Chathurak@hotmail.com
Sri Lanka being placed among those countries
1
285
telecommunication towers, network planning, employees could be more stressed, leading to
network optimisation activities and technical lower levels of performance. This may also
support services to customers‘ requests through impact the delivery commitments of the
customer visits. Consequently, the outsource service provider towards the
telecommunication operator who has to contracting organisation.
provide quality services to its subscribers is
On the other hand, increased work engagement
highly dependent on the outsourcing service
might be the solution to overcome these
provider. Hence, outsourced employees play
challenges considering the positive impact that
important and critical roles in the technical
works engagement generates. Therefore,
service delivery mechanism.
telecommunication operators and outsource
Traditionally, work engagement simply meant service providers ought to be keen on
employment and organisations believed that improving the work engagement of their staff
employees were engaged in their work. to improve or maintain their service levels.
However, over time, this understanding has
Organisations are now seeking to improve the
changed. In a general sense, engagement refers
work engagement of their employees by
to a combination of involvement, enthusiasm,
conducting surveys to figure out what needs to
passion, commitment and dedication
be done (Ketter, 2008). By doing so, Saks (2006)
(Schaufeli, 2012: 3; Schaufeli, 2014: 15). Work
suggests that they might discover a few
engagement means exhibiting these attributes
antecedents of the work engagement that could
towards work.
assist in enhancing engagement. There are
many antecedents of the work engagement, and
There can be multiple benefits to employees
some have been empirically tested (Wollard
and companies from work engagement.
and Shuck, 2011). However, with outsourcing,
Engaged employees experience positive
the main academic attention so far has centred
emotions and enjoy good mental health
on the economics of this activity, and much less
(Schaufeli, 2012; Schaufeli and Salanova, 2008;
attention has been paid to personal or employee
Halbesleben, 2010). At the individual employee
aspects (Raghuram, 2006). Similarly, Purcell
level, a better work engagement may lead to
(1996: 22 cited in Kessler et al., 1999: 5) notes,
better performance, increased output and a
―we lack the research to make definite
lower number of absent days from work
statements about the effect of outsourcing on
(Schaufeli, 2012; Schaufeli, 2014; Rich et al.,
employees.‖ As such, previous research has
2010; Halbesleben, 2010). At the organisational
provided valuable insights into the work
level, this may lead to increased customer
engagement in an in-house context; but only
satisfaction, higher profitability, productivity,
limited studies have been devoted to the
turnover, and safety (Harter et al., 2002; Harter
outsource environment. This study aims to fill
et al., 2009). Therefore, work engagement can
this gap by investigating the work engagement
be identified as a desirable condition for
of outsourced employees.
employees and for the organisation for which
they work (Schaufeli, 2012). Taking into account the above-mentioned
considerations, this study will attempt to
Normally, in-house employees empathise to a
identify antecedents that would have a high
marked degree with their organisations‘ goals,
influence on the work engagement of
are familiar with their organisations, engage in
outsourced employees. Once these influential
direct communication with top management
antecedents of the work engagement are
and are subject to direct supervision by their
identified, telecommunication operators with
managers (Purcell and Purcell, 1998). In
the collaboration of outsourcing service
contrast, outsourced employees are involved in
providers could devise some mechanism to
the service delivery of organisations that avail
reinforce these antecedents in order to improve
themselves of the offerings of third-party
the work engagement of employees.
service providers. They are not part of the
hiring organisation. This means that outsourced Therefore, this study primarily focuses on the
employees are relatively less informed, less research questions:
compensated and operate under less
 What would be the work engagement
supervision.
levels of outsourced employees at present
Therefore, outsourced employees face more in the telecommunication engineering
difficulties or challenges in executing their service sector of Sri Lanka?
work compared to in-house employees. In this  What are the significant antecedents of the
context, there is a high chance that outsourced work engagement in outsourced employees
2
286
in the telecommunication engineering understanding from the viewpoint of
service sector of Sri Lanka? occupational health psychology, an employee‘s
psychological relationship with his work is
visualised as falling in between the endpoints
2. Literature Review of the positive experience of engagement and
negative experience of burnout in a single
2.1 Concept of Engagement at Work
continuum (Maslach and Leiter, 2008). Here,
In a general sense, engagement refers to a the engagement is identified by energy,
display of involvement, absorption, involvement, and efficacy, which are
commitment, passion, enthusiasm, focused considered to be the direct opposites of three
effort, zeal, dedication, and energy of someone burnout dimensions, exhaustion, cynicism, and
or something (Schaufeli, 2014: 15). However, lack of professional efficacy (Maslach and
the traditional view of the work engagement Leiter, 1997, as cited in Maslach and Leiter,
simply meant employment. Organisations 2008). This means that if an employee is low on
assumed that employees were fully engaged at work engagement, he is high on burnout.
their jobs because they happened to be on the
pay role (Schumann, 2010). But recent Schaufeli et al. (2002: 74) define the work
developments in business studies and in engagement ―as a positive, fulfilling, work-
academia have altered this simple related state of mind that is characterized by
understanding of the engagement. vigour, dedication, and absorption.‖ In this
definition, vigour is identified as having a
In the 1990s, a management consulting firm, higher level of energy at work, ability to
Gallup Inc. introduced an employee satisfaction weather-out difficulties at work and
and engagement measurement tool (Harter et willingness to invest effort in work. Dedication
al. 2009). This fuelled much interest on work refers to being strongly involved in work with
related engagement in both business and the feeling of self-importance, enthusiasm,
academia (Schaufeli, 2014). Schaufeli (2014: 17) pride, being inspired by work, and feeling
also attributed this increase in interest partially challenged. Absorption means fully
to a positive psychology movement, where concentrating and being happily occupied in
researchers try to understand the optimal one‘s work. This portrays the work engagement
human behaviour to discover the reasons that as a distinctly different concept from burnout.
enable individuals and organisations to
succeed. Saks (2006: 602) defined employee engagement
as ―a distinct and unique construct consisting of
Employee engagement and work engagement cognitive, emotional, and behavioural
are commonly used concepts in literature. components that are associated with individual
Employee engagement refers to the relationship role performance.‖ This definition is similar to
between the employee and his organisation, the definition of Khan (1990) whose focus was
whereas the work engagement narrows it down on an organisation‘s employees collectively,
to the relationship between the employee and while Saks‘ (2006) focus was on individuals.
his work (Schaufeli, 2014: 15). This study
mainly focused and developed on the concept Schaufeli (2014: 31) claims that most of the
of work engagement, which is a narrower and a academic studies on engagement have utilised
more specific idea. the definition based on vigour, dedication and
absorption. For this research, the same work
The initial reference to the engagement in
engagement definition proposed by Schaufeli et
academia was by Khan (1990: 694) who defined
al. (2002) is adopted.
personal engagement as the ―harnessing of
organisation members‘ selves to their work There are multiple theories available to explain
roles: in engagement, people employ and work engagement behaviour. However, as
express themselves physically, cognitively, Schaufeli (2014: 25) claims, there is no single
emotionally, and mentally during role theory that explains all aspects of work
performances‖. In this view, employees build engagement, but a collection of theories that
an identity for themselves centred on their elaborate on different aspects of work
work, and they put in an effort to develop and engagement. Kahn (1990) in his Need-
maintain this identity. Satisfying model proposed that employees will
be engaged if three psychological needs are
Alternatively, the work engagement is also
met. He identifies these needs as
viewed as the opposite of burnout. In this
3
287
meaningfulness, safety, and availability and to maximise benefits and minimise costs. Saks
suggests that employees ask three questions (2006: 603) suggests that when employees
from themselves relating to work to verify receive benefits such as salary or development
whether these needs will be met. These opportunities, employees also reciprocate in
questions are: (i) Is this work meaningful kind; one way of repayment is on their level of
enough for me to get myself personally engagement. Saks (2006: 603) notes, ―bringing
involved? (ii) How safe will this work be? and oneself more fully into one‘s work roles and
(iii) Am I available for this work? In other devoting greater amounts of cognitive,
words, these three questions assess return on emotional, and physical resources is a very
self-investment, preservation of self-image or profound way for individuals to respond to an
status and possessing of physical or organisation‘s actions.‖ When organisations fail
psychological resources, respectively (Kahn, to provide these reassurances, employees will
1990: 705). Simply put, this theory predicts that tend to disengage themselves from their roles.
when the job is meaningful, the environment is
safe to work in, and personal resources are The Job Demands – Resources (JD-R) theory
available, then employees become engaged proposes that causes of employee well-being
deeply in work. fall under two main categories identified as job
demands and job resources (Bakker and
In the Affective Shift Model, Bledow et al. Demerouti, 2007: 312). Job demands refer to
(2011: 1246) propose that work engagement conditions that generate physiological or
results from an employee moving from a psychological stress such as work pressure and
situation in which negative effects are unfavourable working environment (Bakker
experienced to a situation of high positive and Demerouti, 2007: 312). Job resources are
effects, attempting to explain changes in work those factors that mitigate the physiological or
engagement throughout the day, between tasks psychological stress, stimulate personal growth
or events. Positive affect is a signal to self to or assist in achieving work goals (Bakker and
approach and continue along the line of action Demerouti, 2007: 312). Xanthopoulou et al.
while negative affect interrupts the ongoing (2007, 2009) have expanded this model to
stream of action, disrupts the mental decision- include personal resources. Personal resources
making process and hinders behaviour (Bledow are positive self-evaluations that are linked to
et al., 2011: 1247). Work engagement will be low resiliency; they refer to an employee‘s
if an employee remains in a negative affective conviction that he can successfully control and
state without experiencing positive effects. influence his environment (Hobfoll et al., 2003).
However, employee motivation will improve if Building on above findings, Bakker and
an employee moves to a positive affective state. Demerouti (2008a: 218) propose the JD-R model
of work engagement as shown in Figure 1. This
Social exchange theory proposes that social model shows that job resources and personal
behaviour is the result of an exchange process resources can independently and collectively
Figure 1 - The JD-R model of work engagement (Bakker and Demerouti, 2008a: 218)
4
288
influence work engagement. Bakker and feel more confident in controlling their
Demerouti (2008a: 218) argue that ―job and environment, and increase their pride and
personal resources particularly have a positive confidence in the work they do, leading to a
impact on engagement when job demands are better work engagement (Xanthopoulou et al.,
high‖. This model presents that there is a 2009: 136).
relationship between work engagement and
performance. The JD-R model also includes a In this model, Job demands refer to job aspects
feedback loop through which employees who such as work overload or high time pressure,
are engaged and perform well, can reciprocally which would require employees to exert
influence their own resources. additional effort to achieve work goals and to
prevent performance decline (Schaufeli, 2014:
In this model, the relationship between job 27). These job demands would exhaust the
resources and work engagement could be energy of employees over time and eventually
explained in terms of the motivational process lead to burnout (Schaufeli, 2014: 27). This
(Bakker and Demerouti, 2008a). In accordance negative process is called health impairment
with McClelland‘s theory of needs (1961, as process, which counteracts or mediates the
cited in Robbins et al., 2011), autonomy can be motivational process between resources and
identified as the need for power over oneself. work engagement within the JD-R model.
Social support, feedback and supervisory
coaching can be linked with the need for The JD-R model also presents the relationship
affiliation. When these needs are satisfied, between work engagement and job
employees will be motivated in term of energy, performance. Bakker and Demerouti (2008a)
dedication and focus. In fact, work engagement identified four possible reasons as to why
is the presence of these three qualities engaged employees perform better than non-
(Schaufeli et al., 2002). Further, the job engaged employees. These reasons are, (i)
characteristics model proposes that core job positive emotions experienced by engaged
dimensions such as autonomy or feedback can employees, (ii) better mental health of engaged
lead to a critical psychological state (Hacman employees, (iii) engaged employees to create
and Oldham, 1980, as cited in Robbins et al., own job and personal resources, and (iv)
2011). This psychological state can be engaged employees to induce engagement
considered as work engagement. This is the among other employees (Bakker and
intrinsic motivational relationship between job Demerouti 2008a: 215-217). According to
resources and work engagement. Meijman and Bakker and Demerouti (2008a: 214-215), several
Mulder (1998, as cited in Bakker and studies have provided evidence on the
Demerouti, 2008a) also identify an extrinsic relationship between work engagement and
motivational relationship where a congenial performance.
work environment with supportive colleagues,
guidance from supervisor or feedback would Out of the four models of work, engagement
motivate employees to work hard and was discussed, the JD-R model of work
successfully complete tasks. Multiple studies engagement has received the most empirical
have confirmed this relationship between job support (Schaufeli 2014: 31). Hence, the JD-R
resources and work engagement (Schaufeli and model was adopted for this study.
Bakker, 2004b; Xanthopoulou et al., 2009).
2.2 Measurement of Work Engagement
The relationship between personal resources There are several instruments available for
and work engagement is explained as follows: a measurement of work engagement. The most
person with higher personal resources would widely used instrument to measure employee
have a more positive self-regard and be more engagement in the business domain is Gallup‘s
goal oriented, which in turn would lead to Q12 questionnaire (Schaufeli, 2012: 4). This
higher performance (Judge et al., 2005, as cited questionnaire has items that were developed
in Bakker and Demerouti , 2008a: 213-214). from an actionable standpoint, including items
Xanthopoulou et al. (2009: 136) having such as ―This last year, I had opportunities at
examined the relationship between self-efficacy, work to learn and grow‖ (Harter et al., 2009: 8-
self-esteem, optimism and work engagement 9). The above question assesses opportunities
claim that the motivational process is also for professional development, a job resource
active between personal resources and work under the JD-R model. Therefore, rather than
engagement. They explain that having a high evaluating the work engagement as a
level of personal resources makes employees psychological state, the Q12 questionnaire
5
289
directly measures the antecedents of work goals, perceived organisational support, vigour,
engagement (Schaufeli, 2012: 4). work-life balance, core self-evaluation,
authentic corporate culture, clear expectations,
Bakker and Demerouti (2008b) propose corporate social responsibility, job
Oldenburg Burnout Inventory (OLBI) as an characteristics, job fit, task challenge, manager
alternative measurement of burnout and expectation, manager self-efficacy, work place
engagement. They argue that instead of having safety, work place climate, rewards, supportive
both positively and negatively worded items to organisational culture and use of strengths
measure two dimensions, exhaustion and (Wollard and Shuck, 2011: 433). It is worth
disengagement, the OLBI question items could noting that this extensive list includes the three
be reverse coded to represent two engagement dimensions of work engagement proposed by
dimensions, vigour and dedication (Bakker Schaufeli et al. (2002: 74). This shows that
and Demerouti, 2008b). This follows the different scholars in different contexts have
visualisation of burnout and work engagement considered an array of attributes as possible
as endpoints of a single continuum (Maslach antecedents of work engagement. However,
and Leiter, 2008). testing and assessing all these 41 antecedents
will be an intimidating task for any single
Schaufeli et al. (2002) proposed a 17-item study.
Utrecht Work Engagement Scale (UWES-17)
questionnaire to evaluate the work engagement Schaufeli and Bakker (2004b) found
by assessing the three dimensions, vigour, performance feedback, social support, and
dedication, and absorption. This scale includes supervisory coaching to be important
6 vigour items, 5 dedication items, and 6 antecedents. Saks (2006: 611) found
absorption items (Schaufeli et al., 2002). Bakker organisational support and procedural justice
and Demerouti (2008a: 210) note that multiple to be significant antecedents to employee
researchers in several countries have validated engagement. Xanthopoulou et al. (2007, 2009)
the UWES-17 scale. Later, Schaufeli et al. (2006) found autonomy, social support, supervisory
have developed a shorter version of the UWES coaching, professional development,
scale and have concluded that this new scale organization-based self-esteem, optimism and
can be satisfactorily utilised to measure work self-efficacy to be significant antecedents.
engagement. This compact scale has 3 questions
on each of the 3 dimensions of work For this study, five job resources, viz.
engagement. Schaufeli and Bakker (2004a: 21) autonomy, social support, managerial,
found higher than 0.9 correlations among the supervisory coaching, performance feedback
three dimensions measured with UWES-17 and and professional development opportunities
UWES-9, with the new scale having an internal and four job resources, viz. self-efficacy, self-
reliability measured by Cronbach's alpha at esteem, resilience and optimism were selected
between 0.89 to 0.97 across 25 studies they as possible antecedents.
analysed. For present research, the UWES-9
scale was adopted in view of its alignment with In general sense, autonomy is the freedom that
the work engagement definition and the JD-R an employee has from being governed.
model. Schwalbe (1985: 525) notes autonomy as
freedom of movement, freedom to establish and
2.3 Antecedents of Work Engagement execute plans for task accomplishment, and
freedom from immediate supervision. Robbins
The main purpose of this research revolves
et al. (2011: 91) identify autonomy as the degree
around the antecedents of work engagement.
to which the job provides freedom,
Wollard and Shuck (2011: 433) in their review
independence and discretion in work
of the literature identified a total of 41
scheduling and on deciding the course of
antecedents to employee engagement, 20 of
action.
them individual antecedents and balance 21
organisational antecedents. They found that 11
Etzion (1984, as cited in Brough and Pears,
of 20 individual antecedents and 13 out of 21
2004: 472) defined social support as an informal
organisational antecedents were supported by
social network that provides individuals with
empirical evidence (Wollard and Shuck, 2011:
expressions of emotional concern or empathy,
433). The antecedents with empirical evidence
practical assistance, informational support or
are absorption, dedication, corporate
appraisal. Brough and Pears (2004: 472) identify
citizenship, involvement in meaningful work,
behaviour such as collaborative problem
linkage with individual and organisational
6
290
solving, sharing of information, giving advice feels important at work, then he would be more
to and receiving advice from colleagues as confident in taking up challenging tasks.
workplace social support. When an employee is
faced with high work pressure or a difficult Resilience is an employee‘s ability to face
problem to solve, he could turn to his difficult situations, cope with adverse
colleagues at work who are willing to support. conditions without becoming flustered and
recover from unsettling situations fast (Hornby,
Fournies (1987, as cited in Ellinger et al., 2005: 2005: 1291). When an unfavourable
621) claim that managerial coaching is a development causes a setback to a resilient
leadership style in which constructive feedback employee, he would not continue to be in a
is used to get the most out of employees discouraged state of mind but would recover
through demonstration of respect and quickly and continue to perform well.
appreciation. Managerial coaching helps
employees to recognise their potential, improve In general terms, optimism may be regarded as
their performance, and develop their skills by hopefulness or confidence that the future
empowering them (Ellinger et al., 2005: 622). would be promising (Hornby, 2005: 1067).
The employee would receive proper Optimistic employees will look beyond present
instructions, solutions to problems he faces and difficulties or challenges and remain hopeful.
guidance on how to be efficient through the
supervisory coaching. Further, it also brings As already explained, the JD-R model proposes
clarity to the tasks an employee performs. that these nine resources would assist an
employee to mitigate the adverse impact of job
Performance feedback means providing direct demands on work engagement. The purpose of
and clear information on an employee‘s this study is to explore the relationship between
performance or effectiveness (Robbins et al., work engagement and these nine antecedents
2011: 91). Performance feedback is important if in an outsource service environment.
an employee is to perform as expected by the
organisation, without having to face surprises 2.4 Service Outsourcing
at performance appraisals; it also allows him to Outsourcing is the transfer of an in-house
discuss and find out solutions for difficulties. function to an external service provider, who
then provides the outsourcing company with
Professional development means updating
services at least as good as, or hopefully better
one‘s skill set or acquiring new knowledge.
than the services provided in-house (Godfrey-
When an employee is given the opportunity of
Faussett, 1997: 179). Telecommunication
acquiring new knowledge or mastering a new
operators have been increasingly outsourcing
skill, he will show more competence in
their engineering services to outsource service
performing his job. With advanced skills set or
providers. As Belcourt (2006) point outs, this
knowledge, the employee will be able to face
trend of outsourcing is primarily prompted by
work pressure or manage difficult situations at
the need to reduce the operational costs of the
work more efficiently and effectively.
telecommunication operator in a high price
sensitive market, while there are also other
Self-efficacy is the employee‘s belief that he is
benefits such as the following:
capable of performing a task competently
(Robbins et al., 2011: 80). When an employee
 Outsourcing secondary functions
has high self-efficacy, he would be more
allows organisations to focus more and
confident in his ability to accomplish a task.
better on their core functions
 Easier access to upgraded and the
Self-esteem is the employee‘s own evaluation of
latest technology through outsource
his own worth (Rosenberg, 1965, as cited by
partnership
Robbins et al., 2011: 72). Maslow (1954: 45)
 Improved performance through
asserts that ―satisfaction of self-esteem need
enforcing strict performance standards
leads to feelings of self-confidence, worth,
in outsource contracts
strength, capability, and adequacy, of being
useful and necessary in the world. But  Access to the specialised expertise of
outsourced employees
thwarting of these needs produces feelings of
inferiority, of weakness, and of helplessness.‖
While there are benefits to be reaped through
Stating it differently, when an employee
outsourcing, Belcourt (2006) draws attention to
possesses self-respect, believes in himself and
several risk factors associated with outsourcing.
7
291
Increased inflexibility due to difficulty in Level-5: Managed Service with Managed
changing the terms of the contract, negative Capacity – Outsource service provider takes
impact on employees‘ morale and performance, responsibility for complete service delivery and
and limiting of the organisation‘s ability to developing network capacity infrastructure to
learn and adapt to changes are some of these. achieve business objectives of the
Despite these disadvantages, telecommunication operator.
telecommunication operators in Sri Lanka are
embracing this practice of outsourcing their One telecommunication operator in Sri Lanka
engineering functions. has adopted outsourcing at level-5 of above
classification. Remaining telecommunication
Ramani (2014) has identified five levels of operators have also outsourced their
outsourcing in ascending order of involvement engineering services at level-3 or level-4.
of an outsource service provider as shown in Therefore, there is high involvement of
Figure 2. outsourced employees in the day-to-day
operations of telecommunication operators.
Level-1: Staff Augmentation – Specific skilled Telecommunication operators are therefore
resources to supplement in-house workforce highly dependent on the performance of these
are provided by an outsource service provider. outsourced employees as their numbers are
Activities that outsourced employees engage in substantial and form a vital part of the service
are as instructed by telecommunication delivery to subscribers.
operator and outsource service provider does
not take any risk or responsibility. As discussed, outsourced employees are a key
component of service delivery of
Level-2: Out-Tasking – Outsource service telecommunication operators. However, these
provider only performs certain tasks related to outsourced employees differ from in-house
part of the network element and is responsible employees in several aspects. Purcell and
for those activities. Purcell (1998) note that outsourced staff is
treated differently from in-house staff.
Level-3: Project Based – Outsource service Outsourced staff is paid less, not granted fringe
provider is responsible for the day-to-day benefits and training is kept to a minimum, in
running of a particular project. The contrast to the much higher pay of in-house
telecommunication operators‘ duty is restricted staff, who are entitled to fringe benefits such as
to monitoring and control the project deliveries sick pay and holiday pay. They are also more
are as per service level agreement entered into likely to receive employee training as their
by both parties. organisations are more willing to invest in
them. Outsourced employees who typically
Level-4: Managed Services – Outsource service work alongside the in-house employees are
provider takes complete ownership for well aware of the differential treatment. Hence,
delivery. His scope includes achieving technical they are likely to be in a rather discontented
service delivery and business objectives. state of mind.
Figure 2 - Levels of Service Outsourcing (Ramani, 2014)

8
292
Outsourced employees involved with and personal resources that act as antecedents
engineering service delivery on behalf of to work engagement as identified in the JD-R
telecommunication operators require a high model. Though literature identifies an array of
level of skills and domain expertise to perform antecedents to work engagement, the available
satisfactorily. Whereas, Aron and Singh (2005: literature fails to identify significant
137) point out that outsourced employees antecedents that refer specifically to the
engaged in such demanding work tend to make outsource environment. Therefore, this study
more mistakes in operation compared to in- also attempts to identify the significant
house staff. This is also an indication that antecedents of the work engagement of
outsourced employees are placed at a outsourced employees involved with
disadvantage compared to in-house employees. telecommunication service delivery. By
knowing those significant antecedents,
While telecommunication operators are telecommunication operators and outsource
increasingly dependent on the services of service providers could devise a more effective
outsourced employees, it can be seen that due strategy to enhance the work engagement of
to differences between in-house and outsource outsourced employees, which would, in turn,
employment conditions, outsourced employees improve the performance level of those
are placed in an unfavourable position. employees.
Outsourced employees are faced with 3. Methods

additional challenges or difficulties in their
work environment. To be specific, they are This study adopts a combination of quantitative
faced with high job demands, both and qualitative research methods. Use of
psychological and physiological. According to quantitative and qualitative methods in
the JD-R model of work engagement, a high combination allows the crosschecking and
level of job demands would lead outsourced validating of findings, thus allowing the
employees to lower engagement levels in researcher to have greater confidence in the
comparison to in-house employees, who do not results (Greene et al. 1989: 259; Bryman and
face similar challenges. Bell, 2007: 413). The quantitative study in this
research focuses on establishing the work
However, as Raghuram (2006) highlights, engagement level of outsourced employees and
previous studies on outsourcing mainly on identifying significant antecedents of work
focused on the economic aspects of outsourcing engagement, while qualitative research is used
and much less on the individuals involved. to verify and correlate with findings of the
Purcell (1996: 22, as cited in Kessler et al., 1999: quantitative study.
5) has expressed a similar view stating ―we lack
the research to make definite statements about 3.1 Quantitative Study
the effect of outsourcing on employees.‖ In this
context, even though the JD-R model predicts A quantitative study was conducted by
lower engagement level for outsourced administrating questionnaires to outsourced
employees based on additional job demands employees currently engaged in engineering
they experience while being deprived of service delivery. These employees were
resources, there is no empirical evidence to requested to fill in a structured questionnaire
support this claim. Thus, this study attempts to on a voluntary basis. This study was conducted
fill this gap by estimating the work engagement with one of the larger outsource service
of outsourced employees involved in providers, who provide engineering services to
telecommunication service delivery in Sri telecommunication service providers in Sri
Lanka. Lanka. This company has on its employment
role 281 permanent and contractual staff
As discussed, telecommunication operators are assigned to operations of a single
heavily dependent on outsourced employees telecommunication operator. Due to time and
for delivery of engineering services. resource constraints, it was decided to
Considering the positive impact work administrate the questionnaire to just 50% of
engagement has on the performance of these the population, totalling 140 participants. For
outsourced employees, telecommunication convenience, the questionnaire was
operators would be keen on improving the administrated only to those employees who
work engagement levels. This could be attended monthly review meetings. This
achieved through stimulating the job resources approach assisted in receiving a good response
in terms of completing the questionnaire.
9
293
A structured questionnaire with 77 questions Xanthopoulou et al., 2009). This item scale
was drafted to capture the relevant data on ranges from 1= ‗Strongly disagree‘ to 5=
work engagement and their antecedents. Work ‗Strongly agree‘. Sum of all three items is taken
engagement was measured using the UWES-9 as a measure of opportunities for professional
(Schaufeli et al., 2006). Items were scored on a development.
Likert scale ranging from 0= Never to 6=
Always. Scores of all questions were summed Self-efficacy is captured through the General
up to determine the overall UWES score, where Self-Efficacy (GSE) scale developed by
a high score implies high work engagement. Schwarzer and Jerusalem (1995). GSE consists
Autonomy was measured using a modified of 10 items scored as 1= ‗Not true at all‘ to 4=
version of the Self-Determination Scale (SDS; ‗Exactly true‘. The total score is equal to the
Sheldon et al., 1996; Sheldon, 1995) that was sum of scores obtained for all items. The
designed to assess the extent to which Rosenberg Self-Esteem Scale (RSE) with 10-
employees tend to function in a self-determined items was used to evaluate self-esteem of
way. SDS has two 5-item subscales, one sub- employees (Rosenberg, 1965). Each item was
scale on awareness of oneself and the other on scored using a 4-point Likert scale. For items
perceived choice in one‘s actions. For this 1,3,4,7 and 10, a score of 1 represents ‗strongly
study, a 5-item sub-scale on perceived choice is disagree‘ and 4 represents ‗strongly agree‘,
adopted in the employment context. Each item while for the balance items the scoring is
consists of a pair of statements (statement A reversed. Sum of all items after reversing 2,5,6,8
and statement B), and participants were asked and 9 is computed to arrive at the RSE score. A
to judge the degree to which statement-A is higher RSE score indicates greater self-esteem.
true against statement-B on a 5-point Likert Participants‘ resilience was measured using a
scale. If statement-A strikes as completely true 14-item Resilience scale (Wagnild, 2009) in
and statement-B strikes as completely untrue, which each item was scored on a 7-point scale
the appropriate response would be 1. If the two with 1= ‗totally disagree‘ to 7= ‗totally agree‘.
statements are equally true, the appropriate All scores are summed up to compute the total
response would be 3. A score of each item score.
would be reversed to arrive at SDS value so
that a higher score on every item will indicate a Optimism was measured with the Life
higher level of self-determination. Orientation Test – Revised (Scheier et al., 1994).
This scale is comprised of six items that
A six-item scale that was developed based on measure optimism and four filler items. The
the social support questionnaire-short form four filler items were not considered for
(Sarason et al., 1987) was used to measure social computation. Out of the six main items, three
support for employees at work. Each item is items measure optimism as a positive quality,
scored on a 4-point Likert scale as 1= ‗Not true while the other three items measure pessimism
at all‘ to 4= ‗Exactly true‘. Scores of individual as a negative quality. Participants rate each
items were added up to arrive at the total social item on a 5-point scale ranging from 0= ‗totally
support score. Ellinger et al. (2005) have disagree‘ to 4= ‗totally agree‘. All negative
introduced an exploratory 8-item measure for items were reversed before totalling so that
managerial coaching based on managerial higher scores indicate higher optimism.
coaching behavioural themes. Six items in this
scale focus on supervisory coaching, while the This study attempts to investigate the
balance two items focus on supervisory relationship between work engagement and 9
feedback. Therefore, this scale is used to independent variables. All 9 independent
measure both managerial, supervisory coaching variables and the dependent variable, work
and feedback dimensions in the present study. engagement, are measured on a continuous
Each item is scored on a 7-point Likert scale scale. Hence, Multiple Linear Regression
where 1= ‗Almost never‘ to 7= ‗Almost always‘. technique is employed to investigate the
The scores obtained for question numbers relationship between work engagement and the
1,2,3,6,7 and 8 were taken to calculate the total independent variables.
score of managerial, supervisory coaching, The total number of fully completed
while scores for question numbers 4 and 5 were questionnaires received was 116 out of the 140
summed up to obtain the supervisory feedback distributed, a response rate of 83%. However,
score. Possibilities for professional when checking for outliers while performing
development were measured using a 3-item multiple linear regression analysis, one outlier
scale developed by Bakker et al. (2003, cited in (Participant 89) was identified and excluded
10
294
from the dataset. In the final dataset, 12% A non-probability sample of six participants,
(N=14) were from manager level, 37% (N=42) viz. two managers, two engineers and two
from engineers and balanced 51% (N=59) from technical officers were selected for interviews
technical officers attached to the outsourcing out of those who responded to the initial
organisation. Further, 26% (N=30) of the questionnaire, in order to maximize the validity
respondents have been at the current and reliability of results (Bryman and Bell, 2007:
employment position for less than 2 years, 63% 473). Invitations for the interview were sent to
of them (N=72) from 3 to 5 years and balance all six participants together with a list of five
11% (N=13) have been in their positions for questions. Four out of these six accepted the
more than 5 years. invitation while two of them refused to
ANOVA analysis is performed to examine the participate. All interviews were conducted in a
model fit with regression analysis. Table 1 meeting room at their office at a time
provides the model summary of regression convenient to them, with the interview lasting
analysis. The multiple correlation coefficients approximately 50 minutes. Extensive notes
(R) is the Pearson correlation coefficient were taken during the interviews. However, all
between the predicted scores of work participants had expressed concerns about
engagement by the regression model and actual recording the interviews. Therefore, the
Table 1 - Model Summary of Regression Analysis
R R2 Adjusted R2 Std. Error of the Estimate Durbin-Watson
0.847 0.717 0.687 0.67155 1.835
Table 2 - ANOVA Statistics
Sum of Squares df Mean Square F Sig.

Regression 117.706 11 10.701 23.727 0.000
Residual 46.452 103 0.451
Total 164.157 114
values of work engagement. R value of 0.847 interviews were not recorded.

suggests a strong association. Further, R2 value Notes that were taken during the interviews
of 71.7% indicates that inclusion of independent were read over multiple times to identify a
variables in the regression model caused 71.7% possible logical model or themes. These themes
variability in the work engagement compared were further analyzed with the intention of
to the mean model. Adjusted to reflect the finding the answers to research questions.
variability in the population, the Adjusted R2
value of 68.6% indicates a large size effect 4. Results
(Lakens, 2013). As shown in Table 2, ANOVA
4.1 Descriptive Statistics
predicts the model‘s statistical significance at p
<0.0005. Through these observations, it can be Table 3 presents means, standard deviations,
concluded that the regression model provides a Cronbach‘s alphas and correlations among the
good fit at a statistically significant level. study variables. In this study, Cronbach‘s alpha
3.2 Qualitative Study falls between 0.78 and 0.95, which is within the
acceptable range, indicating that the measures
Qualitative data were obtained from face-to- used are reliable (Tavakol and Dennick, 2011).
face semi-structured interviews conducted with There is a high correlation (r > 0.5) between
five open-ended questions pertaining to the work engagement and four of the independent
focus of interest of this study and in such a way variables with high significance (p < 0.01). Five
that the results could be easily interpreted, of the remaining measures are also sufficiently
despite the lack of flexibility that could be correlated (r= 0.26-0.46) with high significance
caused by novel explanations (Zikmund et al., (p < 0.01). These results suggest that certain
2013: 154). factors identified as possible antecedents of
These five open-ended questions invited work engagement are indeed positively related.
participants to share their own opinions, beliefs Further, it should be noted that these results
and experiences in relation to work also suggest that there are close relationships
engagement in the outsourced engineering between several independent variables. For
service environment and to share their thoughts example, autonomy is positively correlated to
on what they believed to influence the work self-efficacy (r= 0.58, p <0.01) and managerial
engagement. supervisory coaching is also positively
11
295
mean value of 2.54 with a standard deviation of
As given in Table 3, work engagement has a
1.20. Distribution of samples over the work

Level of Work Engagement
Mean SD WES AUT SS MSC FB PD SEF SES RES OPT

WES: Work Engagement 2.54 1.20 0.914
AUT: Autonomy 3.05 0.70 .629** 0.783
SS: Social Support 2.51 1.81 .331** .212* 0.883
MSC: Managerial Sup. Coaching 3.96 1.45 .635** .495** .295** 0.946
FB: Feedback 3.76 1.68 .382** .302** 0.148 .190* 0.838
PD: Professional Development 2.42 0.88 .594** .330** .294** .384** .306** 0.834
SEF: Self-Efficacy 2.41 0.82 .632** .578** .321** .486** .353** .491** 0.937
SES: Self-Esteem 2.41 0.78 .255** .219* 0.074 .240** .251** 0.116 .271** 0.922
RES: Resilience 3.97 1.31 .456** .531** 0.103 .447** 0.146 .384** .343** .233* 0.939
OPT: Optimism 2.13 0.96 .290** 0.082 -0.039 0.01 -0.096 .225* 0.123 -0.088 .226* 0.800
4.2
** Correlation is significant at the P < 0.01 level.
296
12
* Correlation is significant at the P < 0.05 level.
correlated to resilience (r= 0.45, p <0.01). These
findings suggest that there can be a relationship
between job resources and personal resources.
However, no further analysis was performed as
Table 3: Means, Standard Deviations, Cronbach‘s Alphas (on the diagonal) and Pearson Correlations among the study variables (N=115)
this is beyond the scope of the present study.
Work Engagement Level Range Frequency Percent Cum. Percent

1_Very Low <=1.93 38 33 33
2_Low 1.94 – 3.06 22 19.1 52.2
3_Average 3.07 – 4.66 54 47 99.1
4_High 4.67 – 5.53 1 0.9 100
5_Very High >=5.54 0 0 100
Table 4: Work Engagement Levels (N=115)
engagement levels specified by Schaufeli and Similarly, optimism would influence the work
Bakker (2004a: 40) is presented in Table 4. This engagement by 0.97, which is again 3.23 times
shows that 99.1% of samples fall on or below as potent compared to feedback. A unit change
the ―Average‖ work engagement level with in managerial, supervisory coaching would
52% in the ―Very low‖ and ―Low‖ categories. result in a 0.274 change in the work
The data clearly indicate that the work engagement. This is again 2.98 times as
engagement level of outsourced employees in influential compared to feedback. The least
the telecommunication service delivery sector influential antecedent of the work engagement
of Sri Lanka is skewed towards average or is feedback. A unit change in feedback would
lower levels. result in a 0.092 change in the work
engagement.
4.3 Relationship Analysis
Multiple regression analysis permits prediction In conclusion, based on quantitative analysis,
of the value of a dependent variable given the five antecedents that influence work
independent variables. More importantly, in engagement were identified. Of these five
the current research context, it helps to antecedents, four are job-related attributes, viz.
investigate the corresponding changes in the Autonomy, Professional Development,
dependent variable due to variances in Managerial Supervisory Coaching and
independent variables (Keller, 2014). In this Feedback. The remaining antecedent is a
study, the corresponding variation of work personal attribute: Optimism.
engagement is explored in relation to 9
independent variables. 4.4 Qualitative Analysis
Regression coefficients provided in Table 5 Four main themes have emerged as what
indicate that autonomy, professional participants perceived as factors influencing
development, optimism, managerial, work engagement. These themes were further
supervisory coaching and feedback are analysed to identify the factors that impact
statistically significant antecedents that work engagement in their workplace.
influence work engagement. Remaining four
Ambiguity in Instructions and Scope: All four
antecedents are not statistically significant, and
participants have cited the lack of a clear idea
there is no linear relationship between work
about what is expected of each of them and
engagement and those six antecedents. These
ambiguity in instructions as a major cause of
results indicate that a unit change in autonomy
degradation of work engagement. They made
would result in a 0.462 change in the work
statements such as ―[supervisor] tells me to do
engagement. This is the largest influencer of
one thing in front of mobile operator‘s staff and
work engagement as per this qualitative
another privately‖, ―mobile operator expects
analysis. Compared to feedback, the least
me to do certain things, but I cannot do so
influential antecedent, autonomy is 5 times
because of instructions from my company‖ and
more influential. A unit change in professional
―I am afraid to commit to any delivery as I‘m
development would result in a 0.318 change in
not sure how my supervisor would react‖.
the work engagement. This is the second largest
These statements clearly demonstrate the
antecedent, influencing work engagement and
participants‘ lack of freedom to take decisions
is 3.46 times as strong compared to feedback.
Unstandardized Coefficients Infl. Comp. to Feedback

B Std. Error
Autonomy 0.462 ** 0.129 5.02 times
Social Support 0.078 0.086
Managerial Supervisory Coaching 0.274 *** 0.056 2.98 times
Feedback 0.092 * 0.043 1 time
Professional Development 0.318 ** 0.091 3.46 times
Self-Efficacy 0.173 0.109
Self-Esteem 0.096 0.089
Resilience -0.047 0.063
Optimism 0.297 *** 0.071 3.23 times
*** p < 0.0005, ** p < 0.005, * p < 0.05
Table 5 - Regression Coefficients
13
297
or actions on their own. Further, these indicate over them. Telecommunication operator prefers
that there is a lack of proper understanding that senior positions in outsource service team
between supervisor and subordinate. Further, it be filled with their nationality. Outsource
provides evidence of a lack of proper service provider tend to have their own
leadership from the supervisor or senior nationals assigned to senior positions within
management in such a way that subordinates the service team. In this situation, the local staff
are unnecessarily exposed to conflicts between felt neglected because they had been given
the telecommunication operator and outsource lower preference. Through claims such as ―you
service provider. have to be a [nationality] if you are to get
appointed to a senior position‖ or ―only listen
High Work Demand / High Work Pressure:
to or believe what these people say or do,‖
Participants have indicated that they are
participants indirectly highlight the fact that
subjected to high work demands by their own
only foreign nationals are given the
supervisors and by the telecommunication
responsibility or the authority to act as they see
operator staff. Expressions such as ―I‘m asked
fit. The delegation of authority to local
to adhere to impossible deadlines‖ and ―I‘m
employees is perceived as being low. This is an
given impossible tasks to complete‖ reflect the
indication that participants feel the lack of
perception of participants regarding their work
authority delegation to local employees as a
and their reluctance to undertake challenges in
lack of autonomy in relation to work
the work environment. This demonstrates that
engagement. Further, these statements indicate
participants lack the confidence that they can
a connection between perceived social support
undertake and fulfil the assignments entrusted
for local employees and work engagement.
to them. Hence, these statements could be
Local employees seem to sense that there is a
interpreted as evidence of low self-efficacy as
difference in support available to them in
related to the work engagement. It also
comparison with other nationalities working in
highlights the fact that participants lack
the same environment.
sufficient faith in their abilities and in
themselves with respect to completing these The discussion on qualitative analysis can be
tasks. Clearly, these participants associate the summarised as follows:
feeling of lack of confidence or self-esteem with
Theme-1: Ambiguity in Instruction and scope
work engagement.
- Lack of freedom to take decisions or actions
Lack of Training or Development on their own > Autonomy
Opportunities: Participants have a feeling that - Lack of understanding between
those who are assigned to outsource service subordinate and supervisor; lack of proper
team become stagnant, performing the same leadership > Managerial Supervisory
tasks over long periods of time without the Coaching
opportunity for self-development. Statements Theme-2: High work demand/ work pressure:
―same old thing over and over again, nothing - Resistance to undertake challenges in work
new to do or learn‖ and ―technology is environment > Self-Efficacy
changing fast but I‘m working on the old - Lack self-regard or self-confidence in
technology‖ were made regarding the lack of completing assigned tasks > Self-Esteem
opportunity for training and development. Theme-3: Lack of training or development
Participants have realised that exposure to the opportunities:
latest technology and learning and training on - Unavailability of exposure to the latest
the latest technology is lacking when working technology, learning and training on the
in the outsource service delivery domain, latest technology > Professional
which in turn has eroded their enthusiasm for Development
work. All these statements are clear evidence of Theme-4: Preference over Nationality:
participants associating the lack of training and - Perception that only the foreign nationals
development opportunities for work are given the responsibility or the authority
engagement. to act as they see fit > Autonomy
- Difference in support extended to locals
Preference over Nationality: In a multinational
and foreign employees > Social Support
working environment, where the
telecommunication operator is based in one Above identifies the indicators obtained
country, and the outsource service provider through qualitative analysis on associations
originates from another; the local employees perceived between the work engagement and
seemed to feel that preference was given to its antecedents. Based on this, it can be
employees from those two foreign nationalities identified that the qualitative study provides
14
298
evidence that autonomy, social support, evidence to support these assumptions (Saks,
managerial, supervisory coaching, 2006). Secondly, there has been a minimal
opportunities for professional development, investigation into the work engagement in the
self-efficacy and self-esteem are perceived as outsource work environment (Raghuram, 2006).
influential antecedents of work engagement. This study has attempted to academically
explore the antecedents of the work
A combined matrix of antecedents and their
engagement of employees involved in
relationship to the work engagement is shown
engineering service delivery, attached to an
in Table 6. Both the quantitative and qualitative
outsource service provider. The study was built
studies have identified autonomy to be an
on the JD-R model (Demerouti et al., 2001;
influencing factor of work engagement.
Bakker and Demerouti, 2007; Bakker and
Similarly, both studies identify managerial,
Demerouti, 2008a), which presents the concept
supervisory coaching and opportunities for
that work engagement as interactions between
professional development as influencing
job resources, personal resources and job
antecedents. It should be noted that
demands. Five job resources: autonomy, social
opportunities for professional development
support, supervisory coaching, performance
have emerged as the main theme in the
feedback and opportunities for professional
qualitative study and clear evidence is available
development and four personal resources: self-
to support its influence over the work
efficacy, self-esteem, resilience and optimism
engagement. Social support, feedback, self-
were investigated during this study to establish
efficacy, self-esteem and optimism were
their relationship with work engagement.
identified as influential antecedents from either
the qualitative or quantitative analysis, but not This research has resulted in some interesting
Factor Quantitative Observations Qualitative Observations

Autonomy Significant (B= 0.462 **) Evidence of Association
Social Support Not Significant Evidence of Association
Managerial Sup. Coaching Significant (B= 0.274 ***) Evidence of Association
Feedback Significant (B= 0.092 *) No Evidence
Professional Development Significant (B= 0.318 **) Evidence of Association
Self-Efficacy Not Significant Evidence of Association
Self-Esteem Not Significant Evidence of Association
Resilience Not Significant No Evidence
Optimism Significant (B= 0.297 ***) No Evidence
B = Unstandardized Regression Slope Coefficient
*** p < 0.0005, ** p < 0.005, * p < 0.05
Table 6 - Influencing Antecedents of Work Engagement
from both. findings. To start with, the quantitative study

finds the work engagement level of outsourced
5. Discussion employees to be lower than global norms. This
The main purpose of this study is to investigate is not surprising considering that the
the relationship between work engagement and qualitative study points out severe job demands
nine probable antecedents of work engagement. faced by outsourced employees. Such job
This relationship is explored to identify which demands would lead to a degradation of the
antecedents have the more significant and work engagement levels of employees.
larger influences on the work engagement.
This study has revealed that autonomy is the
5.1 Theoretical Implications most influential antecedent of work
There has been a great deal of interest in the engagement. This is in line with the claim by
work engagement in recent years. Bakker and Demerouti (2007: 316) that
Organisations invest in conducting annual autonomy is one of the most important
surveys to investigate work engagement of predictors of work engagement. Bakker and Bal
their employees. (2010: 196) in their study of teachers have also
observed that autonomy is positively related to
Despite all this enthusiasm, linking the work
the work engagement. Xanthopoulou et al.
engagement with organisational performance
(2007: 35 and 2009: 241) have also confirmed
and business results, there is little empirical
that autonomy at work is strongly correlated to
15
299
work engagement. This finding also supports The quantitative study found that there is no
the Karasek (1979) job demand-control model significant relationship between social support
that predicts autonomy, or in other words, and work engagement. This is in agreement
decision latitude, as the most important with Bakker and Bal‘s (2010: 196) finding that
resource available to employees to cope with social support was unrelated to work
job stress. It should also be noted that high engagement. However, the results of the
work demands and high work pressure have quantitative study contradict the finding of
emerged as a theme in a qualitative study. In Xanthopoulou et al. (2009: 241) according to
that context, the job demand-control model whom having supportive colleagues enhances
could be considered to be quite relevant, and so work engagement. The qualitative results of the
it is not surprising to discover that the study present study provide evidence of an
identifies decision latitude, i.e. employees‘ association between social support and the
control over their tasks, as a significant factor. work engagement, in agreement with
Xanthopoulou et al. (2009: 241).
This study also identifies opportunity for
professional development as an influential This study finds optimism to be an antecedent
antecedent of work engagement. This is the of work engagement. Xanthopoulou et al. (2009,
only job resource that emerged as a main theme 241) also state that optimistic employees show a
in the qualitative study. The importance of high level of work engagement. However, in
professional development opportunities with their study Xanthopoulou et al. (2013) found
respect to the work engagement was that in employees facing emotional demands,
resoundingly demonstrated from the optimism did not have any significant
revelations of the participants. Xanthopoulou et relationship with work engagement. The
al. (2007: 135 and 2009: 241) have also quantitative study failed to identify any of the
established that opportunities for professional other personal resources as having a significant
development correlate strongly with work relationship with work engagement. This
engagement. Bakker and Bal (2010: 196) also contradicts with the result of Xanthopoulou et
confirm that any opportunity for development al. (2009, 241) that observed self-efficacy and
is positively related to the work engagement. self-esteem are related to work engagement.
However, the qualitative study agrees with
Similarly, the quantitative study finds evidence Xanthopoulou et al. (2009, 241), as the present
that managerial, supervisory coaching and study found evidence that the work
feedback are antecedents of work engagement. engagement is associated with self-efficacy and
The qualitative study identifies that effective self-esteem.
leadership role of supervisor and
understanding between supervisor and 5.2 Practical Implications
subordinate as important components of Telecommunications operators rely heavily on
managerial, supervisory coaching. services of outsourced employees for the
Xanthopoulou et al. (2007: 135 and 2009: 241) delivery of service to their subscribers. These
have also established that receiving proper outsourced employees are responsible for
coaching has a relationship with work maintaining the network and ensuring good
engagement. Further, Van Dierendonck et al. quality service to subscribers. But this study
(2004) also found that good leadership finds that the work engagement level of those
behaviour has a direct association with responsible for maintaining the network is
subordinates‘ sense of well-being and lower than is desirable. This low engagement
engagement. In contradistinction to above, Saks level would invariably lead to performance
(2006: 610-611) found that there is no significant gaps when executing assigned tasks. As
relationship between supervisor support and Godfrey-Faussett (1997) notes, problems arising
job engagement. In their study, Xanthopoulou due to poor quality or interrupted service could
et al. (2009: 241) also found that high-quality have a serious and direct impact on the
feedback positively correlates with work telecommunication operator. Hence, a lower
engagement. Bakker and Bal (2010: 196) found work engagement level of outsourced
that exchanges with the supervisor, i.e. a employees would be a matter of serious
combined measure of both supervisor feedback concern for telecommunication operators,
and supervisory coaching, positively related to especially in view of the increasing tendency to
the work engagement. outsource more of the engineering services to
third party service providers. At present, labour
laws and contracts between telecommunication
16
300
operators and service providers mainly govern Therefore, this study sample primarily
the financial aspects of outsourcing, represents the employee sub-set, whose work
performance service levels and employee well- engagement is at a low-level and not
being at the point of outsourcing (Godfrey- necessarily outsourced employees in general.
Faussett, 1997). However, there is a minimal This may be due to the error introduced by the
focus on the long-term well-being of convenience sampling technique adopted.
outsourced employees. Considering the Therefore, generalizing the findings of this
findings of this study, telecommunications study to include outsourced employees of the
operators should review the contractual telecommunication service sector of Sri Lanka
frameworks they have with outsourcing service may lead to erroneous conclusions.
providers to ensure that necessary resources are
available to maintain better work engagement This study employed 10 measurement
levels. However, such modification to existing instruments to estimate work engagement and
contracts would certainly come at an increased the antecedents of work engagement. The
cost to the telecommunication operator. researcher was careful to use well-established
and tested measurement tools with the
This is where the second part of this study
intention of employing a reliable scale.
becomes very relevant to telecommunication
Reliability of these scales has been tested and
operators. When telecommunication operators
found to be satisfactory. However, these
attempt to enhance the work engagement of
measurement instruments were developed and
outsourced employees, they should focus on
predominantly verified in western countries.
areas that would have a higher impact on the
Hence, participants may have found it difficult
work engagement. By doing so,
to understand and interpret them. This could
telecommunication operators could gain more
have caused some ambiguities in the responses
in terms of improving the work engagement.
to the questionnaire. Further, this could lead to
This study helps achieve that by identifying
difficulties in interpreting the meaning of
those factors that have the most influence over
culturally determined work engagement scores
the work engagement. For example, a
given by the UWES scale (Shimazu et al., 2010).
telecommunication operator‘s investments in
All attempts to find a locally developed and
improving autonomy at work would be five
tested measurement instruments were
times as effective as the same investment made
unsuccessful.
on feedback process improvement. Based on
the findings of this study, telecommunication
5.4 Recommendation for Future Studies
operators could work out a strategy to bolster
significant job and personal resources in such a In the present study, a high correlation between
way as to put outsourced employees in a strong job resources and personal resources were
position, where they will be able to face the job observed. These inter-correlations between
demands effectively and with much less antecedents are of much academic interest
negative impact on work engagement levels. (Xanthopoulou et al., 2007, 2009; Saks, 2006;
Schaufeli and Bakker, 2004b). This means that
5.3 Limitations of Present Research
there is evidence of a relationship between the
Despite obtaining some interesting results, the motivational process and the health
present study does have certain limitations. impairment process. However, investigating
This study adopts a non-probability sampling the relationships between job resources and
mechanism in this research for its convenience. personal resources or reciprocal relationships is
Even though convenience sampling is widely beyond the scope of the present study. Yet, the
used (Bryman, 1989: 113–114 cited in Bryman data set collected for this research can also be
and Bell, 2007: 198) in the field of business and analysed to find answers to different sets of
management, it introduces a problem in that questions with a much broader academic scope.
with such sampling, it is impossible to Through such an undertaking, it would be
generalize the findings because there is no possible to contribute much towards building a
evidence to indicate what population that the wider knowledge base on the intricacies of
sample represents (Bryman and Bell, 2007: 198). work engagement.
As per its design, this research used
The quantitative study indicates that
convenience sampling and focused on
participants of this research are at the ―very
employees of a single organisation. Further, it
low‖ to ―average‖ level of employee
was evident that the work engagement level of
engagement. Statistically, 99.1% of participants
the study sample was relatively low. Hence,
are at an average or lower level of engagement.
17
301
this study sample is biased towards one end of 11. Ellinger, A. E., Ellinger, A. D., Keller, S.B. 2005.
the work engagement spectrum. All these Supervisory Coaching in a Logistics Context.
concerns lead to a limitation with generalising International Journal of Physical Distribution &
Logistics Management, 35(9): 620 – 636.
of the results of this study. A more
comprehensive study involving a larger sample 12. Fitch Ratings. 2015. 2016 Outlook: Sri Lanka
and including employees of multiple outsource Telecommunications Services. Available at:
organisations could mitigate this limitation. www.fitchratings.lk/download-article-361.pdf;
accessed 30 September 2016.
Future studies should also consider the cultural
and language issues discussed earlier. 13. Godfrey-Faussett, M. 1997. Outsourcing
Questionnaire design in local languages and in Telecommunications Outsourcing a Legal
Perspective. Computer Law & Security Report,
conformity with the local cultural context might
13(3): 176-181.
generate more reliable data for the study.
However, this would require substantial 14. Greene, J. C., Caracelli, V. J., Graham, W. F.
preliminary work developing and testing new 1989. Toward a Conceptual Framework for
measurement tools in local languages before Mixed-Method Evaluation Designs. Educational
employing those in a study aimed at exploring Evaluation and Policy Analysis, 11(3): 255-274.
relationships. 15. Halbesleben, J. R. B. 2010. A Meta-Analysis of
Work Engagement: Relationships with
References Burnout, Demands, Resources, and
Consequences. In: Bakker, A. B., Leiter, M. P.
1. Aron, R., Singh, J. V. 2005. Getting offshoring (eds.), Work Engagement: A Handbook of Essential
right. Harvard Business Review, 83(12): 135-143. Theory and Research. New York: Psychology
2. Bakker, A. B., Bal, P. M. 2010. Weekly Work Press.
Engagement and Performance: A Study among
16. Harter, J. K., Schmidt, F. L., Hayes, T. L. 2002.
Starting Teachers. Journal of Occupational and
Business-Unit-Level Relationships between
Organizational Psychology, 83: 189–206.
Employee Satisfaction, Employee Engagement,
3. Bakker, A. B., Demerouti, E. 2007. The Job and Business Outcomes: A Meta-Analysis.
Demands-Resources Model: State of the Art. Journal of Applied Psychology, 87: 268-279.
Journal of Managerial Psychology, 22: 309-328.
17. Harter, J. K., Schmidt, F. L., Killan, E. A.,
4. Bakker, A.B., Demerouti, E. 2008a. Towards a Agrawal, S. 2009. Q12 Meta-Analysis: the
Model of Work Engagement. Career Development Relationship between Engagement at Work
International, 13(3): 209-223. and Organizational Outcomes. Gallup
Organisation. Available at:
5. Bakker, A. B., Demerouti, E. 2008b. The http://www.aamga.org/files/hr/MetaAnalysi
Oldenburg Burnout Inventory: A Good s_Q12_WhitePaper_2009.pdf; Accessed 21
Alternative to Measure Burnout and March 2016.
Engagement. In J. Halbesleben (ed.), Handbook of
stress and burnout in health care. New York: Nova 18. Hobfoll, S. E., Johnson, R.J., Ennis, N., Jackson,
Science Publishers. A.P. 2003. Resource Loss, Resource Gain, and
Emotional Outcomes among Inner City
6. Belcourt, M. 2006. Outsourcing — The benefits Women. Journal of Personality and Social
and the risks. Human Resource Management Psychology, 84: 632-643.
Review, 16: 269–279.
19. Hornby, A. S. 2005. Oxford Advanced Learners
7. Bledow, R., Schmitt, A., Frese, M., Kuhnel, J. Dictionary, 7th Edition. Oxford: Oxford
2011. The Affective Shift Model of Work University Press.
Engagement. Journal of Applied Psychology, 96(6)
1246–1257. 20. Kahn, W. A. 1990. Psychological Conditions of
Personal Engagement and Disengagement at
8. Brough, P., Pears, J. 2004. Evaluating the Work. Academy of Management Journal, 33(4):
Influence of the Type of Social Support on Job 692 -724.
Satisfaction and Work Related Psychological
Well-Being. International Journal of Organisational 21. Karasek, R. A., Jr. 1979. Job Demands, Job
Behaviour, 8(2): 472-485. Decision Latitude and Mental Strain:
Implications for Job Redesign. Administrative
9. Bryman, A., Bell, E. 2007. Business research Science Quarterly, 24: 285-308.
methods. 2nd edition. New York: Oxford
University Press, Inc. 22. Kessler, I., Coyle-Shapiro, J., Purcell, J. 1999.
Outsoucing and the Employee Perspective.
10. Demerouti, E., Bakker, A. B., Nachreiner, F., Human Resource Management Journal, 9(2): 5-19.
Schaufeli, W.B. 2001. The Job Demands–
Resources Model of Burnout. Journal of Applied 23. Ketter, P. 2008. What‘s the Big Deal about
Psychology, 86: 499-512. Employee Engagement? T and D, 62(1): 44-49.
18
302
24. Keller, G. 2014. Statistics for Management and Department of Psychology, Utrecht University.
Economics, 10th edition. Delhi: Cengage Available at:
Learning India Pvt. Ltd. http://www.wilmarschaufeli.nl/publications/
Schaufeli/Test%20Manuals/Test_manual_UW
25. Lakens, D. 2013. Calculating and reporting ES_English.pdf; accessed 21 March 2016.
effect sizes to facilitate cumulative science: a
practical primer for t-tests and ANOVAs. Front 39. Schaufeli, W. B., Bakker, A. B. 2004b. Job
Psychol. 4: 863. Demands, Job Resources, and their
Relationship with Burnout and Engagement: A
26. Maslach, C., Leiter, M. P. 2008. Early Predictors Multi-Sample Study. Journal of Organizational
of Job Burnout and Engagement. Journal of Behavior, 25: 293–315.
Applied Psychology. 93(3): 498–512.
40. Schaufeli, W. B., Bakker, A. B., Salanova, M.
27. Maslow, A. H. 1954. Motivation and Personality. 2006. The Measurement of Work Engagement
New York: Harper & Row, Publishers, Inc. with a Short Questionnaire: A Cross-National
28. Purcell, K., Purcell, J. 1998. In-sourcing, Study. Educational and Psychological
Outsourcing, and the Growth of Contingent Measurement, 66: 701–716.
Labour as Evidence of Flexible Employment
41. Schaufeli, W. B., Salanova, M. 2008. Enhancing
Strategies. European Journal of Work and
Work Engagement through the Management of
Organizational Psychology, 7(1): 39-59.
Human Resources. In Näswall, K., Hellgren, J.,
29. Raghuram, S. 2006. Individual effectiveness in Sverke, M. (eds.), The Individual in the Changing
outsourcing. Human Systems Management, 25: Working Life. Cambridge: Cambridge
127-133. University Press.
30. Ramani, K. 2014. IT Operations Managed 42. Schaufeli, W. B., Salanova, M., Gonza lez-Roma ,
Services – A Perspective. Mindtree Limited; V., Bakker, A. B. 2002. The Measurement of
Available at: Engagement and Burnout: A Two Sample
http://www.mindtree.com/sites/default/files Confirmative Factor Analytic Approach, Journal
/mindtree-thought-posts-it-operations- of Happiness Studies, 3: 71-92.
managed-services-perspective.pdf; accessed 30 43. Scheier, M. F., Carver, C. S., Bridges, M. W. 1994.
September 2016. Distinguishing Optimism from Neuroticism
(and trait anxiety, self-mastery, and self-esteem):
31. Rich, B. L., Lepine, J. A., Crawford, E. R. 2010.
A Re-Evaluation of the Life Orientation Test.
Job engagement: Antecedents and effects on job
Journal of Personality and Social Psychology, 67:
performance. Academy of Management Journal,
1063-1078.
53: 617-635.
44. Schumann, M. 2010. The New Rules of
32. Robbins, S. P., Judge, T. A., Sanghi, S. 2011.
Employee Engagement: How Business Leaders
Essentials of Organizational Behaviour, 10th
Must Reinvent How they Connect. Management
edition. New Delhi: Dorling Kindersley (India)
Today, 28(1): 44-45.
Pvt. Ltd.
45. Schwalbe, M. L. 1985. Autonomy in work and
33. Rosenberg, M. 1965. Society and the adolescent
self-esteem. The Sociological Quarterly, 26(4): 519-
self-image. Princeton, NJ: Princeton University
535.
Press.
46. Schwarzer, R., Jerusalem, M. 1995. Generalized
34. Saks, A. M. 2006. Antecedents and
Self-Efficacy Scale. In J. Weinman, J., Wright, S.,
consequences of employee engagement. Journal
Johnston, M. (eds.), Measures in health psychology:
of Managerial Psychology, 21: 600-619.
A user’s portfolio. Causal and control beliefs.
35. Sarason, I. G., Sarason, B. R., Shearin, E. N., Windsor, UK: NFER-NELSON.
Plerce, G. R. 1987. A Brief measurement of
47. Sheldon, K. M. 1995. Creativity and Self-
social support: Practical and Theoretical
Determination in Personality. Creativity Research
Implications. Journal of Social and Personal
Journal, 8: 61-72.
Relationships, 4: 497-510.
48. Sheldon, K. M., Ryan, R. M., Reis, H. 1996. What
36. Schaufeli, W. B. 2012. Work Engagement. What
makes for a Good Day? Competence and
Do We Know and Where Do We Go? Romanian
Autonomy in the Day and in the Person.
Journal of Applied Psychology, 14(1): 3-10.
Personality and Social Psychology Bulletin, 22:
37. Schaufeli, W. B. 2014. What is engagement? In 1270-1279.
Truss, C., Delbridge, R., Alfes, K., Shantz, A.,
49. Shimazu, A., Miyanaka, D., Schaufeli, W. B.
Soane, E. (eds.), Employee Engagement in Theory
2010. Work Engagement from a Cultural
and Practice. New York: Routledge.
Perspective. In Albrecht, S. L. (ed.) Handbook of
38. Schaufeli, W. B., Bakker, A. B. 2004a. UWES – employee engagement perspectives, issues, research
Utrecht Work Engagement Scale: Preliminary and practice. Cheltenham, UK: Edward Elgar
Manual. Version 1.1 (December 2004). Utrecht: Publishing Limited.
19
303
50. Tavakol, M., Dennick, R. 2011. Making Sense of
Cronbach‘s Alpha. International Journal of Medical
Education, 2: 53-55.
51. Van Dierendonk, D., Haynes, C., Borrill, C.,

Stide, C. 2004. Leadership Behavior and
Subordinate Well-Being. Journal of Occupational
Health Psychology, 9(2): 165–175.
52. Wagnild, G. M. 2009. The Resilience Scale User‘s

Guide for the US English Version of the
Resilience Scale and the 14-Item Resilience Scale
(RS-14). The Resilience Center, Montana.
Available at:
http://ja.cuyahogacounty.us/pdf_ja/en-
us/defendingchildhood/drcharlesfigley-
scoring-scalesheets.pdf; accessed 15 September
2016.
53. Wollard, K. K., Shuck, B. 2011. Antecedents to

Employee Engagement: A Structured Review of
the Literature. Advances in Developing Human
Resources. 13(4): 429–446.
54. Xanthopoulou, D., Bakker, A. B., Demerouti, E.,

Schaufeli, W.B. 2007. The Role of Personal
Resources in the Job Demands-Resources Model.
International Journal of Stress Management, 14:
121-141.
55. Xanthopoulou, D., Bakker, A. B., Demerouti, E.,

Schaufeli, W.B. 2009. Reciprocal Relationships
between Job Resources, Personal Resources, and
Work Engagement. Journal of Vocational Behavior,
74: 235–244.
56. Xanthopoulou, D., Bakker, A. B., Fischbach, A.

2013. Work Engagement Among Employees
Facing Emotional Demands The Role of Personal
Resources. Journal of Personnel Psychology, 12(2):
74–84.
57. Zikmund, W. G., Babin, B. J., Carr, J. C.,

Adhikari, A., Griffin, M. 2013. Business Research
20
304
The Structure of Land Prices in Colombo and Its

Implications
P. I. A. Gomes and R. K. Hashan
Abstract: Land prices could vary linearly or non-linearly due to several factors such as distance to
the regional and zonal commercial centres and major roads, access road width, retail facilities, health
facilities, schools and so forth. Whether these factors really impact the prices and if to what extent is
unknown. Data for several influencing factors were collected from over 300 land plots on sale from
Homagama, Maharagama, Malabe and Moratuwa for the period 2016 to early 2017. Data collection
was accomplished through printed and online real estate advertisements and direct contact with the
property agents/owners. Data were analyzed using a combination of bivariate and multivariate
analyses. Distance to Colombo and land prices showed a strong correlation in Homagama, where
closer to Colombo, prices increased. All other suburbs prices were rather independent of the distance
to Colombo and showed relative maxima or minima points indicating the impact of within-the-suburb
factors. It was the case for variables that represented socio-economic factors such as distance to nearest
school, hospital, retail facilities and so forth. Access road width was the only variable that showed
uniformity in its impact on prices of all suburbs. Similar results (on important variables and the fact
that Homagama’s pricing behaves differently) were evident from the linear multiple regression, and
detrended correspondence analyses. Other than indicating what factors should be given more
prominence when selecting a land, the results clearly indicated a time lag of development in certain
closely located suburbs. The results of this study could be important to the government and
individuals in deciding their policies and investment decisions, respectively.
Keywords: Colombo suburbs; Distance; Land prices; Major roads
1. Introduction or location [3]. The research pertaining to the

land prices and influencing factors are not as
In traditional societies, the land was considered common as other real estate components such
as a personal/family property and ownership as apartments. However, past quantitative and
was passed from generation to generation, and qualitative research suggests some important
no conversion to monetary terms was factors that may govern land prices in large
observed. The land is a finite resource with geographical areas. Those research have
highly increasing demand. In contrast, the mainly targeted the differentiation and
complementary product of lands, buildings analyses of land prices with land use (e.g. [4]).
due to vertical phasing (thus supply is not Sri Lanka, with the end of civil conflict, is
limited as lands) may not necessarily have a experiencing a booming economy, and an
high demand. Demand for a land could be well expanding real estate market. On one side,
understood by its market price. According to Colombo municipal areas undergo a rapid
Ai [1], the price is the best expression to growth in the condominium market [5]. In
compare land values in a functional market. tandem, Colombo suburbs came up with many
The price of land mainly depends on real estate projects. Obviously, most of the
accessibility to present or future services and lands for sale are from suburbs that contain
its production potential (i.e. ability to use the bare lands (e.g. Homagama, Athurugiriya,
land to produce goods and services, therein by Malabe), and are relatively less populated.
to generate revenue or profits). There is a basic Nevertheless, suburbs which are already
difference between the value and price of a populated, such a
land [2]. Value prescribes an actual benefit of
the land in relation to other similar lands. The Eng. (Dr.) P. I. A. Gomes, BSc(Eng)Hons.,
actual benefits may not necessarily limited to MSc, PhD(Japan), MIE (Sri Lanka), CEng.
Senior Lecturer, Department of Civil
direct monetary benefits, but also includes
Engineering, Sri Lanka Institute of Information
intrinsic and bequest values. Price (market) of a
Technology
land prescribes what a land might be sold for at
R. K. Hashan, BSc(Eng)Hons., Research
a specific period of time. Firstly, land price
Assistant, Department of Civil Engineering, Sri
heavily relies upon its geographical placement
Lanka Institute of Information Technology
305
Moratuwa and Maharagama, too have lands to traditional architectural Vasthu Vidya design.
be sold (e.g. brownfields, old houses), but in However, after preliminary investigations, it
relatively less numbers. was observed certain variables were rather
unresponsive to the land price (e.g. shape of the
This research aims to quantitatively analyse the land, ground water level, etc.).
importance of some factors that are often Unresponsiveness was decided based on the
considered to be vital land transactions. Should strength of coefficient of correlation on a bi-
be noted, it is not easy to give conclusive variate platform. The variables that were
judgement on the behaviour of a variable on considered for the comprehensive data
land prices. As an example, distance to the collection and analyses included: land plot size;
nearest major (e.g. A class) road may have distance to Colombo, distance to the nearest
contrasting impacts on land prices. Noise town (economic center); distance to the nearest
related issues may indicate lands closer to the major road (e.g. A class); access road width;
road to be less expensive, on the other hand, distance to health care and school facilities, and
ease of access may lead to a higher price. Also, domestic plot ratio.
the behaviour of a particular variable could be
changed or influenced by another variable(s). From each suburb, 60-70 samples were
Therefore, standard models or theories will not collected (i.e. 60-70 different locations, and
help in exploring a price and factor based some locations were land sale projects with
study. Hence, using up to date real data several lands, and considered as a single data
collected with data modelling is necessary. point). Data were collected by contacting land
sales advertised (printed and online) by
different agencies, and individuals (owners). It
should be noted all lands were visited by a
2. Materials and Methods
member of the research team, and details on
access road width etc. were taken by in person.
2.1 Study area
Other details such as distance to nearest
services were obtained from aerial maps.
Data from land plots in Moratuwa,
Maharagama, Homagama, and Malabe were The price used in the model is not necessarily
collected during 2016 and early 2017. Other
the advertised price, but the possible realized
than Malabe all other suburbs (study areas) (i.e., actual even if its off the records) price at a
were bound by the respective local authority
time of the deed transfer. Price shown in the
limit (Malabe is part of Kaduwella deed is in many cases the minimum
Municipality). These four areas have different
government valuation, and if used will have
socio-economic status quos. Moratuwa and
wrong implications. Land sales by real estate
Maharagma are two well recognized human
agencies, the advertised price and realized price
settlements since several decades to a few
were the same in almost all cases. However, the
centuries back (i.e. residential land use is high). prices of land sales by individuals did change.
Certain areas of Malabe had settlements since a
We assume the prices obtained were correct as
few decades back, but it was only recently the assurance was given to keep the information
mass settlements started. Homagama, as per
anonymous and the respondents were
Colombo district standards could be considered confident that the data would be used merely
semi to un-urbanized. Therefore, it was
for an academic research. It is assumed, any
expected these four areas to cover a wide range
misinformation (unavoidable in this kind of
of socio-economic aspects, and are
research) is minor and proportionate for all
representative Colombo suburbs.
suburbs.
2.2 Variables (factors) considered and
2.3 Data analysis
sampling details
Bivariate plots were used to check the
A survey conducted for 225 respondents helped
relationships different variables make with the
to shortlist 20 key variables that they
land price (per perch). Then simple multiple
considered important when buying a land.
linear regression
These variables are widely mentioned in land
sale market (including mainstream literature)
and some are considered important in
306
was used to model the price and influencing settlements ready to be sold). It could also be
factors (variables). All variables were checked due to some short comings of not able to
for multi collinearity. First, multiple linear identify some other lands for sale during the
regression analyses were conducted excluding data collection. In Moratuwa, most of the lands
variables with > 10 % variation inflation factor. had old houses (or lands with demolished old
In this step, the variable pairs/groups with houses). Homagama, most of the lands were
multicollinearity were identified and green-field (virgin) lands (i.e., no buildings
accordingly some variables were excluded were constructed before).
before the next modelling. Simple multiple
linear regression was repeated again until all Colombo, the commercial capital (also the
variables modelled were < 5 % variation former official capital) has reputed schools, a
inflation factor. All cases a P < 0.05 was large number of government agencies and
considered as the threshold for significance. commercial entities. It has the most important
SPSS statistical package (IBM V21.0) was used transportation hubs. Colombo has a resident
for the statistical analyses. population of about 0.6 million at a given time.
The floating population (non-resident or
Prices have been arranged in ranges temporary) is about 0.5 million. Because of
considering price and number of lands these reasons lands close to Colombo are
available (e.g. Rs. 0 to Rs. 50,000 per perch), and expensive. When prices were analyzed taking
the respective midpoint (i.e., in this case, Rs. average land prices of suburbs such as
25,000) will assume to represent all values of Dehiwala, Mount Lavinia, Ratmalana and
that range, and the averages of the respective Moratuwa, along the Galle road, this was
dependent variables (e.g. distance to Colombo) clearly evident (data not shown). Therefore, it
were calculated accordingly. was assumed that the distance to Colombo
could be a decisive factor governing the land
3. Results and Discussion price within a suburb as shown in Figure 1.
Equations 1-4 show power functions and the
3.1 Overview of lands sampled and Distance respective coefficient of correlations (R2)
to Colombo on Land Prices developed for each suburb. Homagama clearly
showed a distance dependent price variation,
where closer to Colombo prices were high.
Moratuwa too showed a similar, but a very
weak relationship and the variation was rather
sinusoid. These two contrasting views could be
well understood considering the connectivity to
Colombo. Moratuwa enjoys good connectivity
to Colombo via road (A2), and railway. Both
these modes are well catered by public and
private vehicles/transport modes (also there
are several railway stations within Moratuwa).
In addition, road width and number of rail
tracks are satisfactory. In case of Homagama,
the connectivity is not so good, and the train
services though scheduled, the number of trains
are a few. The data samples managed to collect
from Malabe and Maharagama were not very
Figure 1 – Price variation with distance to suitable to derive a relationship between
distance to Colombo and prices, as many lands
Colombo
had the same distance. However, even with
Many lands that were sampled in Malabe and more or less the same distance, in both suburbs,
Maharagama had more or less the same especially Maharagama showed a huge price
distance to Colombo. As an example, all difference. Therefore, the macroscopic view
sampled land plots in Malabe was about 15 km that prices closer to Colombo is high is
from Colombo. One reason is the shape and the conclusively true, only for Homagama.
orientation of the suburb. Also, it could be due
to localized reasons (e.g. brownfields/old
307
1 --- 3.3 Availability of socio-economic facilities
(1) (retail facilities, schools, medical centres), and
building foot print on land prices
---
(2)
------(3)
-----(4)
3.2 Impact of distance to the nearest major
road on land prices
Figure 3 – Price variation with distance to

nearest school

nearest major road
Homagama and Malabe showed strong

negative correlations between the distance to
the major road and price (Figure 2), as greater
the distance lesser the price. The other two
suburbs, Moratuwa and Maharagama, showed
a quadratic relationship, with a peak (for
Moratuwa the peak was left skewed, i.e.
towards low price side). Therefore, in these two nearest school
suburbs, certain areas showed to have optimum
choices of land. This meant, the price and the The relationships between nearest retail and
distance to nearest road were distorted by medical facilities with the price were similar to
different factors, as such certain land plots the relationship between nearest school and
irrespective of remoteness to a major road price (Figure 3). Only Homagama showed a
showed relatively high prices. The less clear linear relationship (close the socio-
decisiveness of distance to the nearest road in
economic facility greater the price). All other
Moratuwa and Maharagama could be further
explained due to the availability of widespread suburbs the prices were independent of such
(alternative) road network. facilities.
Building foot print and price also showed all

suburbs other than Homagama, to be
independent (Figure 4), and the relationships
were same as with the socio-economic facilities.
308
The main reason for this is due to the centre; and X3 is the distance to the nearest
centralized nature of facilities available in major road.
Homagama and otherwise in Moratuwa and
Maharagama. Importantly, same as in the
relationship with distance to the nearest major
road, these two suburbs showed optimum
locations. This in a way indicated the
accessibility and capacity to satisfy certain
needs outside their municipal boundaries (e.g.
considerable number of students attend to
schools in Colombo).
3.4 Effect of access road width on land prices
Figure 6 shows the detrended correspondence

analysis scatter plot. Samples (data points) with
similar responses are grouped together. The
scatter plot is cluttered and congested due to
the inclusion of large number of data points.
Therefore, small axis lengths were used to
identify the ownership of the data point.
Circled area 1 mainly consisted of Maharagama
samples, whereas 2 mainly consisted of
Moratuwa. Malabe samples were distributed
among these two encircled areas. However,
most of the Homagama samples were located
outside.
Figure 5 – Price variation with the access
road width 3.6 Land prices, influential factors and
The access road width was the only variable government policies
that showed a similar relationship across all
Considering the polarization of suburbs like
suburbs, indicating the common perspective of
Maharagama and Moratuwa, and clear linear
buyers, that the access road width is an
relationship Homagama lands show between
important factor of a land (Figure 5). Perhaps,
price and the respective dependent variables, it
this could be due to convenience in day to day
was obvious Homagama has a different socio-
activities and importantly ease of getting bank
economic signature. This is particularly
loan facilities.
important considering the government policies
3.5 Relating price with variables on a of hosting several IT and higher educational
multivariable approach institutes there. Still in the Sri Lankan context
startups and also higher education
The simple multiple linear regression analyses students/staff prefer domiciling close to major
conducted validated the results of bivariate socio-economic centers. Improved
analyses (Figure 1 to 5) as well as pave the way infrastructure that connects well with Colombo
to identify the significant (P < 0.05) and would make lands in areas like Homagama on
influencing (R2 > 0.4) variables. Equations (5) to par with well-sorted places like Moratuwa and
(8) show the simple multiple linear regression Maharagama.
equations. Where, X1 is the access road width;
X2 is the distance to nearest socio economic
309
considering the relatively weak to moderate
strength of correlations of some suburbs. This
implies there are factors other than the factors
that was considered, also impact the price. As
an example real estate supply could be
controlled by a group of
companies/individuals result in a manipulated
price.
Acknowledgement
Author wishes to acknowledge the two
A
Axis 2
anonymous reviewers for their excellent

xi review, and those who helped in data
s collection.
1
References
1. Ai, P., “Residential land value modelling,”

MSc Thesis, International institute for Geo-
information Science and Earth Observation,
Enschede, The Netherlands 2005.
2
2. Verheye, W. H., (Ed.). (2009). Land use, land
cover and soil sciences. EOLSS Publ..
Axis 1
3. Guidry, K., Shilling, J. D., and Sirmans, C. F.,
“Land-Use Controls, Natural Restrictions ,
Figure 6 –Detrended correspondence analysis and Urban Residential Land Prices,” vol. 29,
scatter plot for all suburbs no. 2, 1999.
Sri Lanka with urbanization should think of 4. Waddell, P., “Modeling urban deveopment
relaxing its bank rules and policies on for land use, transportation and
environmental planning,” APA J., vol. 68, no.
mortgage. Banks consider access road width as
3, pp. 297–314, 2002.
an important factor for loan approvals. As most
suburbs are well connected by public transport. 5. Ariyawansa, R. G., and Gunawardhana, T.,
Also, on call taxi services are available and “Guidelines for Real Estate Research and
resident frequently use those, therefore, it is Case Study Analysis,” Department of Estate
Management and Valuation, University of Sri
rather unnecessary to deprive loan facilities
Jayewardenepura, 2016
based on the access road width. This would
decrease the land prices to a certain extent, as
well as will help to optimize the use of
available land.
5. Conclusion and Future Works
This research observed a suburb dependent

land pricing as its main finding. In the larger
picture Homgama showed a different land
price signature to the other three suburbs. It is
recommended to study the historical variation
of land prices in these suburbs, and fine tune
the results with the incorporation of more data
and variables. This is particularly important
310
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[311 316],2017
2017
The Institution
Lanka
Knowledge Management System for Mobile

Telecommunication Sector in Sri Lanka
Jeewantha Nadeera, Shalanika Dayarathna and Eranda Ihalagedara
Abstract: Sri Lankan mobile telecommunication industry is a fiercely competitive market space
where five key mobile service companies operate. Most operators provide their services island wide;
therefore the coverage footprint is no longer a differentiating factor. On the other hand, knowledge is
a tool, which companies can use to provide a better service quality, which is now in demand with
customers.
In modern context with a dynamic environment, it is crucial to have a proper knowledge management
system in the telecommunication field, because knowledge is the power for the success of many
telecommunication area businesses. Mobile communication is so volatile, and a new technological
revolution takes place in every five years and thus, all mobile telecommunication companies have to
upgrade their networks frequently because of such technological advancement over the years. On
other hand due to globalization and increasing opportunities for individuals, companies face a
challenge to retain best employees within the organization. Now the focus has shifted to keep the
knowledge within the organization even with the movement of skilled labour. When it comes to Sri
Lankan context, Engineers who pass out from leading universities came up with excellent ideas to
improve the performance of telecommunication services, optimize capital expenditure through
creative solutions, and developed new business opportunities by combining unrelated technologies.
However, they might change their job in telecommunication discipline, migrate to other countries, or
upgrade into management field; this will diminish the transfer of tacit knowledge from generation to
generation. It is vital to have a proper tacit knowledge management system to capture and conserve
this valuable situational specific knowledge, and share such knowledge with future engineers. It will
help to create new technical solutions, enhance robustness of the telecommunication network areas,
and improve mobile customer experience.
The objective of this initiative was to study about knowledge management, select appropriate model,
and develop most appropriate knowledge management system for mobile telecommunication sector
in Sri Lanka.
Keywords: Knowledge, management, mobile, telecommunication, model
1. Introduction Even though mobile telecommunication sector

is a service oriented industry, people rarely
This paper is intended to provide the detailed have a platform to share their knowledge.
description of selecting an appropriate Therefore, employees do not get the
knowledge management model and how to opportunity to share knowledge even if they
develop and implement a knowledge wish to do so. Authors have identified one
management system for mobile reason for this as the lack of focus in
telecommunication sector. In addition, implementing knowledge sharing strategies by
suitability of knowledge management system management, mainly because management
for mobile telecommunication sector as well as focus is on competition with rival companies
barriers to implementing such system are Eng. Jeewantha Nadeera, B.Sc. Eng. Hons (University of
discussed. Moratuwa), Manager-Radio Access Operations, Mobitel
(Pvt) Ltd.
2. Suitability
Eng. (Ms.) Shalanika Dayarathna, B.Sc. Eng. Hons
(University of Moratuwa), CIMA Passed Finalist, Radio
Knowledge management systems are common Access Operations Engineer, Mobitel (Pvt) Ltd.
throughout the world, however in Sri Lanka it
is still a rising area. It is even rarer in mobile Eng. Eranda Ihalagedara, B.Sc.Eng.Hons(University of
Moratuwa), Radio Access Operations Engineer, Mobitel
telecommunication sector. (Pvt) Ltd.
311
rather than preserving and sharing of the Producing Technical experts of mobile
knowledge within. The diversity of employee telecommunication industry takes at least five
age is another factor identified to demotivate to seven years with exposure to various areas of
new knowledge sharing strategies. mobile telecommunication. If such an expert
migrate to a different country for higher
Mobile communication sector in Sri Lanka has education, or for a new carrier, or even move to
begun around late 20th century. Thus, every a different company within Sri Lanka, it would
mobile operator group consists of a wider influence the existing organization because they
employee base, which is likely to contain have spent a significant amount to elevate such
employees who are not familiar with computer an expert to the current level by providing
technology and its advanced features. As training and development opportunities.
knowledge management systems are more IT Hence, it is a duty and an obligation of the
based, sometime it could be more complex for expert to share that knowledge with other
employees from older generations. Therefore, employees. Also, if a technical expert move to a
they rarely use complex knowledge different job role, as an example a management
management systems, even if they are willing role, there might be reduced opportunities to
to share their knowledge with younger share such knowledge among the technical staff
generation. Even though, there are such because he/she has to perform various tasks
concerns regarding knowledge management in more related to management area. Thus, from
mobile telecommunication sector, it can be the organization point of view, there is a strong
considered as a highly effective and needed requirement to share such a person‟s
factor for this industry. knowledge, experiences, and new ideas.
Service industry depends on the knowledge Situational knowledge will only be captured at
and experience of its people and loss of highly the incident time. In most of the cases, this
experience or highly intelligent employees knowledge is limited to the people who have
could change the fate of a single company in experienced the incident, even they might have
this sector. Fate of mobile telecommunication forgotten about it after some time. In case a
sector is no different than that. similar incident occurs with a different set of
people, they will have to build up a solution
Currently, mobile telecommunication industry from the sketch. Hence, it is needed to update
in Sri Lanka is in its saturated level. Therefore, such an incident-related knowledge
in order to complete while retaining sufficient immediately to knowledge forum, where it has
profits, companies need to move towards cost happened. To achieve that objective, a proper
cutting strategies. Basic concepts in cost cutting knowledge management platform is required,
strategy are to improve efficiency and as it gives opportunities to share the
effectiveness using same resources, in this case knowledge.
employees. Each operator has several highly
efficient and experienced employees, who has 3. Implementation
less error rate and use shortcuts to complete
more work in less time. That is where Authors have successfully implemented a web
knowledge management system comes in based knowledge management system in one of
handy. Imagine how much company would be the mobile operators in Sri Lanka. They have
able to save if every employee start to work like first performed a detailed analysis of available
that. knowledge management techniques and their
suitability to mobile telecommunication sector.
Another restraining factor is time taken to find
some useful information or old documents. 3.1 Preliminary Research
Even after telecommunication sector moved Initially, research has been conducted on
towards more digital era, documents are rarely different types of knowledge management and
in a central location, thus making it harder and currently available knowledge management
time consuming to find important documents. systems and their features. By understanding
Knowledge management system runs in a types of knowledge management systems, it
single server and act as a central location to was possible to identify the type that matches
keep all knowledge related documents in one the requirement of the mobile
place. In addition, use of complex searching telecommunication industry.
algorithms will make this task much easier and
quicker.
312
Even though there are several knowledge knowledge focused in this paper is tacit
management systems such as, knowledge and explicit knowledge.
 expert systems – perform problem  Tacit Knowledge – the knowledge that is

diagnosis based on experience and pertaining to a
 groupware – facilitate collaboration specific context or incidents. This is difficult
between workers to be documented or transfer to another.
 document management system – share  Explicit Knowledge – the knowledge that
documents and enable search through can be easily put into verbal means or
documents more efficiently documented and hence easy to share.
 decision support system – present
information to users in a manner they can 3.3 Knowledge Management Model
make informed decisions easily This section explains the suggested knowledge
 database management system – store management model for mobile
collection of data telecommunication sector in Sri Lanka.
 simulation systems – model and test real Nonaka‟s knowledge management framework
world scenarios [1] has been used as the basic model in order to
develop a knowledge management model for
it has been identified that features of document Sri Lankan telecommunication sector.
management system, decision support system Leadership, knowledge sharing culture,
and database management system are required socialization, externalization, combination, and
for the knowledge management systems in internalization have been identified as the main
mobile telecommunication sector. Therefore, areas in this knowledge management
following key areas have been identified as framework developed for Sri Lankan
essential implementation steps in knowledge telecommunication sector [2].
management system by authors.
They have introduced this “leadership” as one
 Meta-search software – accepts a single of the key concepts. It has been identified that
query from users and return results from the people are willing to support and contribute
knowledge base. to knowledge management system if
 Knowledge mining software – allows query management (leader) introduces the knowledge
results to be sorted and clusters according management system. Thus, it is required the
to a set of predefined categories. leadership to initiate, implement, and motivate
 Automated categorization tools – apply of people to share the knowledge through a
both available data and incoming streams knowledge management system. Therefore,
of knowledge. following concepts were recommended by
 Knowledge warehouses – contain available authors as required leadership practices in
data with substantially enhances order to introduce a knowledge management
functionality, security. system.
 Forms-base knowledge contribution
schemes – end users can directly input their  Leaders shall create positive synergy
knowledge. among employees across the division
 Knowledge mapping software – pre-build  Encourage employee participation in
knowledge maps designed for specific decision making
tasks.  It is required to absorb uncertainty
 Knowledge directory software – automate currently encountered by employees in the
the process of finding people with specific organization
areas of expertise.
It is required to create a “knowledge sharing
3.2 Basic Concepts of Knowledge culture” in the organization to implement an
“Knowledge” has been defined in many ways effective and efficient knowledge management
by the philosophers since the ancient times. system. In order to form a knowledge sharing
Basically every version of the definition can be culture, management needs to encourage
simply put in to words as “knowledge is the employees to work together more effectively by
understanding of something through collaboration and sharing experiences. People
experience or education”. Further, the are willing to share knowledge, however, they
knowledge can be classified based on different do not have a platform to share that
dimensions. However, the classification of knowledge. In the proposed knowledge
313
management framework, the second core layer During the “externalization” process, it is
would be the knowledge sharing culture. This required to assess the method or process that
is because, in Sri Lankan cultural dimension supports converting tacit knowledge into
analysis, a high level of power distance is explicit knowledge. This is a very important
observed where people believe there is a stage of the knowledge management process
significant power level difference. Therefore, because it is the key factor that drive the
the knowledge management framework in Sri knowledge management system in sustainable
Lankan context and knowledge share culture development in the long run. Hence, it is
should be interconnected with leadership and recommended to introduce the externalization
rest of the knowledge management concepts as concept with following practices in to
in Figure 1. knowledge management framework for mobile
telecommunication sector in Sri Lanka.
 Convenient when accessing and updating

information
 Frequent updating an internal forum to
share new ideas, concepts, and trends
 Having brainstorming sessions when
confronted with a new problem or when a
critical decision has to be made
 Encourage to express their views, argue,
and evaluate new ideas
Figure 1 - Knowledge management
Framework foundation According to Nonaka, the “combination”
mode of knowledge conversion is „a process
Therefore, authors recommend following of assembling new and existing explicit
knowledge sharing culture practices. knowledge held by individuals into a
knowledge system‟ [1]. It is also a process of
 Management encourages sharing quality exchanging, sorting, adding, disseminating,
information sharing, and therefore reconfiguring, different
 Managing excessive competition among the bodies of explicit knowledge among
employees organizational members through documents,
 Allow constructive criticism to find optimal meetings, telephone conversations, and
solutions computerized communication methods.
Nonaka terms the knowledge created through a
“Socialization” refers to how people in the combination process as systemic knowledge. It
organization interact with each other. When it is required to combine various explicit
comes to knowledge management, it is knowledge with individual to single context.
important to have a good relationship and a Hence, it is recommended to introduce the
mutual understanding among organizational combination concept into knowledge
employees, as well as the interaction between management framework for mobile
subject experts with the rest of the employees in telecommunication sector in Sri Lanka with
the organization. Therefore, there is a strong practises such as,
requirement to introducing socialization as a
knowledge management platform for sharing  Combining information such as network
knowledge among people, thus these change history, network issue trouble
socialization practices have been shooting guides, and root cause analysis in
recommended. the knowledge management system
 Combining different individual‟s
 Providing a formal training program when knowledge
employees are recruited to the organization  High level of contribution from technical
 Frequent engagement of newly recruited experts
engineers and technical officers with
technical experts The “internalization” mode of knowledge
 Employees can easily reach a technical conversion is „a process of embodying explicit
expert in the organization knowledge into tacit knowledge‟ [1]. It uses
 Frequent informal team gathering existing knowledge and creates new knowledge
based on the existing knowledge. Generally
314
newly recruited engineers and technical officers searching time and increasing the ability to
access the internal knowledge management cater to specific search requirements. This pre-
system to obtain knowledge. This might be the build knowledge map is used to differentiate
reason for the lack of an integrated knowledge among divisions in the company as well as the
management system. However, organization knowledge categories.
employees are willing to share knowledge,
provided they have a system to share the Knowledge management system was
knowledge. Therefore, a strong requirement developed in order to cater the different
exists from management to introduce such a requirements in different levels of
system and provide leadership. As a management. For example, uploading
recommendation based on this finding, an documents are done by lower management and
organization should introduce knowledge below employees. Therefore, uploading option
management system with following practices. was given within each division. However, in
addition to the search option within each
 Information quality rating mechanism division, another search option is given for
 Easy search mechanism and up to date senior management, which can capture the
information knowledge of every division as a whole. This
 Structured way to retrieve information was enabled to ensure that system support both
 Encourage to build new knowledge based senior and junior level employees
on existing available knowledge requirements.
Finally, considering all factors such as Having a knowledge management system is

leadership, knowledge sharing culture, useless if the users can‟t rely on the system. In
Socialization, externalization, combination, and order for users to reply on the system, they
internalization authors have come up with a need to have a trust towards the content of the
suitable knowledge management framework documents in the systems. A rating system was
for mobile telecommunication sector (Figure 2). introduced to ensure this as shown in Figure 3.
Figure 3 – Rating System implemented within

Knowledge Management System.
In order to encourage people to use the system,

extra steps were need to be taken to ensure
Figure 2 - Knowledge management handling feasibility and ability to force the
Framework for mobile telecommunication system in to the mind of the entire employee
industry in Sri Lanka base in the initial stage. Therefore, system
should be able to operate easily. Authors have
achieved this using several data capturing
3.4 Methodology techniques including forms-base knowledge
Knowledge management system need to be contribution schemes and automated data
implemented in a central location, thus most capturing schemes. They have used several
suitable method was to develop a web based automated data capturing mechanisms to
system that runs in a single server. Therefore, capture performance and system oriented data
knowledge management system was developed while rely on user input for other sources of
using basic web developing languages such as documents.
PPH, CSS, and JS as well as web developing
tools such as XAMPP MySQL database. Some Another important factor is use of suitable data
open source templates [2, 3] were also used to mining software. Final output need to be user
achieve an enriching appearance. friendly and provide effective and efficient
Knowledge management system was searching for end users. Meta search software
developed according to a pre-build knowledge is used in combination with automated
map in order to ensure that the captured categorization tool to achieve this. Meta search
knowledge will be stored with categorization. specification used in the system is shown in
This improves the searching ability by reducing Figure 6.
315
4. https://themewagon.com/themes/free-bootstrap-
3-admin-dashboard-template/ , Visited 5th
December, 2016.
Figure 4 - Meta search specification used in

the Knowledge Management System
Finally, important requirement of any system is

feedback. Therefore, this knowledge
management system has provide the
opportunity for any user to provide their
feedback anonymously.
4. Results
This section discuss the results of the

implemented knowledge management system.
After conducting several training sessions, this

system was initially introduced to one division
within Mobitel, Sri Lanka. After several months,
it was introduced to several other divisions as
well as the next step. As authors continued to
introduce new systems within the Engineering
division, more and more people started to use
and rely on the knowledge management
system.
Currently, this system is being used in several

sub divisions in Mobitel Engineering division as
a platform to store and share knowledge among
divisions as well as within each divisions.
5. Conclusions
It is expected that in the future, this system can

be extended to entire engineering division.
Thus, entire Mobitel team can benefit from
sharing their knowledge while contributing to
Mobitel as well. Hence, this would lead to a
knowledge sharing culture within the
organization.
References
1. Nonaka, I., "A Dynamic Theory of Organizational
Knowledge Creation," Vols. Organization
Science, Vol. 5, No. 1 (Feb., 1994), pp. 14-37, p.
P207, 1994.
2. Nadeera, W. K. J. Knowledge Management

Framework for mobile telecommunication sector
in Sri Lanka, 2016.
3. http://www.marcellop.com/bootmetro/, Visited
14th July, 2016.
316
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[317 325],2017
2017
The Institution
Lanka
Effect of Technical & Non-Technical Competence on

the Job Performance of Private Sector Engineers in Sri
Lanka
With special reference to a group of large scale engineering companies
J.R. Ekanayake
Abstract: In any nation, engineers are expected to play a vital role in economic development
while confronting the prevalent globalized rivalry and hence there is no exception for Sri Lankan
engineers too. Majority of local engineers being employed by the private sector, their contribution for
national economic progress is pivotal. However, it is commonly argued that whether the local
engineers are competent enough to effectively contribute in aforesaid task and also it is observed that
the relevant literature is scarce and hence the knowledge base as applicable to local context is not rich.
The attempt of the paper is to enrich knowledge base of particular disciplinein the local context, by
exploring attributes and characteristics of competences and job performance sought in present-day
engineers and also to investigate their inter-relationship.
This paper will present comprehensive details along with the findings of a methodically performed
study based on a set of appropriate data (about competence & job performance) primarily collected by
administering a suitably designed questionnaire. The study consisted of the detail facts and figures of
both statistical and descriptive analysis. Ultimately, the paper presents the multiple dimensions of
technical competence, non-technical competence and job performance as applicable to engineers. The
findings of statistical analysis reveal that the average level of technical competence, non-technical
competence and job performance are moderate. Further, this study statistically demonstrates, strong
relationships in job performance vs technical competence, job performance vs non-technical
competence and job performance vs overall competence.
Keywords: Engineers, Technical Competence, Non-technical Competence, Job Performance,

Statistical Analysis
1 Introduction require both strong technical knowledge and

skills and an understanding of relevant
It is obvious that, in any nation including Sri environmental, social, economic and cultural
Lanka, engineers play a vital role in all contexts [5].
nationally important sectors; it may be
agriculture, industry or service. This significant Above phenomenon is also applicable to local
nature of engineers’ contribution for national engineers who are predominantly employed by
economies makes it much important to explore private sector organizations which operate in a
and study about the attributes of competences globalized business environment. Hence it is
sought in engineers and their effect on job obvious that contemporary Sri Lankan
performance. engineers requires not only technical excellence,
but also some additional, non-technical
With prevailing nature of rapidly advancing competences. Irrespective of the nature of local
technical environment and speedy establishment, as a result of the enduring
globalization, there exists a considerable phenomenon of globalization local companies
consensus that the modern engineering also confront a severe rivalry in their business.
profession requires not only technical
Eng. J. R.Ekanaykae, B.Sc. Eng. (Hons), MBA, C. Eng.,
excellence, but also some additional, non-
MIE(SL), MASME (USA), MIET (UK), Senior Lecturer in
technical competences.The problems faced by Mechanical Engineering, IESL College of Engineering.
engineers today are increasingly complex and
317
The companies who try to achieve competitive group of large scale engineering companies and
advantage in an environment with continuing thereby to make recommendations to enhance
and enhancing competition, certainly demand the individuals’ job performance and
their employees (engineers) to be competent in organizational performance.
various facets in addition to essential technical
competencies. But, it is argued that whether 2 Research Questions and
local engineers are competent enough to meet Objectives
this demand of the industry.
The research questions and objectives are as
As written by Wijesiri[11], the overwhelming below.
consensus among employers is that, “too many 1. What are the attributes and characteristics of
graduates lack critical-thinking skills and the the concept of competence and Job
ability to communicate effectively, solve performance in relation to engineering
problems creatively, work collaboratively and profession?
adapt to changing priorities. In addition to 2. How does competence influence job
these "soft skill" deficits, employers are also performance of engineers?
finding that young people lack the technical, or 3. What are the recommendations to improve
"hard", skills associated with specific jobs”. job performance of engineers?
Based on the above research questions,
So, engineering graduates being a subset of
following research objectives are formulated.
graduates produced by the system of local
1. To study and clarify the attributes and
universities, with equal standards and funding,
characteristics of the concept of competence
governed by the same entity (i.e. by the
and job performance in relation to
University Grants Commission (UGC) under
engineering profession.
the Ministry of Higher Education and
2. To analyse the relationship between
Highways), engineering graduates can’t be
competence and job performance
expected to have an exception.
3. To make recommendations to improve job
Based on author’s 1st hand experience, it is performance of engineers.
understood that even at a highly technical
working environment as in the case of his
previous employment, also with sound
3 Methodology
technical competences, an engineer cannot
This study adopts a commonly used scientific
become successful and climb up on the ladder
research method, where following hypotheses
in his career, unless he or she owns sufficient
with relevant to the topic of interest are made
level of non-technical competences.
based on the supporting evidences of available
Further it is evident that, under prevailing literature.
economic situation and job market in Sri Lanka,
Hypothesis 1: High level of technical competence
the majority of graduates including engineering
results in high job performance.
graduates are employed by the private sector
[1].So it is justified to do a research on this issue Hypothesis 2: High level of non-technical
and it can be considered as a pertinent issue to competence results in high job performance.
the majority of private sector engineers
graduated predominantly from local Hypothesis 3: High level of technical competence
government universities. together with high level of non-technical competence
will result in high job performance.
As implied by the title itself, this study intends
to identify the attributes of competence (i.e. the Subsequently those hypotheses are tested and
dimensions of technical and non-technical proved by statistical analysis of gathered data
competence) and their influence on the job relevant to the study.
performance as applicable to Sri Lankan private
sector engineers with special reference to a
318
The target population of this research study is special interest to engineers and engineering
the engineers of local private sector companies. profession.
A random sample consisting of fifty engineers
employed in several private sector engineering Based on the comprehensive literature survey it
companies (includes large scale manufacturing, could be identified that engineering is no
design & engineering, construction, operations longer a matter of just engineering, as the
& maintenance and services sector companies) profession of engineering and the roles of
was selected from the entire population and the engineers have changed rapidly over the past
relevant data were collected through this few decades. The need to educate “holistic
sample by administering an appropriately engineers” or “global engineers” is widely
designed questionnaire. acknowledged. Previous studies suggest that
present days’ engineers need to be competent
The questionnaire consisted of three main in both technical and non-technical
sections to gather data about demographic competences in order to be successful in their
information, competence and job performance. job roles and to progress in career.
Demographic data includes, age, sex,
qualifications, designation, level of While taking various explanations and
employment and experience. Competence data definitions about competence in to account,
were gathered under the two categories of measurable dimensions of technical
Technical and Non-Technical competence. Job competences were derived from the studies of
Performance data were collected under the sub OECD [8] and Nguyen[7]. They are;
categories of Task Performance, Contextual Competence and updated knowledge in
Performance and Adaptive Performance. computer science and technology, Competence
Except the demographic data, all others were in mathematics skills and problem-solving
collected through closed ended statements skills, Competence in Engineering & Science
(representing measurable attributes) which fundamentals Competence in Engineering
were to be responded with a degree of Practice, Competence in invention and
agreement measured with 5-point Likert’s scale development of new products/solutions.
of 1-5 from every employee’s point of view
Similarly, the measurable dimensions of non-
Finally, the analysis was carried out based on technical competences were derived from the
hypothesis deductive method using state of the study of Heuristic Model of Engineering
art software “IBM SPSS Statistics Ver.24”. For Nontechnical Competences [5] and those are;
that purpose of analysis, the average level of Professional ethics, Personal competences
competence and job performance of each Interpersonal competences, Leadership,
respondent were calculated. Further, to find out management, and administrative competences,
the behaviour of dependent variable according Innovation and entrepreneurial competences
to the movement of independent variables, two and Law and legal system competences.
statistical methods namely the Correlation
Measurable dimensions of Job Performance
Analysis and Linear Regression Analysis were
were identified from the study of Linda
used.
Koopmans [4]and Sabine Sonnentag J. V.[9].
These dimensions are categorized under three
4 Literature Review
main dimensions and listed below namely;
There have been numerous researches done by Task Performance, Contextual Performance and
scholars to identify and define the Adaptive Performance. Each of these
“competence” (or “competency”), “job dimensions are consist of multi-dimensions.
performance” and to establish a relationship
In various studies, around the world they
between them. In this study the literature
highlight about the strong relationship between
review was narrowed down to the journals and
competences and job performance of engineers.
publications which focus on competences and
With the speedy globalization and resultant
job performance dimensions particularly with a
319
severe rivalry faced by contemporary Table 1 - Cronbach's Alpha - Questions to
organizations, there is a serious consensus that check Technical Competence
engineers need both technical and non-
Reliability Statistics
technical competences in almost equal
Cronbach's
proportions. Mare Teichmannstates that “the
Alpha Based on
“new century engineer” is expected to be
Cronbach's Standardized
technically competent, globally sophisticated,
Alpha Items N of Items
culturally aware, innovative and
entrepreneurial, nimble, flexible and mobile” .650 .664 5
[5].
Accordingly, there exists considerable Table 2 - Cronbach's Alpha - Questions to

consensus that the modern engineering check Non-Technical Competence
profession requires not only technical
Reliability Statistics
excellence, but also some additional, non-
Cronbach's
technical competences.
Alpha Based on
Cronbach's Standardized
5 Analysis and Discussion
Alpha Items N of Items
As the first step in the process of the analysis, .704 .701 6
the reliability analysis of the questionnaire was
carried out by computing “Cronbach’s
Table 3 - Cronbach's Alpha - Questions to
Alpha”.The Demographic Data, Competence
check Technical and Non-Technical
Data and Job Performance Data which were
Competence together
collected by administering the questionnaire
were subjected to a broad analysis which Reliability Statistics
consist of both descriptive and statistical Cronbach's
analysis.The details of the analysis are Alpha Based on
comprehensively presented in following Cronbach's Standardized
paragraphs. Alpha Items N of Items
5.1 Reliability Analysis of Questionnaire .811 .813 11
The reliability analysis of the questionnaire was
done by computing the “Cronbach’s alpha”.
Table 4 - Cronbach's Alpha - Questions to
The “Cronbach’s alpha” reveals the reliability,
check Job Performance
or internal consistency, of a set of questions
designed to check anyparticular concept or
principle[6]. Here, the reliability of any given Reliability Statistics
measurement refers to the extent to which it is a Cronbach's
consistent measure of a concept. Questionnaire Alpha Based on
used in this study consisted of main three Cronbach's Standardized
sections to collect 1) Demographic information, Alpha Items N of Items
2) Data about competence and 3) Data about job .869 .865 17
performance. Among these three sections,
section 2) and 3)were used to gather data about
variables considered in this study and hence, As presented in above tables, for all sets of
the questions only in these sections are questions Cronbach’s Alpha was found to be
associated with answers on Likert scale. between 0.65 and 0.90. Therefore, it was found
that the entire questionnaire is reliable enough
to serve the desired purpose of data
gathering[2].
320
5.2 Descriptive Analysis Table 5 - Descriptive Statistics of Technical
As per the findings of descriptive analysis, it Competence and Job Performance
can be resolved that the majority of private
sector engineering jobs are occupied by middle
Descriptive Statistics
aged engineers. Hence this study will
contribute significantly to enhance the job Std.
performance of young &middle-aged engineers Mean Deviation N
who are considered to be more productive in Technical 3.4520 .38926 50
science and engineering tasks [3], and thereby Competence
to enhance the organizational performance and Job Performance 3.3506 .37884 50
economic contribution.
Further the descriptive analysis suggests that

Table 6 - Descriptive Statistics of Non-
the majority of engineers in targeted population
Technical Competence and Job Performance
are only BSc. Eng. qualified while a very less
are qualified with post graduate and non- Descriptive Statistics
technical qualifications. Further it was revealed Mean Std. Deviation N
that the, engineers’ turnover rate is less. This Non-technical 3.1630 .42769 50
can be a result of good remuneration they are Competence
being offered by the present employer. When Job Performance 3.3506 .37884 50
they are remunerated well by the present
employer, they don’t tend to look out for new
jobs and therefore their interest on post As summarized in above both tables 5 and 6, it
graduate and non-technical qualifications could is evident that the Technical Competence, Non-
be very less. Ultimately all these matters get Technical Competenceand the Job Performance
linked together and results in poor career of the respondents are at Moderate Level (i.e.
progress. because 2.5 ≤ Mean ≤ 3.5 and SD < 0.43)[10].
5.3 Statistical Analysis 5.3.2 Correlation Analysis

Univariate Analysis was carried out to find out Correlation analysis was carried out in order to
the average levels of competence and job determine what kind of relationship exists
performance. Co-relation and regression between variables. The results of correlation
analysis were carried out in order toformulate analysis of “Technical competence (TC)” vs
the relationship between variables and test the “Job performance (JP)” and “Non-technical
validity of collected data, investigate the competence (NTC)” vs “Job performance(JP)”
significance and test the hypothesis. are as below.
5.3.1 Univariate Analysis Table 7 - Correlation of Technical Competence

Univariate analysis, revealed that in average and Job Performance
engineers’ technical competence, non-technical
competence and also the job performance are Correlations
only at moderate level. This also can be a result
TC JP
of lack of post graduate and non-technical
TC Pearson Correlation 1 .769**
qualifications and finally this also will badly
affect in long-term career progress while Sig. (2-tailed) .000
limiting engineers only to lower and middle N 50 50
management levels. JP Pearson Correlation .769** 1
Sig. (2-tailed) .000
N 50 50
**. Correlation is significant at the 0.01 level
(2-tailed).
321
Table 8 - Correlation of Non-Technical Case B: Non-Technical Competence (NTC) vs
Competence and Job Performance Job Performance (JP)
JP = βo + β1 (NTC) + error … (2)
Correlations Case C: Technical Competence (TC) together
NTC JP with Non-Technical Competence (NTC) vs Job
NTC Pearson Correlation 1 .756** Performance (JP)
Sig. (2-tailed) .000 JP = γo + γ1 (TC) + γ2 (NTC) + error ... (3)
N 50 50 By ANOVA test for the linear regression it was
JP Pearson Correlation .756** 1 found that, there is a relationship between the
Sig. (2-tailed) .000 dependent variables and the independent
N 50 50 variables in all above cases. Since the
**. Correlation is significant at the 0.01 level probability of the F statistic (p<0.000) is less
(2-tailed). than to the level of significance (0.01), the
research hypothesis that there is a relationship
According to above presented summarized between the variables is supported and null
results of correlation analysis, it is obvious that, hypothesis that coefficient of correlation is
a strong positive correlation exists between equal to 0 is rejected.
Technical Competence (TC) and Job
Performance (JP) with r = 0.769, p < 0.01. The coefficients;αo, α1, βo, β1γo, γ1andγ2 of
Similarly, it is observed that, a strong positive above equations were found using regression
correlation exists between Non-Technical analysis.
Competence (NTC) and Job Performance (JP)
Hence above equation (1) can be re-written as
with r = 0.756, p < 0.01 (r = 0.5 to 1.0; A Strong
follows.
Positive Relationship as referred to V.J.Delima,
[10]). Since both technical and non-technical JP = 0.767 + 0.748 (TC) … (4)
competences are positively correlated with job
performance and significant at 0.01 level (2- The coefficient for the independent variable
tailed), it can be resolved that, the linear “Technical Competence (TC)" found to be 0.748,
relationship in the sample data is statistically indicating a positive relationship with the
significant and therefore the relationship dependent variable “Job Performance (JP)”.
between these variables can be fairly modelled Thus, higher numeric values for the
by linear regression. These results conclude independent variable “Technical Competence
that, enhancement of either technical or non- (TC)" are associated also with higher numeric
technical competences will result in higher job values for the dependent variable “Job
performance of engineers. In comparison to the Performance (JP)”.
level of influence by non-technical competence
on job performance, it can be identified that the Therefore, the statement of hypothesis, high
technical competence influences slightly more technical competence results in high job
on the job performance of engineers. This is performance is proved.
because of the slightly high correlation
Table 9 - Model Summary - Case A
coefficient between technical competence and
job performance than the that of non-technical Model Summary
competence and job performance. Adjusted Std. Error of
Model R R Square R Square the Estimate
5.3.3 Regression Analysis
1 .769a .591 .583 .24470
Research hypothesis was tested using the
a. Predictors: (Constant), Technical
regression analysis and following three
regression equations have been developed. Competence
As per model summary above, the coefficient of
Case A: Technical Competence (TC) vs Job determination (R2) of 0.591 illustrates that;
Performance (JP) 59.1% of the total variation in “Job
JP = αo + α1 (TC) + error … (1) Performance” can be explained by the linear
322
relationship between “Technical Performance” and that for the other independent variable
and “Job Performance” (as described by the “Non-technical Competence (NTC)” found to
regression equation (4) stated above). The be 0.380. This illustrates that there is a good
value of 0.591 obtained for coefficient of positive relationship of both independent
determination indicates that, the regression line variables with the dependent variable “Job
represents the data reasonably well. Performance”. This implies that, higher
numeric values for both independent variables
Similarly, above equation (2) can be re-written "Technical Competence" and “Non-technical
as follows. Competence” or either of them will result in
higher numeric values for the dependent
JP = 1.234 + 0.669 (NTC) … (5)
variable “Job Performance”. Furthermore, the
The coefficient for the independent variable higher coefficient (0.460) associated with
“Non-technical Competence (NTC) " found to Technical Competence implies that the effect of
be 0.669, indicating a positive relationship with Technical Competence on Job Performance is
the dependent variable “Job Performance (JP)”. higher than the that of Non-Technical
Hence, higher numeric values for the competence.
independent variable “Non-technical
The coefficient of determination (R2) of 0.687
Competence" are associated also with higher
shows that; 68.7% of the total variation in “Job
numeric values for the dependent variable “Job
Performance” can be explained by the linear
Performance”.
relationship as described by the regression
Therefore, the statement of hypothesis, high equation (6) stated above. The value of 0.687
non-technical competence results in high job obtained for coefficient of determination
performance is proved. indicates that, the regression line represents the
data reasonably well.
Table 10 - Model Summary - Case B
Therefore, the statement of hypothesis, both
Model Summary higher technical and non-technical competence
Adjusted Std. Error of results in high job performance is proved.
Model R R Square R Square the Estimate
1 .756a .571 .562 .25071 6 Conclusions and
a. Predictors: (Constant), Non-technical Recommendations
Competence
The findings of the descriptive analysis suggest
The coefficient of determination (R2) of 0.571 that, the majority of engineers are not qualified
shows that; 57.1% of the total variation in “Job with the post graduate and non-technical
Performance” can be explained by the linear qualifications and also they are not much
relationship between “Non-technical attentive of the career prospectus with a new
Performance” and “Job Performance” (as employer. In response to these findings it is
described by the regression equation (5) stated recommended for engineers to take up relevant
above). Further, the value of 0.571 obtained for post graduate and non-technical studies on
coefficient of determination indicates that, the time and to look out for better job positions
regression line represents the data reasonably with new employers and thereby to maximize
well. the opportunities of securing more senior
management positions.
Finally, above equation (3) can be re-written as
follows The results of statistical analysis (Univariate
analysis) revealed that, the average level of
JP = 0.562 + 0.460 (TC) + 0.380 (NTC … (6)
technical competence (Mean = 3.45), non-
The coefficient for the independent variable " technical competence (Mean = 3.16) and job
Technical Competence (TC) " found as 0.460, performance (Mean = 3.35) of private sector
engineers are moderate. Therefore, it is
323
required to further enhance both the technical am standing with respect to the level of
and non-technical competences and also the job technical competence and non-technical
performance of private sector engineers. competence. Then having done the self-
assessment with respect to competence, they
As per the statistically significant findings of shall concentrate on developing both
quantitative analysis (i.e. by Correlation and technical and non-technical competences
Regression Analysis), it is evident that high consistently and continuously.
competence results in high job performance of
engineers in local private sector. Further, the Acknowledgement
outcomes of comprehensive literature review
clarified that, the competence as applicable to Author wishes to thank Mr. Deepal
engineers consist of main two dimensions; BataduwaArachchi and Dr. Nalin Abeysekara
technical competence and non-technical for advising throughout the process of
competence. The comprehensively performed performing this study. Further the author
quantitative analysis revealed that both types of conveys heart felt gratitude to Dr. K. E. D.
competences are almost equally important to Sumanasiri for providing very much needed
achieve high level of job performance, but with encouragement and guidance during the
a slight higher importance of technical process of preparing this paper.
competence as applicable to engineering
profession. In affirmative to these key findings References
of quantitative analysis, following
1. Ariyawansa, R. G. (2008). Employability of
recommendations are made to intensify the Graduates of Sri Lankan Universities, Vol.2,
competences of local engineers. Hence those No.1. Sri Lankan Journal of Human Resource
recommendations will pave the path towards Management.
enriched job performance of engineers and
2. Goforth, C. (2015, November). Using and
thereby to uplift the overall performance of
Interpreting Cronbach’s Alpha. Retrieved from
private sector companies.
University of Virginia Library, Research Data
Services + Sciences:
 It is recommended to review and upgrade
http://data.library.virginia.edu/using-and-
the academic programmes of local
interpreting-cronbachs-alpha/
engineering universities to broaden the non-
technical content, while strengthening the 3. Institute of Medicine (US) Committee on the
technical core of course curriculum. Long-Run Macroeconomic Effects of the Aging
 It is suggested to introduce and conduct U.S. Population. (2012, December). Aging and the
Macroeconomy: Long-Term Implications of an Older
more and more education and training
Population. Retrieved from National Centre for
programmes by IESL at a subsidized price
Biotechnology Information, U. S. National
(or if possible free of charge for members) to
Library of Medicine:
enhance both technical and non-technical https://www.ncbi.nlm.nih.gov/books/NBK144
competence of local engineers. These 283/
programmes should be made attractive by
making the content opportune, globally 4. Linda Koopmans, M. C. (2011, August).
Conceptual Frameworks of Individual Work
cognizant, industry oriented and interactive
Performance; A Systematic Review. Journal of
(or student-centred learning type).
Occupational and Environmental Medicine ; Vol:53.
 It is proposed for local private organizations
to stress on appropriate competency 5. Mare Teichmann, V. P. (2013).A Heuristic Model
management models and to concentrate on of Non-technical Competences for Engineers.
competency-based approaches to enhance
6. Mohsen Tavakol, R. D. (2011). Making sense of
job performance of their engineers
Cronbach’s alpha. International Journal of Medical
(employees). Education.
 It is recommended for individual engineers
to make a self-assessment to check where I
324
7. Nguyen, D. Q. (1998). The Essential Skills and
Attributes of an Engineer: A Comparative Study
of Academics, Industry Personnel and
Engineering Students. Global J. of Engng. Educ.,
Vol. 2, No.1.
8. OECD. (2009). A Tuning-Ahelo Conceptual

Framework Of Expected/Desired Learning Outcomes
In Engineering.Organisation for Economic Co-
operation and Development.
9. Sabine Sonnentag, J. V. (2008). Job

Performance. In J. Barling, Micro Approaches
- SAGE Handbook of Organizatioanl Behaviour
- Vol.1 (pp. 427-447). Los Angeles.
10. V.J.Delima. (2016). Influence of Employees’

Functional Competencies on Employees’
Job Performance: A Special Reference to
Insurance Companies in Batticaloa District.
Journal for Studies in Management and
Planning, e-ISSN: 2395-0463, Volume 02 Issue
6, 188-195.
11. Wijesiri, L. (2016, March 5). Plugging Sri

Lanka's skills gap. Daily News.
325
pp. [1 --10],
[327 332],2017
2017
©
© The Institution
Lanka
Artificial Bone Substitutes and Bone Tissue

Engineering as Alternative Methods for Natural Bone
Repair and Regeneration in Severe Traumas and
Complex Clinical Conditions of Bones
B.R.M.G.W.U.B. Balagalla
Abstract: Human bone tissue has ability to develop and regenerate itself throughout the life.
However, there is a need of alternative methods for natural bone repair and regeneration in severe
traumas such as terrible bone fractures, infections, tumors and skeletal abnormalities. Additionally,
the rate of bone disorders are increasing rapidly. Thus. , it is necessity to conduct research on
alternative cures for complex clinical conditions which compromises the natural repair process. One
such method is bone grafting. However, there are some drawbacks which limit the grafting
treatments. Researchers are working on improving more alternative methods such as artificial bone
substitutes and bone tissue engineering. Biomedical engineering plays a major role in developing
biocompatible, non-toxic biomaterials for artificial bone substitutes and testing their physical
parameters to estimate the ability to mimic the natural bone. The latest trend of alternative methods
for natural bone repair and regeneration is bone tissue engineering. Currently, bone tissue engineering
is developing rapidly and in the near future it will become the major solution for severe traumas and
complex clinical conditions.However, it is important to conduct more researches to find out the best
scaffold materials, study them in micro and nano structural level, study the in-vitro cell developing
and maintaining the physical parameters and growth factors in cell culture to obtain the bone tissue
required. This review emphasizes the importance of artificial bone substitutes and bone tissue
engineering, alternative methods for natural bone repair and regeneration in severe traumas and
complex clinical conditions of bones, the current development and future directions.
Keywords: Biomedical Engineering, Artificial Bone Substitutes, Bone Tissue Engineering

1. Introduction women. United States estimated that 4 in 10
women after age of 50 experience hip, spine or
Bone tissue develops and remodels continuously wrist fracture. Furthermore, some fractures are
itself throughout human life and it repairs and life threatening or have some lifetime risks.
regenerates itself during bone fractures. It is a Additionally, genetic diseases caused by
complex physiological process which can be mutations such as deletion, insertion, RNA
seen during normal fracture healing. However, splicing, single base substitution cause some
in severe traumas such as terrible bone fractures, common ontogenesis diseases. One such
infections, tumors and skeletal abnormalities significant problem is osteoporosis. It is
and complex clinical conditions such as estimated that around 10 million and 33.6
avascular necrosis, atrophic non-unions and million individuals over age 50 suffer from hip
osteoporosis. Moreover, in some cases such as osteoporosis and low bone mass/ osteopenia
avascular necrosis, atrophic non-unions and respectively in US. The risk of osteoporosis and
osteoporosis the regenerative process is osteopenia is increasing and by 2020 1 in 2
compromised. To overcome these clinical amaricans has the risk of having osteoporosis in
problems of impaired and insufficient bone any site of skeleton. Additionally, calculated
regeneration, bone grafting, bone tissue percentage of osteoporosis risk for
engineering and artificial bone substitutes are postmenopausal White women is 35
used[19]. percent[4,5].
2. Significance Eng. (Miss.) B.R.M.G.W.U.B. Balagalla, B.Sc. in

The risk of bone fractures are increasing rapidly Biomedical Engineering (General Sir John Kotelawala Defence
University), Demonstrator, Department of Electrical and
due to less exercises, lack of nutrition and Electronics, Faculty of Engineering, University of Sri
accidents. It is estimated that 1.5 million Jayewardenepura, Sri Lanka.
individuals suffer bone fractures in each year.
Risk of fractures is greater in elderly people and
1
327
3. Bone Grafting Techniques 3.2.1 Advantages of Allogenic Bone Grafting
Bone tissue grafting is a commonly used surgical
Allogeneic bone grafts are histocompatible in
procedure which is used as a treatment for many
most of the cases. Moreover, there are different
bone defects. There are three types of grafting
forms of allogenic bone grafts are available
techniques, autograft, allograft and xenograft.
according to the host-site requirements namely
demineralized bone matrix, morcellised and
Autografting is re-implanting a tissue by another
cancellous chips, corticocancellous and cortical
similar tissue taken from the same individual.
grafts, osteochondral and whole-bone segments.
Allografting is replacing a tissue by another
similar tissue taken from a different individual 3.2.2 Disadvantages of Allogenic Bone Grafting
in the same species. Xenografting is replacing a
tissue by another similar tissue taken from a Allogenic bone grafting is more susceptible to
different individual in a different species. immune-rejections and pathogen transmission.
Nevertheless, autografting and allografing is the Moreover, allografts show reduced
commonly used techniques in bone grafting. osteoinductive properties because donor grafts
Xenografting is not popular in bone grafting due are devitalized via irradiation or freeze-drying
to the biocompatibility, mechanical and chemical processing. Another problem associates with the
dissimilarities and immune rejections. allogenic bone grafting is unmet supply to the
great demand. To be exact, there is a shortage in
3.1 Autologous Bone Grafting allograft bone grafts. Though the allogenic bone
grafting costs less comparatively to the
Autologous bone grafting is known as the gold autologous bone grafting, it costs a significant
standard method because of it’s advantages over amount[9,10].
other existing methods[6].
3.1.1 Advantages of Autologous Bone Grafting

4. Artificial Bone Substitutes
Artificial bone substitutes appeared in the
Autologous bone grafts are biocompatible and clinical set-up as a solution for the
non-immunogenic. They provide all the physical disadvantages of gold standard method and
and chemical properties required as a bone graft allogenic bone grafting. There are continuous
material. In contrast, they are osteoinductive, researches going on to find the ideal materials
osteogenic and osteoconductive. Moreover, it for bone replacement with adequate mechanical
can be easily re-vascularized and it combines properties and biocompatibility which can be
with the recipient site within less time period. provided upto demand.
Furthermore, possibility of pathogen
4.1 Different types of Bone Substitutes
transmission is rare in autologous bone grafting.
4.1.1 Ceramics
3.1.2 Disadvantages of Autologous Bone
Grafting Ceramics can be categorized as bioabsorbable
ceramics, bioactive ceramics and bioinert
Autologous bone graft is harvested from the
ceramics based on the chemical reactivity and
same patient. Thus, it requires a second surgery
porosity. Mostly they are derived from coral
in the donor site. It may result in significant
which has a biological pore interconnectivity
morbidity or deformity in donor site. Possibility
mesh network and hemihydrate of calcium
of excessive bleeding, inflammation and
sulphate known as Plaster of Paris which is
infection is high as there are two surgical sites in
prepared by heating gypsum.
the same patient. If the defected site is larger
Bioabsorbable ceramics is the first and most
than feasible or available donor sites, it is
clinically used synthetic material in bone
impossible to perform autologous bone grafting.
replacement. They bind to the host cells well
Additionally, autologous bone grafts are
and bioabsorption rate depends on the
comparatively expensive[7,8,12]
osteogenesis properties arises due to the
3.2 Allogenic Bone Grafting chemical composition, crystal structure and
Allogenic bone grafting is the most commonly surface area. For an example, tricalcium
used second bone grafting technique. phosphate resorbed 10–20 times faster than
hydroxyapatite. Tricalcium phosphate can be
totally resorbed or converted into
hydroxyapatite which may remain in the body
significantly.
2
328
Porosity of the ceramics is important in grafting. delayed degradation of coralline-derived
The optimal osteoconductive pore sizes of hydroxyapatite is a most common problem
ceramic bone grafts is 150 and 500 μm. Injured occurs.
bone regenerates itself through pores during Bioglass contains SiO2 less than 60%, high NaO2
healing osteogenic process and the synthetic and CaO. The ratio of CaO/P2O2 is almost
material is reabsorbed. However, the intrinsic similar to natural bone. The surface of the
strength of the callus site can be compromised bioglass becomes highly reactive and forms
due to weak synthetic ceramics and increased hydroxyl carbonate apatite when it is exposed to
porosity can cause significant loss in an aqueous medium. Formation of hydroxyl
compressive strength. rapidly is needed to make a good bond with
Hemihydrate of Calcium Sulphate which is host tissues as bioglasses are not reabsorbable.
generally known as Plaster of Paris is the first Bioglasses are used for the treatments of
bone substitute used. It is reabsorbable ceramic medium size bone defects due to the less
with a high reabsorbable rates and indefinite mechanical strength.
shelf life. It does not compromise osteogenesis Bioinert ceramics are biocompatible material
or increase the risk of infection when used in with highest mechanical strength which do not
infected cavities. Additionally, it is easy to react with the host tissue. Often, the
synthesize and sterilize. It allows drug composition of bioinert ceramics contain metal
administration and very inexpensive. oxides such as zirconia, alumina or titanium.
Nevertheless, Plaster of Paris does not provide However, the applications of bioinert ceramics
internal strength or support. Accordingly, it can are limited due to less bonding capabilities with
be used to fill small bone defects only. bone[1].
Calcium phosphate is an osteoconductive
bioabsorbable ceramic composed of 4.1.2 Polymers
hydroxyapatite, TCP or combination of the two.
However, mixture of hydroxyapatite and TCP is Nonbiodegradable polymers such as polymethyl
used most of the time clinically. It is obtained by methacrylate have been used as osteoconductive
sintering calcium phosphate salts at high artificial bone substitutes. Activity of PMMA is
temperatures to evaporate water and high similar to bioinert ceramic and bioglasses.
pressure compaction produces the powder Though it does not bind to the new bone, it is
which can be used to mold into pellet. Calcium adhesive to the bone surface without
Phosphate achieves higher percentage of it’s interrupting the regeneration. PMMA is selected
compressive strength within 24h after injected as an artificial bone substitute over a decay due
into bone cavity and more stable due to high to it’s osteoinductive properties yet it is weak.
density. Mechanical properties of porous However, polymers may even inhibit new bone
calcium phosphate is similar to cancellous bone growth and normal healing. Partially resorbable
once they have been incorporated and polymers such polyglycolide and poly-L-lactide
remodeled. Possibility of early adverse effects are introduced to overcome the problems such
namely inflammation and foreign-body as inhibition of new bone growth and healing
responses to calcium phosphate are process. Polymers have a great advantage over
comparatively low. But the disadvantage is that ceramics as they can be used in large bone
it cannot be used where significant torsion fracture fixation and use as substances
bending or shear stress is required due to being carriers[,211,12]
a brittle material with insufficient tensile
strength. 4.1.3 Metals
Calcium carbonate skeletal structure of sea coral
is used to synthesize hydroxyapatite by a A number of metals and their alloys, namely Fe,
hydrothermal exchange method which replaces Cr, Co, Ni, Ti, Ta, Mo, and W have been used as
the original calcium carbonate with a calcium an artificial bone substitutes for years. The first
phosphate replicate. Structure of hydroxyapatite metallic implant developed known as Sherman
is more similar to human cancellous bone. Vanadium Steel is used as bone fracture plates
If porites are used as the synthesis and screws. Metals have good mechanical
material, microstructure structure becomes more properties such as high durability and high
similar to interstitial cortical bone. However, strength. Moreover, they are readily available
hydroxyapatite can be used only to fill the for the surgery and comparatively low cost.
defects upto 7-8cm as it does not contain the However, there is a great problem related with
intrinsic strength required. Additionally, corrosion in metals which may weaken the
3
329
implant and cause toxic reactions. Additionally, proper functioning of cells and the new bone
metals are affected by electromagnetic formation.
environment. But Ti and it’s alloys are corrosion Scaffolds should be highly porous and highly
resistive and are not affected by the interconnected to provide mass transport, cell to
electromagnetic environment yet very cell communication and vascularization. They
expensive. should be hydrophic to allow the cell
The common advantages of artificial bone attatchment. Their anisotropic properties
substitutes are that they are readily available promotes cell adhesion, proliferation and
upto the demand, FDA approvals are provided differentiation. Moreover, for an optimum
and less ethical considerations has to be taken performance, they should be available in a
into account. Nevertheless, there are drawbacks variety of shapes, sizes and structures according
such as low mechanical strengths, high foreign to the requirements. It is important to have the
body immune reactions and toxicity[3]. mechanical properties similar to those of the
host and it should be structurally stable.
Additionally, scaffold material should be a
5. Bone Tissue Engineering
biodegradable, biocompatible and non-toxic.
Bone tissue engineering is the process of
There are different types of scaffold materials,
culturing osteogenic cells within three-
namely bioactive ceramics, biological polymers,
dimensional scaffolds, under conditions
synthetic polymers and hybrid materials.
supporting bone formation to obtain fully
Bioactive ceramics are used over many years.
viable, biological bone grafts in vitro. Osteogenic
They can be divided into two as natural and
cells, scaffolds, combinations of biochemical and
synthetic based on the origin. Coralline
biophysical signals can be given as the key
hydroxyapatite is an example for natural origin
components of bone tissue engineering[5,13,14].
and synthetic hydroxyapatite and β-tricalcium
phosphate can be given as examples for
5.1 Osteogenic Cells synthetic origins. There are many advantages of
There are differet types of osteogenic cells used bioactive ceramics namely, having high
in bone tissue engineering namely, embryonic modulus non-toxicity, good osteo-integration,
stem cells, bone marrow-derived mesenchymal supporting osteoblast adhesion and
stem cells, umbilical cord-derived mesenchymal proliferation. However, they are brittle and has
stem cells and human amniotic fluid-derived a poor fatigue resistance. Additionally, if the Ca
stem cells. and P concentrations are increased, it may result
Stem cells appears as the most valid and more in cellular death. Moreover, the degradation rate
promising solution in bone tissue engineering. of bioactive ceramics is difficult to predict.
Embroyonic stem cells has nearly unlimited self- Biological polymers are extensively studied for
renewal capability and the capacity to bone tissue engineering as they have many
differentiate via precursor cells. Disadvantage of advantages such as low immunologic response
embryonic cells is the tumorigenesis. and have a good bioactive behavior. They
Mesenchymal stem cells can be isolated from the interact with the host tissue properly. However,
intended recipient very easily. Therefore, there the degradation rate cannot be controlled and
is a great advantage of minimized immune mechanical characteristics of biological
rejection. In addition, it has a high polymers do not match those of the bone.
differentiation and proliferation capabilities. Examples for biological polymers are collagen,
Mesenchymal stem cells have capability of self- chitosan and silk.
renewal and multi-lineage differentiation. Synthetic polymers are biocompatible and have
Human amniotic fluid-derived stem cells is potential of delivering drugs and growth
isolated from amniotic fluid after routine factors. The degradation rate of synthetic
amniocentesis. It expresses both embryonic and polymers can be controlled. Additionally, it’s
adult stem cell properties and highly mechanical properties can be tuned according to
multipotent. Another advantage is that, it does the requirement. Their capability of the cell
not create tumorigenic reactions. attachment can be improved by negatively
charged chemical groups. Moreover, they can be
fabricated into complex shapes[15,16].
5.2 Scaffolds
Composites consist of 2 phases, matrix phase
In the bone tissue engineering, 3D scaffolds play
and dispersed phase.These materials are mostly
an important role. There are many properties
used to achieve additional good properties. For
that should be fulfilled by a good scaffold for
an examples, polymer and ceramic composites
4
330
give the toughness of a polymer with the 6. Conclusion
stiffness of a ceramic. Moreover, There is a great need of alternative methods for
biocompatibility can be enhanced by using natural bone repair and regeneration due to the
composites[17,18]. severe traumas and complex clinical conditions.
There are many alternative methods are
5.3 Biochemical and Biophysical Signals introduced. However, there are several
Bioreactors and the growth factors are the most
drawbacks and limitations in each methods.
important signaling molecules in bone tissue
engineering. According to the review, researches should be
Bioreactors are the dynamic cell culture systems done in every alternative methods for the
which allows more control to generate larger further development. Tissue engineering will be
volumes of cells. They enhance the mass transfer the prominent solution for the need of
of nutrients and gases and metabolites to alternative methods for natural bone repair and
regulate the size and structure of the tissue regeneration due to the severe traumas and
being generated. complex clinical conditions in the near future.
Examples for the growth factors are recombinant
human BMPs (rhBMPs), BMP-2, and BMP-7. References
They direct blend or coat on the scaffold or it
adhere to the cell culture medium during the 1. Takao, Y. (1994) Bone bonding behavior and
development of the cell/scaffold construction. clinical use of A-W glass-ceramic, in Bone
Grafts, Derivatives and Substitutes (Urist, M. R.,
O’Connor, B. T., and Burwell, R. G., eds.),
5.4 Advantages of Bone Tissue Engineering
Butterworth-Heinemann Ltd, pp. 245–259.
In bone tissue engineering, the possibility of
pathogen transmission is zero and the host has 2. Hench, L. L. (1992) Bioactive bone substitutes, in
to undergo fewer surgeries. Possibility of Bone Grafts and Bone Graft Substitutes (Habal,
immune rejections are rare. Moreover, it is M. B. and Reddi,A. H., eds.), Saunders,
advantageous as the host does not need to wait Philadelphia, pp. 263–275.
for a donor. It is a permanent solution and
3. Park, J.B. and Lakes, R.S., 2007. Metallic implant
helpful in severe traumas. It avoids the
materials. Biomaterials, pp.99-137.
permanent damages occurs to the donor site.
There is a possibility of correcting incurable 4. Kuivaniemi, Helena, Gerard Tromp, and Darwin
genetic defects with the aid of bone tissue J. Prockop. "Mutations In Collagen Genes:
engineering. Causes Of Rare And Some Common Diseases In
Humans". FASEB Journal (1991).
5.5 Disadvantages of Bone Tissue Engineering
The most prominent problem related to the bone 5. Laurencin CT, Ambrosio AM, Borden MD,
tissue engineering is difficulty of constructing Cooper JA. Tissue engineering: orthopedic
suitable scaffolds. Moreover, it is difficulty applications.Annu Rev Biomed Eng. 1999;1:19–
46.
obtain the full functionality of the bone. It is
comparatively high cost. Still there are fewer 6. St John TA, Vaccaro AR, Sah AP, Schaefer M,
technologies are available. Berta SC, Albert T, et al. Physical and monetary
costs associated with autogenous bone graft
5.6 Future Strategies of Bone Tissue harvesting. Am J Orthop (Belle Mead NJ)
Engineering 2003;32(1):18–23.
Biomedical engineer has a great role to play in 7. Younger EM, Chapman MW. Morbidity at bone
bone tissue engineering developments. It is graft donor sites. J Orthop Trauma.
1989;3(3):192–195.
important to carry out research to figure out the
new techniques and matrix materials will be 8. Banwart JC, Asher MA, Hassanein RS. Iliac crest
developed to enhance the properties of needed bone graft harvest donor site morbidity. A
bone tissue and scaffolds. In the future, there is a statistical evaluation. Spine (Phila Pa 1976)
possibility in reprogramming of adult cells to 1995;20(9):1055–1060.
form induced pluripotent stem cells. Moreover,
9. Delloye C, Cornu O, Druez V, Barbier O. Bone
it is important to look into cost-effective
allografts: What they can offer and what they
methods as bone tissue engineering has many cannot. J Bone Joint Surg Br. 2007;89(5):574–579.
advantages yet very expensive.
10. Tomford WW, Starkweather RJ, Goldman MH.
A study of the clinical incidence of infection in
5
331
the use of banked allograft bone. J Bone Joint
Surg Am. 1981;63(2):244–248.
11. Glowacki, J., Jasty, M., and Goldring, S. (1986)

Comparison of multinucleated cells elicited in
rats by particulate bone, polyethylene, or
polymethylmethacrylate. J. Bone Miner. Res. 1,
327.
12. Laurencin CT, Khan Y, El-Amin SF. Bone graft
substitutes. Expert Rev Med Devices.
2006;3(1):49–57.
13. Cordonnier, T., Sohier, J., Rosset, P. and Layrolle,

P. (2011). Biomimetic Materials for Bone Tissue
Engineering - State of the Art and Future Trends.
Advanced Engineering Materials, 13(5), pp.B135-
B150.
14. Amini, A., Laurencin, C. and Nukavarapu, S.

(2012). Bone Tissue Engineering: Recent
Advances and Challenges. Critical Reviews� in
Biomedical Engineering, 40(5), pp.363-408.
15. Ulery BD, Nair LS, Laurencin CT. Biomedical

applications of biodegradable polymers. J Polym
Sci B Polym Phys. 2011;49(12):832–864.
16. Rossi, F., Santoro, M. and Perale, G. (2013).
Polymeric scaffolds as stem cell carriers in bone
repair. Journal of Tissue Engineering and
Regenerative Medicine, 9(10), pp.1093-1119.
17. Karageorgiou V and
Kaplan D. Porosity of
3D bio-material
scaffolds and
osteogenesis.
Biomaterials.2005;26(27)
:5474–5491.
18. Stevens, B., Yang, Y., Mohandas, A., Stucker, B.

and Nguyen, K. (2008). A review of materials,
fabrication methods, and strategies used to
enhance bone regeneration in engineered bone
tissues. Journal of Biomedical Materials Research
Part B: Applied Biomaterials, 85B(2), pp.573-582.
19. Dimitriou, R., Jones, E., McGonagle, D. and V

Giannoudis, P. (2011). Bone regeneration: current
concepts and future directions. [online] BioMed
Central. Available at:
http://bmcmedicine.biomedcentral.com/articles
/10.1186/1741-7015-9-66 [Accessed 27 Oct.
2016].
6
332
Annual Sessions of IESL, pp. [333
[1 - 10], 20172017
- 340],
Condition Assessment of Power Transformers using

Dissolved Gas Analysis
W.T.D. Samarasingha, J.R.S.S. Kumara and M.A.R.M. Fernando
Abstract: Power transformer is one of the key elements in an electrical power network.
Therefore, in order to ensure reliable supply, proper condition assessment program of power
transformers is an essential requirement. In the power transmission network of the Ceylon Electricity
Board, there are over 160 numbers of power transformers (132/33 kV and 220/132/33 kV) in
operation which are distributed in grid substations all over the country. They are from different
manufacturers having different ratings and in the wide range of service life. This study has been
carried out in order to seek the possibility of assessing the condition of power transformers based on
the results obtained from Dissolve Gas Analysis. All the available Dissolve Gas Analysis data of the
power transformers in grid substations were collected and analyzed in order to assess the condition of
those power transformers. A health index (a numerical value which described the condition of the
transformer) was calculated for each transformer based on Dissolve Gas Analysis test results. Several
special cases were selected and further analyzed in order to investigate the potential of DGA for
predicting of condition of power transformers. Other test results such as winding resistance and
excitation current were also used in these case studies. Finally, it was identified that, Dissolve Gas
Analysis can be effectively used in condition assessment of power transformer.
Keywords: Condition Assessment, Power Transformers, Dissolved Gas Analysis (DGA)
1. Introduction serious faults, while some of gasses are

considered as not high risk indicators but
Dissolved Gas Analysis (DGA) is a widely used should be at alarmed depend on their quantity
tool in transformer condition monitoring which has been formed [1].
can be used to analyze the gas components
dissolved in the insulating oil. Electrical The gases listed above are generally considered
disturbances and thermal decompositions can as combustible gases (except CO2). Presence of
be identified as two common causes of gas these gas components can be due to thermal,
formation in energized transformers. electrical, or corona faults inside a transformer.
The rate at which each of these key gases is
Insulating mineral oil inside a transformer is a produced basically depends on the fault
mixture of different hydrocarbons, and the temperature and volume of material. Because of
decomposition processes for these the volume effect, amount of gases that a large
hydrocarbons against a thermal or electrical volume of insulation at moderate temperature
fault are different to each other. Chemical produced would be moreover equal to the
reactions occurred at fault/stressed conditions amount of gases produced in a smaller volume
are involving in breaking of carbon-hydrogen at a higher temperature. Therefore, the
and carbon-carbon bonds [1]. During this concentrations of the individual gas component
process, active hydrogen atoms and found in transformer insulating oil may be
hydrocarbon fragments are formed. directly used and trended to evaluate the
Combination of few of these fragment may thermal history of the transformer in order to
form hydrogen (H2), methane (CH4),ethylene
(C2H4),ethane (C2H6) and acetylene(C2H2). In Eng. W.T.D. Samarasingha, AMIE(SL), B.Sc. Eng. (Hons)
addition to that thermal decomposition of (Peradeniya), Electrical Engineer, Ceylon Electricity Board.
cellulose and electrical faults produce methane Eng. (Dr)J.R.S.S. Kumara, AMIE(SL), SMIEEE, B.Sc.
(CH4), hydrogen (H2), carbon monoxide (CO), Eng.(Hons) & M.Phil (Peradeniya),Tech. Lic. & PhD.
and carbon dioxide (CO2). In DGA, these gases (Chalmers), Sweden, Senior Lecturer, Dept. of Electrical &
Electronic Engineering, University of Peradeniya.
are generally referred to as key gases and
Eng. (Prof.) M.A.R.M. Fernando, C.Eng., FIE(SL),
measured in Parts Per Million (PPM). Some of
SMIEEE, B.Sc.Eng.(Hons) (Peradeniya), Tech Lic. (KTH),
these gasses are considered as high risk Sweden, Ph.D.(Chalmers), Sweden, Professor, Dept. of
indicators as they are generated only under Electrical & Electronic Eng.., University of Peradeniya.
1
333
evaluate any potential faults inside the
transformer and hence asses condition of the
transformer[2].
Among many diagnostic techniques developed

for the interpretation of these gases below listed
methods and techniques can be identified as the
widely used once in transformer condition
diagnostics [3].
1. Key Gas Method

2. Roger’s Ratio Method
3. IEC Basic Ratio Method
Figure 1 - Relationship between gas
4. Duval’s Triangle Method
constituent vs. fault temperature [2,6]
5. ANSI/IEEE C57.104
Figure 1 shows the relationship between gas
2.1 Key Gas Method
constituent vs. fault temperature. Study of these
In this method, type of fault is identified based
relationships reveals that gases are generated in
on the percentage amount of a specific gas.
following order with temperature increase
Table 1 summarizes the diagnostic criteria for
H2→CH4→C2H6→C2H4→C2H2 [2], [6].
key gas method. The percentage amount of gas
is an approximated figure based on total
2.2 Roger’s Ratio Method
dissolved combustible gases (TDCG) [3-5]. For
Roger’s ratio method can be identified as an
example, large amount of H2with little amount
extension of Doernenberg’s ratio method. It
of other hydrocarbons are produced by low
described three ratios: R1, R2& R5 as defined
intensity Partial Discharge (PD) or corona.
below.
Further, relationship between fault temperature
R1 = CH4/H2 ... (1)
and gassing characteristics is described in the
thermodynamic model proposed by Halstead
R2 = C2H2/C2H4 …(2)
[7]. This model assumes that all hydrocarbons
in the oil are decomposed into same product
R5 = C2H4/C2H6 … (3)
and each of these products are in equilibrium
with all the other products. According to the
As shown in Table 2, there are 5 types of faults
model, the generation rate of each gas can be
are described in Roger’s ratio method.
calculated at any temperature.
Table 2 - Diagnostics Technique for Roger’s
Table 1 - Diagnostic criteria of the key gas
Ratio Method [6]
method [3-5]
Case R2 R1 R5 Fault
Fault Key Criteria Gas
Gas percentage 0 < 0.1 0.1–1.0 <1.0 Normal
Arcing C2H2 High H2 and C2H2, H2: 60% Low energy
1 < 0.1 <0.1 <0.1 partial
low CH4 and C2H4, C2H2: 30%
if cellulose discharge
involvedCO, CO2 2 0.1–3.0 0.1–1.0 >3.0 Arcing
Corona (PD) H2 High H2, moderate H2: 85% Low
CH4, low CH4: 13% 3 < 0.1 0.1–1.0 1.0–3.0 temperature
C2H6,C2H4, if thermal
celluloseinvolvedC Thermal fault
4 < 0.1 >1.0 1.0–3.0
O, CO2 below 7000C
Overheating C2H4 High C2H4, C2H4: 63% Thermal fault
5 < 0.1 >1.0 >3.0
of oil lowC2H6, moderate C2H6: 20% above 7000C
CH4,H2& CO
Overheating CO High CO & CO2. CO: 92%
of cellulose
2
334
Table 3 - Diagnostics technique for IEC Basic Ratio Method [1,4,6]
Range of ratios Code
C2H2/C2H4 CH4/ C2H4/
H2 C2H6
R2 R1 R5
<0.1 0 1 0
0.1 to 1.0 1 0 0
1.0 to 3.0 1 2 1
>3.0 2 2 2
Fault Types Typical phenomena
Normal (No fault) 0 0 0 Normal aging
Partial discharge of 0 (but not 1 0 Discharge in gas filled cavities resulting
low energy density significant) from incomplete impregnation or high
humidity
Partial discharge of 1 1 0 Same as PD of low energy density but
high energy density leading to tracking or perforation of
cellulose
Discharge of low 1 to 2 0 1 to 2 Continuous sparking in oil or breakdown
energy of oil between solid insulation
Discharge of high 1 0 2 Discharges with power follow-through.
energy Arcing of oil between windings or coils or
between coils and ground.
Thermal fault of low 0 0 1 General insulation conductor overheating
temperature < 1500C
Thermal fault of low 0 2 0 Local overheating of the core, increasing
temperature range hot spot temperatures, shorting links in
150-3000C the core, overheating of copper due to
Thermal fault of 0 2 1 bad contacts/ joints, circulating currents.
medium
temperature range
300-7000C
Thermal fault of 0 2 2
high temperature
>7000C
Roger’s ratio method has not defined zones are defined based on amount of
minimum gas level requirements regarding CH4,C2H2 and C2H4and each zone is
applicability of the method. However, it is corresponds to a particular fault.
suggested that the method be applied only
when normal gas levels are exceeded [6]. The Triangle coordinates corresponding to
DGA results in PPM are calculated as follows:
2.3 IEC Basic Ratio Method
IEC Basic Ratio Method is further extension of C2H2% = 100 * x / (x + y + z) ….(4)
the Roger’s Ratio method. IEC 60599: “Mineral
oil-filled electrical equipment in service - C2H4% = 100 * y / (x + y + z) … (5)
Guidance on the interpretation of dissolved
and free gases analysis” discusses a ratio CH4% = 100 * z / (x + y + z) … (6)
method to analyze the concentrations of
dissolved gases which can be interpreted to Where, x is C2H2 inppm, y is C2H4 in ppm
diagnose the condition of oil-filled electrical andz is CH4 in ppm. These coordinates are
equipment in service. Diagnostics as per IEC plotted in the Duval’s Triangle which is in
60599 is summarized under Table 3 [1], [4], [6]. Figure 2to evaluate the fault condition [5],
[7].This method has proven to be accurate and
2.4 Duval’s Triangle Method dependable over many years and is now
A triangular graphical representation was gaining in popularity.
used to analyze the gas concentration inside
the transformer. As illustrated in Figure 2,7
3
335
power transformers are 220kV, 132kV and
33kV.All these power transformers are
ONAN/ONAF type with On Load Tap
Changer (OLTC). Further, all these
transformers are oil immersed type and
suitable for outdoor operation.
MYRKOS portable dissolved gas analyzer

manufactured by Morgan Schaffer was used to
analyze the dissolved gas content in the
transformer insulating oil. In this process, first,
recommended amount of insulating oil
quantity is sampled to a shake test syringe.
Then, the oil sample is shaken as per the
Figure 2 - Duval’s Triangle [5,7]
recommendation given by the instrument
manufacturer in order to extract the dissolved
2.5 ANSI/IEEE C57.104
gas components from liquid to air media.
Four condition levels have been introduced in
Then the extracted gas amount is fed to the
order to assess the level of risks to
transformers. Table 4 list the dissolved gas instrument as per the instruction given by the
instrument manufacturer. The gases are
concentration levels allocated for each
condition [2]. evaluated using chromatography method in
two columns inside the instrument and a test
report is prepared for injected gas sample.
Condition 1: TDCG below 720 indicates the
transformer is healthy. If any individual
These test results were analyzed to evaluate
combustible gas exceeds specified levels,
the present health condition of the
further investigations are recommended.
transformer. Some transformers were tested
for several times to monitor the generation of a
Condition 2: TDCG within 721-1920 indicates
particular gas component(s).
condition of the transformer is acceptable. Any
individual combustible gas exceeding
specified levels should prompt additional
4. Analysis of DGA Test Results
investigation. It is recommended to monitor
DGA test results are analyzed for transformer
the behavior of gas generation over the time.
condition diagnostics. As per Duval’s
suggestions [8],90% typical gas levels were
Condition 3: TDCG within 1921-4630 indicates
calculated for each and every gas component.
poor condition of the transformer. Immediate
A comparison between calculated 90% gas
action should be taken to establish a trend over
levels for CEB network with threshold gas
the time to monitor the behavior of gas
levels stated in ANSI/IEEE C57.104 and
generation. Faults are probably present.
threshold gas levels stated in IEC 60599 are
summarized in Table 5.
Condition 4: TDCG exceeding 4630 indicates
very poor condition and immediate attention
When considering gas levels in Table 5, 90%
is required to avoid possible failure of the
typical gas levels of H2, CH4, CO, C2H4 and
transformer.
C2H2 are substantially matched with IEEE or
IEC standard. 90% typical values of CO2 and
3. Tested Power Transformers C2H6 show a clear deviation from relevant
values in IEEE or IEC standard.
In this study, DGA tests were performed on
161 nos. of power transformers in Ceylon Gas levels stated in ANSI/IEEE C57.104 and
Electricity Board (CEB) which are spread all IEC 60599 are only recommended values
over the country. They are from different which can be used to evaluate the health
manufactures, with different characteristics condition of a randomly selected power
and in the wide range of age groups (1 to 49 transformer. But, when it considers the
years range). Operating voltages of these
4
336
Table 4 - Dissolved gas concentration level allocated for each condition in ANSI/IEEE C57.104 [2]
Status Dissolved key gas concentration limits [μL/L (PPM)]
H2 CH4 C2H2 C2H4 C2H6 CO CO2 TDCG
Condition 1 100 120 1 50 65 350 2500 720
Condition 2 2501- 721-
101-700 121-400 2-9 51-100 66-100 351-570
4000 1920
Condition 3 701- 401- 571- 4001- 1921-
10-35 101-200 101-150
1800 1000 1400 10000 4630
Condition 4 >1800 >1000 >35 >200 >150 >1400 >10000 >4630
Table 5 - Comparison of 90% gas levels for CEB network with threshold gas levels stated in
ANSI/IEEE C57.104 and IEC 60599
Gas Threshold gas levels Threshold gas levels stated 90% gas levels for CEB
Component stated in IEC 60599 in ANSI/IEEE C57.104 network
H2 60-150 100 68
CH4 40-110 120 112
CO 540-900 350 710
CO2 5100-13000 2500 4921
C2H4 60-280 50 33
C2H6 50-90 65 273
C2H2 3-50 2 2
Table 6 - Threshold gas levels for power transformers in CEB network and Maximum detected gas
levels in CEB network
Gas Component H2 CH4 CO CO2 C2H4 C2H6 C2H2
Threshold level 70 110 710 5000 35 270 2
Maximum detected gas level 2737 7883 2282 11406 9956 3293 47
Table 7 - 11 nos. of cases that, all three diagnostics methods give same result for a given
transformer and calculated health index for these transformers
Transformer Duval's Triangle Roger's Ratio IEC Basic Ratio HI
Kelanitissa TR02 Thermal fault (300°C Thermal fault Thermal fault (300°C - 1.2
- 700°C) (below 700°C) 700°C)
Seethawaka TR02 Thermal fault (above Thermal fault Thermal fault (above 1.2
700°C) (above 700°C) 700°C)
Pannipitiya TR03 Low energy electrical Arcing Low energy electrical 1.6
discharges discharges
Pannipitiya TR02 Thermal fault (above Arcing High energy electrical 1.8
700°C) discharges
Galle TR04 Thermal fault (above Thermal fault Thermal fault (above 1.9
700°C) (above 700°C) 700°C)
New Anuradhapura High energy electrical Arcing High energy electrical 1.9
TR02-Y discharges discharges
Veyangoda MT02 Thermal fault (300°C Thermal fault Thermal fault (300°C - 2.05
- 700°C) (below 700°C) 700°C)
Galle TR01 High energy electrical Arcing High energy electrical 2.25
discharges discharges
Kotugoda TR01-B Thermal fault (300°C Thermal fault Thermal fault (300°C - 2.35
- 700°C) (below 700°C) 700°C)
Veyangoda MT03 Thermal fault (above Thermal fault Thermal fault (above 2.35
700°C) (above 700°C) 700°C)
Biyagama TR01-B Thermal fault (above Thermal fault Thermal fault (above 2.9
700°C) (above 700°C) 700°C)
5
337
batch, the level of dissolved gas components particular transformer is in worst condition in
may different from those recommended values the sense of DGA. The minimum possible HI
due to difference in loading pattern, weather value is 1.0 which indicates that the particular
condition and fault levels. Therefore, threshold transformer is in satisfactorily normal
values defined in Table 6 were used for condition in the sense of DGA. Health Index
diagnostic of transformers in CEB network. value was calculated for all the power
transformers subjected to this study. Further,
Transformers having at least one gas justification was carried out with calculated
component higher than the threshold level Health Index values with available diagnostic
were further diagnosed as per Duval’s results of transformers.
Triangle, Roger’s Ratios and IEC Basic Ratios
methods. In this analysis, there were 11 nos. of Based on the variation of health index value, it
transformers that, all three diagnostics is clear that the severity of fault condition
methods give same result as shown in Table 7. becomes higher with increase of the health
The output result “Arcing” which described in index value. Most of the transformers which
Roger’s ratio method can be mapped with have health index value less than 1.15,
“high/ low energy electrical discharges” or “normal” condition or “low temperature
“thermal faults with temperature above 7000C” thermal faults” are dominated in other
in other two diagnostics methods. diagnostics techniques. For transformers
which have health index value in between 1.15
Further, high concentration of CO2 can be and 1.8, thermal faults in mid temperature
observed due to atmospheric air leakages ranges, low density electrical discharges and
through gaskets or breathers of the partial discharges are dominated in other
transformer. Therefore, concentration of CO2 diagnostics techniques.
cannot be considered as a strong parameter in
diagnostics of transformers. Table 8 - Score and weight defined for each
gas component
5. Health Index based on DGA Score
Gas 1 2 3 4 Weight
A Health Index (HI) was defined for selected H2 < 70 71 to 701 >1800 2
transformers. Equation 07 was used to 700 to
calculate the health index [9-10]. 1800
CH4 < 111 401 >1001 2
∑ 110 to to
….(7) 400 1000
∑
CO < 711 931 >1751 2
Where, HI: Health Index, Si: Score defined for 710 to to
each gas depends on concentration, Wi: 930 1750
Weighting factor allocated for each gas andi: CO2 < 5001 8001 >10000 1
Relevant gas component. 5000 to to
8000 10000
To define a score system for each gas C2H4 < 35 36 to 91 to >181 4
component, the threshold gas levels for CEB 90 180
network and maximum detected gas level for C2H6 < 271 401 >801 3
each gas component in CEB network (Table 6) 270 to to
and gas levels allocated for each condition 400 800
(condition 01 to 04) in ANSI/IEEE C57.104 C2H2 < 2 3 to 11 to >35 6
standards were taken into consideration. The 10 35
score system is defined in Table 8.
For transformers which have a health index
Allocating of a weight factor for each gas value greater than 1.8, high temperature
component was basically depending on gas thermal discharges and high density electrical
generating temperature described in Figure 1. discharges are dominated in other diagnostics
techniques. Therefore, health index value can
As per the score values assigned for each gas be further used to evaluate the fault conditions
level, the maximum possible HI value that can which are previously associated with
obtained is 4.0 which indicates that the transformer as previous fault conditions
6
338
directly related to the present condition of the regular basis on power transformers which
transformer. Therefore, the conditions of the were identified as critical based on the health
transformer gets deteriorate with the increase index. The duration of such two DGA tests
of the health index value. shall be two weeks, one month or three
months which depends on the criticality of the
Accordingly, 161 nos. of power transformers transformer.
were tested for DGA and each power
transformer were assigned with Health Index. It can be concluded that proposed health index
Number of power transformers allocated for based on DGA is an effective tool to evaluate
several ranges of Health Indexes is the condition of a power transformer.
summarized in Table 9. Continuous monitoring of DGA will lead the
development of a database of health condition
Table 9 - Number of power transformers vs. of power transformers which is very helpful
Health Index for a utility to manage and plan its future
Range of Health No. of Power developments.
Index Transformers
1.0 – 1.5 146 References
1.5 – 2.0 09
2.0 – 2.5 05 1. International Electro-technical Commission,
2.5 – 3.0 01 “Mineral oil impregnated electrical equipment in
3.0 – 3.5 00 service – Guide to the interpretation of dissolved
3.5 – 4.0 00 and free gases analysis”, IEC 60599, 1999.
2. Institute for Electrical and Electronics

In this sense, Biyagama TR01 – B, Veyangoda Engineers,“IEEE Guide for the Interpretation of
MT03, Kotugoda TR01 – B, Pannala TR01, Gases Generated in Oil-Immersed Transformers”,
Galle TR01, Veyangoda MT02, New IEEE Std C57.104–1998, 2009.
Anuradhapura TR02 – Y, Galle TR04,
Pannipitiya TR02, Mathugama TR01 can be 3. Jasim, S. Y. and Shrivastava, J. V., “Dissolved
identified as most critical 10 transformers (in Gas Analysis of Power Transformers”,
condition based) in CEB transmission network. International Journal of Electrical and
Electronics Engineering Research (IJEEER),
pp2., December 2013.
6. Conclusion
4. Borakhade, S. A., and Kunjal Jane, “Dissolved
This study has been carried out to seek the Gas Analysis in Transformer using Three Gas
possibility of assessing the condition of power Ratio Method and Fuzzy Logic based on IEC
transformers based on a health index defined Standard”, SSRG International Journal of
using results of Dissolve Gas Analysis. Electrical and Electronics Engineering (SSRG-
IJEEE) – volume 2 Issue 4, pp 2,April 2015.
All the power transformers in CEB
5. David Woodcock, William, H., Bartley, M.
transmission network were tested for DGA, Duval, J. Dukarm, A. de Pablo and Lindgren,
and the test results were analyzed based on the S. R., “Serveron White Paper: DGA Diagnostic
guidelines given in international standards. Methods”, Serveron Corporation, USA, pp 4,
Further, threshold limits for gas components 2007.
were defined by taking not only the threshold
limits given in international standards but also 6. Zhenyuan Wang, “Artificial Intelligence
the distribution of each gas concentration in Applications in the Diagnosis of Power
Transformer Incipient Faults”, Faculty of the
CEB network. Accordingly, a health index (a
Virginia Polytechnic Institute and State
numerical figure varies from 1 to 4) was
University, Virginia, pp 9 – 11, August 2008.
defined for each transformer. Based on the
health index values, conditions of the 7. M. Duval, “A Review of Faults Detectable by Gas-
transformers were evaluated and some in-Oil Analysis in Transformers”, IEEE Electrical
transformers were found which are in high Insulation Magazine, Vol. 18, No. 3,pp 8 –
critical stage. 17,2002.
It is recommended to carryout DGA at least 8. Duval, M., “Calculation of DGA Limit Values
and Sampling Intervals in Transformers in
once a year for each power transformer.
Service”, IEEE Electrical Insulation Magazine,
Further, it is required to carry out DGA test in Vol. 24, No. 5, pp 7 – 13, 2008.
7
339
9. Naderian, A., Piercy, R., Cress, S., Wang, F.,
Service, J., “An Approach to Determine the
Health Index of Power Transformers”, 1988,
Electrical Insulation, IEEE International
Symposium, pp 2, July 2008, DOI:
10.1109/ELINSL.2008.4570308.
10. Jahromi, A. N., Piercy, R., Cress, S., Service, J.

R. R., Fan, W., “An Approach to Power
Transformer Asset Management Using Health
Index”, IEEE Electrical Insulation Magazine,
Vol. 25, No. 2, pp 2, 2009.
8
340
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[341 348],2017
2017
The Institution
Lanka
Battery - Supercapacitor Hybrid Energy Storage

Control for Electric Vehicle
W.M.J. Madhusankha, G.A.T.P. Madhuranga, S.A.S. Indika,
B.G.L.T. Samaranayake, P.J. Binduhewa and J.B. Ekanayake
Abstract: The battery, the dc bus, the dc to ac inverter and the electrical machine are the main
components of a conventional fully electric vehicle (EV) powertrain. However, it severely limits the
absorption of regenerative electrical energy during the down-hill motion and braking due to the low
power density of batteries. Therefore, researchers work on incorporating Supercapacitors, which has
high power density even though energy density is lower than that of an EV battery, in harnessing the
regenerative energy. The range of a light EVs, with limited battery capacity due to weight restrictions,
can be significantly increased by harnessing the regenerative energy. In order to integrate the
Supercapacitors and the battery, a three-port dc-dc converter is used which is realized by simple
augmentation of two non-isolated dc-dc converters. The power flow of this converter is obviously bi-
directional. The ports are connected to the battery, Supercapacitors and the dc bus respectively.
In this paper, the average model of the non-isolated dc-dc converters were derived and thereby the
small signal model of the three port bi-directional dc-dc converter was obtained. The system is
presented in the linear state space format and classical control system design tools are used to derive
the stability based minimum and maximum voltage and current limits of the converter. The controller
of the non-isolated dc-dc converter between the Supercapacitors and the dc bus is used to control the
dc link voltage while the controller of the other dc-dc converter between the battery and the dc bus is
used to control the battery current. Both the controllers use simple Proportional Integral (PI)
controllers. The powertrain is tested in simulation using standard drive cycles to show the
effectiveness of adding the Supercapacitors.
Keywords: Battery, bi-directional dc-dc converter, electric vehicle, Supercapacitor.
1. Introduction between battery and supercapacitor (SC). These

strategies include fuzzy logic, rule-based
Energy storage system has a major role in control, model predictive control,etc [1 – 7]. The
electric vehicles. All the vehicles have a battery basic idea of all these control strategies is that
with the unique set of requirements to battery supports low frequency power
maintain, but the common disadvantage to all component and SC supports high frequency
of them is the limited life time of the battery. power component momentarily. In this paper, a
The EV battery supplies the instantaneous simple novel control strategy is proposed for
power demand, which most of the time, gives battery and supercapacitor hybrid energy
rise to random sudden changes in the battery storage system (HESS). The proposed method is
current. The high current derivatives lead to based on Kalman Filter [8]. Some parameters of
shorten the battery life time [1]. Coping up with the battery and supercapacitor have to be
this is a serious concern among the EV constrained within certain ranges to meet the
manufacturers [1]. Therefore, it is worthy to objective of maintaining a long life.
find a solution to extend the battery life time, The parameters are namely the State of Charge
while delivering the equivalent or better
performance. In order to extend the battery life
time by reducing the battery current Mr. W.M.J. Madhusankha, Mr. G.A.T.P. Madhuranga
derivatives, a hybrid energy storage system & Mr. S.A.S. Indika are undergraduates;
consisting of a battery and a supercapacitor Eng. (Dr.) B. G. L. T. Samaranayake, AMIE(Sri Lanka),
with novel control strategy, has been proposed. MIEEE(USA), B.Sc.Eng. (Hons) (Peradeniya), Tek. Lic., (KTH)
Ph. D. (KTH), is a Senior Lecturer;
In our study, we have used two bi-directional
Eng. (Dr.) P. J. Binduhewa, AMIE(Sri Lanka), MIEEE(USA),
dc-dc converters coupled with the battery and
B.Sc.Eng.(Hons) (Peradeniya), Ph.D. (Manchester), is a Senior
the supercapacitor. This configuration is also Lecturer;
aimed at decreasing the life cycle cost of the Eng. (Prof.) J. B. Ekanayake, C.Eng. (SL & UK), FIE(Sri
energy storage system of the EV. Lanka), FIET(UK), SMIEEE (USA), B.Sc. Eng. (Hons)
(Peradeniya), Ph.D.(UMIST), is a Chair Professor
In the past work, different control strategies In the Department of Electrical and Electronic Engineering,
have been adopted to control power sharing Faculty of Engineering, University of Peradeniya.
1
341
(SOC) and State of Health (SOH). The SOC and ̌ = current of low voltage side of converter
the SOH can be defined [3] as follows: ̌ = Inductor current of the converter
( ) | ( )| ̌ = voltage of low voltage side of converter
=− ….(1) ̌ = voltage of high voltage side of converter
( ) ( ) =voltage of the high voltage side at initial
= ….(2) condition
( ) ( ) = inductor current at the initial condition
= ….(3) ̌ = duty ratio of buck converter
̌ = duty ratio of boost converter
where,
() = Current of the battery = initial duty ratio of the buck converter
= initial duty ratio of the boost converter
= Capacity of the battery
= leakage resistance of the supercapacitor
= Life span of the battery
= internal resistance of the supercapacitor
() = Current of the capacitor = resister in the converter
= Voltage of the capacitor = capacitance of the supercapacitor
= Capacity of the supercapacitor. = low voltage side capacitor
= high voltage side capacitor
These parameters are used in deriving the L = inductor in the converter
control decisions presented towards the later
sections of the paper. This paper is organized 2.2 Supercapacitor with Buck Converter
as follows: in section 2, the theoretical models Model
of the hybrid energy storage system is The linearized governing equations averaged
illustrated. Section 3 presents the proposed over one switching period of the bi-directional
controller. In section 4, the simulation results dc-dc converter, where the power flow takes
are reported and the conclusions are presented place to the supercapacitor side from the dc bus
in section 5. side as shown in Figure 3 (converter operating
in the Buck mode) can be derived as
2. General Information
̃
=-⌊ ⌋̌ + ̌
The arrangement of Bi-directional dc-dc
converter, battery, dc bus, supercapacitor and + ̌ ...(4)
power demand is shown in Figure 1.
̃
L =D ̌ +V ̌ − ̌ -R ̌ …(5)
BUCK H p
̃
C = ̌ -I ̌ -D ̌ …(6)
H IND BUCK
̃
C =̌ -̌ …(7)
L
Figure 1 - Block diagram of the powertrain
2.1 Supercapacitor Model

Figure 2 shows the supercapacitor model used
in this paper. In order to represent the capacitor
internal resistance and leakage resistance of the
SC respectively, series and parallel resisters are
added to the ideal capacitor model.
Figure 3 - Model Diagram of Buck Converter

with Supercapacitor.
The above equations were presented in state

space format with state vector being
̌ ̌ ̌ ̌ and the input vector
Figure 2 - Supercapacitor model
The variables used in bidirectional converter being ̌ ̌ as:
with Supercapacitor model are:
2
342
̌̇
̌̇
̌̇ ̌̇
̌̇ ̌̇
[ ]
− − ̌̇
̌
− ̌ ̌̇
= [ ]
− ̌ ( )
− −
[ ̌ ] ̌
[ −⌊ ⌋] − ̌
= ( ) +
̌
̌
+ [ ] [ ̌ ]
− ̌ [ − −⌊ ⌋]
[ ]
and ̌
̌ [ ]
− − ̌
̌ [ ]
Y= .
̌
[ ̌ ] 2.4 Battery̌ Model
Figure 5 showš the battery model used in this
Y=
2.3 Supercapacitor with Boost Converter paper. It is the ̌Thevenin equivalent model of a
[ ̌ ] reduces the complexities of
battery. This model
Model
The linearized averaged model equations of the modeling the HESS.
same bi-directional dc-dc converter with the
power flowing from the supercapacitor side to
dc bus side as shown in Figure 4 were derived
as
̃
=-⌊ ⌋̌ + ̌ -
̌ (8)
̃ Figure 5 - Battery model

̌ =R ̌ +L + (1-D ) ̌ - V ̌ …(9)
p B H
The variables used in bidirectional converter
̃ with battery model are:
C =- ̌ -I ̌ + (1-D ) ̌ …(10) ̌
H IND B = current of low voltage side of converter
̌ = Inductor current of the converter
̃
̌ = voltage of low voltage side of converter
C =̌ - ̌ …(11)
L ̌ = voltage of high voltage side of converter
= initial voltage at high side
= inductor current at the initial condition
̌ = duty ratio of buck converter
̌ = duty ratio of boost converter
= initial duty ratio of the buck converter
= initial duty ratio of the boost converter
= internal resistance of the battery
= resister in the converter
= low side capacitor
= high side capacitor
L = inductor in the converter
Figure 4 - Model Diagram of Boost Converter
with Supercapacitor 2.5 Battery with Buck Converter Model
The linearized governing equations averaged
The above equations were presented in state over one switching period of the bi-directional
space format with state vector being dc-dc converter, where the power flow takes
̌ ̌ ̌ ̌ and the input vector being place to the battery side from the dc bus side as
[ ̌ ̌ ] as: shown in Figure 6 (converter operating in the
Buck mode) can be derived as
3
343
̃
…(12) The latter equations were presented in state

L =D ̌ +V ̌ - ̌ - R ̌
BUCK H p
̃ space format with state vector being
C =- ̌ + ̌ ….(13) ̌ ̌ ̌ and the input vector
L
̃ being ̌ ̌ .
C = +̌ - I ̌ -D ̌ …(14)
H IND BUCK
̌ =R ̌ …(15) ̌̇ − − −
( )
B ̌
̌̇ = − [ ̌ ]
( ) ̌
̌̇
[ ] [ ]
̌
+[ ][ ]
̌
̌
Y= [ ̌ ]
Figure 6 - Model Diagram of Buck Converter ̌
with Battery
The above equation set can be presented in 3. Control of Converters
state space format with state vector being 3.1 Basic controllers for supercapacitor
̌ ̌ ̌ and the input vector being and battery
̌ ̌ .
In order to implement the PI controller a
̌̇ − − linearized model was obtained by the actual
̌
̇̌ system model as described Section 2. In this
= − [ ̌ ] topology, the dc-dc converter connected to the
̌
̌̇ [−
Supercapacitor acts as a voltage regulator and
[ ] ]
the dc-dc converter connected to the battery
̌ acts as a current controller. To implement PI
+[ ][ ] controllers for these two cases, it is necessary to
̌
− have an idea about the poles of system. PI
̌ controllers were constructed using pole zero
cancelation method while placing the poles as
Y= [ ̌ ]
required. Time constant and damping ratios are
̌
the main parameters of the controllers.
2.6 Battery with Boost Converter Model
The linearized averaged model equations of the
In Figure 08, Idemand is the total demand current
same bi-directional dc-dc converter with the
of the system. Ifeedback is the feedback from the
power flowing from the battery side to dc bus
power converter. The error is fed into the PI
side as shown in Figure 07 were derived as
̃ controller and the control signal (duty ratio) of

̌ =R ̌ +L + (1-D ) ̌ -V ̌ …(16) converter were calculated. Vref is equal to the
p B H
bus voltage (48V). The PI controller of the
̃ converter with the Supercapacitor regulates the
C =- ̌ -I ̌ + (1-D ) ̌ …(17) dc bus voltage. Vref is fixed at 48V and the
H IND B
̃ Vfeedback sense the bus voltage and error was fed
C =̌ - ̌ …(18) into PI controller. Hence PI controller calculate
L
̌ = -R ̌ …(19) the duty of the converter and fed it to the
B
converter.
Figure 8 - Block Diagram of HESS Controller

Figure 7 - Model Diagram of Boost Converter
with Battery
4
344
Idemandis not the exact demand current. The open loop control and the corresponding
demand current(Idemand) flows through the output is shown in Figure 9.
Kalman filter algorithm which has a variable
gain according to the SOC of the A step is applied with duty ratio 0.716 to
Supercapacitor. However, dc bus voltage achieve the 48V at the output side of the
fluctuates due to the inequality of the demand converter in the Boost mode. The converter is in
current and the supply current. Supercapacitor open loop control and the corresponding
acts as a voltage regulator and restores the bus output is shown in Figure 10.
voltage to the set point (48V), irrespective of
whether it injects or draws current from the dc Next a PI controller was introduced to the
bus. supercapacitor connected bi-directional dc-dc
converter. A step of 48 V is applied as the
3.2 Power sharing between supercapacitor reference input and the output of the Buck
and battery mode is shown in Figure 11.The same
Currents components with high derivatives configuration in Boost mode is shown in Figure
(high frequency components) of the demand 12.
current, causerapid degradation of battery life.
Simple one dimensional Kalman filter
algorithm is utilized to share the power
between supercapacitor and the battery. So that
high frequency current is fed by supercapacitor
and the low frequency components of current is
fed by the battery. Following shows the simple
Kalman filter algorithm used in this Rise time: 0.00073 s
application. Settling time: 0.016 s
I(k) = I(k-1) …(20)
P(k) = P(k-1) + Q …(21) Figure 9 - Open loop step response of

K(k) = P(k)(P(k) + R)-1 …(22) Supercapacitor connected bi-directional dc-dc
I(k) = I(k-1)+K(k)*(Idemand(k)–I(k)) …(23) converter in the Buck mode.
P(k) = (1 – K(k))*P(k) …(24)
where,
I(k-1) = Initial current value
I(k) = Estimated current value
Idemand(k) = Demand current
Rise time: 0.00076 s
P(k-1) = Initial error value Settling time: 0.012 s
P(k) = Initial estimated error value
K(k) = Estimated Kaman gain
Q = Process noise covariance
R = Sensor noise covariance Figure 10 - Open loop step response of
Supercapacitor connected bi-directional dc-dc
converter in the Boost mode.
4. Simulation Result & Discussion
In this section, we present the simulation
results of the open loop models of the
supercapacitor connected bi-directional dc-dc
converter, battery connected bi-directional dc-
dc converter, closed loop control of the same
and the operation of the HESS.
4.1 Results of Supercapacitor connected

bi-directional dc-dc converter
A step is applied with duty ratio 0.618 to
achieve the 48V at the output side of the Figure 11 - Closed loop step response of the
converter in the Buck mode. The converter is in supercapacitor connected bi-directional dc-dc
converter in Buck mode.
5
345
Next a PI controller is added to the buck mode
converter and the rated current step of -208.35
A was applied as the reference input. The
Rise time: 0.00381 s corresponding output is shown in Figure 15.
Settling time: 0.018 s convertor.
Finally, the process is repeated in the Boost
mode and the corresponding output is shown
in Figure 16.
Rise time: 1.08 x 10-6 s

Figure 12 - Closed loop step response of the Settling time: 6.02 x 10-6 s
supercapacitor connected bi-directional dc-dc
converter in Boost mode.
4.2 Results of Battery connected bi-

directional dc-dc converter
The same procedure as in the previous section
is followed for the battery connected bi-
directional dc-dc converter.
A unit step was applied with unity duty ratio to Figure 15 - Close loop step response of the
achieve rated maximum current at the battery battery connected bi-directional dc-dc
side of the converter. The converter is in Buck converter in the buck mode.
mode and open loop control and the
corresponding response is shown in Figure 13.
Figure 16 - Close loop step response of the

battery connected bi-directional dc-dc
converter in the buck mode.
Figure 13 - Open loop Step response of the
battery connected bi-directional dc-dc 4.3 Hybrid Controller Simulation Results
converter in the Buck mode.
This result implies the hybrid current controller
The corresponding Boost mode operation is of the system works to its bi-directional
shown in Figure 14.
maximum current demand. Step of -208.35 to
208.35 was applied to the hybrid current
controller of the system as input and both buck
and boost converters shares the current
demand as shown in Figure 17.
Figure 18 shows the drive cycle simulation

results for the first 70 s of the Cranfield drive
cycle. Beyond 60 s, during de-acceleration, the
Battery connected converter comes in to buck
mode to store the energy in the battery.
Figure 14 - Open loop Step response of the
battery connected bi-directional dc-dc Switching between buck and boost of the
converter in the Boost mode. battery convertor for the ARTEMIS drive cycle
is shown in Figure 19.
6
346
Figure 17 - Step Response of Battery Hybrid
Controller
Figure 19 - Current demand being served by

the battery connected dc-dc converter.
Figure 18 - Current Demand during the

Cranfield drive cycle
4.4 Total system simulation results for

ARTEMIS urban drive cycle
Figure 20 - Current sharing between battery
Figure 20 shows the total demand current, and and supercapacitor
its split between the battery and
supercapacitorby using the Kalman filter
proposed in this paper.
Figure 21 shows the improvement of Battery
State of Health (SOH) in the Battery
Supercapacitor Hybrid Energy Storage System
as compared to the situation with Battery alone
case without SC. The SOH reduction in the
Battery is approximately 33% in the former case
compared to the latter case for the 900 s drive
cycle considered. Hence the HESS contributes
to significantly lower the SOH reduction, which
on the other hand means a significant
improvement in the Battery useful life.
Figure 21 - Improvement of Battery SOH
7
347
4. Omar Laldin, MazharMoshirvaziri, and
Olivier Trescases, Predictive Algorithm for
Optimizing Power Flow in Hybrid
Ultracapacitor/Battery Storage Systems for
Light Electric Vehicles, IEEE Trans. On Power
Electronics, vol. 28, no. 8, pp 3882 – 3895, Aug.
2013
5. Sathish Kumar Kollimalla, Mahesh Kumar

Mishra, and N. Lakshmi Narasamma, Design
Figure 22 - Variation of SOC of the
and Analysis of Novel Control Strategy for
supercapacitor
Battery and Supercapacitor Storage System,
IEEE Trans. On Sustainable Energy, vol. 5, no.
5. Conclusions 4, pp 1137 – 1144, Oct. 2014
Proposed control strategy is based on the 6. BranislavHredzak, Vassilios G. Agelidis, and

linearized model of the HESS. The Kalman filter Minsoo Jang, A Model Predictive Control
is used to filter out the rapid changes and System for a Hybrid Battery-Ultracapacitor
Power Source, IEEE Trans. On Power
obtain the reference for battery connected dc-dc
Electronics, vol. 29, no. 3, pp 1469 – 1479, Mar.
converter. The remaining power is supplied by
2014
the Supercapacitor.
7. Daniel J. Auger, Maxime F. Groff, Ganesh
Based on the SOC of Supercapacitor, the Mohan, Stefano Longo, Francis Assadian,
optimum power sharing between Impact of battery ageing on an electric vehicle
Supercapacitor and Battery is assured. It will powertrain optimization, Journal of
also ensure that high frequency current sustainable development of energy, water
components will not flow through the battery and environment, vol. 2, issue 4, pp 350-361,
2014
so that the SOH reduction of battery is
minimized. Voltage of the bus bar is in a fixed
8. Ramsey Faragher, Understanding the Basis of
value since Supercapaciror ensure the voltage the Kalman Filter Via a Simple and Intuitive
regulation of the bus bar so that the inverter Derivation, IEEE Signal Processing
input voltage is a constant. Magazine, pp 128 – 132, Sep. 2012
Acknowledgement
Authors would like to acknowledge the

University of Peradeniya Research Grant
URG/2016/32/E for supporting this research.
References
1. Ziyou Song, Heath Hofmann, Jianqiu Li, Jun
Hou, Xuebing Han, Minggao Ouyang, Energy
management strategies comparison for
electric vehicles with hybrid energy storage
system,www.elsevier.com/locate/apenergy.
2. Jesús Armenta*, CiroNú~nez, Nancy Visairo,

Isabel L_azaro, Anadvanced energy
management system for controlling the
ultracapacitor discharge and improving the
electric vehicle range
3. Mid-Eum Choi, Seong-Woo Kim, and Seung-

Woo Seo, Energy Management Optimization
in a Battery/Supercapacitor Hybrid Energy
Storage System, IEEE Trans. On Smartgrid,
vol.3, no.1, pp 463 – 472, Mar. 2012
8
348
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[349 353],2017
2017
The Institution
Lanka
Pitch Control of a Twin Rotor System using Error

Dynamics Based Nonlinear Controller
Yasitha Liyanage, Jagannath Wijekoon, Shirantha Welikala and
Lilantha Samaranayake
Abstract: In this paper we address the pitch axis control of the standard twin rotor multi-input-
multi-output (MIMO) control system problem. Dynamic model of the pitch control system is derived
using basic laws of physics, which results in a nonlinear plant model.
Some of the plant nonlinearities are able to linearize, while some, such as squared input, are not. In
the conventional approach, this problem has been solved by linearizing the possible nonlinearities,
while neglecting the others, at the cost of reduced performance. Most importantly, the latter partially
linearized approach demands linearized plant model at every operating point, which makes the
controller implementation complicated when considering a wide dynamic operating range.
In order to overcome limitations of the partially linearized approach, we derive a nonlinear controller.
In that we use the tracking error dynamics to arrive at a compromise between the tracking
performance of the pitch axis control and the smoothness of the actuator input.
The plant and the nonlinear controller have first been simulated in the MATLAB/SimulinkTM
environment and secondly, it has been implemented in hardware to control the pitch of an actual
Twin Rotor MIMO system.
The performance is assessed using Integrated Absolute Error (IAE) and Accumulated Squared Input
(ASI) representing the energy input. The performance of the controller is compared with a PID
controller tuned to the partially linearized plant model and the results show approximately 50%
improvement both in terms of IAE and ASI.
Keywords: Error dynamics, linearized plant model, nonlinear control, pitch control.
1. Introduction Several models have been proposed for the

twin rotor MIMO system. The conventional
Recent times the development of several approach is to linearize the nonlinearities while
approaches for controlling the flight of air neglecting some which in turn reduces the
vehicle such as helicopter and unmanned air performance of the system. Furthermore, the
vehicle (UAV) has been studied frequently. The plant model of the system has to be linearized
modeling of the air vehicle dynamics is a highly at every operating point in the partially
challenging task due to the complicated linearized model. Thus it complicates the
nonlinear interactions among the various implementation of the controller in a wide
variables and also there are certain states which operating range.
are not accessible for the measurement. This paper proposes a nonlinear controller to
The twin rotor multi input multi output system the pitch control based on state feedback
(TRMS) is an experimental set-up that linearization technique in order to overcome
resembles with the helicopter model. The TRMS the limitation caused by partially linearized
consist of two rotors at each ends of the model.
horizontal beam known as main rotor and tail Mr. Yasitha Liyanage, Student Member IEEE, Student
rotor which is driven by a DC motor and it is Member of IESL, B.Sc.Eng. (Hons) Peradeniya, Temporary
Instructor, Department of Electrical and Electronic
counter balanced by a pivoted beam. The TRMS Engineering, Faculty of Engineering, University of
can rotate in both horizontal and vertical Peradeniya.
direction. The main rotor generates a lift force Eng. Jagannath Wijekoon, B.Sc.Eng.(Hons) Peradeniya,
due to this the TRMS moves in upward Temporary Instructor, Department of Electrical and
Electronic Engineering, Faculty of Engineering, University
direction around the pitch axis. While, due to of Peradeniya.
the tail rotor TRMS moves around the yaw axis. Eng. Shirantha Welikala, Student Member IEEE, AMIE
However TRMS resembles with the helicopter (SL), B.Sc.Eng. (Hons) Peradeniya, Research Assistant,
Department of Electrical and Electronic Engineering,
but there is some significant differences Faculty of Engineering, University of Peradeniya.
between helicopter and TRMS. In helicopter, by Eng. (Dr.) B. G. L. T. Samaranayake, Senior Member
changing the angle of attack controlling has IEEE, AMIE(SL), B.Sc.Eng.(Hons.) Peradeniya, Tech. Lic
been done, while in TRMS it has been done by & Ph.D. KTH Sweden, Senior Lecturer, Department of
Electrical and Electronic Engineering, Faculty of
changing the speed of rotors. Engineering, University of Peradeniya.
1
349
The plant is first simulated in Friction forces momentum (M B ) can be
MATLAB/SimulinkTM environment and estimated using the following equation,
secondly, it was implemented in hardware to ̇ ̇ …(4)
control the actual system. The performance is
assessed using Integrated Absolute Error (IAE)
and Accumulated Squared Input (ASI) The gyroscopic momentum (M G ) can be
representing the energy input. described by the following equation,
̇ ….(5)
This paper is presented as follows.
Methodology section describes the plant model, The motor and the electrical control circuit is
nonlinear control derivation, and performance approximated by a first order transfer function
evaluation scheme. It is followed by the thus in Laplace domain the motor momentum
implementation of the simulation model and is described by,
experimental setup. The result and discussion
section compare the results of simulation ….(6)
results and experimental results. Conclusions Where u1 is the input signal.
are presented in the final section.
Similar relationships can be derived to the
2. Methodology horizontal plane motion. The net torque
produced in horizontal plane motion is
2.1 Plant Model described by the following equation,
The Twin Rotor MIMO system has two degrees ̈ …..(7)
of freedom namely, the vertical axis and
horizontal axis. The mechanical unit is Where,  is the angular acceleration in the
presented in Figure 1. horizontal plane and MR is the cross reaction
momentum.
The nonlinear static characteristic of the main

rotor (M2) given by,
…..(8)
Frictional forces momentum is calculated as the

same way as the main rotor
̇ ̇ …..(9)
The cross reaction momentum MR is described

Figure 1 - Twin Rotor MIMO System Model. by,
Using basis laws of physics, the model of the …..(10)

Twin Rotor MIMO system can be derived as
follows. By applying the Newton’s second law The dc motor with the electrical circuit is given
of motion to the vertical movement, it can be by,
obtained that
̈ ….(1) …..(11)
Where, ̈ is the angular acceleration in the Table 1 - TRMS Model Parameters [6]
vertical plane. M1 is the nonlinearity caused by
the rotor and it is estimated as a second order Parameter Value
polynomial. A torque is induced due to this. I1 - moment of inertia of 6.8×10-2 kgm2
…(2) vertical rotor
I2 - moment of inertia of 2×10-2 kgm2
The weight of the system produce a horizontal rotor
gravitational torque around the pivot point – static characteristic 0.00135
(MFG) which is given by, parameter
…(3) – static characteristic 0.0924
parameter
– static characteristic 0.02
2
350
parameter Therefore the control law can be written as,
– static characteristic 0.09
parameter ̈ .(16)
– gravity 0.32 Nm ̇
momentum Such that error dynamics become linear and in
– friction 6×10-3 Nms/rad the form of controllable canonical form. Here c1
momentum function and c2are constants which can be adjusted to
parameter have desired pole locations for second order
– friction 1×10-3 Nms2/rad linear error dynamics.
momentum function ̇ ….(17)
parameter ̇ .…(18)
– friction 1×10-1 Nms/rad
momentum function Error dynamics can be written as,
parameter ̇
[ ] [ ] [ ] …..(19)
– friction 1×10-2 Nms2/rad ̇
momentum function The characteristic equation of error dynamics
parameter can be written as,
– gyroscopic 0.05 s/rad ….(20)
momentum parameter By placing the poles on desired locations we
– motor 1 gain 1.1 can find out the values for and .
– motor 2 gain 0.8
– motor 1 1.1 2.3 Performance Evaluation Scheme
denominator Derived nonlinear controller is compared
parameter against conventional PID controller designed
– motor 1 1 using classical techniques. In order to compare
denominator the performance of the nonlinear controller two
parameter methods are used. They are Integrated
– motor 2 1 Absolute Error (IAE) as defined in equation 21
denominator representing the tracking performance and
parameter Accumulated Squared Input (ASI) defined in
– motor 2 1 equation 22 representing the energy input to
denominator the plant. The performance of both the partially
parameter linearized PID controller and the nonlinear
controller are evaluated and compared using
– cross reaction 2
the latter indexes.
momentum parameter
– cross reaction 3.5 ∫| | ….(21)
momentum parameter
– cross reaction -0.2
momentum gain ∫| | ….(22)
2.2 Nonlinear Controller Derivation

According to the plant model the horizontal 3. Implementation
plane motion is given in the equation 1 and the
control input to the vertical plane motion is 3.1 Software Model
which is given in the equation 2. Prior to implementing in hardware, the
Consider the error dynamics of the system complete system consisting of the Twin Rotor
where MIMO system and the controller are is
….(12) implemented in MATLAB/SimulinkTM
Here is the desired yaw angle of the system. environment. The performance of the new
̇ …(13) controller is tested for tracking and energy
Then, consumption. The simulation of the PID setup
̇ ̈ ̈ ….(14) is presented in Figure 02 and the simulation
The ̈ can be derived from equation 7. Hence ̇ setup of the nonlinear controller is presented in
can be written as, Figure 3.
̇
….(15)
̈
3
351
Figure 2 - Simulation Model of the PID Controller
Figure 3 - Simulation Model of the Nonlinear Controller
3.2 Experimental Setup Table 2 - Comparison of the results of PID

Both the PID controller and the nonlinear Controller and the Nonlinear Controller
controller ran on the experimental setup and
the performance of the both systems were Index PID Controller Nonlinear
compared. Controller
IAE 323.2708 46.0165
4. Results and Discussion ASE 10507 3910.9
4.1 Simulation Results Figure 4 represents variation between actual

Table 2 compares the simulation results of the pitch angle and the desired pitch angle with the
PID controller and the nonlinear controller. PID controller. ( Kp=5 Ki=8 Kd=10)
4
352
0.8
(a) 5. Conclusions
Actual Pitch Angle
0.7
Desired Pitch Angle
0.6
The conventional PID controller is

Pitch Angle/rad
implemented on a linear system where the
0.5
nonlinearities are neglected through

0.4
0.3
0.2
assumptions. In the nonlinear controller design
0.1
all the nonlinearities of the system were taken
0
0 10 20 30 40 50
Time/s
60 70 80 90 100
into consideration and they were cancelled off
in controller design process. Therefore
(b)
compared to the PID controller the system’s
0.3
Actual Pitch Angle

Desired Pitch Angle
performance was increased and the energy
0.2
input of the system was reduced.

Pitch Angle/rad
0.1
References
-0.1
-0.2
1. Pandey, S. K., and Laxmi, V., “Optimal Control

-0.3
of Twin Rotor MIMO System Using LQR

-0.4
0 10 20 30 40 50 60 70 80 90 100
Time/s
Figure 4 – Simualtion Results of the PID Controller Technique,” in Computational Intelligence in Data
(a) and Simualtion Results of the Nonlinear Mining - Volume 1, vol. 31, L. C. Jain, H. S.
Controller (b) Behera, J. K. Mandal, and D. P. Mohapatra, Eds.
New Delhi: Springer India, 2015, pp. 11–21.
4.2 Experimental Results
2. Pandey, S. K., and Laxmi, V., “Control of Twin
Table 3 compares the experimental results of
Rotor MIMO System using PID Controller with
the PID controller and the nonlinear controller. Derivative Filter Coefficient,” in Electrical,
The experimental results of the PID controller is Electronics and Computer Science (SCEECS), 2014
presented in Figure 5.(Kp=5 Ki=8 Kd=10) IEEE Students’ Conference on, 2014, pp. 1–6.
Table 3 - Comparison of Experimental Results 3. Prasad, G. D., Manoharan, P. S., and

of PID Controller and Nonlinear Controller Ramalakshmi, A. P. S., “PID Control Scheme for
Twin Rotor MIMO System using a Real Valued
Index PID Controller Nonlinear Genetic Algorithm with a Predetermined Search
Controller Range,” in Power, Energy and Control (ICPEC),
IAE 28.3148 21.2901 2013 International Conference on, 2013, pp. 443–
448.
ASE 1300.8 1054.5
4. Ramalakshmi, A. P. S., and P. S. Manoharan,
0.8
Actual Pitch Angle

(a) “Non-Linear Modeling and PID Control of Twin
Desired Pitch Angle
Rotor MIMO System,” in Advanced
0.7
Communication Control and Computing

0.6
Pitch Angle/rad
Technologies (ICACCCT), 2012 IEEE International

0.5
Conference on, 2012, pp. 366–369.

0.4
0.3
0.2
0.1 5. Wen, P. and Li, Y., “Twin Rotor System

0
0 10 20 30 40 50 60 70 80 90 100 Modeling, De-Coupling and Optimal Control,”
Time/s
in Mechatronics and Automation (ICMA), 2011
0.7
Actual Pitch Angle

(b) International Conference on, 2011, pp. 1839–1842.
0.6
Desired Pitch Angle
6. “TRMS 33-949S User Manual.” Feedback

Pitch Angle/rad
0.5
0.4
Instruments.
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100
Time/s
Figure 5 – Experimental Results of the PID

Controller (a) and Experimental Results of the
Nonlinear Controller (b)
5
353
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[355 362],2017
2017
The Institution
Lanka
Museum Navigation System with Beacons

N.D.C.D. Kannangara and W.D.S.S. Bandara
Abstract: Bluetooth Low Energy (BLE) is a new specification of Bluetooth available for all new
smart-devices. It is a new low power consumption technology aimed to transmit a small amount of
data. In addition, to be able to transmit data, BLE can be used to locate objects. iBeacon® protocol,
which uses BLEframe to navigate and find the artefacts. This is a new technology, and as yet it is
unclear how it will be used. In conventional museums, information or explanation of the exhibited
artefacts is presented by using panels or leaflets. Some museums are complicated places to navigate
within. They may not have a clear path through the museum and may offer many alternatives. The
problem is complicated even further given the fact that a visitor usually has limited time for a visit
and may miss some important place. However, such paper maps are inconvenient and not easy to use,
especially when a group of visitors is visiting the museum together. The objective of this project was
to implement an Android app ―Museum Navigation System‖ based on Bluetooth Low Energy
beacons. Here an Android app illustrates approximate location and information on the smart device
by using Received Signal Strength Indicator (RSSI) power and BLE frame (Universally Unique
Identifier (UUID), major and minor). In the developed system three beacons were clustered to find the
position. The system starts with a scan for beacons and then saving the RSSI power values in an array.
Then sort the RSSI power in the descending order. RSSI-based triangulation method is used to find the
position in the cluster. When the user moves to the next cluster region, it identifies route by the
scanning. Then the user coordinates of the position and direction on the map are displayed by the
Museum Navigation System app.
Keywords: Bluetooth Low Energy, BLE proximity location, Beacons, RSSI power, UUID,
Triangulation.
1. Introduction This is a new technology, and as yet it is

unclear how it will be used. Therefore the aim
In conventional museums, information or of the project is to develop a museum
explanation of exhibited materials is presented navigation system with beacon’s delivering
by using panels or leaflets. Some museums are context-based information in museums.
complicated places to navigate within. They
may not have a clear path through the museum 2. Literature Review
and may offer many alternatives. The problem
is complicated even further given the fact that a Beacon is a small computer. Its 32-bit ARM®
visitor usually has limited time for a visit and Cortex CPU is accompanied by the
missed some important place. The classic accelerometer, temperature sensor, and what is
navigation aid is the paper map of the museum, most important—2.4 GHz radio using
which is based on the museum floor plan and Bluetooth 4.0 Smart, also known as BLE or
enables the visitor to orient themselves and find Bluetooth low energy. Beacon other application
the way in the museum. However, such paper is retail, stadium and airport navigation
maps may be inconvenient and not easy to use, systems. But those applications are in trail
especially when a group of visitors is visiting condition.
the museum together. Current mobile
technology opens new possibilities for Biggest Beacon manufacturers are BlueCats,
supporting indoor navigation. BlueSenser, Estimote, Gelo, Kontakt and
Sensorberg.
Beacons are small wireless sensors that
communicate with Bluetooth-enabled smart
devices such as smartphones or tabs by Eng. N.D.C.D. Kannangara, B.Tech.( Hons) in Eng.
continuously advertising their location using a (OUSL), Dip in Software Eng.
Bluetooth low energy radio transmitter. In turn,
smart devices monitor the received signal Eng. W.D.S.S. Bandara, B.Sc.(Eng.) Hons.(Peradeniya),
M.Eng (AIT) Lecturer, Department of Electrical &
strength indication (RSSI) and determine the Computer Engineering, The Open University of Sri Lanka.
device's proximity to the beacon.
1
355
Beacons are low energy Bluetooth device. 3.3 Build communication links
Therefore, it uses Bluetooth protocol stack to Beacon frames are broadcasted by the Access
connect network and communicate like any Point. it is in Basic service set of infrastructures.
network, Bluetooth has protocol layers that The Android platform consists Bluetooth
define the network and allow it to function. network stack, which allows a device to connect
wirelessly and exchange. The Android
2.1 Similar museum navigation systems Bluetooth APIs provided to access the
application through Bluetooth. The beacon
Augmented Reality Technologies introduced a frame is one of the frames in IEEE 802.11.
system with a head-mounted display (HMD)
which is introduced by Noboru Koshizuka. The 4. Conceptual Design
issues and disadvantages of this system are
high cost, difficult to use for everyone, limited Museum Navigation System is a novel system
no of HMD to use and need a person for an which gives the navigation informationinside a
explanation. museum to use with a smart device. Here the
developed Android application illustrates
PDMA(Personalized Digital Museum Assistant) navigation route and the artefact information
is a tool and implementation of relatively high- on the smart device by using RSSI power,
speed wireless LANs which is introduced by UUID, major and minor of the beacon frame.
Ken Sakamura. Issues and disadvantages are
high cost and limited no of PDMAs.
Start
3. Methodology
Beacon’s configuration Initialize boundary

positions.
Design a mobile App

Read UUID, Major,
Minor and RSSI
Build communication links power
Figure 1 – Project steps
3.1 Beacon’s configuration Check UUID and

get RSSI power to
The paired Bluetooth Beacons transmit with decide cluster’s Tx
exact intervals (e.g. at about 100ms or 1 s). positions.
Beacons transmit their ID, which is contained in
three parts, they are UUID (organization),
Major (a specific group of items) and Minor Calculate distance
(specific item). For example, whole museum to Tx positions.
area and location identify by UUID, a large
group of items identify by major (King’s
jewellery), and the specific beacon identifies by
Calculate the x
a minor(crown).
and y
coordinates and Display Beacon
3.2 Designing the Mobile Application
moving information on
direction app
Bluetooth Low Energy technology is known as
Bluetooth Smart technology. It support from
Android 4.3 Jelly Bean version. Android
smartphone requirements are greater than Display current Check Major
Android version 4.3 and Bluetooth Smart radio position on map and Minor
unit (v4.0)
Figure 2 – Main flow chart of the app
2
356
The power of a Beacon’s signal is captured experiments were carried out to get the
byRSSI and Measured Power. Broadcasting combination. However the, Bluetooth
Power is the power which it transmits its signal indicators fade with distance under these
by beacon BLE module. Admin of the Beacon is environmental influences. Let
able to change RSSI power. If the power is high,
the range of the beacon is large, and it has a Therefore:
high ability of stable the signal, on the other ( )
hand, battery life is shorted. where is the received RSSI power at .
Here three Beacons Cluster is used to find the 4.2 Improved Algorithms for Beacon
position. The first scan for beacons and then Navigation system
RSSI power values are saved in an array. Then
they are sorted ascending order of RSSI power. The algorithm is based on three Beacons which
RSSI-based triangulation method is used to find can communicate with the smart device and
the position in the cluster and when the user those Beacons measures the signal coming from
moves to next cluster region it identifies by the smart device.
scan route. Then it identified new beacon and
previous beacon use to find anew position in
the new region. Scan run within constant time
and update beacons initial position with user
position on the map.
Figure 4 – Measuring the signal
Let beacons coordinates as ( ), ( )

and ( ) and then
( )
( )
APx
( )
Where r is represent as:
Figure 3 – Cluster plan
Solving (1) – (2) and (2) – (3), the above
4.1 Bluetooth RSSI vs Distance
equations reduced to two equations as:
( ) ( )
Anti et al. design and implementation of a
Bluetooth Local Positioning Application in ( )
which the RSSI power level is converted to ( ) ( )
distance model as follows:
( )
( ) Let:
( ) ( ) ( ) ( )
( ) ( )
Where is the transmit power; and

So that;
are the antenna gains and is the total
( )
antenna gain: , is light ( )
speed, is the central frequency( )n is
the attenuation factor (2 in free space), and is Finally, the coordinate of the device being
the distance between transmitter and receiver tracked is:
in meters. Then
( ) ( )
( )
Due to multipath and noise, the relationship
between and distance will differ from The actual motion is the very important things
above. Further the, more environmental factors in order to improve model accuracy. In this
too effect for the RSSI power. In the case several case, higher order equation is used to
3
357
determine the distance that user moves in the 4.4 Android RSSI power test application
time period. (Scan time duration is used).
Android test application used to get
power to find true values. Therefore
( )
considering three beacons and found true
( ) values to simulate and to get the results.
Where,
– distance, – velocity, – acceleration, – Block diagram for the RSSI testing app on
variable acceleration and – rate of variable Android device.
acceleration.But in this application 3rdorder
equation is used to indicate motion. For the
velocity
( ) ( )
( ) ( )
For the acceleration Figure 6 – RSSI test application block diagram
( ) ( )
( ) ( )
5. Implementation
For the variable acceleration
( ) ( )
( ) ( ) The methodology used in this project, as well as
Assuming all beacons transmit the same signal the positioning algorithms,has been covered
power and therefore the average of velocity, too. All these things are connected together and
acceleration and etc. can be obtained by implemented into a navigation system which is
substituting equation 8 and 9.Get the values of consisting of a group of configured beacons and
which are related to and finally calculate an Android application of the final product.
position of and . Every dashed values are
previous values and scan time. The system works like this: at first, few beacons
are positioned as boundary and other used as
Error of the motion and distance ~ artefact transmitters. The app scans beacons
( ) and received their RSSI readings. And calculate
( ) the distance between the user and each beacon.
The pointer rotates to indicate the user moving Then app calculated and coordinates to
direction by using current and previous indicate direction of the user moving by
coordinates. substituting distance. After that beacons’ and
the user’s position displayed by app, as well as
( ) ( )
the distance between them, are shown in the
map.
Figure 5 – Direction of pointer
4.3 Error calculation
It is used to find the error between actual point

and calculated point. It is used a measure of the
differences between values of predicted and
actually observed. the standard deviation of the
differences between predicted values and
observed values represented by it.
Figure 7 – Screen prints of app
∑ √( ) ( )
4
358
5.1 Testing environment 5.4 User’s movement
The app was tested inside a part of Now the application estimates the distance of
department with the distance of three beacons, the user each of the beacon and indicates on the
separated by approximately , were map by displaying a black arrow with a man,
positioned. The map of the area as well as the such as navigate pointer, user and user
position of the beacons is shown in figure. movement. Indicate the user navigation area.
The setting of the beacons and position According to the calculation, the user location,
environment is important to increase the testing distance to the closest beacons and the round
value the accuracy of the system. Because of its beacons how as accurately as possible on the
capability to calculate the distance accuracy map. However, the user position is located
with low distance and high power transmit relative to the configured beacons is shown in
beacons. This situation will give an idea of how figure 6.4
beacons should be configured regarding their
output power. Which is affected the range of The thing is important to have in mind that the
receive readings collected by the app. For pointer is roughly showed the user position.
example, if the area is less than in this The accuracy of the distance calculations differs
case no problem of using several beacons from the pointer the actual position of the
operating with 0 dBm transmit power with beacon. Because of multipath receive and noise
around range. This will only effect of the level. On the other hand environment facts
beacons battery life and affect the accuracy of effect to the system. This inaccuracy may be
the distance calculations as values will tolerated due to the position. It is not too far
vary. It is able to increase update speed by from reality, and the user can get an idea of
reducing transition frame intervals. where he/she is.
5.2 The Android Application
At the start, the dialogue is displayed on the

screen to get the permission from the user to
turn on the Bluetooth adapter. On the other
hand, the dialogue is professional and quickly
starts the scan.
Figure 9 – Screen prints of navigation app
Figure 8 – Bluetooth permission request 6. Simulation Results

dialogue
Here used to get distance from measured RSSI
5.3 Scanning power and accuracy of the improved algorithm
of the Beacon navigation system for simulation
The appscans BLE devices and highest 3 purposes. To get the measured RSSI values a
power in every scan are passed through an separate ―OUSL App by Chanuka‖ Android
algorithm of positioning. Then from that app was used.
previous and current values calculate direction
of moving and display on the map. And context The RSSI power values were collected by
base data display on the screen by using major changing Beacons and getting 20 values. Then
and minor. decided the mean value for each considered
distance.
AMATLAB Simulation was used to plot the
data and to obtain the results.
5
359
Table 1 – Static result for data
Coefficients (with 95% confidence

Distance bounds): Goodness of fit:
Adjusted
(m) P1 P2 SSE: R-square: R-square: RMSE:
0 -0.003 -55.07 11.79 0.0005 -0.055 0.81
(-0.068, 0.06294) (-55.9, -54.23)
0.5 -0.0097 -85.55 10.49 0.006 -0.049 0.763
(-0.072, 0.0524) (-86.3, -84.8)
1 -0.027 -90.62 11.31 0.04 -0.012 0.793
(-0.092, 0.037) (-91.4, -89.84)
1.5 -0.015 -95.08 9.584 0.017 -0.038 0.73
(-0.075, 0.0436 (-95.1, -94.37)
2 0.044 -97.22 8.441 0.13 0.086 0.685
Figure 10 – Matlab simulation screenshot (-0.0114, 0.1002) (-97.8, -96.55)

3 0.0067 -97.82 5.72 0.005 -0.049 0.564
(-0.039, 0.0527) (-98.3, -97.27)
4 0.012 -98.93 9.104 0.01 -0.04 0.711
Signal (-0.045, 0.069) (-99.6, -98.23)
Strength 5 0.01504 -99.86 4.05 0.035 -0.018 0.474
(dBm)
(-0.024, 0.05368) (-100, -99.39)
6.1 Decision for getting value of RSSI

power at 1m
It is very important to get ―A― value for
logarithm function.
Signal
Streng
th
(dBm)
No. of sample
Figure 11 - RSSI power at 0 m
Signal
Strength
(dBm)
No. of sample
Figure 14 – Analysis of fit 1m for data set
Therefore,
( )
( )
No. of sample
From the above data, ; which is the
Figure 12 – RSSI power vs distance
attenuation constant and it is depend on the
environment.
Linear model Poly1: ( )
Signal 6.2 Simulation Results
Strength
(dBm)
Distance (m)
Figure 13 – Signal Strength vs Distance

Figure 15 - Actual path of Smart device
6
360
6.5 Future works
Furthermore, the developed app will improve

and will eventually be published in the
―playstore‖. Future work will focus on
multipath receiving and improve the app
functionality. Making the system stability for
different environments will be a top priority to
increase results of the distance calculations
previously this system can only relate to a very
specific testing environment.
Figure 16 - Points measured by Smart device 7. Conclusions
iBeacon consist with apple simple ios and

software protocols. In this case, BLE interface
and linking problems were faced. No iBeacon
support libraries were released for Android.
But it was available for ios.
According to the well-established concept of

outdoor navigation, indoor navigation must
still face many challenges. For example is the
limited technologies that can be used for it, as
well as the accuracy. Furthermore, how to build
indoor navigation systems available to the
market is still the challenge.
Figure 17 - Final points
6.3 Root Mean Square Errors The system will be more useful and will have
RMSE of reference = 12.1741 access to these, and they will become part of the
everyday life.
6.4 Product Test Result
The product was tested at Research Laboratory, Acknowledgement

Electrical and computer department.
Authors wishes to pay their gratitude to the
faculty of Engineering Technology, the
Department of Electrical and Computer
Engineering, the Department of Mathematics
and Philosophy of Engineering, The Open
University of Sri Lanka for the facilitate they
received in carrying out this project work.
References
1. Heydon, Robin, 2013, Bluetooth Low

Energy:The Developer’s Handbook, New
Jersey: Pearson.
2. Gupta, Naresh, 2013, Inside Bluetooth Low

Energy,U.S.A: Artech House .
3. Townsend, K., Cufi, C., Akiba and Davidson,

R., 2014, Getting Started with Bluetooth Low
Energy, Califonia: O’Reilly Media.
4. Bluetooth, 2016. Core specification v4.1.

Figure 18 - Screen prints of product test [online] Available at: <
7
361
https://www.bluetooth.org/en-
us/specification/adopted -
specifications>[Accessed 3 December 2013].
5. Bluetooth, 2016. Supplements to the Core

Specification (CSS) v6. [online] Available at: <
https://www.bluetooth.org/en-
us/specification/adopted -
specifications>[Accessed 14 July 2015].
6. Infoworld, 2014. What you need to know about

using Bluetooth beacons. [online] Available
at:<
http://www.infoworld.com/article/2608498/
mobile-apps/what-you-need-to-know-about-
using-bluetooth-beacons.html>[Accessed 22
July 2014].
7. Umbel, 2016. 15 Companies From Airports to

Retail Already Using Beacon Technology.
[online] Available at:<
https://www.umbel.com/blog/mobile/15-
companies-using-beacon-technology>.
8. Bluetooth, 2016. Core System Architecture.

[online] Available at:<
https://developer.bluetooth.org/TechnologyO
verview/Pages/Core.aspx>[Accessed 2016].
8
362
pp. [1 -- 10],
[363 372],2017
2017
© The
The Institution
Lanka
Adaptation of Scasa Technique for Surge Protective

Devices for Equatorial Belt Countries
J.P.D.S. Athuraliya, N.L. Rathnayake, N.A.A.N. Dilrukshi, P. Mahadevan, R.
Gihara, T M.S. Dias, S.R.J.S. Bandara, H. Wijesooriya and Nihal Kularatna
Abstract: Modern electronic systems are very susceptible to failures due to high energy
transients superimposed on the power input or the signal ports. In equatorial belt countries like Sri
Lanka, lightning activity is high and it causes severe damage to equipment. To safe-guard electronics
from such high energy transients, surge protection devices (SPD) rated to absorb repeated surges are
required. Traditionally, SPDs use cascaded multiple non-linear devices (NLD) to clamp the transient
overvoltage created by a transient. However, typical NLDs have energy absorption ratings based on
short–term basis, and they have a limited life time.
Supercapacitor assisted surge absorber (SCASA) technique is a patented novel technique, where a
very large capacitor’s continuous energy absorption capability given by ½ CV^2 could be effectively
coupled with NLDs and filters and a multi-winding magnetic component for a low component count
based SPD where repeated transients can be absorbed.
In adopting this technique for absorbing severe transients with several orders of Joules dumped into a
SPD, on a repeated basis, selection of the supercapacitor and the multi-winding magnetic core is a
very critical design-task. Since the combination of the leakage and magnetizing inductances of the
multi-winding magnetic core plays a dominant role, selection of the core is critical.
In absorbing repeated Class-B type surges, in equatorial belt locations like Sri Lanka, specific attention
is required to come up with a reliable SPD design and this paper presents the design details and test
results for an early-version of a prototype design of a differential-mode surge protector based on the
SCASA technique. Experimental results generated using a lightning surge simulator with surge
capability up to 6 kV/3 kA is used to validate the results. Performance of this technique with
optimized magnetics is compared with some selected commercially available surge protectors sold in
Sri Lanka. Test results clearly indicate, the developed device has a much higher energy absorption
capacity than tested commercial products and can be used in commercial surge protectors.
Keywords: Lightning Protection, Supercapacitors, Metal Oxide Varistors, Bidirectional Break-

over Devices
1. Introduction Eng. J.P.D.S. Athuraliya, C.Eng., MIESL, MIET, MSc, BSc Eng
(Hons), Director (Electronics Engineering), Electronic and
Microelectronic Division, Arthur C Clarke Institute for Modern
Lightning is a huge discharge originating in Technologies(ACCIMT), Katubedda,Sri Lanka
cumulonimbus clouds. As seen in figure 1, Eng. N.L. Rathnayake, BSc(Eng) in Electronic and
Telecommunication, MEng, Senior Research Engineer, Electronic and
they can occur within a cloud, between clouds, Microelectronic Division, Arthur C Clarke Institute for Modern
down to earth or just expend their charge in Technologies(ACCIMT), Katubedda,Sri Lanka
air. Fortunately in the tropics, only about 10 % Eng. N.A.A.N. Dilrukshi, BSc(Eng) (Equivalent) Electronic
Communication and Computer Eng , PG Dip (Industrial Automation),
of the discharges terminate on the earth or AMIESL, Research Engineer, Electronic and Microelectronic Division,
earth bound objects. However, even this 10% Arthur C Clarke Institute for Modern Technologies(ACCIMT),
can cause a disaster, and these disasters have Katubedda,Sri Lanka
Eng. P.Mahadevan, BSc(Eng) Electronic and Communication Eng
occurred in a cyclical pattern in the world[1]. (Hons) , AMIESL, Research Engineer, Electronic and Microelectronic
Division, Arthur C Clarke Institute for Modern
Technologies(ACCIMT), Katubedda,Sri Lanka
Eng. R Gihara ,BSc. Electrical and electronics Eng. (Hons) , AMIESL,
Research Engineer, Electronic and Microelectronic Division, ACCIMT,
Katubedda,Sri Lanka
Mr. T.M.S. Dias, NCT(Elect), Engineering Assistant, Electronic and
Microelectronic Division, Arthur C Clarke Institute for Modern
Technologies(ACCIMT), Katubedda,Sri Lanka
Mr. S.R.J.S. Bandara, IVQ Level 06, MIET, Engineering Assistant,
Electronic and Microelectronic Division, Arthur C Clarke Institute for
Modern Technologies(ACCIMT), Katubedda,Sri Lanka
Ms. H Wijesooriya, BSc Physical Science ,Engineering Assistant,
Electronic and Microelectronic Division, Arthur C Clarke Institute for
Modern Technologies(ACCIMT), Katubedda,Sri Lanka
Eng. (Prof.) N Kularatna, BSc Eng(Hons), DSc, C.Eng, FIENZ,FIET,
Figure 1 - Propagation of lightning channel FIE, SMIEEE,Associate Professor in Electronic Engineering, The
University of Waikato, New Zealand
363
Lightning causes damage to buildings, during monsoon lightning times in equatorial
electronic and other equipment as well as belt countries like Sri Lanka and many other
causing injury and even death to people and parts of the world due to their failure of
livestock. This may be compared to typical components, such as capacitors. The use of
temperate climates where the thunder days high value capacitors (supercapacitors) could
may be low, around 25 or 30 per year [2]. Since probably overcome some of these problems as
the majority of high technology specialized they reduce the spikes.
military, communications, navigational and
A supercapacitor (SC) is a high-capacity
switching equipment is designed and
generally manufactured in these temperate electrochemical capacitor, with capacitance
values much higher than other capacitors, but
countries, scant regard is often paid to the
need to protect this equipment from the in lower voltage limits. It has the ability to
absorb high-voltage (HV) transient surges with
devastating effects of lightning strikes whether
they be direct or indirect. For this reason a short-duration occurrence[4].
lightning protection against both direct and
Early researchers[5] have shown that a
indirect lightning strikes at critical
communications and navigational aid sites in supercapacitor is not destroyed by the
repeated application of HV transients and the
tropical regions of the world should perhaps
be mandatory[2]. gradual voltage rise across terminals after each
hit is in the order of millivolt This also
In most tropical countries, lightning and storm indicates that the device still retains its
activity is high, compared to the more capacitive behaviour and not adversely
temperate regions of the world. For example in affected by the transient HV at the terminals.
the equatorial belt, ten degrees north and Therefore a supercapacitor can be used to
south of the equator, thunder day statistics absorb part of the surge energy in the SPD
may vary from 150 to 200 days per year[2] . during a transient traveling through the
incoming mains or the telecom/data circuits.
In order to protect buildings against lightning,
a structural protection system should be
installed at the building[3] . These do not protect
2. Methodology
domestic appliances which can usually be 2.1 Background
protected from lightning currents by
unplugging them from service lines during Researchers have shown[5] that supercapacitors
thunderstorm periods. However, such an act is could have continuous energy storage
not realistic in most of the industrial capabilities in the range of energy carried in a
andservice sectors as even a short period of transient surge into an electrical circuit. They
out-of-operation could cost the company a few are comparable with the transient energy
million rupees. Therefore in such cases, absorption capabilities of non-linear devices
lightning surges should be prevented from used in typical surge protectors such as metal
entering the building, or in particular sensitive oxide Varistors (MOVs) and bidirectional
equipment[3]. break-over diodes (BBDs) , coupled with LC-
type filter stages.
This is done by installing surge protection
devices (SPDs) to the power and However, at present commercially available
communication lines[3]. In the case of surge supercapacitors have very low DC voltage
protection, both the quality of the product and ratings, such as less than 4 V for single-cell
the engineering of installation are equally devices. (Ex: 2.5V 1 F, 2.7V 5F). This voltage is
important. It should be remembered that the far below the instantaneous voltages occurring
provision of lightning protection both against on the AC mains. Given this problem, a surge
direct strikes and indirect effects will only protector cannot just substitute a
improve lightning immunity[3]. supercapacitor for a MOV or any other non-
linear devices. Thus the necessity of testing is
It is unlikely that 100% protection can ever be very important with the instantaneous voltage
achieved even by proper design at an early developed across the supercapacitor sub
stage before equipment installation. It can both circuit for the entire design.
reduce later costs and substantially improve
protection in the longer term. Most common
surge protectors consist of non-linear devices
in the market and they do not work well
364
2.2 Design Approach Table 1 - Characteristics of MOVs
System design was done to achieve the Type of MOV

expected results and there by selecting the Epcos B722
Parameter
suitable components after analyzing their – S20 PANASO
characteristics. NIC
Voltage Rating V AC 230 V 230 V
Tests were carried out as per the IEC 61643-
Voltage Rating V DC 300V 300V
11[6] with the 1.2/50 voltage - 8/20 current
combination wave generator (EM TEST/UCS Clamping Voltage Vc 595V 595V
500-M) which has a maximum transient Peak Surge Current @ 8kA 10kA
voltage up to 6 kV and the maximum current 8/20µs
up to 3 kA. Testing flow chart of the voltage Operating -40°C -40°C
protection level of the class III type SPDs is Temperature Min
shown in figure 2. Operating 85°C 85°C
Temperature Max
Peak Energy 130J 255J
(10/1000uS)
2.3.1.1 Voltage buildup across MOVs
By applying 1 kV to 6 kV surges from the

lightning surge simulator, different clamping
output voltages could be observed.
Table 2 - Voltage build up across MOVs
Output Clamping Voltage (V)
Applied
+/- 20%
Surge
Epcos – B722 PANASONIC
Voltage (V)
S20
1000 510 570
4000 630 710
6000 690 760
Figure 2 - Flow chart of testing the voltage

protection level
Source : IEC 61643-11
Figure 3 – Voltage build up across Epcos-S20

2.3 Selection of components
2.3.1 MOVs
Two types of MOVs which have following

characteristics were tested using the lightning
surge simulator (UCS 500-M).
Figure 4 –Voltage build up across B722

Panasonic
365
Therefore, Epcos – S20 MOV which has the
clamping voltage nearly 600 V, was selected
for this design topology.
2.3.2 Supercapacitors
2.3.2.1 Voltage build up across super

capacitors
Two types of supercapacitors were tested by
applying different levels of transient voltages.
Terminal voltage developed across Figure 6 – Voltage build up across 5F-2.7 V
supecapacitor after several strikes were SC
measured using a multimeter. The test data set
provides some valuable insight in estimating These results indicated that the supercapacitor
the capabilities of these new supercapacitors to is not destroyed by the repeated application of
withstand surges and transients, which in turn HV transients and the gradual voltage rise
could lead to non-traditional applications[5]. after each hit is in the order of millivolt. Also
they still retain its capacitive behaviour and
not adversely affected by the transient HV at
(a) 1F – 2.5V SC (B Series)– Initial voltage the terminals[5]. Therefore, 1F-2.5 V
0.20mV supercapacitor which has the highest voltage
rise with 1 ohm resistor was selected as a
Table 3 - Voltage build up across 1F-2.5 V SC supercapacitor sub circuit for this design.
No. of mV (for mV (for mV (for
strikes 1.5 kV) 4.5 kV) 6.0 kV) 2.3.3 Magnetic Core
5 22.5 58.3 76.43
2.3.3.1 Selection of magnetic core
10 43.7 78.2 145.3
In an inductor, the core provides the flux
15 64.1 90.6 209.3
linkage path between the circuit winding and a
20 83.5 110.2 269.2 non-magnetic gap, physically in series with the
core. Virtually all of the energy is stored in the
gap. These cores approach the ideal magnetic
material characteristic – square loop with
extremely high permeability (60,000), high
saturation flux density (0.9 Tesla = 9000 Gauss)
and insignificant energy storage.
Unfortunately, resistivity of these metal alloys
is quite low. To minimize losses due to
induced eddy currents, these cores are built up
with very thin tape wound laminations[7].
Figure 5 – Voltage build up across 1F-2.5 V Ferrites are ceramic materials made by
SC sintering a mixture of iron oxide with oxides
or carbonates of either manganese and zinc or
(b) 5F - 2.7V SC (Maxwell)– SC – Initial nickel and zinc. MnZn ferrites are used in
voltage 0.26 mV applications up to l or 2 MHz and include the
power ferrite materials used in switching
Table 4 - Voltage build up across 5F-2.7 V SC power supplies. The permeability of power
No. of mV (for mV (for mV (for ferrite materials is in the range of 1500 to 3000
strikes 1.5 kV) 4.5 kV) 6.0 kV) (relative). As shown in the low frequency
5 3.7 10 22.8 characteristic of Fig. 7, a ferrite core will store a
10 4.1 15 44.9 small amount of energy, as shown by the areas
15 4.3 20 66 between the hysteresis loop and the vertical
20 4.5 24 88 axis. The permeability is high enough to keep
the magnetizing current at a generally
acceptable level in transformer applications[7].
366
a semiconductor device such as bidirectional
break-over device[8].
3.2 Main Configuration& Its Operation
Designed circuit of the surge protective device
for the differential mode based on the
supercapacitor concept is shown in Figure 8.
Figure 8 – Circuit Diagram – Differential Mode

Figure 7 – Ferrite Core Characteristic
Figure 8 – Design Circuit Diagram
Composite powdered-metal cores, such as
powdered iron, Kool Mµ, and Permalloy In this circuit, typical NLDs such as a MOV is
powder cores do store considerable energy. combined with a magnetic component and an
However, energy is not stored in the very high SC-based sub circuit. However, compared to a
permeability magnetic metal portions of the typical surge arrestor without supercapacitors,
composite, but in the non-magnetic regions where NLDs are placed directly across the
between the magnetic particles in the binder pairs of wires such as the neutral and the live
that holds the cores together[7]. (differential mode) or neutral or live and
Essentially, these composite cores store their earth(common mode) as shown in figure 9.
energy in a non-magnetic gap that is
distributed throughout the entire core. These Active
cores are manufactured and categorized by
Load
their effective permeability. Different effective Differential mode
permeability in the range of 15 to 200 (relative) Neutral

are achieved by varying particle size and the
amount of magnetically inert material in the Common mode
composite mix. Their relatively low Ground
permeability results in high magnetizing

current and energy storage undesired in a
Transformer[7]. Figure 9 - Differential and common mode
Thus, Powdered core which has more energy surges
storage capability and having relatively low
permeability has been selected for this design. As seen in figure 8, one end of the MOV is
placed after the primary winding of the coil
3. System design and the other end is connected to the return
wire. A surge occurring at the mains input
3.1 Design Overview
results in a voltage spike. At the instant the
surge voltage increases above the MOVs firing
The implemented supercapacitor assisted voltage the high instantaneous current flows
surge absorber device was developed by using through the primary of the coil which results
2.5V/1F SC, 1Ω resistor, transformer and a in an induced voltage in the secondary of the
non-linear device(MOV). SC’s continuous coil in the opposite direction. In order to
surge energy absorption capability given by 1⁄2 achieve this, it is necessary for the primary and
CV^2 could be effectively used with several the secondary coils to be wound in the
other components such as MOVs, LC filters opposite direction. This drop in voltage across
and a multi-winding magnetic components. the coils will assist in reducing the voltage
This magnetic part works as a transformer across the critical load that needs to be
when a surge travels through the power line protected. The turns ratio can be adjusted to
and fires a nonlinear device such as a MOV or suit the voltage requirements of the super-
capacitor based sub circuit[8] .
367
4.1.1 Math lab simulation results for the
4. Results and analysis sub circuit combinations
 Sub-circuit voltage variation for
In general practical circuit developments for
different capacitors
low voltage surge protector for equipment are
based on designing the protective circuit as an
add-on block to the input wiring of the
equipment. This block is used to attenuate the
incoming transient using passive series
impedances or bypassing the surge currents
and absorbing the transient energy which is
done using non-linear devices such as GDTs,
MOVs, TVS, spark gaps, fuses, etc.
In the usual surge protector devices, the high Fi
transient energy can cause these non-linear Figure 11 - Sub-circuit voltage variation for
devices to gradually deteriorate and different capacitors
eventually fail. If this energy can be taken out When capacitor value increased, voltage across
from the non-linear devices their life time will sub circuit was decreased.
be greatly enhanced.  Sub-circuit voltage variation for
different resistors (R+1F sc)
This paper presents the results of the use of
supercapacitor based circuit together with
non-linear devices to not only absorb the
transient energy, but also to control the output
clamping voltage to a lower value to give
better protection for the equipment.
Since the protection circuit is for class III SPD,
the equipment should be subjected to a 6 kV.
Thus testing has been carried out at both 4 kV
and 6 kV in the comparison. Figure 12 - Sub-circuit voltage variation for
different resistors
4.1 Impact of The Supercapacitor Sub When resistor value increased, voltage across
circuit and The Magnetic Component sub circuit was increased.
As indicated in Figure 10, a supercapacitor-  Sub-circuit voltage variation for
based subcircuit configuration could have few different combinations (R+C, C+L,
possible variations. R+C+L)
C C C
C L
R L
Figure 10 – Sub circuit combinations

Figure 13 - Sub-circuit voltage variation for
SC-based sub circuit receives the voltage different combinations
difference between primary and secondary Vsub(R+C)<Vsub(C+L)<Vsub (R+C+L)
coils. It creates a circulating current through Therefore, sub circuit with resistor, capacitor &
the sub circuit and absorbing part of the surge inductor can have the ability of minimize the
energy into the supercapacitor. With this output voltage at the load end compared with
ability of supercapacitor, transient surge other two combinations.
energy absorbed by the NLD is significantly
reduced[9]. The matlab simulation results for
the sub circuit combinations is mentioned
below.
368
4.1.2 MATLAB simulation results for the Vs = Vp+Vsub
complete circuit
By applying surge input of 6 kV, the primary
and secondary current variation of the
magnetic core is shown as in figure 14.
Vs(max) = 821 V
Figure 18 - Secondary voltage variation

Ip(max) = 2523 A Is(max) = 87A
Voc = Vsurge-Vs
Figure 14 - Primary and secondary winding
current variation
Vp = Vsurge - VNLD
Voc(max) = 667 V
Vp(max) = 794 V Figure 19 -: Open circuit voltage (No-load)
Figure 15 - Primary voltage variation
4.2 Energy Calculation
4.2.1 By using powdered core as a transformer
By applying 4 kV and 6 kV, total energy at the
input and the energy absorbed by the MOV
iscalculated using the comma-separated
values(CSV) from the oscilloscope trace.
4.2.1.1 For 4 kV Surge
Vsub(max) = 87 V
Figure 16 - Sub-circuit voltage variation
Vsub = Isub*Zsub
Figure 20 – Current, voltage & power

waveforms across MOV
Vsub(max) = 87 V
Figure 17 - Sub-circuit voltage variation
369
Figure 21 – Total current, voltage & power
Figure 25 – Total power at the input
waveforms at the input
Power Across MOV - 4 kV Total Energy = 30.8 J

1.20E+06
1.00E+06
Output clamping voltage = 720 V
8.00E+05 Energy across MOV = 25.2 J
Energy absorbed by the supercapacitor
Power(watt)
6.00E+05
4.00E+05 subcircuit = (30.8-25.2) J

2.00E+05
= 5.6 J
0.00E+00
0.00E+00
7.50E-07
1.50E-06
2.25E-06
3.00E-06
3.75E-06
4.50E-06
5.25E-06
6.00E-06
6.75E-06
7.50E-06
8.25E-06
9.00E-06
9.75E-06
1.05E-05
1.13E-05
1.20E-05
1.28E-05
1.35E-05
1.43E-05
1.50E-05
1.58E-05
1.65E-05
1.73E-05
1.80E-05
1.88E-05
1.95E-05
-2.00E+05
Time (s)
4.3 Energy Comparison of Powdered &
Ferrite cores
Figure 22 – Power across MOV Table 5 - Energy levels of powdered/ferrite

Total Power - 4 kV cores
1.40E+06
1.20E+06
1.00E+06
Energy Absorbed by SC
Power(watt)
Surge Sub-Circuit (J)

8.00E+05
6.00E+05
Voltage (V) Powdered

Ferrite Core
4.00E+05
2.00E+05
Core
4000 2.6 1.4
0.00E+00
0.00E+00
7.50E-07
1.50E-06
2.25E-06
3.00E-06
3.75E-06
4.50E-06
5.25E-06
6.00E-06
6.75E-06
7.50E-06
8.25E-06
9.00E-06
9.75E-06
1.05E-05
1.13E-05
1.20E-05
1.28E-05
1.35E-05
1.43E-05
1.50E-05
1.58E-05
1.65E-05
1.73E-05
1.80E-05
1.88E-05
1.95E-05
-2.00E+05
Time(s)
6000 5.6 5.1
Ferrite core stores less energy from the source

which will be absorbed and dissipated by the
Figure 23 –Total power at the input
SC sub-circuit. Hence, powdered core has
Output clamping voltage = 660 V
more capability of absorbing surge energy
Total Energy = 17.2 J
than the ferrite cores.
Energy across MOV = 14.6 J
Energy absorbed by the
supercapacitorsubcircuit = (17.2-14.6) J 5. Prototype implementation of
= 2.6 J the proposed system
4.2.1.2 For 6 kV Surge
5.1 Differential Mode
Power Across MOV - 6 kV
2.00E+06
1.80E+06 Final prototype implementation of the
1.60E+06
1.40E+06 supercapacitor concept in a differential-mode
protector circuit is shown in figure 26. The SC
Power (watt)
1.20E+06
1.00E+06
8.00E+05 subcircuit is formed by a 1 Ω resistor and a 1
6.00E+05
4.00E+05 F/2.5 V SC in series. The Epcos S-20 MOV
2.00E+05
0.00E+00
with a maximum clamping voltage 595 V
together with SC subcircuit-created
0.00E+00
7.50E-07
1.50E-06
2.25E-06
3.00E-06
3.75E-06
4.50E-06
5.25E-06
6.00E-06
6.75E-06
7.50E-06
8.25E-06
9.00E-06
9.75E-06
1.05E-05
1.13E-05
1.20E-05
1.28E-05
1.35E-05
1.43E-05
1.50E-05
1.58E-05
1.65E-05
1.73E-05
1.80E-05
1.88E-05
1.95E-05
-2.00E+05
Time (s) input/output voltages as shown in table 5.
Figure 24 –Power across MOV
370
When applied input surge voltage of 6 kV,
energy absorbed by the supercapacitor
subcircuit with magnetic core is nearly 40% of
total energy. Hence, supercapacitor subcircuit
plays the role of absorbing part of the surge
energy superimposed on the incoming pair of
wires and increase the overall life time of the
SPD device. Although the clamping voltage of
SPD without supercapacitor is nearly 890 V,
the supercapacitor assisted surge protective
Figure 26 - Proposed design of SPD
device has a 690 V at its output.
(Differential Mode)
As the test results clearly indicate that the
The output clamping voltage for the different
supercapacitor assisted surge protective device
voltages are tabulated as follows.
forms the basis of an entirely new on non-
traditional applications that will yield not only
Table 6 - Output voltage at load end
longer life, but also a lower clamping voltage
Applied Surge Output Voltage At across a critical load to be protected. This
Voltage (V) Load End L-N (V) proves that the new topology could be used as
2000 600 a base technique to develop full-scale common
and differential mode surge capable, fully
4000 675
versatile, commercial surge protectors with
6000 690
better performance than traditional surge
protectors with higher component counts.
Overall performance of the supercapacitor
6. Conclusions technique with optimized magnetics is
The research study has been carried out to practically used to safeguard electronic
determine the best configuration of a systems against transient over-voltage related
supercapacitor (SC) to absorb excess energy power quality issues. Further, this research can
during a transient. The study has shown that be developed to analyse the overall circuit, in
the large value supercapacitors are able to not order to predict its theoretical performance in
only handle large amounts of energy but also detail[11].
their large time constants permit this stored
energy to be released at a slow rate permitting Acknowledgement
a longer life to the surge protective device.
Compared to typical non-linear devices used The authors would like to acknowledge the
in surge protective device (SPD) circuits, staff of the Electronic and Microelectronic
supercapacitors are capable of storing a large division, ACCIMT for the great support given
amount of energy despite their very low DC during the design and development of the
voltage rating.In the usual surge protector device.
devices, the high transient energy can cause
these non-linear devices to deteriorate and References
eventually fail. If this energy can be taken out
from the non-linear devices their life time will 1. https://www.britannica.com/science/lightnin
be greatly enhanced. g-meteorology
The test results indicate that the supercapacitor
is not destroyed by the repeated application of 2. Phillip R Tompson BE(Hons)FIE(Aust)CPEng
MIEE, “PRINCIPLE OF LIGHTNING
high voltage transients and the gradual
PROTECTION”,2008.
voltage rise after each hit is in the order of
millivolt. Also it still retain its capacitive 3. Chandima Gomes, “On the Selection and
behaviour and is not adversely affected by the Installation of Surge Protection Devices in a TT
transient high voltage at the terminals[5]. Wiring System for Equipment and Human
Therefore , 1F-2.5 V supercapacitor which has Safety” , 4th February 2011, Elsevier Journal.
the highest voltage rise with 1 ohm resistor
was selected as a supercapacitor sub circuit for 4. https://en.wikipedia.org/wiki/Supercapacitor
this design.
371
5. N. Kularatna, J. Fernando, S. James, A. Pandey, 17. https://www.maginc.com/Products/Powder-
“Surge capability testing of supercapacitor Cores/Learn-More-about-Powder-
families using a lightning surge simulator”, Cores.aspx,Visited, 10th April 2016.
IEEE transactions on Industrial Electronics,
2011, 58, 10 ,pp.. 4942 – 4949.
6. IEC61000-4-5 : Testing and measurement

techniques - Surge immunity test, 2005-11
7. Texas Instrument, reference with “Magnetic

Core Properties," originally titled "An Electrical
Circuit Model for Magnetic Cores. " Unitrode
Seminar Manual SEMIOOO, 1995.
8. Fernando, Jayathu, Nihal Kularatna, Howell

Round, and Sadhana Talele. "Implementation
of the supercapacitor-assisted surge absorber
(SCASA) technique in a practical surge
protector", IECON 2014 - 40th Annual
Conference of the IEEE Industrial Electronics
Society, 2014.
9. Jayathu Fernando, Nihal Kularatna “A

supercapacitor based enhancement technique
for stand-alone surge protection circuits”
School of Engineering, The University of
Waikato Hamilton, New Zealand.
10. Nihal Kularatna, “Energy Storage Devices For

Electronic Systems-Rechargeable Batteries and,
School of Engineering, The University of
Waikato Hamilton, New Zealand, 8th
September 2014.
11. Fernando, Jayathu, and Nihal

Kularatna."Supercapacitor assisted surge
absorber(SCASA) technique: Selection of
supercapacitor and magnetic components",2014
IEEE Energy Conversion Congress and
Exposition (ECCE), 2014.
12. MdHisamHanapei and MohdRezadzudin

Hassan, “Satellite Communication Equipments
Reliability And Lightning Surge Measurement
Results”, Lightning Protection And EMC
Unit,CEEM'2006/Dalian 3A1-05.
13. Kostas Samaras, Chet Sandberg, Chris J.Salmas

and Andreas Koulaxouzidis, “Electrical Surge-
Protection Devices for Industrial Facilities”,
IEEE Transactions on Industrial Applications,
vol.43,No.1,January/February 2007.
14. http://news.nationalgeographic.com/news/20
13/11/131102-lightning-deaths-developing-
countries-storms, Visited, 15th January 2016.
15. http://batteryuniversity.com/learn/article/w
hats_the_role_of_the_supercapacitor, Visited,
5th March 2016.
16. http://www.capacitorguide.com/supercapacit
or,Visited, 5th March 2016.
372
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[373 378],2017
2017
The Institution
Lanka
Detecting Key Actors in Multiple SIM Networks Via

Centrality Measures and Mobile Service usage
Behaviors
A.J.M. Korale and N.S. Weeraman
Abstract: The ubiquity of mobile service and the need for communication among small groups
of individuals for personal and business use has prompted widespread registration of multiple
Subscriber Identity Modules (SIM) under the ownership of a single individual. Operators desire to
identify the key individuals in a multiple SIM network so that promotional messages and offers can be
directed effectively and efficiently to the right customer where it will have the most impact and
greatest chance of propagation through word of mouth.
In this paper we propose a novel algorithm that identifies key actors in a multi SIM network by way of
their intra SIM call patterns, mobility information, demographic characteristics of the registered user
and other information relating to the value added service (VAS) subscription portfolio. In this
approach we associate a set of centrality related features with each SIM in the network based on the
incoming and outgoing call patterns, their duration, the time of such calls, the location and mobility
information of the SIM estimated at the time of the calls, the intra and inter operator traffic, average
revenues and other ancillary information related to the age and gender of the registered user.
The algorithm provides a view of the hierarchical structure of the multiple SIM network and insights
in to the degree of importance of the individual SIMs and their relationships with each other. We also
develop a set of criteria based on the features and detected leaders to help in the identification and
validation of the key actors. These features and criteria also help to rank the SIMs in order of
importance helping to prioritize an operator’s approach to each individual SIM. They also help group
multi SIM networks together according to a certain similarity criteria.
The detected key actors in the network are validated by matching their SIM with those SIMS
registered for other services within the other operator’s line of businesses such as contact numbers
registered for customer services and other high value uses of the SIM such as roaming and
international calls.
Keywords: Centrality, Social Network, Mobile Telecommunications, Eigen Vector Centrality,

Multi-SIM Networks
1. Introduction small multi SIM network. The attributes of

what constitutes a leader may depend on such
The ubiquitous availability of mobile services considerations as inter network call activity
from a multitude of mobile service operators level, mobility, time and duration of calls in
has resulted in the widespread registration of addition to what other products and services
multiple mobile telecommunications service are used from the service portfolio of the
connections under the ownership of a single mobile operator.
individual. It is usually the case that subscribers In the Sri Lankan mobile telecommunications
tend to register multiple service connections market an individual is required to register the
under a single network operator as the cost of connection under ones National Identity Card
calls to other networks is significantly higher (NIC) for security reasons.
than calls within a single network. Thus while
Eng. (Dr.) A.J.M. Korale, C.Eng., MIE(SL), B.Sc. Eng.
these multiple Subscriber Identity Module (Wisconsin), M.Sc. Eng. (Columbia), Ph.D. Eng. (Imperial
(SIM) networks usually belong to a single College), MBA.(Sri Jayewardenepura), Senior Technology
network operator the identification of the key Consultant, Millennium Information Technology.
actor in the multi SIM network poses a Mr. N.S. Weeraman, AMIE(SL), B.Sc. Eng. (Moratuwa),
significant challenge to the operator. The key MBA. (Sri Jayewardenepura), Manager, Business
actor in this respect refers to the leader of the Intelligence, Dialog Axiata PLC.
1
373 ENGINEER
And mobile operators also place additional organization. Mobile operators leverage this
restrictions on the number of simultaneous or opportunity by selling potential advertising
multiple connections an individual can register “space” in the SMS domain to clients who wish
under a single NIC. In the case of the operator to reach large groups of individuals in
Dialog Axiata, an upper limit of five particular demographics.
simultaneous connections per individual (NIC)
is placed on the subscriber base. 2. Mobile operator specific criteria
that help in detecting and validating
These restrictions are to ensure that the mobile
key actors in multiple SIM networks
connections held are ones that serve a genuine
purpose and are ones that will be in regular
Certain mobile operators are in many lines of
use. In these multiple SIM networks separate
business and in the case of Dialog a quad play
billing arrangements will be made for each
operator, has mobile telephone services,
connection under a particular NIC with
broadband internet, fixed line and television
detailed call analysis available for each
service in its service portfolio. This provides a
connection but the ownership rests with a
unique opportunity for the operator to reach
single individual and therefore so lies with the
out to its customer base in one line of business
responsibility for making payments with
to promote its other lines of business to
respect to the multiple connections.
subscribers who haven’t yet availed themselves
the opportunity of trying out the other products
Business and other mercantile organizations
on offer. When properly channelled, offers from
have recourse to multiple connections through
one line of business can be promoted to
corporate accounts where many hundreds of
consumers in its other lines of business
connections can be obtained under the
presenting a unique cross sell opportunity.
ownership of a corporate. In the case of such
corporate multi-SIM networks it is clear where
This also provides the operator with a way of
the ownership lies and any promotional
determining key actors in their subscriber base
messages or material on special offers can be
by identifying those mobile service contact
delivered to the relevant representative of the
numbers left by individual customers which are
corporate. Also the number of such corporate
associated with different product lines of the
networks is relatively small in number so that
company. For example an individual customer
an individual account manager from the mobile
contact number associated with a Dialog
service operator can be assigned to handle all
Television service subscription may provide
the affairs relating to that corporate multi SIM
clues to the principal actor in a multi SIM
network.
network consisting of multiple mobile
connections owned by an individual. If one of
Returning to the problem of the personal five
the multiple SIM numbers owned by a
connection multi-SIM networks under the
particular customer matches a customer contact
ownership of a single owner, identification of
number left with another Dialog product such
the key actor in such networks poses a major
as a TV, broadband or fixed line connection it
problem for mobile operators. It may be that
may provide a means of identifying the main
such multi SIM networks are used within a
actor of that multi-SIM network.
single family or in a small business or used
both as a personal communication network
While in this paper we do not rely exclusively
between ones family members and with those
on such heuristics to identify key actors in
employees of a small private enterprise. In such
multi SIM network, we however do use them to
a scenario the operator cannot pay individual
validate the key actors found via our novel
attention to the many thousands of such
algorithm for detecting such key individuals.
networks and requires an automated means of
identifying the key individual at the centre of
Due to the particular nature of the mobile
the multi-SIM network.
telecommunications service usage profile of
individual customers we also may use other
Operators therefor wish to have an efficient
mobile service usage patterns, usage contexts
means of reaching the key actors in a network
and properties associated with mobile service
via an algorithm that can dispatch promotional
assigned to a particular SIM in our quest to
materials and offers via electronic means such
validate the key actors determined via the
as SMS. SMS based advertising is now a key
algorithm.
part of the digital advertising strategy of any
ENGINEER 2
374
It is also very likely that the SIM that is used for “well connected”. In this respect we determine
high value services like International Direct the SIM that has the most number of
Dialling (IDD) is in the hands of the key actor of connections and connections of connections.
the multi SIM network. Thus by monitoring the
IDD service usage of each of the SIMs an idea to We believe a well-connected individual has
the key individual of the network may be many connections and also employ the thinking
arrived at. In this scheme it is the outgoing call that the connections of the well-connected
pattern more than the incoming call pattern individual must themselves be well connected
that will be most indicative. As it is the individuals. Using network theory and
outgoing calls that are directly billed to a heuristics based on the incoming and outgoing
particular connection while incoming calls even call pattern of each of the members on the five
from overseas are received without cost to the member multi SIM network we are able to
particular connection. determine a set of properties or statistics for
each node (or SIM) in the multi SIM network.
We may also use the overseas roaming profile These statistics then allow us to pick the node
of a subscriber (or SIM) to hazard a guess as to or SIM most “outstanding” in the network.
the importance placed on a SIM in a multiple
SIM network. A customer that has activated The outgoing and incoming calls between the
roaming on a SIM or takes a particular SIM as nodes (actors) of the multi SIM network are
his roaming connection when traveling used to populate two adjacency matrices. In
overseas clearly places a priority on that SIM. In one, the adjacency matrix is created using the
such a scenario we have grounds to believe that number of calls between the actors of the
the key individual of the network is the one network in a directed network configuration. In
behind the use of that SIM. the other, the adjacency matrix is populated
with the call durations between two actors in
Mobile service operators place internal the network, in a directed network
rankings and priority handling tags on configuration.
individual SIMs based on an internal
assessment of the value of the customer to the 3.1 Eigen vector centrality [1]
network. In these schemes customers and SIMs The magnitude of the elements of the principal
associated with them are given high priority in Eigen vector of the adjacency matrix of the
customer care and are eligible for discounts and multi SIM network when sorted in descending
special offers on products and services on offer. order of magnitude indicates in that order the
This type of priority determination is made on most well connectedness of actors in the
the basis of the revenue contribution of each network. Thus at the head of the sorted Eigen
SIM (subscriber), the payment pattern which vector lies the most well-connected actor and
displays a history of prompt and on time our prime contender for the position of key
settling of dues and subscription to numerous actor of the multi SIM network.
and high value products and services from
across the quad play portfolio. In light of this a The principal Eigen vector also accounts for the
SIM ranked highly in the internal customer connections of connections of the actors in the
ranking is also bound to be a key actor in a network and so indicates the heuristic that the
potential multi SIM network. most well-connected actor is also the one who
has the most well-connected friends
All of these considerations can be used in the (connections). Employing the idea that the
validation of potential key actors in the multi importance of a connection is related to how
SIM networks determined through the many other important connections he has
proposed novel key actor detection algorithm. (employing the concept of having friends in
high places)
3. Overview of the key actor
detection algorithm In our framework we utilize both types of
adjacency matrices discussed above and
The key actor detection algorithm operates by determine key actors based on their
processing the internetwork call pattern of a connections (the number of calls) and the
multi SIM network consisting of a maximum of strength of their connection (the duration of the
five SIMs. In our quest to detect this key actor calls).
we employ the reasoning that the key It is quite frequently the case that some pre-
individual in a network is the one who is most filtering of calls is carried out prior to being
3
375 ENGINEER
considered for inclusion in the adjacency duration of a call for it to be considered valid
matrix. This is to eliminate those calls that may for consideration in the radius of gyration
not be the result of strong relationships measure.
between the calling and called party. In light of
this we place a lower limit of 30 seconds call 3.5 Time stamp of incoming and outgoing
duration for a call to be considered valid and calls to assess level of activity in making and
for inclusion in the adjacency matrix. receiving calls during the day
We also observe the number of calls coming in
3.2 In-Degree to and leaving a particular actor across the time
This measures the number of calls coming in to of day by dividing the day in to periods 00:00-
the node or calls received by the actor. In 8:00, 8:00-16:00 and 16:00-00:00.
directed networks in-degree is considered more
important than out-degree. The page rank This measure is also an indicator which may
algorithm also uses a similar heuristic which tell us information relating to which calls may
we too employ in our model [1]. be related to official purposes and which calls
may relate to more personal purposes as it is
We also consider the duration of incoming calls likely that calls received after business hours
to a network node (actor). This is essentially to are generally due to personal reasons.
determine if there is a relationship between the
number of calls received and the total duration As activity level is also an ancillary property of
of those calls. For it is likely that a key actor a node in the network and not directly related
receives a large number of calls and also that to the relationships between actors we do not
those calls are of a relatively long duration place a limit on the duration of a call for it to be
when compared with the subordinate calls in considered valid for consideration in the
the network. activity during the day measure.
3.3 Out-Degree Clear outliers in this field also indicates a key

This measures the number of calls going out of actor in the network as the results will show.
the node or calls made by the actor.
4. Results and Validation
We also consider the duration of total calls
made by a network node (actor). For it is likely In the following figures and table we present
that a key actor in addition to receiving a large some example results to demonstrate the
number of calls also makes a fairly large working of the algorithm to detect the key
number of calls in return or in response to those actor(s) in a multi SIM network.
calls. We may also surmise that those calls are
of a relatively long duration when compared Table 1 – Multi SIM Call Node Properties.
with the subordinate calls in the network. Out In
Node In Out In Out calls calls calls
Cell Cell
No Degree Degree MOU MOU 0-8 8-16 16-24
Change Change
3.4 Radius of Gyration
This is a measure of the degree of mobility of a 1 51 0 4,025 - 24 21 9 63 29
SIM (actor). It is generally believed that a key 2 101 139 3,731 11,954 40 38 10 149 117
actor may have a large radius of gyration.
Meaning that in a day the calls taken and 3 9 17 671 748 7 5 1 12 15
received by a key actor may happen in many 4 65 93 5,440 3,441 29 16 3 82 73
more locations than that when compared to a

5
subordinate SIM in the network.
42 19 2,884 608 6 6 5 52 16
We measure this mobility by determining the In Table 1, we present the network statistics for
number of times the serving cell of the each node in a multi SIM network to
subscriber (SIM) changed during the day. We demonstrate how the properties of the
do this by noting the number of changes in the internetwork call patterns can be used to detect
cell ID through which a call was made or the most well-connected (important) actors in
received. As mobility level is an ancillary the network. The 30 second call duration filter
property of a node in the network and not had been applied to the degree calculations but
directly related to the relationships between was not applied in determining the mobility
actors as evidenced by the calls between each of level or time of day call activity levels.
the SIMs, we do not place a limit on the
ENGINEER 4
376
With reference to Table1, the first column gives durations depicted in Figure 3 also confirm the
the respective node number or identity of the earlier finding that node 2 is most central and
actor, the second and third columns the number that nodes 4 and 1 are perhaps next most
of incoming and outgoing calls to each node, important. This finding is also validated by the
the fourth and fifth columns present the total Eigen vector centrality measure and by the
duration of calls (minutes of use) coming into node level statistics found in Table 1.
and leaving a node. Columns 6 through 7 give
the number of times the serving sell of the actor
Node 2
changed when receiving or making a call which 7

7
is a measure of his mobility. Columns 8 44 26
11
Node 3
through 10 give an indication of how busy the 8
actor is in those hours giving an idea of the
1
number of calls received or made that fall in to 1

Node 5
those time intervals. 1

62 83
2
8
From the table we can clearly conclude that 1
actor 2 is the leader of the network or is the

Node 4 5
most well-connected as well as most active 1
member of the network. Due to these reasons

Node 1
we consider node 2 to be the most important in
the network and consider it the key actor of the Figure 2 – Call based Multi SIM network
five sim network.
Node 2
The Figure 1 depicts the principal Eigen vector 602

308
of the call based adjacency matrix. It too 3784
Node 3
2236
352
reinforces the finding that node 2 is the most 352

32
central in the network. We also note that the
Eigen vector centrality measure can also be 44
37
Node 5
5332 3071
used to rank the centrality of each of the nodes 64

296
in the network. Thus it is a measure that can

serve to prioritize the actors in the network in
44 Node 4
160
order of importance. Eigen vector centrality is 37
the principal feature used to identify the central

actor in the network while the radius of Node 1
gyration and time of calls are of lesser

Figure 3 – Call duration (minutes of use -
importance. We give higher priority to the
MOU) based Multi SIM network
centralities obtained for the network based on
the number of calls and then for the network
based on call duration and consider radius of
5. Conclusions
gyration and time of calls as the next most
important criteria. Through this modelling we have shown that we
Principal Eigen Vector
are able to detect key actors in multi SIM
0.8
networks. We are also able to rank the degree of
0.7 well-connectedness or in this case a degree of
0.6
importance by using the Eigen vector centrality
measure as well as the tabulated properties of
0.5
individual nodes that corresponds well with
0.4 the Eigen vector centrality measure.
0.3
We were also able to independently verify that
the key actors detected were in fact important
0.2
0.1 actors in the network by using the criteria

0 outlined in section 2 which acted as another
level of validation for our findings.
1 1.5 2 2.5 3 3.5 4 4.5 5
Node Number
Figure 1 – Principal Eigen Vector of Call based

Adjaceny Matrix References
The internetwork diagram of call patterns of 1. Newman, MEJ, Networks, Oxford
Figure 2 and internetwork diagram of call University Press, 2013, pp 109-149.
5
377 ENGINEER
2. Smilkov etal, Identifying communities by
influence dynamics in social networks
3. home.dei.polimi.it/matteucc/Clustering
tutorial_html/cmeans.html, extracted on 10
April 2017.
ENGINEER 6
378
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[379 384],2017
2017
The Institution
Lanka
Intelligent Pest Repellent System for Sri Lankan

Farming Industry
P.N. Karunanayake, W.H.H.N. de Soysa, J.A.C.L.C. Jayasundara,
Y.G. Wanniarachchi and A.P.I. Karunarathne
Abstract: Sri Lanka has a culture based on agriculture which has been descending from
centuries. Currently in Sri Lanka cultivation activity from ploughing to harvesting and packing being
carried out by machinery with less amount of labour. But even with such machines involved in the
field, farmers find it very difficult to protect their crops from pests which cause maximal damages
when strict supervision is absent. Though insects and rodents can be controlled by using pesticides,
larger pests such as wild boars, warthogs, porcupines and birds cannot be controlled using such
methods since it requires manpower to keep them away from the crops. The absence of advanced
technology involved in this area has led the farmers to face many difficulties such as profit loss, the
requirement of excess manpower to protect the crops from pests and usage of illegal and inhumane
pest attack prevention mechanisms.
A novel approach is introduced as a solution for the mentioned difficulties. Wireless sensor network is
used to detect pests in the field, and once a pest is detected, an appropriate repellent system which is
harmless to animals and the environment is activated by the system. Emitting ultrasonic soundwaves
and high-intensity flash lights are used as the repellent systems. Since there is a probability of sensor
node failures, central node evaluates a pattern of the arrival of pests. When the central node detects a
failure of one or more sensor nodes, it activates the repellent system automatically according to the
evaluated arrival pattern of the pests and notifies about the node failure to the owner. Therefore this
system is focused on the implementation of user-friendly, energy efficient, intelligent autonomous
pest repellent system.
Keywords: Intelligent, Wireless Sensor Networks, Pest Repellent, Autonomous, Pattern Recognition
1. Introduction As a result, pest controlling has become major

problem leading to economic, social and ethical
Among the various agricultural sectors in Sri issues in the Sri Lankan society such as,
Lanka, the primary form of agriculture is rice • Animal poacher.
production. Rice is the single most important • Animal attacks on humans.
crop occupying 708,000 hectares of land in Sri • Usage of illegal pest control
Lanka by 2016 [1]. mechanisms.
• People being victims of their own traps
In the past, the farmers used to the spent high ( ex: Trap Guns )
amount of man power for cultivation, but • Crop Losses
nowadays with almost every cultivation The goal of this research project was to focus on
activity from ploughing to harvesting and an effective, user-friendly, humane, energy
packing being carried out by machinery efficient and cost-efficientpest repellent system
resulting minimization of labour requirement. that can be used by Sri Lankan farmers. From
The farmer is being transformed to a part time all the pests that harm paddy fields in Sri
manager who coordinates various activities. Lanka, wild boar and sparrows are the pests
But even with such machines involved in the which destroy the paddy mostly according to
field, Sri Lankan farmers find it very difficult to the farmers (apart from insect pests). Therefore
protect their crops from pests which cause this work only focused on repelling wild boars
maximal damages if strict supervision is absent. and sparrows.
Though insects and rodents can be controlled Eng. P.N. Karunanayake, M.Eng.(SL), B.Sc. Eng.
using pesticides, larger pests such as wild (Moratuwa), Department of Electrical, Electronic &
boars, warthogs, porcupines and birds cannot Telecommunication Engineering, General Sir John
be controlled using such methods since it Kotelawala Defence University, Sri Lanka.
Mr. W.H.H.N. de Soysa,
requires manpower to keep them away from Mr. J.A.C.L.C. Jayasundara,
the crops. Mr. Y.G. Wanniarachchi
Mr. A.P.I. Karunarathne
1
379
The autonomous pest repellent system was the range of frequencies that only targeted pests
designed by using wireless sensor networks can hear. When the power of the sound wave is
(WSN) incorporating pattern recognition sufficiently large, the pests cannot withstand it.
algorithm. Several techniques were used to Thereby the sound wave repels them away
increase the life time of the system since from the area without affecting the
monitoring a paddy field should be done for environment and non-target organisms,
few months in one cultivation season. including man. The second type of devices
The remainder of this paper is organized as which are electromagnetic devices emit
follows. Section 2 discusses on pest repellent electromagnetic waves, inimical to pests. The
systems, wireless sensor networks and its advantages of these devices over other pest
applicability for a pest repellent system. Also, it controlling devices are lower cost,
discusses on the behaviour of the targeted environmental friendly and no known threat to
pests, wild boars and sparrows. Section 3 the humans[2]. Therefore with the technological
discusses the node construction which is used advancement, it can be seen that electronic pest
for the pest repellent system. Under section 4, repellent systems can be developed reducing
the operation of the system is discussed. Section the disadvantages and the risks of conventional
5 discusses incorporated pattern recognition pest repellent methods.
technique. The user interface and the SMS alert
system is discussed in the section 6. Section 7
discusses the results obtained, and section 8 2.2 Wireless Sensor Networks (WSN)
provides the conclusion. Future improvements
are discussed under section 9. Wireless sensor networks are applied in almost
every application in nowadays to make life
2. Literature Review easier for the human. WSN systems have many
advantages over wired networks such as easy
2.1 Pest Repellent Systems deployment, and less amount of maintenance.
But WSNs have many limitations such as
Conventional pest control mechanisms that are limited amount of power, security, scalability,
used in Sri Lankan farm fields can be named as, routing, node failures, data integrity,
scarecrows, bells, electric fences, light bulbs, Localization, etc. According to the application,
horns etc. These mechanisms are proven to be WSN should be designed considering all of its
very inefficient and primitive. Among these, limitation to obtain its maximum output.
electric fences are used to repel considerably
large animals but dangerous for both humans WSNs have been widely used for precision
and animals. Even though electronic pest agriculture [3]. There are many goals for using
controlling mechanisms have a risk of danger, precision agriculture as follows.
farmers tend to use them to save their crops or
paddy.  To reduce Nitrogen, Methane and
Carbon and other dangerous gases and
Examples for electronically powered pest liquid emission.
repellent are electronic pest chasers, electronic
 To manage the available resources
pest deterrent, electronic pesticides. These are
according to the requirements of the
generally known as electronic pest control
crops.
devices. These repellent methods are more
 To improve the output of the crops by
popular since their nature ofenvironmental
monitoring their growth.
friendliness over the conventional pesticides.
Two types of electronic pest control devices are
available. The high-frequency devices and the These goals can be accomplished by using
electromagnetic are the two types of devices. WSNs. Although many research work has been
They are popular for their reliable operation done for precision agriculture and to control
and availability. High-frequency devices insect pests [4], [5], [6], research work for pests
operate by transmitting sound waves greater such as wild boars together with birds such as
than 20,000 Hz. Although some animals such as sparrows has not been done.
dogs, bats, rodents, birds and wild boars can
hear well in the high-frequency range, human
ear lacks the capacity to hear such sound.
Therefore high-frequency devices are designed
and constructed to emit sound waves within
ENGINEER 2
380
2.3 Behavioral Patterns of Wild Boars and 3.1 Sink Node
Sparrows
The Sink node consists of following main
Since this research project only focused on wild components:
boars and sparrows, their behavioural  SIM800L GSM Module
information is important to design the repellent  SD Card module
system.  FM Receiver
 LCD Panel
Wild boars are highly capable of adapting to  Arduino Nano
the environment and capable of learning from  12V Battery
the environment. Usually, at night time they  8 Way Relay
attack to paddy fields and destroy it for their  Solar Panel (10W)
food. The traditional method of repelling them  Real Time Clock
is waving a flash light or a torch towards them.  PIR Sensors
Within few seconds after a torch is flashed,  FM Receiver
wild boars leave the paddy. But if they  High-Frequency Speakers
understand flashing the torch is occurred only  High intensity LED
during a certain period of time, they change the
 Solar charge controller and Frequency circuit
time of their attack by learning the pattern.
Therefore there are many disadvantages in this During the daytime, birds are the most
method.
likely to invade the crops. Therefore thehigh-
frequency speaker unit will be activated
1. The farmer needs to be awake and throughout the day. Day and night are
watchout throughout the night every distinguished by using a real time clock
day resulting in a requirement of man attached to the sink node. At night the crops are
power in higher quantity. prone to be attacked by larger animals such as
2. If only flashing is done only for a wild boars and porcupines. Therefore both the
certain period of time, wild boar speaker unit and the light unit will be in
realizes it with their capabilities and operation during the night till dawn. The light
will visit the paddy at a different time. unit is switched off during the day time to save
energy. Figure 1 illustrates the sink node. The
Using this method is environmentally friendly. solar panel was used to power up the sink
Therefore designing a system which identifies node. During the night time,12V battery was
the wild boar and flashing a light towards them used to provide energy.
will provide an echo friendly system which
saves a considerable amount of man power.
Sparrows damage the paddy fields at the day

time. The traditional method of repelling them
from harming the paddy is using a whistle
which emits a sound wave higher than
20,000Hz.
This method is also environmental friendly. But

it requires a higher amount of man power and
time. Figure 1 - Sink node
3. Device Construction 3.2 Sensor Node

Two main units are being used in the system
1. Sink node Sensor node consists of following
2. Sensor node components.
Each unit has distinctive characteristics which  PIR sensors
enable them to communicate with each other  Pulse Generator circuit
and execute operations as given by the sink  Battery level indicator LED Circuit
node.  6V Battery
 FM Transmitter
3
381
Sensor nodes are simplified versions of the sink phone number within a short period of time.
node. These nodes remain active all the time The user can then inspect the sensor node and
and are vigilant about the movements in the do the necessary actions.
surrounding. Each node consists of 3 PIR
sensors mounted on its head which cover a Operation of the system is categorized as
whole range of 360 degrees. “Real mode” and “Demo mode”. The purpose
of these two modes is for real time
deployment and for demonstration and testing
purpose respectively. In “Real mode” the
system functions considering night time as
from 1800 hours to 0600 hours. The “Demo
mode” is for testing and demonstration
purpose. Hence night time is considered as 12
minutes.
5. Pattern Recognition
During a sensor node failure, until the node is

replaced, any animal can disturb the paddy in
Figure 2 - Sensor node the area which the failed sensor node
provided coverage. Hence it will damage the
4. Operation paddy. To minimize the damage, sink node
identifies a pattern of the arrival and the time
Sensor nodes are placed to cover the area of the of arrival of wild boars.
paddy field. For this research, nodes were The Main purposes of Pattern Recognition are
placed within one hop distance from the sink as follows,
node (Multi hop scenario is not considered). • To Reduce excess energy consumption
The nodes are placed in a known location. Each • Function autonomously without any
sensor node is given an individual address and human interaction.
ID. Node placement information is fed to the • Determine the times that the animals
sink node to identify the direction of the arrival visit the field and repel them from entering
of the pest. Once a movement is detected by the the field.
sensor node, it sends a data packet to the sink • If a secondary node fails, the pattern
node to indicate the presence of the pest. When recognition mechanism will still be in
a packet is received, the sink node activates operation using the previously recorded
either its flash light system or the speaker data. The base station can, therefore, carry
system which is directed towards the direction on normal functions in lieu of the failed
which the movement was detected. The high- secondary node. This enables the system to
frequency speaker System emits a range of function smoothly and provides time for the
frequencies ranging from 20 kHz above in user to look into the failed node and replace
order to repel any animal that is in its way. The it.
flash light unit flashes the light during the night
time to repel the wild boars. Once packets are received informing the arrival
of pests, the information such as the direction
If the sink node keeps on receiving signals from and the time of arrival is stored in the sink
a certain stable target, it will automatically send node.This informationis collected for three
an SMS to the user notifying about the presence consecutive days. Then sink node calculates a
of an unknown animal or pest which cannot be pattern of the arrival of pests with the direction
sent away from the system. and the time. This pattern is stored in the sink
The sensor node sends a beacon periodically to node and once a node failure is detected. The
the sink node to provide information about the lighting system is activated according to the
presence of the sensor node. By using this pattern, and the sound system is activated in a
beacon, the sink will be able to detect new random manner for a period of the 40s. Since
nodes as well as the failure of the sensor nodes. flash light requires more power, by using the
If the sensor node ends its life time, it is pattern can reduce the time the flash light is
identified by the sink node. When node failure activated. Hence can reduce the power
is detected, an SMS is sent regarding the node consumption of the system.
failure with the node’s ID to the assigned
ENGINEER 4
382
For this pattern recognition mechanism, The interface which interacts with the sink node
theonly arrival of wild boars was considered only is used to input the phone number which
since the arrival of birds have a random SMS alerts should be sent. Also mode selection
pattern of arrival. When the sink node detects between “Real Mode” and “Demo Mode” is
a node failure, it informs the user by sending selected using this interface.
an SMS and activates the flash light systems
according to the recognized pattern. Therefore The second interface is the graphical user
the time window that was identified as the interface (GUI). The GUI is used to analyze the
most number collected data from the nodes. The GUI was
Figure 3 – Analysis of collected data
of the arrival of wild boars, the flash light created using Visual Basic.net. Figure 3
system will be activated. The objective of this illustrates the analysis of collected data.
technique is to repel the wild boars even the
sensor nodes are not activated for a period of SMS messages are sent from the sink node to an
time with lower power consumption. assigned number alerting about the condition
of the nodes and the detection.
The identified pattern is used for seven days.
Three types of messages are being programmed
After seven days, anew pattern is identified.
to be sent with three different alerts. They are,
The pattern needs to be trained again since
1. Battery failure and Low battery levels of
wild boars are able to identify patterns very
specific sensor nodes.
easily. If the animals recognize that the flash
2. Malfunction of sink node.
light is activated only in a particular period of
3. Uncontrollable animal behaviour.
time, it will change the arrival of time.
Therefore continues training of the pattern is If the sensors keep on reporting detections for
important. more than 5 minutes continuously, the base
station will send a message to inform the
6. User Interface and SMS Alerts presence of unwanted animal so that the user
can inspect.
The system has two main user interfaces. One
interface interacts with the sink node only, and
the other interface interacts with all the nodes.
5
383
7. Results A prototype design was made using WSN and
tested thoroughly. The system can be
visualized through a simple software which
Test results were analysed under two operating
was built using Visual Basic.Net to analyse the
modes namely,
collected data. A GUI enables the user to
 Real Time Mode
operate the system easily making it possible to
 From 0600 hours to 1800 hours
let anyone operate the system.
 From 1800 hours to 0600 hours
 Demo Mode SMS alerting system is activated with the
sensor networks simultaneously. This service
The test environment was selected considering enables the user to remotely monitor the
the habituation and the density of the birds. agricultural field and the equipment.
Although lighting system was not able to test
Therefore a system which is sustainable,
with real wild boars, motion detection, lighting
autonomous and user-friendly was developed
system, user interfaces and the SMS alert
as a pest repellent system from this research
mechanism were tested and provided more
project.
than 95% results.
Acknowledgement
8. Improvements
Authors wish to acknowledge the guidance
Since the prototype was not tested in a real
provided by the Department members of
paddy field for a longer period of time, it can be
Electrical, Electronic & Telecommunication
tested and identify the alteration which needed
Engineering - KDU.
to be done.
The prototype used single hop WSN. Therefore
coverage area by the sensor nodes is lower
References
compared to actual paddy fields. Therefore by 1.http://www.statistics.gov.lk/agriculture/Paddy%
using multihop sensor networks, this system 20Statistics/PaddyStats.htm, Visited 14th May 2017.
can be improved.
A cloud based data upload system can be 2. O. O. D. et. al. Ibrahim A. G., “Electronic Pest
implemented for the data to be stored and Control Devices  : A Review of their Necessity ,
referred later. The user will be able to analyse Controversies and a submission of Design
and use the recorded data from anywhere, Considerations,” Int. J. Eng. Sci., pp. 26–30, 2013.
anytime via the internet.
A smart user interface can be implemented 3. Azfar, S., Nadeem, A. and Basit, A. “Pest
where the user can easily add / Remove nodes detection and control techniques using wireless
from the field and change the topological sensor network: A review,” J. Entomol. Zool.
deployment of the nodes. This will enable Stud. JEZS, vol. 3, no. 32, pp. 92–99, 2015.
anyone to be able to arrange and assign the
desired number of nodes at desired locations. 4. C. Scientist, “Low Cost Sensor Based Embedded
A daily status report of the activities can be System for Plant Protection and Pest Control,”
arranged as per the user’s requirement. pp. 179–184, 2015.
9. Conclusion 5. Aqeel-Ur-Rehman, Abbasi, A. Z., Islam, N. and

Shaikh, Z. A. “A review of wireless sensors and
This work is focused on autonomous, efficient networks’ applications in agriculture,” Comput.
and an intelligent system for pest repellent in Stand. Interfaces, vol. 36, no. 2, pp. 263–270, 2014.
various disciplines of agriculture but mainly
focused on paddy fields. The system consists of 6. Akyildiz, Ian F., Weilian Su, Yogesh
integrated hardware components and an Sankarasubramaniam, and Erdal Cayirci.
intelligent system which uses pattern "Wireless sensor networks: a survey." Computer
recognition algorithms to operate. An initial networks 38, no. 4 (2002): 393-422.
survey was carried out to get an overall idea
about the requirement of such a system to the 7. Schlageter, A. and Haag-Wackernagel, D. “A
Sri Lankan farming industry and to identify the Gustatory Repellent for Protection of Agricultural
key point which should be targeted in Land from Wild Boar Damage: An Investigation
developing such a system. on Effectiveness,” J. Agric. Sci., vol. 4, no. 5, pp.
60–68, 2012.
ENGINEER 6
384
Land Cover Classification and Sub Component Analysis

Using Hyperspectral Imagery
A. M. R. Abeysekara, T. S. J. Oorloff, S. S. P. Vithana, H. M. V. R. Herath,
G. M. R. I. Godaliyadda and M. P. B. Ekanayake
Abstract: This paper presents a classification of the land cover under soil, vegetation and water bodies
in a strip along the north eastern region of Sri Lanka, using hyperspectral image data obtained by the Earth
Observing (EO-1) satellite‟s Hyperion sensor. Using hyperspectral imagery in land-cover mapping is
beneficial as it could capture the fine details of the region under consideration, which may not be achieved
by other image processing techniques. In this paper, mathematical concepts and statistical tools such as
Principal Component Analysis (PCA) and concepts of euclidean geometry have been used to develop an
algorithm which classifies each pixel of the image under the above mentioned classes. In the process of
classification, the pixels that are not classified with a clear cut boundary initially are identified as „mixed‟
pixels, and are subjected to a separate algorithm in order to identify the percentages of fractional
components of such pixels. Finally, the image is remapped according to the class that each pixel belongs to,
and the mixed pixels are shown separately according the intensities of their fractional components.
Keywords: Hyperspectral Imagery, Hyperion, Principal Component Analysis, fractional components
1. Introduction 2. Area of Study and the Dataset

A collection of high resolution monochromatic As shown in the Figure 1, the area of interest of
pictures representing the reflectance values our study comprises of a strip along the North
corresponding to a broad range of wavelengths is a Eastern region of Sri Lanka.
hyperspectral image. Reflected radiation at a series
of narrow and contiguous wavelength bands is The Hyperion images of the aforementioned
measured in hyperspectral images. To identify the region seen in Figure 1 (a) and of geographical
characteristic information of a given pixel, an coordinates as in Table I, had been acquired on the
appropriate frequency band can be used. 17th of September, 2005 starting at 04:40:57 and
ending at 04:45:35 in local time (+5:30 GMT). Two
Land cover mapping using hyperspectral image cameras, VNIR (Visible and Near-Infrared) and
data has both advantages and disadvantages. Since SWIR (Short Wave Infrared) had been used to
the finest details of the images are available, the obtain the hyperspectral images of 242 bands
accuracy of classification is high when the ideal spanning across a range of wavelengths, from
spectral bands are chosen. The disadvantages of 355.59nm to 2577.08nm [5]. However, 44 spectral
using Hyperspectral Imaging (HSI) data for land bands have not been calibrated and cannot be used
cover mapping are, redundancy of data, for data analysis.
complexity in computations and excessive
Ms. A. M. R. Abeysekara, B.Sc. Eng. Undergraduate at the
memory usage. [1]–[3]. Department of Electrical and Electronic Engineering, University of
Peradeniya.
The classification of the pixels under soil, water Mr. T. S. J. Oorloff, B.Sc. Eng. Undergraduate at the Department
of Electrical and Electronic Engineering, University of Peradeniya.
and foliage, HSI data which was obtained by the Ms. S. S. P. Vithana, B.Sc. Eng. Undergraduate at the
Hyperion Earth Observing - 1 (EO - 1) sensor [4] Department of Electrical and Electronic Engineering, University of
has been used. The sensor which was used in Peradeniya.
imaging is capable of resolving spectral bands Eng. (Dr.) H. M. V. R. Herath, C. Eng., SMIEEE, MOSA,
MIE(SL), B.Sc. Eng. (Hons) (Peradeniya), MS (Miami, US), Dr. -
from 0.4 to 2.5 µm with a 30 meter resolution and it Ing. (Paderborn), Senior Lecturer at the Department of Electrical
has the ability of imaging a 7.5 km by 100 km land and Electronic Engineering, University of Peradeniya.
area per image, while providing detailed spectral Eng. (Dr.) G. M. R. I. Godaliyadda, B.Sc. Eng. (Hons)
(Peradeniya), Ph.D. (NUS, Sg), AMIE (Sri Lanka), Senior
mapping across all channels with high radiometric Lecturer at the Department of Electrical and Electronic
accuracy. The telescope provides for two separate Engineering, University of Peradeniya.
grating image spectrometers to improve signal-to- Eng. (Dr.) M. P. B. Ekanayake, B.Sc. Eng. (Hons) (Peradeniya),
noise ratio (SNR). Ph.D. (Texas Tech, US), AMIE (Sri Lanka), Senior Lecturer at the
Department of Electrical and Electronic Engineering, University of
Peradeniya.
385
The Geo „TIFF‟ images hence obtained was of
resolution 7481 X 1851 pixels of which each pixel
covers a geographical area of 30m X 30m. The data
included in the images were radiance values of the
respective wavelengths scaled by 40 for images
obtained through the VNIR camera in the range of
wavelengths from 400 nm to 1400 nm (bands 1 -
70) and by 80 for images obtained by the SWIR
camera in the range of wavelengths from 900 nm
to 1700 nm (bands 71 - 242) [6].
The true colour image, showed the region consists

of foliage, water bodies and soil mainly.
3. Preprocessing
Preprocessing is deemed to be an important stage

prior to any data analysis with the intention of
providing qualitative and more productive data
[7]. Preprocessing included extraction of data in a
useful manner while removing the uncalibrated
(a) (b) (c) sets of data as indicated in [5] and conversion of
Figure 1 - (a) The geographical location of the the radiance values of the Geo „TIFF‟ images to
study area (b) True color satellite image (c) The reflectance.
area chosen for classification.
3.1 Extraction of Data
Table 1 – Latitudes and Longitudes of the Chosen It is well evident that the useful information in the
Region. image is limited to a much less region compared to
Latitudes Longitudes the total size of the image (approximately 14%) as
/◦N /◦E seen in Figure 1 (b). Hence in order to improve
Upper left corner 10.146810 81.251940 efficiency and to the memory optimization, the
Upper right important data were extracted in the following
10.133729 81.319533 manner.
corner
Lower left corner 8.137243 80.818856
MATLAB® variable files (.mat) containing arrays
Lower right
8.124515 80.886088 of size 262 x 251 each, were created for each
corner
information containing row of the image in the
structure shown in Table 2.
Table 2 – The Structure of Data Files. Finally the uncalibrated data were removed
creating a final matrix of size 262 X 207.
Col Availability of Neighbouring Spectral
no. pixel information information
3.2 Conversion of Radiance to Reflectance
N N E S S S W N 1 2 - 242
E E W W
In Hyperspectral image analysis reflectance values
1st
of closely lying bands are used and hence the
index radiance values obtained from the raw data set
had to be converted to reflectance values prior to
2nd
index any process of analyzing. The conversion process
was done using [4],
3rd
index
... (1)
-
262nd
where,
index = unitless planetary reflectance,
= spectral radiance at the sensor's
aperture,
386
= Earth-Sun distance in astronomical
units,
= Mean solar exoatmospheric
irradiances,
By substituting the spectral radiances obtained

from the extracted data upon scaling with the
scaling factors provided for the respective range of
wavelengths, (1/40 for the first 70 spectral
components and 1/80 for the rest) the earth-sun
distance from, mean solar exoatmospheric
irradiances and the Solar zenith angle based on the
location, date and time of which the images were
captured, obtained from [8], the respective
reflectance values were obtained. Since the image Figure 2 - Spectral signatures of the three classes
spans across a small region and as the zenith angle soil, vegetation and water bodies, using the
doesn‟t vary considerably in the range of interest, training sample data in the original space.
the zenith angle was assumed to be constant In order to find the transformation matrix for the
throughout the region of interest. dimensionality reduction, a training sample of 60
pixels were chosen, 20 from each class [12]. This
selection was done with the aid of google maps
4. Algorithm and the image processing tools available in
MATLAB®. The spectral signatures of the three
Using the pre-processed Hyperspectral Image data classes based on the training sample data
obtained as mentioned in the above section, the (considering the original spectral space) are shown
land cover in a strip of the north eastern region of in figure 1. The reduced space was obtained by
Sri Lanka was classified under „soil‟, „vegetation‟, using the PCA equations [3], [10], [11] given by,
„water bodies‟ and „mixed regions‟. A total of
43975 pixels were individually classified with the …(2)
aid of mathematical and statistical concepts and
the tools available in MATLAB®. The pixels that ∑ …(3)
did not have a prominent class were classified as
„mixed pixels‟ and the percentage of each end …(4)
member (soil, vegetation and water) was
calculated for such pixels. where,
= covariance matrix, (198 198),
The pre-processed Hyperspectral data contained = expected value operation,
reflectance values of 198 spectral bands for each = vector representing the normalized 198
pixel. These pixels were represented as points in a spectral band information (reflectance) of a
198 dimensional space, in which the dimensions pixel, (198 1),
corresponded to the reflectance values of each = vector representing the mean spectral
spectral band. Since the algorithm requires information (normalized reflectance) of all
normalized data for the classification to be pixels, (198 1),
completely based on the spectral characteristics of = number of pixels in the training sample
the image, the spectral signatures [9] of each of the (60),
pixels were shifted downwards, to have a mean of = eigenvector of the covariance matrix,
zero and were divided by the standard deviation (198 1),
of the reflectance values of each pixel. = eigenvalue of the covariance matrix.
As the large amount of data availability caused The first 20 eigen vectors corresponding to the
redundancy as well as complexity, along with the largest 20 eigen values were stacked as rows in the
ability to detect fine features of the images, the 198 transformation matrix that transforms the original
dimensional space was transformed into a 20 198 dimensional space into the 20 dimensional
dimensional space which maximized the scatter of space with maximum pixel scatter. The
the pixels using principal component analysis transformation matrix is given by,
(PCA) [3], [10], [11].
...(5)
where,
387
- Transformation matrix, = unit vector representing the reference
- eigenvectors of the spectral signature of vegetation
covariance matrix. = unit vector representing the
reference spectral signature of water
All the pixels including the training sample, were
transformed to the new reduced dimensional Since the requirement is to obtain a measure of
space using the transformation, affinity, the reciprocal of the distances was
considered for the classification. The vector
… (6) representing the percentage affinity of a given
pixel, to the reference spectral characteristics of the
where, three classes is given by,
= Matrix representing the pixels (in
columns) in the original space (198x43925),
= Matrix representation of pixels (in
columns) in the reduced space (20x43925).
… (9)
The pixels are now represented as points (vectors
directing to specific points) in a 20 dimensional
space. The direction of each of these vectors
characterizes the spectral behavior of the pixels
compared to the magnitude of the vectors, as the where,
magnitudes of the reflectance values could differ = percentage affinity vector and the
due to the effects of the environment, sensing other terms are the same as before.
equipment etc. Based on that conclusion, the
direction of the vectors was selected as the basis of The unmixed pixels were selected using a
the classification algorithm. For this, the vectors threshold for the largest affinity percentage as
representing all the pixels, including the training follows.
sample, were divided by its magnitude to convert
them into unit vectors. Now all the pixels are …(10)
represented as points (unit vectors directing to
each of the points) on the surface of a sphere of Where is the maximum affinity percentage of the
unit radius, in a 20 dimensional space. The affinity vector .
equation for obtaining the unit vectors is given by,
If 50%, the pixel under consideration was
… (7) classified under the class which has the maximum
affinity with it.
where, If 50%, the pixel was labeled as „mixed‟ and the
= Unit vector in the reduced space percentage of each end member (soil, vegetation
= Original vector in the reduced space and water) was obtained by the percentage affinity
| = Magnitude of the original vector in vector .
the reduced space
The reference spectral signatures (reference 5. Results and Discussion

vectors) for the three classes were chosen to be the
three mean unit vectors obtained by each of the 20 A portion comprising of 43925 pixels, of the image
pixels in the training sample, from each class. The of a strip along the North Eastern region of Sri
difference between each vector representing a Lanka, taken by the Earth Observing - 1 satellite‟s
pixel on the unit sphere and the three reference Hyperion sensor was initially classified under soil,
vectors were calculated to form a distance vector water and vegetation, based on the HSI data of the
for each pixel, which is given by image. An algorithm based on classical feature
reduction techniques, was used for the
… (8) aforementioned classification. The results of the
classification are tabulated in Table 3 and
where, represented using a pie chart as in Figure 3 and the
= the distance vector recreated image generated is shown in Figure 4 of
= unit vector representing a given pixel which soil is represented in red, vegetation in
= unit vector representing the reference green and water in blue which is comparable with
spectral signature of soil the original true colour RGB image in Figure 1 (c).
388
Figure 6, 7 and 8 respectively.
Table 3 - Number of Pixels belonging to each
Class
Substance Soil Vegetation Water
Number of 13236 27026 3663

pixels
Figure 5 - Mixed pixels identified in the

Recreated Image
Figure 3 - Percentage of each substance in the

chosen part of the image
Figure 6 - Percentage of soil in each mixed pixel
Figure 4 - Recreated image based on the results of

Classification
Figure 7 - Percentage of vegetation in each mixed
pixel
However, it was evident that most of the pixels
had more than one substance within the pixel. This
is expected, as the size of a pixel correspond to an
area of 30m x 30m. Even though the pixels have
more than one substance, the substance that covers
the majority of the area of the pixel was taken as
the class of that specific pixel in this step. This
practical interpretation corresponds to the shortest
euclidean distance between the pixel and the three
pure substances in the feature space.
Subsequently, the pixels which didn‟t have a

majority percentage of affinity greater than 50%, Figure 8 - Percentage of water in each mixed
was considered to be mixed, that contained pixel
substances belonging to more than one class. The
mixed pixels hence identified, are shown in Figure
5 in yellow. The percentage of each end member;
soil, vegetation and water are depicted through
389
6. Conclusion and SAR in remote sensing,” ISPRS J. Photogramm.
Remote Sens., vol. 93, pp. 112–122, 2014.
Since hyperspectral image data contains fine data
4. “USGS EO-1.” [Online]. Available:
which is indeed necessary for a fine classification,
https://eo1.usgs.gov/. [Accessed: 11-Jan-2017].
HSI image data analyzing approach results in
better classification. The image considered in this 5. Simon, K. “HYPERION LEVEL 1GST (L1GST)
paper consisted of 198 spectral bands which had PRODUCT OUTPUT FILES DATA FORMAT
calibrated data. Hence, it is possible to select the CONTROL BOOK (DFCB) Earth Observing-1
most useful spectral bands for better classification. (EO-1).” USGS Center for Earth Resource
However, excess information could also be Observation and Science (EROS) Sioux Falls,
challenging, especially in cases with computational South Dakota, Apr-2006.
limitations. In this paper, an algorithm which is
based on the classical dimensionality reduction 6. Griffin, M. K. ,. Hsu, S. M., Burke H., Orloff, K.
algorithm PCA and transformation of the spectral S. M., Upham, C. A. and Misra, B. “Examples of
EO-1 Hyperion Data Analysis,” Lincoln
data of each pixel onto a unit sphere have been
Laboratory, MIT, Lexington, Massachusetts,
used in a logical manner as explained in the body HTAP-21, Jan. 2005.
of the paper. The classification algorithm in
classifying the pixels was a combination of PCA 7. Vaiphasa, C. “Consideration of smoothing
and a comparison of mean spectral signatures of techniques of hyperspectral remote sensing,”
the training data set with unlabeled pixels based ISPRS J. Photogramm. Remote Sens., vol. 60, no. 2,
on euclidean geometry focusing on the directions pp. 91–99, 2006.
of the vectors solely. The classification process
resulted in an image similar to the true RGB 8. N. US Department of Commerce, “ESRL Global
image. It could further be concluded that, the part Monitoring Division - Global Radiation Group.”
of the image of a strip in North Eastern Sri Lanka [Online]. Available:
https://www.esrl.noaa.gov/gmd/grad/solcalc/.
is mainly covered by soil, followed by foliage
while water bodies are much less in the region of
9. Filipczuk, P., Krawczyk, B. and Woźniak,
consideration. Considering the pixels that were “Classifier ensemble for an effective cytological
classified as mixed, it could be concluded that the image analysis,” Pattern Recognit. Lett., vol. 34, no.
mixed pixels tend to appear mainly around 14, pp. 1748–1757, Oct. 2013.
borders of clustered pixels belonging to the same
class. 10. Tyo, J. S., Konsolakis, A., Diersen, D. I. and Olsen,
R. C. “Principal-components-based display
Acknowledgement strategy for spectral imagery,” IEEE Trans. Geosci.
Remote Sens., vol. 41, no. 3, pp. 708–718, Mar. 2003.
We are immensely grateful to the United States
11. Raiko, T., Ilin, A. and Karhunen, J., “Principal
Geological Survey (USGS) for making available the component analysis for large scale problems with
Hyperspectral images obtained by the EO - 1 lots of missing values,” presented at the 18th
satellite‟s Hyperion sensor and other related data European Conference on Machine Learning
which were used in applying the algorithm (ECML 2007), Warsaw, Poland, 2007.
discussed in this paper.
12. Tharshini, G., Dinesh, G., Godaliyadda, G. M. R. I.
and Ekanayake, M. P. B. “A Robust Expression
References
Negation Algorithm for Accurate Face
Recognition for Limited Training Data,” presented
1. Clark M. and Kilham, N. “Mapping of land cover at the 10th IEEE International Conference on
in Northern California with simulated Industrial and Information Systems, ICIIS - 2015,
hyperspectral satellite imagery,” ISPRS J. Colombo, Sri Lanka, 2015.
Photogramm. Remote Sens., vol. 119, pp. 228–245,
2016.
2. Ntouros, K. D. Gitas, I. Z. and Silleos, G. N.

“Mapping agricultural crops with EO-1 Hyperion
data,” presented at the Hyperspectral Image and
Signal Processsing: Evolution in Remote Sensing,
2009. WHISPERS ‟09. First Workshop on,
Grenoble, France, 2009, pp. 1–4.
3. Zabaiza J. et al., “Novel Folded-PCA for improved

feature extraction and data reduction with HSI
390
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[391 399],2017
2017
The Institution
Lanka
Spatial Modulation for Indoor MIMO Visible Light

Communication System
U.G.R.K.W.S. Palitharathna, A.G.D.U.K. Samarasinghe, Y.M.L.D. Wijayarathna,
S.A.H.A. Suraweera and G.M.R.I. Godaliyadda
Abstract: Power light emitting diodes (LED) are widely used in interior lighting systems due to
the high energy efficiency and convenience of use. In addition, visible light communication (VLC)
systems use light emitting diodes for communication purposes exploiting their high frequency
capability. Though VLC systems are capable of achieving higher speeds, implementing multiple-input
multiple-output (MIMO) systems is challenging due to co-channel interference. This paper presents
the design and implementation of a VLC MIMO system using spatial modulation technique to
suppress the co-channel effect of parallel VLC Links. Proposed controlling system can be used to
obtain the optimum performance.
Keywords: Visible light communication, spatial modulation, multiple-input multiple-output

system.
1. Introduction VLC to be used in indoor wireless

communication systems and for underwater
The need for a new medium of wireless communication [8].
communication has emerged to satisfy the
increasing demand of faster and more secure Even though there are lot of advantages, VLC
wireless communication, as the radio spectrum has several challenges too. Effect of ambient
is already crowded [1]. In order to fulfil these light and other light sources, enabling duplex
requirements, visible light communication communication, non-line of sight effect, and
(VLC) technology is being developed to enabling multiple input multiple output
modulate and transfer data using visible light (MIMO) communication are the major
where light emitting diode (LED) is used as the challenges to be addressed [9,10,11]. Using an
transmitter, while photodiode is the receiver. adaptive algorithm to remove time varying DC
offset voltage component at the receiver,
As a result of latest technology, power LEDs are ambient light effect can be removed. Various
widely used in interior lighting systems due to frequency selective filters are being used to
their high power handling capability, high remove flickering effect, fluorescent lamp effect
energy efficiency, high reliability, and long etc. Among various methods that have been
lifetime than normal LEDs [2,3,6]. In addition, presented, most practical methods to enable
high frequency capability of LED enables the duplex communication is using visible light as
LED to be modulated with high rate digital downlink while infrared is up link.
data. Exploiting these inherent features, MIMO technique is used to have multiple VLC
existing LED lighting systems like interior links in the same indoor environment to
lighting systems, traffic lights [4], and LED achieve a network of communication channels.
display units can be adopted to use VLC. Also MIMO can increase the performance of the
system.
VLC systems use visible light spectrum which Mr. U.G.R.K.W.S. Palitharathna, B.Sc. Eng.
is from 380nm to 780nm in the electromagnetic (Peradeniya), Temporary Instructor, Dept. of Electrical &
spectrum for data transmission. Since this range Electronic Engineering, University of Peradeniya.
Mr. A.G.D.U.K. Samarasinghe, B.Sc. Eng. (Peradeniya),
of frequencies doesn‟t interfere with radio Verification Engineer, Synopsis Lanka (Pvt) Ltd, Sri Lanka.
frequency (RF) communications, VLC is a better Mr. Y.M.L.D. Wijayarathna, B.Sc. Eng. (Peradeniya),
solution for RF spectrum overcrowding. VLC is Engineer, SysmexEngineering (Pvt) Ltd, Peradeniya.
more secure since it is confined to the area Eng. (Dr.) S.A.H.A. Suraweera, B.Sc. Eng. (Peradeniya),
Ph.D. (Monash), Senior Lecturer at Dept. of Electrical &
where transmitted light signal reached. Visible Electronic Engineering, University of Peradeniya.
light has smaller wavelength compared to RF Eng. (Dr.) G.M.R.I. Godaliyadda, B.Sc. Eng.
which makes VLC systems to operate even at (Peradeniya), AMIESL(Sri Lanka), Ph.D.(NUS), Senior
Lecturer at Dept. of Electrical and Electronic Engineering,
1Gbps [5,6,7]. These advantages have made University of Peradeniya.
1
391
Even though there are various techniques are
being used to achieve MIMO, spatial 2.1. Theoretical Background
modulation [12, 13, 14] that also use the
position of the transmitter for the LED was assumed to be a Lambertian emitter
communication can be efficiently used for where light intensity is constant along a circle
VLC.Even though spatial modulation is mainly and Lambertian emitter model was used to
used for RF, it is not commonly used with VLC. derive a mathematical equation for the power
gain of the VLC channel for the geometric
In this paper,the design and implementation orientation shown in Figure 2 as follows [8]:
methodology of spatially modulated VLC
MIMO system is described. A VLC MIMO ℎ= = cosm(a)cos(b) ... (1)
system was implemented using eight power
LEDs as transmitters and eight photodiodes as
receivers. Furthermore, spatial modulation Where Lambertian order is given by,
technique and special lens arrangement was
introduced to mitigate co-channel interference m =− ..(2)
and increase robustness of the system.
Proposed controlling system of the lens
Next,the maximum angle of the receiver with
adjustment was used to optimize the
the vertical axis was derived to identify the
performance of the system for dynamic spatial
geometrical limitations of the VLC link.
coordinates of the transmitter and the receiver.
Following sections of this paper will describe K=𝑐𝑐−tan-1( ) … (3)

the design and implementation of the system.
Basics of implementing a single VLC link are
presented under Section 2. Next, under Section
3, spatial modulation is introduced. Under
Section 4,design and implementation of spatial
modulated MIMO system are introduced.
Finally, under Section 5, results for the
experimental setup are introduced.
2. Basics of the VLC Link
Design and implementation aspects of a single

Figure 2 – Geometry of the VLC link.
directional VLC link arepresented in this
section. In spite of having a transmitter and a
Here,A is the detector physical area of the
receiver as in every communication system,
photodiode, d is the distance between the
VLC uses a power LED as transmitter and a
transmitter and the receiver, a is the angle of the
photodiode as the receiver. Light emitted from
line-of-sight (LOS) and the normal to the
LED is modulated according to the data signal.
receiver, b is the angle between LOS and the
The major sub modules of the VLC system are
vertical axis, c is the power angle at the
shown in the Figure 1.
receiver, K is the maximum tilting angle of the
photodiode.
Transmitter Receiver
2.2. Transmitter
Data Source Data Sink
The main purpose of the VLC transmitter is to
convert digital data signal into variation of light
Receiver intensity. Modulated digital data stream to be
LED Driver transmitted is 0-5 V signal from the
Circuit
microcontroller. Since the power output from
the microcontroller is not enough to drive a
power LED, a LED driver circuit using
LED Photodiode 2N2222[15] power transistors is used to bridge
the data source and the 1W power LED
Figure 1 –Block diagram of the VLC link.
2
392
transmitter due to high power handling and measured for 200 kHz pulse stream by
high frequency capabilities of transistors. increasing the distance between transmitter and
On off keying (OOK) can be easily used for receiver in steps. Moreover, to observe the
VLC systems. But due to variations of the data effect of BER having several LEDs at the
signal, the light intensity is changed with time transmitter, the number of LEDs at the
making unnecessary fluctuations of light in transmitter were increased up to 3 by placing
indoor conditions. Pulse width modulation them in a circle with equal distances.
(PWM) is a good replacement for OOK. But
sincepeak-to –average power ratio (PAPR) of According to the results as shown in Figure 4,
PWM is 50%, pulse position modulation BER is higher for longer distances. With the
(PPM)which has a PAPR of 75% is used. In increment of number of LEDs, BER can be
PPM, the light intensity is kept at a higher and reduced greatly. To achieve longer distance
constant value which is convenient for human communication larger number of LEDs are
eye. Four PPM patterns were selected to needed.
represent two bit data packet as shown in
Figure 3. 1.E+00
1 LED 2 LEDs 3 LEDs
1.E-01
1.E-02
1.E-03
1.E-04
BER
1.E-05
1.E-06
1.E-07
1.E-08
1.E-09
10
30
50
70
90
110
130
150
170
190
Figure 3 – PPM bit patterns. Distance (cm)
2.3. Receiver Figure 4 – BER vs. distance for VLC link.
SFH 203P photodiode is used at the receiver as 3. Spatial Modulation

its spectral range of sensitivity is from 400nm to
1100nm which is in visible light range and it
In MIMO VLC system multiple transmitters are
has high frequency capabilities [15]. Since the
transmitting and multiple receivers are used to
output of the photodiode is a current signal, it
receive signal on the same air channel [12, 13,
was converted to a voltage signal using a trans-
14]. Through this, parallel VLC links can be
impedance amplifier. The voltage output of the
implemented to transmit different data streams.
trans-impedance amplifieris fed into a
Even though phase shift keying (PSK), PPM can
microcontroller to extract transmitted data.
be used for multi-channel systems, they cause
Moreover, moving average and median filters
bandwidth to be reduced. To mitigate this effect
are used at the microcontroller to remove noise
spatial modulation technique can be used.
components. DC voltage component due to the
Spatial modulation technique for four
ambient light effect is removed by subtracting
transmitters and four receivers are explained in
the minimum voltage from the received signal.
Figure 5.
Finally PPM signal was demodulated to get the
original transmitted data signal.
2.4. Performance of the VLC link
Experiments were carried out to quantify the

performance of the VLC link using bit error rate
(BER) which is the most commonly used
method in communication systems. BER was
calculated taking the ratio between the
numbers of bits received with an error and the
total bits transmitted. Transmitter and receiver
were placed in a vertical axis and BER was
3
393
control system to control the position of the lens
is implemented.
Figure 5 – Spatial modulation.

Four receivers R1, R2, R3, and R4 receive Figure 7 – Experiment setup.
different data signals transmitted from T1, T2, 4.1. Transmitter
T3, and T4 respectively. As an example optical
lens focus light from T1 on R1 as shown in Data that need to be transmitted are sent to the
Figure 5. Due to the spatial distribution, R1 microcontroller using transmitting personal
receives light signal only from T1. The computer. In the microcontroller, data signals
interference from other transmitters are are separated into eight channels and
avoided using the optical lens arrangement. modulated using PPM and are sent to the
This method is used to differentiate distinct corresponding LED driver which is same as a
light signals from spatially distributed VLC single directional VLC link. Corresponding
transmitters in indoor environment. LED transmitters on the LED panel as shown in
Figure 8 transmit the light signal. Each
transmitter uses three power LEDs to have a
4. Implementation of the Spatial
higher light intensity at the receiver. The LED
Modulated MIMO system transmitter array was made in such a way that
the x and y coordinates of a particular
Experiment setup for VLC MIMO link was transmitter and the vertical distance of the
designed and implemented to be able to handle transmitter from the receiver also can be
eight different VLC channels. Block diagram of
the MIMO system is shown in Figure 6.
changed for experimental purpose.

Figure 6 – Block diagram of experiment setup.
Figure 8 – LED transmitter array.
The transmitter consists ofa power source, a
microcontroller, LED driver circuit, and an
array of eight LED transmitters each having
three LEDs. The receiver consists of a power
source, a microcontroller, trans-impedance
amplifier, and eight photodiodes. In between
the transmitter and the receiver an optical lens
arrangement is used with the capability of
moving along the vertical axis. Further, a
4
394
4.2. Receiver
The receiver of the MIMO system has a

photodiode bed which consists at least eight
photo diodes to receiver transmitted signals as
shown in Figure 9.
Figure 10 – Double convex lens arrangement.
Here O’ is the position of the LED transmitter, I’

is the position of the photodiode receiver, F is
the focal point of the optical lens, d0is the
distance from LED panel to optical lens, diis the
distance from photodiode panel to optical lens,
f is the focal distance, h0is the distance from axis
of the optical lens to the LED, hiis the distance
from axis of the optical lens to the photodiode.
Two equations can be derived using the
Figure 9 – Photodiode array.
principles of optical lens as follows.
Photodiode grid is made in such a way that the
x and y coordinates of a photodiode can be … (5)
changed. Those photodiodes are connected to a
special receiver circuit with trans-impedance
amplifiers for each photodiode separately. …(6)
Outputs of the photodiodes are connected to an
analog to digital converter. Digitized signals are Here f is known. For a particular spatial
fed into microcontroller to demodulate and distribution of transmitter and receiver, since
separated data into each channel. After that the angle between transmitter and receiver is
those separate data signals are sent to data known, h0 and hi can be easily determined. Also
sinking personal computer using serial d0 + di is constant and can be determined. Using
communication from microcontroller to the these two equations exact positions of the
computer. transmitting LED, receiving photodiode, and
optical lens can be modelled. Position of the
4.3. Optical lens optical lens is fixed for optimum performance
of the system.
In spite of having eight transmitters and
receivers, VLC needs modulation method to 4.4. Control system
operate 8 channels at the same time since each
channel make co channel interference from one Even though optimum lens position for static
to other. In spatial modulation double convex receiver can be calculated using above
optical lens is used to reduce the co channel equations, when transmitter to receiver length
interference. In addition to the power level changes optimum lens position changes.
spatial modulation uses spatial distribution of Special control system is designed that can be
transmitter and receiver. It can be easily adopted to use with dynamic receivers to adjust
derived using principles of optical lens. the position of the lens before transmitting data.
Focusing mechanism of double convex lens are In this control system first transmitter transmit
shown in Figure 10. predefined bit sequence with ones and zeroes
and optical lens moves using two servo motors
from bottom to top while receiver calculates the
output peak to peak voltage and stores in an
array. Then maximum value among them is
selected. Again lens moves from top to bottom.
Once the maximum voltage with threshold
value is reached it stops. Two limit switches are
used to detect upper most and lower most
5
395
limits of the controller. Following algorithm is
used on a separate microcontroller to detect the
optimum position for the optical lens at a given
instant.
Start
Send bit pattern „01‟

from transmitter
Move lens upward Sample voltage at receiver
Calculate peak to peak voltage
Store calculated values Figure 12 – Experiment setup.
Here X is the distance from axis of the optical

No If lens is at lens to transmitting LED, L1 is the vertical
the top
distance between transmitter and receiver, L2 is
the distance between receiver and optical lens.
Yes D is the distance between axis of the optical
lens
Determine the maximum peak to lens to receiving photodiode when peak voltage
is at
peak voltage
the
is detected.
top Since at an indoor lighting systems transmitters
Yes position
Move lens to the optimum are fixed, X was assumed to be fixed at 6cm.
Also experiment was done for the static
End receiver assuming L1 to be 50cm. random data
stream was sent from the transmitter at 200
Figure 11 – Algorithm to detect the kbps data rate.
optimum lens position. The measured output peak to peak voltage at
the output of the receiver R2 circuit with
5. Experiment on performance changing D for fixed L2 value at 10cm is as
shown in Figure 13.
Even though spatial modulation method using
1.5
optical lens enables multiple VLC links to be

operated at same indoor environment without
an interference, there are some disadvantages
Output Voltage, Vout(V)
caused by this methodology. Specially, optical

1
lens should be adjusted by the controlling

system to get the light signal properly on to the
effective area of the receiving photodiode.
0.5
Otherwise there won‟t be any output at all.

Also the distance between transmitter and the
receiver, the distance among transmitters, and
0
the distance among photodiodes are important 0 1 2 3 4 5

D (cm)
6 7 8 9
factors to be considered. With proper

Figure 13 – Output peak to peak voltage (V) at
performance analysis these effects can be R2 vs. D (cm).
identified. According to the Figure 13, for a fixed position
of the optical lens there is a peak value at the
To identify effect of different parameters on graph. It shows only at a particular D value it
performance of the system set of experiments receives data. For L2 at 10cm this optimum
were done changing the parameters of the value of D is 2cm. At that D value the
setup as shown in Figure 12. performance of the system is very high. But
with slightly changed D value output peak to
peak voltage is low which results to have more
errors at the receiving signal.
6
396
Next the D value was fixed at its optimum Above results are important when designing a
position which is 2cmfor this setup and L2 was spatial modulated MIMO system. Designing
changed.The output peak to peak voltage was parameters and dimensions of the system
observed for different L2 values. Observed should be accordingly.
results are shown in Figure 14.
In this system BER is related to output peak to
1.4
peak voltage. But for proper analysis BER value
need to be observed. For the implemented
Output Voltage, Vout (V)
1.2
system BER was measured at different L2
1 distances to ensure the performance of the
0.8 system. Figure 16 shows the BER of the VLC
0.6
MIMO system at different lens positions.
Transmitter to receiver distance was set to 1m.
0.4
X was set to 6cm and D was at 2 cm.
1.00E+00
0.2
0
0 5 10 15 20 25 30 35
L2 (cm)
Figure 14 – Output peak to peak voltage (V) 1.00E-01
vs. L2 (cm).
BER
According to the Figure 14 graph has a sharp
peak at 10cm. Which means for fixed D value at 1.00E-02
2cm, light signal focused on to the photodiode

when L2 is 10cm. at slightly different L2 this
system doesn‟t receive data well. This is a very
1.00E-03
important observation for this system. For a 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
L2 (cm)
static receiver L1 is fixed. In practical situations
D cannot be changed. Observing Figure 13 it Figure 16 – Bit error rate vs. L2 (cm).
was set to 2cm where L2 can be adjusted within
the systems physical limits. Using this results According to the Figure 16, optimum BER can
for particular indoor environment D can be be observed when light is focused on to the
fixed considering Figure 13. For dynamic effective area of the photodiode.
receivers, optical lens need to be adjusted to get
the maximum output voltage at the receiver.
When designing a particular receiver for a

spatially modulated VLC MIMO receiver, to
limit its physical dimensions behaviour of L2
with D needs to be identified. For different L2
values D values at maximum output peak to (a) (b)
peak voltage was observed by changing L2 Figure 17 – Light patterns at the receiver
values. Observed results are as in Figure 15. (a) Focused, (b) Not focused.
According to the Figure 17 when lights are

9
focused on to the photodiodes sharp light
8
pattern can be observed. It can be clearly seen
7
that there is no interference when focused.
6 When it is not properly focused light patterns
D (cm)
5 interfere with each other resulting higher BER

4 at each receivers.
3
2
6. Future work
1
0
0 5 10 15 20 25 30 35 For robust and continuous communication,
L2 (cm) adaptive algorithm for controlling system need
to be implemented to optimize the spatially
Figure 15 – D (cm) vs. L2 (cm). modulated MIMO system for continuously
changing spatial coordinates of the receiver.
7
397
When extending this system for a mobile References
receiver closely placed photodiode array with 1. Karunatilaka, D, Zafar, F., Kalavally, V., and
more photodiodes need to be implemented to Parthiban, R., 'LED Based Indoor Visible Light
detect the small spatial variations of the Communications: State of the Art', IEEE
receiver. Communications Surveys & Tutorials, vol. 17,
In this paper, the presented spatial modulation pp. 1649 -1678, third quarter 2015.
2. Gancarz, J., Elgala, H., and Little, T.D.C.,
technique is applicable to downlink
“Impact of Lighting Requirements on VLC
communication. However, in a complete Systems,” IEEE Communications Magazine, vol.
communication system, both downlink and 51, no. 12, pp. 34-41, Dec. 2013.
uplink needs to be operational. For the uplink
infrared or RF can be used. 3. Komine, T., and Nakagawa, M., 'Fundamental
analysis for visible-light communication system
Moreover, the transmitters and receivers where using LED lights', IEEE Transactions on
designed as 2D arrays. However, there could be Consumer Electronics, vol. 50, no. 1, pp. 100-107,
effects due to the 3D geometry of transmitters Feb. 2004.
and receivers. To obtain the optimum
4. Cailean, A., Cagneau, B., Chassagne, L., Topsu,
performance of the system, geometric aspects of S., Alayli, Y., and Blosseville, J., 'Visible light
transmitters and receivers need to be further communications: Application to cooperation
analysed. between vehicles and road infrastructures', in
Proc. IEEE Intelligent Vehicles Symposium, Alcala
In practical indoor environments, multipath, de Henares, Spain, 2012, pp. 1055-1059.
reflection, and refraction effect can occurdue to
shiny surfaces. To this end, suitable filtering 5. Azhar, A., Tran, T., and O‟Brien, D., “A
methods are needed to remove the noise added Gigabit/s indoor wireless transmission using
MIMO-OFDM visible-light communications,”
due to the above effects.
IEEE Photon. Tech. Lett., vol. 25, no. 2, pp. 171–
174, Jan. 2013.
7. Conclusion
6. O. Bouchet et al., "Visible-light communication
In this paper, we presented a spatial system enabling 73 Mb/s data streaming,"in
modulation technique which can be used to Proc.IEEE Globecom2010 Workshops, Miami, USA,
implement parallel VLC channels in the same Dec. 2010, pp. 1042-1046.
indoor environment without any interference
7. Grobe, L., Paraskevopoulos, A., Hilt, J., Schulz,
from adjacent links. Although the implemented
D., Lassak, F., Hartlieb, F., Kottke, C.,
system works perfectly up to 200 kHz at a Jungnickel, V., and Langer, K., 'High-speed
distance of 2m using general purpose visible light communication systems', IEEE
electronics, using more sophisticated Commun. Magazine, vol. 51, no. 12, pp. 60-66,
electronics, the system can be made to operate 2013.
at a higher data rate.
8. Wang, C., Yu, H. Y., and Zhu, Y. J., "A Long
The results of the paper demonstrate that co- Distance Underwater Visible Light
Communication System With Single Photon
channel interference can be mitigated using the
Avalanche Diode," IEEE Photonics Journal, vol. 8,
developed spatial modulation technique. Also
no. 5, pp. 1-11, Oct. 2016.
due to the use of optical lens arrangement to
focus the light beam on each photodiode, effect 9. O'Brien, D. C., Zeng, L., Le-Minh, H.,
from other light sources can be reduced. Faulkner, G., Walewski, J. W., and
Randel, S., "Visible Light Communications:
This VLC MIMO technique can be used to Challenges and Possibilities", in Proc. IEEE
implement an indoor wireless VLC MIMO PIMRC 2008, Cannes, France, Sept. 2008, pp. 1.5.
system where several LED transmitters are
10. Jovicic, A., Li, J., and Richardson, T., 'Visible
mounted on the ceiling while receiving device
Light Communication: Opportunities,
is attached to the mobile device or personal
Challenges and the Path to Market', IEEE
computer on the use. Receiving device consists Commun. Magazine, vol. 51, no. 12, pp. 26-32,
an optical lens and several photodiodes where 2013.
light signals from different transmitters can be
focused on to different photodiodes to reduce 11. O'Brien, D., 'Visible Light Communications:
co-channel interference. Challenges and Potential', IEEE Photonic Society
24th Annual Meeting, Arlington, USA, 2011, pp.
365-366.
8
398
12. Di Renzo, M., Haas, H., Ghrayeb, A., Sugiura, S.,
and Hanzo, L., "Spatial Modulation for
Generalized MIMO: Challenges, Opportunities,
and Implementation," Proceedings of the IEEE,
vol. 102, no. 1, pp. 56-103, Jan. 2014.
13. Kumar, C. R., and Jeyachitra, R. K., "Power

Efficient Generalized Spatial Modulation MIMO
for Indoor Visible Light Communications," IEEE
Photonics Technology Letters, vol. 29, no. 11, pp.
921-924, June 2017.
14. Alaka, S. P., Narasimhan, T. L., and

Chockalingam, A., "Generalized Spatial
Modulation in Indoor Wireless Visible Light
Communication," in Proc.IEEE Global
Communications Conference (GLOBECOM), San
Diego, USA, Dec. 2015, pp. 1-7.
15. Edirisinghe, E. S. S., Karunarathna, P. H. R. S. S.,

Dissanayake, D. M. T. B., and Godaliyadda, G.
M. R. I. "Design and implementation of a bi-
directional visible light communication system,"
in Proc. IEEE 10th International Conference on
Industrial and Information Systems (ICIIS),
Peradeniya, Sri Lanka,Dec. 2015, pp. 519-524.
9
399
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[401 409],2017
2017
The Institution
Lanka
Energy Management System of a Solar Driven DC

System
H.M.A.A.B. Herath, K.K.S.U. Kodithuwakku, D.R.D.S. Dasanayaka,
P.J. Binduhewa, J.B. Ekanayake and S.M.K.B. Samarakoon
Abstract: Today in the power sector the focus is shifting towards distributed power generation
in the mode of renewable sources instead of fossil fuels. Solar photovoltaic is an attractive renewable
energy source for Sri Lanka considering the location. Generation of solar power has some limitation
with the time of the day as it only produces power in the day time. So, it is not going to be match with
the load pattern of the consumer. There by this creates a need of a battery-storage for the non-power
generation time. This sort of a system with a battery-storage creates a list of challenges which include
intermittency of source, time of the day prices, sizing of solar panels and battery, limitations of
charging and discharging rates of the battery. The task of the proposed system is to maximize the
savings for the power consumer using some methodology. In order to do so, this paper introduces an
algorithm for an energy management system of a Solar driven dc system. In fact this paper presents
the decision making side of an Energy Management System which also includes a physical layer
comprise of Power electronic converters. Outputs of the results of the system are fed into that
controlling system’s controllers as inputs. The charge controller of the battery uses this algorithm for
making decision on charging or discharging of the battery or keeping it in the idle mode. Then the
system commands when to use the grid and when to net meter the excess power. At the initial stage
the algorithm is simulated in MATLAB with the use of realistic load and solar generation patterns. In
the later stage a sensing circuit is developed to do a hardware simulation of the system. The proposed
system with the algorithm helps in effectively deriving the potential benefits of a solar system with
battery storage by providing an optimum usage routine.
Keywords: Algorithm, Solar, Battery, Load, Energy, Power, Management, MATLAB, Raspberry
Pi
1. Introduction panels [3]. Today there are many solar power

users use battery storages. Just having battery
Today all the Governments in the world are in storage doesn’t give the consumer all the
search of ways to shift more of their power available benefits of a solar power system. This
generating sources towards renewable energy is where an Energy Management System (EMS)
sources. With the increased awareness of the is required to gain those available benefits.
depletion of traditional energy sources and The EMS should be with an effective algorithm
environmental damage caused by increased which considers all the aspects related to
carbon dioxide emissions from coal-fired power gaining the benefits through the system. The
generation, the use of renewable energy has base of the EMS should be cost as the amount
become the goal for energy development. Many of savings going to matter at the end of the day.
countries have set a goal of increasing the usage So, initially the per unit power generation cost
of renewable energy above 20% of their total of each source was required to be calculated.
power consumption by the year 2020 [1]. Out of Then the technical specifications of each power
the renewable power sources solar energy is generating source need to be considered.
more straight forward, flexible and much easier Considering both of these aspects at a given
to set up the power generation as well [2]. So, time period Energy Management System
roof top solar turned out to a popular energy should recommend the power sources usage to
generation source. This also mainly due to the fulfil the load requirement.
fact that roof top solar panels doesn’t require a Mr. H.M.A.A.B. Herath, B.Sc. Undergraduate at Dept. of E&E
separate area and it can be easily installed on Engineering, Peradeniya.
the roof of the power consumer’s house or any Mr. K.K.S.U. Kodithuwakku, B.Sc. Undergraduate at Dept. of
E&E Engineering, Peradeniya.
sort of building. As the Solar power does not Eng. (Mrs.) D.R.D.S. Dasanayaka, B.Sc. Undergraduate at
exist throughout the day there should be a Dept. of E&E Engineering, Peradeniya.
Dr. P.J. Binduhewa, B.Sc. Eng. (Peradeniya),
proper management which utilize this Senior Lecturer at Dept. of E&E Engineering, Peradeniya.
renewable source generated power in the best Eng. (Prof.) J.B. Ekanayake, B.Sc.Eng.(Peradeniya),
possible manner. In order to do so battery Professor at Dept. of E&E Engineering, Peradeniya.
Eng. (Dr.) S.M.K.B. Samarakoon, B.Sc.Eng.(Peradeniya),
storage need to be installed along with the solar Senior Lecturer at Dept. of Computer Engineering, Peradeniya.
1
401
This Paper presents an Energy Management 3. Development of load profile, PV
System considering the above mentioned profile and selecting Battery storage
aspects. This study was done to implement the
system on a Sri Lankan household consumer’s 3.1 Typical Load profile for a Domestic user
power facility. As the electricity tariff
household time based tariff was considered [4]. Power consumption of a medium type house in
Initial implementation of the algorithm was Sri Lanka is considered to develop the load
done in MATLAB and the expected results profile in the proposed system. Table 1 shows
were achieved. Then as the start of the the electrical loads in the house and their power
hardware implementation the algorithm was ratings. For this analysis, it was assumed that
implemented on a Raspberry Pi and the system dc-loads with power ratings given in Table 1
gives a video output. This paper includes the are used in the considered domestic house.
work procedure up to that point. Here the
raspberry pi model B was used. The MATLAB Table1 - Power Ratings of Electrical Loads
simulation results and hardware results are
included in this paper. Appliance Power Rating
Refrigerator 32W
2. Proposed distribution board and
Rice cooker 630W
integrated Energy Management
Television 45W
System
Heater 1200W
The Architecture of the power electronic Blender 110W
interface is shown in Figure1. PV panel (Solar Laptop 40W
Panel), battery Storage and the utility supply 30W CFL (2 Nos) 60W
are the power sources included in System and a
16W CFL (1 Nos) 16W
household user’s case is considered.
20W CFL (2 Nos) 40W
11W CFL (2 Nos) 22W
Following assumptions were made when

developing the load profile. Even though the
appliance like refrigerator, washing machine
has varying power usage throughout the using
time, an average on-off cycle for an appliance is
considered [5]. A weekday load profile was
considered. Further it was assumed that the
people in the house start up their activities at 5
am and ends activities at 10.30 pm. Abnormal
weather, national or international events were
Figure 1 - Architecture of the System not considered in this study [6]. Figure 2 shows
the load profile we have developed for a typical
The final output of our System is a set of Sri Lankan household power consumer.
decision to the Power Electronic Converters 1.4
Daily Load Profile
(PEC) which controls the power flow of the

power sources whether to turn ON or OFF. At 1.2
the start of the day a set of these decisions are

given to the converters to follow throughout
1
Load Requirement(kW)
the day for an expected load and power 0.8

generation pattern. Then the System monitors
the decision set throughout the day and if a 0.6
sudden change from the expected values is

there, System is developed to react and change 0.4
the set of decisions for the day from that point. 0.2
0
0 5 10 15 20
Time (h)
Figure 2 - Developed Load Profile
2
402
3.2 PV system power profile manufacturer was used. Here, one average
temperature value for the whole day is
Photovoltaic system is the local power considered, deviations of temperature from that
generation source in the house. It was assumed average value is not considered. The rated
that a rooftop 2 kW photovoltaic system is temperature of the PV panel is 25°C. The power
mounted. In this study, it was assumed that generation of the PV panel is calculated
photovoltaic system consists of an inbuilt through the following equations.
maximum power point tracker thus it always
operates at its maximum power point [3]. So, For temperature, T>25
any expected voltage or current variation in PV Power=Ix2x (1-TE)(T-25) …..(1)
panel for the input power for our system is not For temperature, T=25
considered. In order to generate the Power Power=Ix2 … (2)
generation of the PV panel an irradiance level For temperature, T<25
curve for a day is required. Irradiance is the Power=Ix2x (1+TE)(25-T) … (3)
solar power received by a surface for unit area.
By considering Sri Lanka’s earth positioning an Figure 4 shows PV power generation profile for
irradiance curve for a day in Sri Lanka was a cloudy day with 25°C average temperature.
obtained by applying Sri Lanka’s latitude
values to the Solar irradiance calculation PV Power Generation Profile
0.9
equation set [7],[8]. The chosen day for the
calculation was June 1st, the middle day of the 0.8
year. As Sri Lanka is a country near to the 0.7
equator, the day of the year doesn’t matter 0.6
much for the average irradiance curve for a day

PV Power (kW)
0.5
[7]. To account the effect of cloud cover for
irradiance a sky clearness factor was used,
0.4
which is the actual solar irradiation, divided by 0.3
solar irradiation from a completely clear sky [9]. 0.2
In our study, 3 cases considered as 100%, 60% 0.1

and 40%. That means in our Energy
0
Management System we are considering 3 0 5 10
Time (h)
15 20
types of days with 3 types of average cloud

levels. This sky clearness factor causes to give
Figure 4 - PV Power Generation Profile
wrong peak solar power generation values for a
given time, but this gives an approximately
PV power generation cost is calculated by the
correct PV energy value for the day. The
following equation.
MATLAB plots of these are shown in Figure 3.
Cost= (Initial Installation Cost + Power …(4)
Electronic Converters Cost) /
Daily Irradiance Profile
1.4 Total Expected Power Units
Sunny Day
1.2
Normal Day
Cloudy Day
Selected PV panel generates 228kWh average
per month and the guaranteed life time is 25
1
years, Installation and power electronic
converters cost is LKR 650,000/= [10]. The
Iradiance(kWm-2))
0.8
calculation gave the answer as LKR 9.50/=.
0.6
3.3 Battery Storage

0.4
Most of the batteries used for PV systems are

0.2
Lead acid batteries since they are cheap and
0 they have a higher depth of discharge (DOD).
0 5 10 15 20
Time (h) Various types of Lead-acid batteries are
Figure 3- Daily Irradiance Profile of Sri Lanka available in the market and we have looked in
to common electrical parameters of these
Using the Irradiance curve, the relevant power batteries which include Battery Voltage,
generation curve was obtained. For that Charging and Discharging current, Number of
temperature coefficient (TE) of the PV panel cycles during the lifetime, Operating
which is specified by the PV panel temperature, Capacity of the battery etc.
3
403
State of Charge (SOC) is an important factor
required for the EMS as it decides whether to
charge or discharge a battery. The SOC
estimation methods available in the literature
are open circuit voltage method, Specific
gravity method, SOC estimation by battery
impedance, Coulomb counting method [11]. In
open circuit voltage method load needs to be
disconnected before taking the measurement
and it’s not possible. Taking the specific gravity
measurement of electrolyte is not practical at
load conditions. To decide variation of battery
Figure 6 - Battery cost variation with DOD
impedance parameters with SOC needs lot of
experiments. But coulomb counting method is
much more practical to use in the proposed 4. Development of the Algorithm
system since it determines the SOC by
integrating the charging and discharging This Proposed algorithm consists of cost and
current over the time which is more simple and technical conditions. The main base is cost and
accurate. main focus is cost reduction and increasing of
savings. Initially we have defined the elements
Unit Battery Price/ Energy per cycle/ …(4) in our algorithm and those are of two kinds as
Cost= No.of cycles fixed and variable. Fixed elements are initiated
before the start of the algorithm and the
variable elements either generated or changed
By referring data sheets of industrial solar with in the algorithm.
batteries, it was found that cycles of the battery
are inversely proportional to the DOD. At Fixed Elements:
higher DOD, total cycles of the battery are less Ebat,full Battery Storage Capacity in
and at lower DOD, total cycles of the battery kWh
are high [11]. According to the data obtained by Ebat,min Minimum allowable battery
the data sheet of DISCOVER 12VRE-300TF-L storage level in kWh
Battery, plot in Figure 5 was obtained in Ichrg,rated Battery charging current in A
MATLAB simulations. In this study the effect of Idchrg,rated Maximum Battery discharging
temperature is not considered. current in A
ChrgEff Battery charging Efficiency
DChrgEff Battery Discharging Efficiency
Cbat 1 kWh generation cost of
Battery in LKR
Vbat Battery Storage Voltage in V
Variable Elements:
t time interval
Psolar,t Generated Solar power at t in kW
PLoad,t Load power requirement at t in kW
Ebat,t Battery Storage level at t in kWh
Ichrg,t Battery charging current at t in A
Idchrg,t Battery discharging current at t in A
Figure 5- Battery cycles variation with DOD Cutility 1 kWh cost of the Utility Supply at t
in LKR
By considering all of the above conditions, it
was found that best DOD to operate the battery Calculations:
is 40% (60% of SOC). Figure 6 shows the plot =(Psolar,t -PLoad,t) x
obtained for the unit cost variation with DOD Ichrg,t
ChrgEff/Vbat … (5)
after a MATLAB simulation and it shows that =(PLoad,t- Psolar,t) x
cost is minimum at 40% DOD and the unit cost Idchrg,t
DChrgEff/Vbat ….(6)
is LKR 16.50/= [12]. (200Ah Exide battery, LKR =Ebat,t + Ichrg,t x
31500/=, at 40% DOD) Ebat,t+1
Vbat x hours … (7)
4
404
=Ebat,t - Idchrg,t x Load requirement and PV power generation for
Ebat,t+1
Vbat x hours (8) the considered time period and the current
energy level of the battery storage are the
Figure 7 shows the algorithm we have inputs of the algorithm.
developed for our Energy management System.
Figure 7 - Algorithm (1) For the Energy Management System
Out of the three available power sources it was the battery is not fully charged excess PV
evaluated that PV generation cost is the lowest. power is used to charge the battery. If that
So, PV source stands above the rest two in any charging current is higher than the rated
case. At the initial tier load and the PV power is battery charging current the remaining power
compared and if PV power is greater total load is supplied to the utility. At the initial tier if the
of that interval is fulfilled by the PV power. In load is higher than the PV power hers would be
the next stage battery storage state is a PV power supply deficit. In order to cover
considered and if it is fully charged the excess that deficit next two options are considered. Per
PV generated power is supplied to the utility. If unit cost of both sources are compared and for
5
405
the whole day battery is a fixed value, but the above the minimum allowable storage level
utility cost changes with the time period [4]. battery is used to supply the power.
That variation is shown in the Table 2.
If the required current to the load is higher than
Table 2 - Time of Day Tariff rates the battery’s rated discharging current utility
supply is turned on and it is a situation where
all the three supplies are with ON state
Time Per kWh decision. This algorithm runs for total number
Period/ cost in of time intervals for the day and Power source
hrs LKR profiles of each source are generated. This
2230-0530 13 happens at the start of the day and this sets the
0530-1830 25 system a path to follow throughout the day.
1830-2230 54
If there is a sudden change in the value of the
If the utility is cheap the required power is expected PV power value or Load value, EMS
fulfilled by the Utility. If the battery is cheap has an algorithm to react and to set up a new
the battery storage status is checked and if it is decision set. Figure 8 shows that algorithm.
Figure 8- Algorithm (2) For the Energy Management System update for sudden Source changes
This algorithm was developed considering the 5. Results and Discussion

fact that the only data affect to the next time 5.1 MATLAB Simulation
cycle of the system is battery storage level. Both
the load and the PV power values are For the simulation of the algorithms, the time
independent of the future or past interval interval duration was considered as 5 minutes,
Values. So, what this algorithm does is update which means 288 time intervals for a day. The
the change to the system, consider the battery two algorithms were simulated in MATLAB
storage level as the initial storage value and run and the results are displayed in a plot. Figure 9
the earlier algorithm for the rest of the day as shows the implementation of the first algorithm
the way did at the start of the day. which depicts the Power source and load
profiles set by the Energy Management System
at the start of the day for the Normal Day Case.
25oc was considered as the average
temperature. Battery or Utility power is not
required in the middle part of the day and they
6
406
are only ON in early morning and the later part 5.2 Raspberry pi Implementation
of the day. Because of the utility cost variation
with time of the day preference for battery and The System Algorithm was also implemented
utility usage has varied and it is clearly shown on a Raspberry pi and Figure 11 shows a photo
in the plot. of the setup. Here raspberry pi is considered as
our EMS and its output is displayed on the
Daily Power fulfilment Summary monitor. Figure 12 shows the Graphical
1.4
Solar Power
interface developed to get a data entry from the
Power Generation,Load,Battery and Utility Supply(kW)
Load user. If there is a considerable change from the

1.2
Battery Supply Normal day routine for the next day the user
Grid Supply
1
can enter those specific cases to the system and
he/she can build load pattern for the next day.
0.8
0.6
0.4
0.2
0
0 5 10 15 20
Time (h)
Figure 9- Algorithm (1) MATLAB Simulation
Figure 10 shows the implementation of the

Figure 11 - EMS setup on Raspberry pi
second algorithm in MATLAB. Here sudden
changes to the solar power source and load
were given (At 1300h PV power -400W, Load
Requirement +400W/at 2100h Load
Requirement -300W) and the plot for the day
was obtained. These changes were given to the
case we have considered in implementing the
first algorithm. A comparison of the two plots
figure 8 and figure 9 shows how our energy
management system have reacted to a sudden
changes in conditions. In the above expected
profile battery is OFF at 1300h, Utility is ON at
2100h. In the new profile battery is ON at
1300h, Utility is OFF at 2100h.
Figure 12 - Load Profile User Interface
Daily Power fulfilment Summary with Sudden Changes
1.4
Solar Power In this interface the major appliances of the user
Power Generation,Load,Battery and Utility Supply(kW)
Load
1.2
Battery Supply is given and the expected using time intervals
Grid Supply can be selected. The interface also shows the
load pattern as well as the expected power
1
0.8
usage in values and the expected cost. So, the
user can rethink on any of the plans on the
0.6 power usage for the next day.
0.4 Implementation of the EMS algorithm on

Raspberry pi was done and a screen shot of the
video output obtained. It is shown in Figure 13.
0.2
0
Here the raspberry pi gives a real time output
0 5 10
Time (h)
15 20
of the System implementation the output is the
four live plots of the power sources and the
Figure 10 - Algorithm (2) MATLAB
load.
Simulation
7
407
Solar and Utility User
Cost= Utility Usage x Time Based
Utility Cost + PV generation x PV
unit- Units to utility x Net
..(12)
Accounting Rate
In this comparison payment for a unit of a net

accounting customer was taken as LKR 22/=,
which is the rate which is only valid for the first
7 years of connection. Here, the cost for a
month period was calculated and plotted. This
comparison was done to a user with a load
pattern which consumes 6 kWh per day, 180
kWh per month. Figure 14 shows the cost
Figure 13- Real time Algorithm
comparison and it clearly depicts that both
Implementation
types of EMS user have a considerable saving
over the user who only uses the utility.
These plots were plotted in real time and
showed how the system shifts the sources
within a day cycle. In the video output 24 hour
Cumilative Distribution of Cost for a month
14000
day was scaled down to a 24 second time With EMS(NM)

Only Utility (Time Based)
frame. Any time interval with the power value 12000
With EMS(NA)
of any source is greater than 0 in these plots Solar and Utility

10000
means that the system has considered ON
Cost of the System/LKR
decision for that source and if any power value 8000

is 0 means that either that power source is
turned OFF by the system or it is not available. 6000
5.3 Cost Comparison

4000
2000
A Cost comparison was done to show the
potential benefits of our energy management 0
0 5 10 15 20 25 30
System. For this comparison 4 cases were Day
considered as for a normal utility user, normal Figure 14- Monthly Cost Comparison
solar power user, our EMS user with net
metering connection and our EMS user with a
6. Conclusions and Future Work
net accounting connection. Equations for the
cost calculations are as follows.
In this study, an algorithm for an Energy
management system was presented which
Normal Utility User
controls the power system of a household with
Cost= Total Power Usage x Time Based
..(9) a utility connected solar power system and
Utility Cost
Battery storage. Software level implementation
Net Metering User
was completed and the hardware
Cost= PV generation x PV unit Cost + ..(10)
implementation would be developed around
Battery Usage x Battery Unit Cost
the Raspberry Pi with modelled PV panel and
load modules. A Real system linked with the
Net Accounting User
physical layer which sends decisions to the
Cost= Utility Usage x Time Based
converters is to be developed. An algorithm
Utility Cost + PV generation x PV
which predicts the load pattern of the next day
unit Cost + Battery Usage x ..(11) is to be added to the system. Adding the case of
Battery Unit Cost - Units to
the power cut situation to the algorithm is
utility x Net Accounting Rate
under study and it would be added to the
algorithm in the future. For further
optimization of the system use of a FPGA
instead of the Raspberry pi is to be considered.
8
408
References
11. Chang, W. Y., «The State of Charge
1. Yu-Kai Chen, Member, IEEE, Yung-Chun Wu, Estimating Methods for Battery: A Review»,
Chau-Chung Song, and Yu-Syun Chen “Design Int. Sch. Res. Not., vol. 2013, page953792,
lug. 2013.
and Implementation of Energy Management
System With Fuzzy Control for DC Microgrid
12. https://www.quora.com/How-is-the-cost-
Systems” in IEEE Transactions on Power
Electronics, VOL. 28, NO. 4, April 2013. for-batteries-per-kWh-calculated-Energy-
storage-battery-companies-get-prices-of-
800-per-kWh-for-their-cells-and-Tesla-says-
2. Sonal Gaurav, Chirag Birla, Aman Lamba, S.
their-Powerwall-costs-about-350-per-kWh-
Umashankar, Swaminathan Ganesan, School of
How-are-these-figures-calculated-with-
Electrical Engineering, VIT University, Vellore,
632014, India, Schneider Electric, Bangalore, grid-electricity-prices-around-0-12-per-
560072, India,” Energy Management of PV - kWh, visited 27th May 2017.
Battery based Microgrid System”, SMART GRID
Technologies, August 6-8, 2015.
3. Nasim Jabalameli, Student Member, IEEE, Sara

Deilami, Student Member, IEEE, Mohammad
A.S. Masoum, Senior Member, IEEE, Farhad
Shahnia, Member, IEEE “Rooftop PV with
Battery Storage for Constant Output Power
Production Considering Load Characteristics” in
2013 8th International Conference on Electrical
and Electronics Engineering (ELECO).
4. http://www.pucsl.gov.lk/english/news/press-
release-3/, Visited, 24th May 2017.
5. Manisa Pipattansomporn, Senior Member, IEEE,

Murat Kuzlu, Member, IEEE, Safiur Rahman,
fellow, IEEE, and Yonael Teklu, Member, IEEE
“Load Profiles of major household appliances
and their demand response opportunities”.
6. Luo Chuan and Abishek Uku, Senior Member,

IEEE “Modeling and Validation of Electrical
Load Profile of Residential building in
Singapore”.
7. Anna Petrasova, Brendan Harmon, Vaclav

Petras, Helena Mitasova of Center for
Geospatial Analytics, North Carolina State
University, Raleigh, NC, USA , “Tangible
Modeling with Open Source GIS”, Chapter 8 –
“Solar Radiation Dynamics”.
8. http://www.pveducation.org/pvcdrom/calcul
ation-of-solar-insolation, Visited, 11th May 2017.
9. Exell, R.H.B., King Mongkut’s University of

Technology, Thonburi “The Intensity of Solar
Radiation”-2000.
10. http://www.jlankatech.com/net-metering-
users/, Visited, 24th May 2017.
9
409
410
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[411 419],2017
2017
The Institution
Lanka
Laser Ranging Based Intelligent System for Unknown

Environment Mapping
T.H.M.N.C. Thelasingha, U.V.B.L. Udugama, E.M.S.P. Ekanayake,
G.M.R.I. Godaliyadda, M.P.B. Ekanayake, B.G.L.T. Samaranayake and
J.V. Wijayakulasooriya
Abstract:Autonomous systems are used in many applications like unknown terrain mapping,
Reconnaissance, Explorations etc. in areas of impossible human access. They require an accurate and
practically realizable Intelligent Navigation System(INS), which can handle the complexity in the
environment and necessarily be ready for mobile platform implementation. Also, a method of error
free localization is a major requirement for the practical realization of such a system. This work
outlines implementation of a system, simple enough to be implemented on a mobile platform, focused
on provision of a logical and computationally efficient algorithm which uses measurements from a
Laser Range Finder (LRF) as input. Usage of LRFs for mapping is a trending research area in contrast
to the usage of Ultrasonic Sensors with errors like cross talk and specular reflections or vision systems
that are power hungry. As, the LRFreadings too contain some noise due to reflectivity problems,
methods were developed for pre-processing the measurements. INS analyses the data spectrum
intelligently to determine the obstacle free path that should be traversed, which will in turn enable the
mobile platform to canvas the whole environment and construct the map. Also, the localization
methods have been implemented through usage of similarity transform and particle filter. The
conceptual design was simulated in PythonTM on WindowsTM and was tested for robustness in
artificially generated environments. Then the system was emulated in PythonTM with real
environment data fed from the LRF in real-time. Then it was implemented in the embedded system
level in a Raspberry pi3 TM on a 3WD Omni-directional mobile platform, and tested for real
environments.Also, the separately running localization algorithm has been used to correct the
generated map and the position of the platform. The system was able to generate a complete two-
dimensional map of the locale accurately. A quantitative justification of the proposedmethodology is
presented in a comparative analysis of the execution time.
Keywords: Intelligent Navigation, LIDAR, Autonomous Mapping, Monte Carlo SLAM
1. Introduction the computational power required will make

them less feasible for mobile platform
Environment mapping has been a key fact in implementation. In this work, a low cost LRF
endeavours like space explorations,
Mr. T.H.M.N.C. Thelasingha, B.Sc. Eng. Undergraduate
excavations, disaster relief and reconnaissance at Dept.of Electrical &Electronic Engineering, university of
applications. For that automation is highly Peradeniya, Sri Lanka
essential when such tasks are carried out areas Mr. U.V.B.L. Udugama, B.Sc. Eng. Undergraduate at
Dept.of Electrical &Electronic Engineering, University of
with restricted human access. Mapping an Peradeniya,Sri Lanka
unknown environment autonomously has been Ms. E.M.S.P. Ekanayake, B.Sc. Eng. Undergraduate at
an engineering problem that many researches Dept.of Electrical &Electronic Engineering, University of
Peradeniya,Sri Lanka
have been focussed[1][2] on for years. For it to
Eng.(Dr.) M.P.B. Ekanayake, B.Sc.Eng. (Peradeniya),
besuccessful, development has to be done in Ph.D. (Texas Tech,US), AMIE(Sri Lanka),Senior Lecturer
many fronts. at Dept.of Electrical &Electronic Engineering, University of
Initially a correct sensing method has to be Peradeniya,Sri Lanka
Eng. (Dr.) G.M.R.I. Godaliyadda, B.Sc. Eng.
implemented to sense the geometry of the (Peradeniya), Ph.D. (NUS,Sg), ,AMIE (Sri Lanka), Senior
environment. Many of the published work Lecturer at Dept.of Electrical &Electronic Engineering,
include the use of RADARs[3], SONARs[4], University of Peradeniya,Sri Lanka
Eng.(Dr.). B. G. L. T. Samaranayake, BSc Eng.
Vision systems and Laser Range Finders
Peradeniya, Tech. Lic& PhD KTH(Sweden), Senior Member
(LRFs)[5]. The use of methods like and IEEE, AMIE(SL), Senior Lecturerat Dept.of Electrical
SONARs are convenient for simple distance &Electronic Engineering, University of Peradeniya,Sri
sensing but whenconsidered for applications Lanka
Eng. (Dr.) J.V. Wijayakulasooriya, B.ScEng,
like mapping they contain much noise. (Peradeniya), Ph.D. (Northumbria,UK),MIE (Sri Lanka),
Although vision sensors provide more accurate Senior Lecturer at Dept.of Electrical &Electronic
and vivid data spectrum of the environment, Engineering, University of Peradeniya,Sri Lanka
1
411 ENGINEER
[6]has been used as the distance sensor. A high- orientation of it will be computed and fed to the
end sensor like Microsoft KinectTM will render navigation system in order to find the direction
much accurate data but to keep the it should travel in the next moment. But the
implementation realizable a simple sensor has errors in the measurements and actuation are
been selected. always possible. Also, the encoder based
Another major problem is the method of position is prone to errors due to wheel
traversing the unknown area. Various slipping and data transmission errors. Hence,
approaches like mobile robots and quad- for correct localization and to provide a
rotors[7] are possible solutions. But considering correction for the map data generated from the
the accuracy and convenience in positioning LRF, a separately running simultaneous
and manoeuvrability an omni-directional localization and mapping (SLAM)[12]
3WD mobile platform[8] has been used as the algorithm based on Monte Carlo approach has
mapping agent in this work. It has been also been developed. It uses the LRF sweeps
developed completely in house. And its size from short time frames when the agent
and the instant mobility in any directiondue to traverses in its trajectory and correct the
omni directional nature has allowed it to encoder based position estimate in real time.
traverse many complex environments The combined implementation of the above
effectively. A state feedback linearizing three aspects namely, the hardware setup (omni
nonlinear controller has been implemented so directional mobile platform and the LRF), the
that the platform can adjust itself to a given navigation and path planning algorithm and
reference coordinate and orientation. the SLAM algorithm has allowed to develop a
Additionally, a convenient sensor has been successful attempt in generating a 2D overlay
selected and a method of motion is developed. of an unknown obstacle filled environment. For
Most approaches lack a correct navigation the task of extension of the 2D map to a
system[1] where the agent can be guided complete 3D map, incorporating a vision
through the obstacles in the environment so system would be an effective approach. Further
that it canvases the whole area. Obstacle research in this work would include such a
detection[9]and traversable space identification combination.
methods[10] have already been developed.
Also, algorithms for finding the minimum cost 2. The Hardware Implementation
path has been developed for some time[11].
Figure 2 – The complete embedded system

architecture
Figure 1 – The Practical Realization
2.1 The Omni Directional Mobile
But an approach to intelligently analyse the Platform
data from the distance measurements and
decide the direction to be explored is a The omni-directional mobile platform is the
requirement. Hence a separate navigation practical realization on which the proposed
algorithm has been developed which will work has been implemented. It has been
analyse the information spectrum from the designed and constructed completely in house.
sensor and decides the path to be taken by the A separate embedded system architecture has
mobile platform to create a complete 2D been developed, so that the control algorithms
overlay of the locale. and also sensor fusion algorithms can be
Here, to construct themap, the mobile platform implemented on general purpose micro
traverses the area as guided by the navigation controllers. Also, the electronics of the system
algorithm. And also the current position and including the electronics of motor controlling
ENGINEER 2
412
and other auxiliary systems has been The laser sensor works by projecting a pulse
implemented on a single board with sensor stream of lasers around and measures the time
modules. it takes to reflect back. The distance to the
Here, it is to be noted that the usage of GPS is reflection point can be accurately calculated (to
avoidedas in the scenarios like explorations and 1mm) from the time of flight thesensor then
excavations which the equipment is developed encodes the distance data into a packet and
for, would not guarantee the availability of GPS forward it to the LIDAR controller. The data
everytime. An accurate estimation of the packets are then decoded in the controller and
position and orientation of the mapping agent relayed through the serial port up to the host
(pose or state) is obtained by fusing data from device which is the embedded system that the
three different sensors that detect the Intelligent Navigation System
orientation. The following sensors were used: a (INS)implemented on. As the percentage error
gyroscope (MPU6050), a compass (HMC5883L) is small, in distances less than 3m, the LRF
and optical shaft encoders (OSE) (model - measurement can be used as accurate
E50050030) attached to the wheels. To fuse the estimation of distance.Separate method has
data from the above sensors, a centralized been developed to mitigate the noise in the LRF
Kalman filter (CKF) was implemented in a measurements. Most of it was due to
general-purpose microcontroller. As thereflectivity effects on the object surfaces and
odometrical measurement system is coupled also the variation of the rotational speed of the
with an inertial measurement unit and a LRF. To prevent overlapping of data points and
compass the accuracy of the position estimate incorrect angle measurements, A PID controller
has been improved much, while the system is implemented in the XV LIDAR controller to
moves in different motion conditions like regulate the rotational speed of the sensor to
slipping and jerky motion. As controller the optimum speed of 200 rpm, where the
strategy, a nonlinear state linearization device can send out a distance data point per a
controller has been implemented on the main degree of rotation.Many of the reflectivity
micro controller.so that the ODR platform can distortions could be rectified by sampling
be guided along any given trajectory while through multiple realizations. Here it was
following a given orientation profile. about five realizations, considering speed of
operation and the power consumption.
2.2 The Laser Range Finder (LRF) Data A main capability that the system was intended
Processing to poses was its ability to be versatile and to be
dynamically adaptive to any mobile platform.
The LRF used, has a Neato laser sensor[6]
Hence, at the scanning step, the LRF is powered
driven by the Xv-11 LIDAR controller. The used
ON only at the scanning moment otherwise it is
LRF is a popular low-cost range finder
powered OFF, such that lesser power is
compared to others in the market and has been
consumed in the case of mobileimplementation.
used for many research and experimental
applications. Although the accuracy is much
Figure 4 – A Polar LRF Sweep of 360o
3. The Intelligent Navigation

Figure 3 – The Error Characteristic of LRF System
higher than SONAR, or simple vision systems,
it also has some deviations in the measured The Intelligent Navigation System (INS) is
distances. The error characteristics of the LRF is the entity which analyses the data from the
shown in the Fig. 3. sensors and directs the mobile platform to the
necessary place to explore [17]. The INS was
run for the simulation environment in Fig. 5(a)
3
413 ENGINEER
N6
N3
N2 N2G2 N2G1
N2
N1
(b) (c) (d)

(a)
N13
N7G1
N7
N2
N2 N2G1 N2 N2G1
(f) (g) (h)

(e)
Figure 5 – Simulation of Navigation Algorithm
and the result was generated in the steps shown the first scan. It is a 300o scan ±150o around the
in the Fig. 5(b) to Fig. 5(h). It should be noted agents heading direction. In Fig. 5(d), it can be
that the anchor points where the laser sweeps seen that the agent has now travelled up to the
are done, are named as „nodes‟ and respective node N6 and further to node N7 in Fig. 5(e).
nodes, which the agent can traverse to and 3) Step 03: It can be observed that direct
from each other, are called as „neighbours‟. The travelling from N2 to N7 is possible. Hence,
procedure of the INS can be analysed through those two nodes should be neighbours. Also, it
following explained steps, can be noticed that a scan down towards N2
1) Step 01: All the storage classes and data from N7 is not necessary as the whole
structures are initialized and current position environment in between the nodes have been
and orientation is read. A node is created in the already mapped through the scans done at the
current position namely labelled as N1 in respective nodes. Hence, to handle those kinds
Fig.5(b). (if this is not the first node, new node of situations a separate 'neighbour
is also added as a neighbour for the previous identification' is done as,
node and vice versa) scanning is done for 360o A.) If any two nodes are closer than twice the
initially because agent has to localize and scan range, the whole area in between nodes
decide which direction to explore. The list of can be assumed to be explored. Hence above
distance data per degree is obtained from the situation occurs.
LRF. B.) Each gap of the corresponding two nodes
2) Step 02: Next the gaps and the walls are considered and checked if they intersect the
should be correctly identified. Gap points are line joining the two nodes. They are set as
selected as data points with a relative distance explored gaps as there is no new un-explored
from the agent,which is longer than a given data in side those gaps, which are in between
length. Because those places can be assumed as known nodes.
places where it can be expected to find new C.) Also if the widths of those gaps are larger
information about the environment. The than the size of the agent, they can be travelled
distance data is analysed to classify them as through. Hence, the corresponding nodes are
walls or gaps. At the end of each classification, considered neighbours of each other (where
its properties (width and centroid) are node to node travelling is possible).
calculated. Here in the example, a large gap is After the above analysis, the agent is now
detected from the scan at N1 hence the agent is directed towards N7G1 because that is the
directed towards the centroid of it. Hence the information lacking area in the map. But INS
scanning is done again at the node N2, where 2 has now learned that there is a path available to
gaps have been detected. Here in Fig.5(c), it can N2 from N7, without travelling or exploring
be seen that the agent has been directed any gap between them.This is kept track by
towards the minimum width gap first which is constantly updating the travel cost map from
N2G2. This approach is justified in the latter each and every node to current node. The cost
part of the document. Here it should be noted of travel (distance) is calculated to each node
that the scanning is done forward biased after from the current node, through the
ENGINEER 4
414
neighbouring nodes. The minimum cost, uses the data from the LRF and compute
unexplored node is selected and it is analysed necessary corrections needed for the position of
to find the minimum width unexplored gap in the agent as well as the generated map.
its corresponding gaps. But the gap should be Apart from the above approaches, the INS has
wide enough so that it can be traversed been developed on several logics that improves
through. The minimum cost unexplored node the mapping time. one is the scanning method.
obviously is the node where the agent is Initially at the agents first scan of the
currently positioned. But if it does not contain environment, it is done for a full circle of 360o,
any unexplored gaps, means a dead end. Then so that the agent can localize itself and correctly
the INS intelligently decides to travel to the decide that where should it explore first. But
minimum cost node instead of just the scans followed after the initiation are done
backtracking to find an unexplored gap. (This forward biased around ±150o around the
minimum cost node is the first approach and heading direction. This approach limits the
the narrowest gap first approach is justified in agent of collecting existing information again in
next section) the area it just explored and keep the
4) Step 04: In the next step, Fig. 5(f) after information spectrum forward biased.
agent has explored the upper part of the map, Another approach is the continuously updated
at node N13, it has decided through the above cost of travel (distance) map of the nodes. If
explained process in step 03, to go back to there are no gaps to explore in the current node
unexplored node N2 to the gap N2G1. As of the traditional approach is to backtrack and
now the INS knows the path from N7 to N2, find the unexplored gaps in the past nodes
hence it does not unnecessarily backtrack, but according to the order of traversing. But in the
travels through middle part of the environment proposed INS next node to explore is decided
to node N2 as observable from Fig. 5(g). The by analysing the cost of travel to the nodes and
minimum cost path is found to the selected selecting the unexplored node with minimum
node employing the Dijkstra‟s Shortest path cost of travel. This comes much effective in
algorithm[11]. The reference point to travel is complex environments, when there exist two or
generated for the agent to follow the path more paths for node to node traversing, which
intended. After node N2, it explores another are identified through the process of 'neighbour
unexplored gap at N3 and finishes up mapping identification' explained in the above section.
as in Fig. 5(h) Hence, if the agent has found a lesser cost path
The algorithm has been implemented so that it to a past node, in its path planning, it will
uses minimum amount of external complex effectively utilize the found path instead of
libraries. Also, program loops have been simple backtracking. It can be seen clearly in
avoided as much as possible and functions the example in Fig. 5, where a path has been
have been implemented so that it will be faster found from N2 to N7 without an explicit scan.
in execution. Conditional switches have been Another approach towards improving mapping
used to handle errors caused by data efficacy is selection of the minimum width gap
dependencies. The lesser processing power to explore first. After selecting the respective
needed and the usage of simple functions and node to travel as explained above, a gap need to
libraries have made it much more feasible to be selected in the respective node.
implement the system on a mobile platform
with limited resources.
4. Improvement of Mapping
Efficacy
When the efficacy of mapping is considered,

two key factors are the accuracy of the map and
the time it takes to map the whole area. The
accuracy of the map highly depends on the
accuracy of the LRF measurements and also
correct estimation of current position relative to
the locale. The control strategy deployed in the
agent will determine the accurate positioning of Figure 6 – INS execution time comparison
the platform to the coordinates directed by the
INS. Hence to correctly localize the agent a Here it‟s done under the general assumption
special SLAM algorithm has been employed. It that in the real world a smaller gap is much
5
415 ENGINEER
probable to lead to a convergence and reach a traveling on curved paths. The detected feature
dead-end, unlike a wider gap. Hence, the positions can be used for
minimum width gap is selected to explore correction.Correspondences between erroneous
before heading out in to the wider area where object positions and previously determined
the probability of convergence is much lower. positions of the same object are found through
As a quantitative analysis of the efficacy, figure similarity transform and optimize the values of
6 shows a comparison of the execution times of the transformation from using recursive
the INS in „Min. gap vs. Max. gap first‟ method to minimise the error of the similarity
approaches for five test environments. From the transform.The transform can be found as;
comparative analysis, it can be concluded that
the min. gap first approach is an efficient
… (1)
method of exploration.
5. Localization - SLAM
Equation (1) can be optimized by minimizing
The accurate localization or else knowing the the sum of absolute error. The determined
exact position is critical in the mapping task. transform isthen used remap the objects and
As the existing GPS systems provide accuracy correct the position as shown in the Fig.7. (a)
of few meters and also fails in indoors, SLAM and (b) shows how the objects have been
approach has been utilized as a localization remapped and agents path has been corrected.
method[16]. Although the ussage of extended
Kalman filter for SLAM is popular[13],due to its
high computational complexity, solutions based
on similarity transforms and particle filter
(Monte Carlo Localization)[14][15] based
methods have been used in thiswork.
In both methods, encoder readings are used to
estimate the position, then the LRF data is used
to correct the position estimate. Finding the
relevant features/objects using LRF is done
using differentiation.
(a) Before (b) After

– calculated Path
– Reference path
– observed features
– Reference objects
Figure 8 – Applying Similarity Transform
5.2 Particle Filter (Fast SLAM)
The major drawback of the similarity transform

method is the difficulty to converge on
- LIDAR measurement, index
optimised parameter value set in real-time.
vs. range in mm
Hence, another method based on Bayesian
– Differentiated signal
filtering called Particle filter has been used.
– Identified Objects/Features
Particle filter SLAM [15] is a better solution, as
the convergence rate is very high and with the
probabilistic inferences to the estimation, it will
Figure 7 – Feature Identification
address the uncertainty in the LRF too.Initially,
it is hypothetically assumed that the agent will
5.1 Similarity Transform
be anywhere with any orientation.Then
This method is based on finding an Position estimate is computed using the density
optimized parameter estimation for the of the particles, representing the position with
transformation, to transform the agent back to orientation. This is a two-step process,
its correct position. It has been assumed that the
erroneous localization is occurred when it is
ENGINEER 6
416
(a)Before applying correction (b) correction applied (c) Estimated path by average particle
density
Figure 9 – Applying the Particle Filter
A). Prediction step- Particles are generated to 6. Conclusion
represent the position, through iterative
sampling of a normal distribution This work has been focussed on finding a
simple and practically realizable solution tothe
… (2)
engineering problem of constructing a map of
Then the particles are translated an unknown environment. In the approach
with the aid of the applied control signal and proposed, a low-cost laser range finder (LRF)
then resampledusing resampling wheel has been selected as the sensor due its high
technique. accuracy along with simplicity when compared
It is a method that gives a higher chance to with traditional ultra-sonic sensors and
choose a particle from a positionwhich has a complex vision systems. Methods have been
higher weight (weighting is done in the developed for correcting the distortions and
correction step) rather than others. acquiring the data from the LRF. But any
…(3) convenient sensor capable of producing an
accurate distance data spectrum can be easily
B). Correction Step- This is done by weighting used to replace the LRF.
the particle by finding the match between
obtained measurement and the position of the
particle.This will reduce the outliers with the
time and quickly converge to the actual
path.Drawn samples with probability α
…(4)
= error in the distance

= error in orientation
After the calculations,
…(5)
Figure 10 –The Test Environment and
Generated 2D Map.
As the hardware platform, a 3WD Omni

Through the above process, a particle filter directional mobile platform has been chosen
SLAM can be implemented.A much accurate due to its convenient manoeuvrability into any
position estimation is rendered through that direction intended. To guide it towards
approachas seen from the Fig.8. information gap in the data and to provide
optimum path planning a separate navigation
algorithm has been implemented. Key features
like, „Smallest gap first approach‟ and the
„Intelligent neighbour identification process‟
7
417 ENGINEER
has allowed the mapping efficacy to be at Conference on Control, Automation, Robotics and
optimum. Vision (ICARCV), 2016, pp. 1–6.
Due to the errors in the encoder based 6. Konolige, K., Augenbraun, J., Donaldson, N.,
Fiebig, C., and Shah, P., “A Low-Cost Laser
position measurement of the system, a
Distance Sensor,” in 2008 IEEE International
parallelly running simultaneous localization Conference on Robotics and Automation, 2008, pp.
and mapping (SLAM) algorithm has been 3002–3008.
developed. Similarity transform based 7. Zhou, S., Kang, Y., and Shi, X., “Indoor
approach and Monte-Carlo method (Particle Localization, Navigation and Mapping for
filter) has been experimented and the Particle Quad-Rotor,” in Proceedings of 2014 IEEE
filter based Fast SLAM has been implemented Chinese Guidance, Navigation and Control
on the system. Conference, 2014, pp. 2669–2674.
Altogether taken the combined work on above 8. Abeysekara, A. H. A. D., Liyanage, D. P.,
Welikala, W. R. C. B. S., Godaliyadda, G. M. R.
approaches has enabled to create a simple and
I., Eakanayake, M. P. B., and
rapidly realizable solution to the problem of Wijayakulasooriya, J. V., “Depth Map
environment mapping. Although specific Generation for a Reconnaissance Robot via
hardware set-up is considered in this work, the Sensor Fusion,” in 2015 IEEE 10th International
developed algorithms are quickly adaptable to Conference on Industrial and Information Systems
any other hardware combination. Also, as (ICIIS), 2015, pp. 320–325.
further improvements if a suitable sensor like a 9. Catapang, A. N., and Ramos, M., “Obstacle
vision system is available, it would be Detection using a 2D LIDAR System for An
convenient to extended the 2D map, generated Autonomous Vehicle,” in 2016 6th IEEE
International Conference on Control System,
through the proposed approach to a complete
Computing and Engineering (ICCSCE), 2016, pp.
3D map of the locale. 441–445.
10. Plonski, P. A., Hook, J. V., Peng, C., Noori, N.,
Acknowledgement and Isler, V., “Navigation around an
Unknown Obstacle for Autonomous Surface
Vehicles using a Forward-Facing Sonar,” in
This research work has been hosted and
2015 IEEE International Symposium on Safety,
supported by the Department of Electrical and Security, and Rescue Robotics (SSRR), 2015, pp.
Electronic Engineering, Faculty of Engineering, 1–7.
University of Peradeniya, Sri Lanka. 11. Dijkstra, E. W., “A Note on Two Problems in
Connexion with Graphs.,” Numer. Math., vol.
References 1, pp. 269–271, 1959.
12. M. T. Sqalli et al., “Improvement of a Tele-
1. Hatao, N., Okada, K., and Inaba, M., Presence Robot Autonomous Navigation
“Autonomous Mapping and Navigation Using SLAM Algorithm,” in 2016 International
System in Rooms and Corridors for Indoor Symposium on Micro-NanoMechatronics and
Support Robots,” in 2008 IEEE International Human Science (MHS), 2016, pp. 1–7.
Conference on Mechatronics and Automation, 13. Saman, A. B. S. H. M., and Lotfy, A. H., “An
2008, pp. 840–845. Implementation of SLAM with Extended
2. Ip, Y. L., Rad, A. B., and Wong, Y. K., Kalman Filter,” in 2016 6th International
“Autonomous Exploration and Mapping in An Conference on Intelligent and Advanced Systems
Unknown Environment,” in Proceedings of 2004 (ICIAS), 2016, pp. 1–4.
International Conference on Machine Learning and 14. Aini, F. R. Q., Jati, A. N., and Sunarya, U., “A
Cybernetics (IEEE Cat. No.04EX826), 2004, vol. Study of Monte Carlo localization on Robot
7, pp. 4194–4199 vol.7. Operating System,” in 2016 International
3. Yuan, S., and Schumacher, H., “Vehicle Conference on Information Technology Systems
microwave radar system for 2D and 3D and Innovation (ICITSI), 2016, pp. 1–6.
environment mapping,” in 2015 IEEE MTT-S 15. Zhao, L., Fan, Z., Li, W., Xie, H., and Xiao, Y.,
International Conference on Microwaves for “3D Indoor Map Building with Monte Carlo
Intelligent Mobility (ICMIM), 2015, pp. 1–3. Localization in 2D Map,” in 2016 International
4. Dam, R. R., Biswas, H., Barman, S., and Conference on Industrial Informatics - Computing
Ahmed, A. Q., “Determining 2D Shape of Technology, Intelligent Technology, Industrial
Object using Ultrasonic Sensor,” in 2016 3rd Information Integration (ICIICII), 2016, pp. 236–
International Conference on Electrical Engineering 240.
and Information Communication Technology 16. Galgamuwa, G.I.R.K., “A Novel Scan Matching
(ICEEICT), 2016, pp. 1–5. Algorithm for Slam Problem,” M.Sc. Eng. thesis,
5. Forouher, D., Besselmann, M. G., and Maehle, Department of Electrical and Electronics
E., “Sensor Fusion of Depth Camera and Engineering, Faculty of Engineering, University of
Ultrasound Data for Obstacle Detection and Peradeniya, Peradeniya, Sri Lanka, January 2017
Robot Navigation,” in 2016 14th International 17. Jayasuriya, D.M.N., Liyanage, W.L.Y.S., Herath,
H.M.A.S., Godaliyadda, G.M.R.I., Ekanayake,
ENGINEER 8
418
M.P.B., and Wijayakulasooriya, J.V.,
“Intelligent Navigation System for Mapping
Unknown Environments,” in 2016 IEEE 8 th
International Conference on Information and
Automation for Sustainability (ICIAfS 2016), Galle,
Sri Lanka,December 2016.
9
419 ENGINEER
pp. [1 -- 10],
[421 428],2017
2017
© The
© The Institution
Engineers,SriSri
Lanka
Lanka
A Novel Handover Mechanism for Visible Light

Communication Network
M.A.N. Perera, N.G.K.R. Wijewardhana, A.A.D.T Nissanka, S.A.H.A. Suraweera
and G.M.R.I. Godaliyadda
Abstract: Visible light communication (VLC) is an emerging technology and considered as an

alternative to overcome some of the disadvantages of radio frequency communication technology in
an indoor environment. However, the line of sight nature of the technology limits the user mobility
and create new challenges to provides seamless network coverage under user mobility scenarios. In
VLC multi access points and multi cell based network, co channel interference (CCI) between
neighbor cells limits the overall performance. Therefore, by following the already existing VLC
handover systems, this statistical parameter based novel handover method is designed. This paper
proposes a handover mechanism for indoor VLC systems by introducing cell ID bits and statistical
kurtosis values of the cell ID waveforms as metric for the handover initiation. As advantage of the
method, effects from the CCI can be eliminate when measuring signal strengths in the signal
overlapping cell boundary area. Also the handover criteria are adoptive to the different ambient
lighting conditions compared to existing pre-configured light intensity threshold based handover
systems. With the use of bit error rate (BER), the experiment results showed that Kurtosis value of the
cell ID waveforms can be used as metric to initiate network handover in indoor VLC systems
Keywords: Visible light communication, handover, co channel interference, Indoor VLC
1. Introduction In order to use VLC in mobile network

systems, it is necessary that this system can
Visible light Communication (VLC) has provide uninterrupted high speed connectivity
become a popular research area during recent in presence of user mobility scenarios. To use
years, due to the lack of bandwidth in the VLC systems as an alternative for RF
frequency spectrum for radio frequency (RF) communication systems, it should be able to
communication [1,2]. The bandwidth initiate handovers between access points to
frequency spectrum of visible light is larger provide seamless connectivity for the mobile
than that of RF bandwidth. RF bandwidth users [4]. Usually in RF communication
ranges from 3 kHz to 300 GHz and VLC systems handover is initiated on the basis of
spectrum offers a bandwidth range from 400
Mr. M.A.N Perera, B.Sc.Eng.(undergraduate) Department
THz to 800 THz. There are several other
of Electrical and Electronics Engineering, Faculty of
advantages in VLC compared to RF based Engineering, University of Peradeniya.
mobile communication systems. Unlike RF
mediums, visible light cannot penetrate walls Mr. N.G.K.R. Wijewardhana, B.Sc.Eng.(undergraduate)
Department of Electrical and Electronics Engineering,
in indoor environments. This characteristic
Faculty of Engineering, University of Peradeniya.
provides an inherent network security for
indoor environments. The use of visible light Ms. A.A.D.T Nissanka,
as a carrier of data reduces the VLC systems B.Sc.Eng.(undergraduate)Department of Electrical and
Electronics Engineering, Faculty of Engineering, University
from electromagnetic interference and
of Peradeniya.
associated health concerns. Therefore, VLC
systems can be implemented on places where Dr. S.A.H.A. Suraweera, B.Sc.Eng.(Peradeniya), Ph.D.
RF communication systems have (Monash University), Senior Lecturer, Department of
Electrical and Electronics Engineering, Faculty of
disadvantages. Such as hospitals, factories,
Engineering, University of Peradeniya.
under water, inside airplanes etc. Also VLC
facilitates the reuse of existing light Eng. (Dr.) G. M. R. I. Godaliyadda,
configuration for communication. [3] Since the B.Sc.Eng.(Peradeniya), Ph.D. (National university of
Singapore), Senior Lecturer, Department of Electrical and
characteristics of reusability and abundant
Electronics Engineering, Faculty of Engineering, University
spectrum compared to RF, VLC systems can be of Peradeniya.
built with lesser efforts and at a lower cost.
421
physical parameter measurements, such as systems. In [10] authors propose two
received signal strength. To define optimal handover mechanisms for overlapped uniform
criteria for handover, there are several factors VLC cells and non – overlapped spotlight VLC
taken into consideration. such as energy cells. In overlapped uniform VLC cell, the user
consumption, positioning, end user quality of performs intensity test and detects the
service etc. presence of overlapped area [11]. It initiates
the handover when intensity reach the
This paper proposes a handover method for threshold level. In spotlight VLC cell method,
indoor multi cell based VLC networks. Cell ID there is buffer memory allocated to save data
bits and statistical parameters are introduced on mobile user to avoid connectivity
to eliminate the CCI. In indoor VLC systems, interruptions in dead zones. In [12], there is an
Intensity level can vary quickly due to reasons optical at to cell network proposed which can
such as receiving height, receiving orientation, reduce the co channel interference by using
background light intensity etc. As an joint transmission method from multiple
advantage of using statistical parameters, access points. However, as the authors
impact from above mentioned incidents can be mentioned the downside of joint transmission
minimized compared to other intensity systems is that they need extra signaling
threshold based handover methods. According overhead. In a cell based network, using same
to the purposed methodology, received visible frequency in adjacent cells causes co-channel
light cell ID wave form is analyzed interference [13]. This significantly reduce the
statistically. Handover initiation is decided network performance to a mobile user in
when statistical parameters reached to their overlap region. Therefore, interference
relative maximum value. Since there isn’t any coordination techniques are used to eliminate
comparison between pre-configured constant CCI. In [14] CCI is eliminated by splitting cells
threshold values, the system can be designed into clusters. Inside of a cluster, frequency
to adopt in different uniform lighting resources are used in orthogonal manner.
conditions.
The rest of the paper is organized as follows.

Section 2 describes the existing methodologies 3. System Model
for VLC networks and their drawbacks. The
system model is explained in section 3. Section The system model consists of multiple
4 gives the experiment details and observed transmitters and multiple receivers. The block
results. Finally, conclusions are given in diagram of a single data link is shown in Fig. 1.
section 5.
Presently there are several technologies
designed for achieving seamless connectivity
for mobile users [5,6,7]. One of the most
widely recognized standardized approaches to
VLC is the exploitation of IEEE 802.15.7
standard for visible light data communication.
IEEE 802.15.7 [8] supports high-data-rate
visible light communication up to 96 Mb/s by
fast modulation of optical light sources which
may be dimmed during their operation. IEEE
802.15.7 provides dimming adaptable
mechanisms for flicker-free high-data-rate
visible light communication.
In [9] vertical handover initiating algorithm is Figure 1 – Block diagram of system model.
proposed for VLC – LTE Hybrid
communication system. since this method 3.1 Transmitter
designed for hybrid communication system, it
The transmitters were designed to provide
cannot apply on handovers in VLC only
good indoor illumination in addition to data
422
transmission. Data communication by using
higher illumination intensities can minimize
the distortions to the transmitted data in the
transmitting medium.
The Prototype transmitter unit consists of LED

driver and power LED panel. Since power
LEDs have the high speed switching capability
and higher illumination intensity compared to
the traditional LEDs, 5x 7w Chips on Board
type (COB) cool white power LEDs were used
in order to achieve large distance data
transmission. COB power LED contain more
light sources in the same area than standard
LEDs could occupy. This will result a greatly Figure 3 – Prototype transmitter
increased illumination output per square with multiple power LEDs.
meter.
3.2 Receiver
In order to handle required current for the
LED panel, TTL compatible metal–oxide– Typically, in indoor environments, there is a
semiconductor field-effect transistors highly possibility of distorting and interfering
(MOSFET) were used. Transistor switching the incoming signal wave due to the DC
circuit was used for the LED driver. The components from ambient light and rectified
MOSFETs were operated by switching signal sine wave emitted from filament, fluorescent
from the Arduino output. The switching signal lamps. [16] In order to minimize the impact
carries the data which was needed to be from background disturbances, two
transmitted. The block diagram and the photodiodes (PD) with noise cancellation
prototype transmitter are shown in Figure 2 & method was used in the receiver unit. Two
Figure 3. transimpedance amplifiers were used to
convert PD’s output current in to voltage
signal. DC shift corresponding to ambient
noise and light sources were reduced by using
differential amplifier in the receiver. Also a
reflector fixed on signal receiving PD to
improve the signal receiving vertical distance.
Switching Signal The reflector directs the signal receiving PD to
Arduino transmitter direction. Therefore, it limits the
signal receiving PD scope and minimize the
LED Driver exposure of the ambient noise. To further
DC POWER LED PANEL
MOSFET minimize the DC shifts of the incoming
waveforms, a high pass filter was used in the
receiver. The prototype receiver is shown in
Figure 4.
Figure 2 – Block diagram of prototype

receiver.
Figure 4 – Prototype receiver.
423
Then, received data is transmitted to receiver’s
ADC unit to sample the waveforms at higher
sample rate. Block diagram of the prototype
receiver is shown in Figure 5.
Signal + Noise
Photo diode 1
Noise reduction ADC

Amplifier
Differential amplifier sampling
Noise
Photo diode 2
Figure 5 – Block diagram of prototype

receiver.
Figure 6 – VLC system with cell ID waveforms.
 Handover initiation decision is based
Two ARDUINO UNO boards were used to the on statistical characteristics of the ID
communicate with the prototype VLC system. wave form. According to the Fig. 6,
Rx3 receives superimposed signals
3.3 Handover Initiation
which have equal signal strengths on
In VLC system model can be described as two cell bits while, Rx5 and Rx1
illustrated in Fig. 6. To initiate handover, receive only one strong signal on cell
system follow few steps. bits. The waveform shape depends on
the received signal strength of the
 A data structure is defined to transmit transmitted cell ID. Received signal is
data from transmitter to receiver. First shown in Figure 7.
two bit of the data structure is used for
cell ID bit. each transmitter adds its
cell id bits before data transmission
[17]. The cell ID bits are transmitted by
OOK (On Off Keying) modulation.
Fig. 7 illustrates the received data
frame structure.
 Then from the receiver, received wave

form shape is analyzed. Mobile user
separates the cell id segment and data
segment of the data structure. Then it
calculates the kurtosis value of the cell Figure 7 – Received data structure.
ID waveform to estimate position. As
shown in figure, different shapes of 3.4 Mathematical Background
cell id waveforms received for
The shape of cell ID waveform varies with the
different user positions. Fig. 7 received cell ID signal level strengths of
illustrates the received data frame adjacent transmitters. The variations of this
structure. type waveforms can be analyzed by
calculating the kurtosis value of the waveform.
User position for the handover initiation can
be estimated based on the kurtosis values. For
the calculation of the kurtosis, conditional
probability density function (PDF) of the
waveform was constructed using sampled
424
data of the received waveform. In this handover method, combination of
Mathematically, the sampled values kurtosis and bit error rate is used to initiate
constructed from the received waveform from handover, hence a more reliable outcome is
the PD, can be written as achieved. Bit error rate is calculated from the
equation,
….(1)
where is the sampling time and is the

sampled value.
The sample values are mapped to a histogram 4. Experiment & Results

to construct the PDF of distribution.
4.1 System Setup
Kurtosis is a measure of whether the data are
heavy-tailed or light-tailed with respect to a The experiment was carried out in an area of
normal distribution. The kurtosis decreases as dimensions 1.6 m x1.6 m. The experiment was
the tails become lighter and it increases as the carried for two transmitters and one mobile
tails become heavier. Kurtosis for a discrete set observer. Observed user moving path across
of data is boundary of the neighboring cells is shown in
Fig. 8. Two transmitters are named as T1 and
T2 in the figure. BER was calculated by using
 Y 
N
4
Y /N 8 x 8 grid with resolution of 0.2m x 0.2m
i
Kurtosis = i 1
…(2) placed over the area. Transmitters were fixed
s 4 2m above from receiver plane. Transmitters
were the only light sources at the experiment
Where Y is sampled data point, ̅ is mean of premises.
sampled data, S is standard deviation and N is
the sample size.
If the kurtosis is close to 0, then a normal
distribution is often assumed. These are called
mesocratic distributions. If the kurtosis is less
than zero, then the distribution is light tails
and is called a platykurtic distribution. If the
kurtosis is greater than zero, then the
distribution has heavier tails and is called a
leptokurtic distribution [13].
Kurtosis is a component of both tailedness and Figure 8 – Path of the mobile user
peakedness of a distribution. It is basically
because kurtosis represents a movement of 4.2 Results
mass that does not affect the variance. In the
case of positive kurtosis, where heavier tails To observe the receiving cell ID waveform
are accompanied, can be modeled to the patterns, 0 1 and 1 0 bit sequences transmitted
waveform received for the cell identification as cell ID from T1 and T2 respectively. Fig. 9
on handover region. If the distribution is more shows the received Cell id waveforms when
concentrated towards the tails, then the moving along A to M. Random walking
variance will also be larger. To leave the effect pattern across the two transmitters was taken
of variance unchanged, standardized kurtosis to consideration when creating the observing
is taken with respect to variance. Hence a high path. From position E to H user change the
value for kurtosis is obtained at the handover moving direction and from H onwards it
position. It is also recognized that although continues to go across the two transmitters.
tailedness and peakedness are often both Calculated kurtosis values variation for the received
components of kurtosis, kurtosis can also cell ID waveforms is given in Fig. 10. From the
reflect the effect of primarily one of these received waveforms shown in Fig. 9, it can be
components, such as light tails. Thus, for observed that waveforms corresponding to position
symmetric distributions, positive kurtosis (E) and (I) have the highest kurtosis values
indicates an excess in either the tails, or compared to other positions.
peakness which is used to identify the receiver
location.
425
Figure 9 – Received cell ID wave form for different positions.
Figure 10 – Kurtosis values for

different positions.
Figure 11 – BER for TX1 & TX2 in different positions of the grid.
426
Measured BER on different positions of the 8x8 introduced and received waveforms were
grid is shown in Fig. 11. Each coloured contour analyzed statistically. The kurtosis values and
lines represents a different BER value observed bit error rates from two transmitters were
in the area. considered in the experiment and the results
showed that the higher kurtosis values can be
4.3 Handover Based on Results obtained when multiple transmitters have
equal visible light signal intensity. Also low
Fig.12 shows the BER variation with kurtosis BERs were observed at the high kurtosis
along the A-M path. Position (E) and (I) have positions. Therefore, it can be showed that
the minimum BER in both TR1 and TR2. kurtosis values of the cell ID can be used as
Positions directly under the transmitters have handover decision making metric for the
low kurtosis values compared to the other indoor VLC networks. The system can be
positions. Such as position A, B, C, G, J, K, L, further improved in combination with other
M. Therefore, it can be clearly observed that devices such as CCTV video cameras in order
positions which have higher CCI results high to track the walking patterns of the users and
Kurtosis values. According to the results optimize the handover decision criteria.
position (I) and (E) which have equal signal
strengths from the TR1 & TR2, can be used as a
handover initiation positions. However, since References
user change its moving direction, handover 1. H. Elgala, R. Mesleh, and H. Haas, “Indoor
initiation is not required in position E. optical wireless communication: potential and
state-of-the-art,” IEEE Communications
Magazine, vol. 49, Sept. 2011.
2. P. H. Pathak, X. Feng, P. Hu and P. Mohapatra,

“Visible light communication, networking, and
sensing: A survey, potential and challenges,”
IEEE communications surveys & tutorials, 17(4),
pp. 2047-2077, 2015.
3. A. Jovicic, J. Li, and T. Richardson. "Visible

light communication: opportunities, challenges
and the path to market." IEEE Communications
Magazine vol. 51, pp. 26-32. Dec. 2013.
4. Pollini, Gregory P. "Trends in handover

Figure 12 – Kurtosis values and
design." IEEE Communications magazine 34.3
BER variations for different
(1996): 82-90.
positions
5. Dinc, E., Ergul, O., & Akan, O. B. (2015,
Therefore, experiment results shown in Fig. 12 September). Soft handover in OFDMA based
concludes that higher kurtosis values of the visible light communication networks. In
cell ID bits directly related to the minimum Vehicular Technology Conference (VTC Fall), 2015
BER positions of the multiple access points. IEEE 82nd (pp. 1-5). IEEE.
Results of the prototype experiment
6. Xiong, J., Huang, Z., Zhuang, K., & Ji, Y. (2016).
demonstrate that kurtosis parameter can be
A cooperative positioning with Kalman filters
used as metric for handover position together
and handover mechanism for indoor
with minimum bit error rate.
microcellular visible light communication
network. Optical Review, 23(4), 683-688.
5. Conclusions
7. Hien, Dang Quang, and Myungsik Yoo.
This paper presents a novel handover "Handover in outdoor Visible Light
mechanism for VLC network in order to Communication system." Information
mitigate CCI in neighbor cells and detect the Networking (ICOIN), 2017 International
accurate signal strengths of the neighbor Conference on. IEEE, 2017.
transmitters in order to initiate handovers and
achieve seamless coverage for the mobile 8. Rajagopal, Sridhar, Richard D. Roberts, and
users. To perform that, cell ID bits were Sang-Kyu Lim. "IEEE 802.15. 7 visible light
427
communication: modulation schemes and
dimming support." IEEE Communications
Magazine 50.3 (2012).
9. Liang, Shufei, et al. "A novel vertical handover

algorithm in a hybrid visible light
communication and LTE system." Vehicular
Technology Conference (VTC Fall), 2015 IEEE
82nd. IEEE, 2015.
10. Vegni, Anna Maria, and Thomas DC Little.

"Handover in VLC systems with cooperating
mobile devices." Computing, Networking and
Communications (ICNC), 2012 International
Conference on. IEEE, 2012.
11. *Wu, Y., Yang, A., Feng, L., Zuo, L., & Sun, Y.
N. (2012). Modulation based cells distribution
for visible light communication. Optics express,
20(22), 24196-24208.
12. Chen, Cheng, Dobroslav Tsonev, and Harald

Haas. "Joint transmission in indoor visible light
communication downlink cellular networks."
Globecom Workshops (GC Wkshps), 2013 IEEE.
IEEE, 2013.
13. Cui, Kaiyun, Jinguo Quan, and Zhengyuan Xu.

"Performance of indoor optical femtocell by
visible light communication." Optics
Communications 298 (2013): 59-66.
14. Marsh, Gene W., and Joseph M. Kahn.

"Channel reuse strategies for indoor infrared
wireless communications." IEEE Transactions on
Communications 45.10 (1997): 1280-1290.
15. Bui, T. C., Kiravittaya, S., Nguyen, N. H.,

Nguyen, N. T., & Spirinmanwat, K. (2014,
October). LEDs configuration method for
supporting handover in visible light
communication. In TENCON 2014-2014 IEEE
Region 10 Conference (pp. 1-6). IEEE
16. Zhao, Yan, and Jayakorn Vongkulbhisal.

"Design of visible light communication receiver
for on-off keying modulation by adaptive
minimum-voltage cancelation." Engineering
Journal 17.4 (2013).
17. Bao, X., Zhu, X., Song, T., & Ou, Y. (2014).
Protocol design and capacity analysis in hybrid
network of visible light communication and
OFDMA systems. IEEE Transactions on
Vehicular Technology, 63(4), 1770-1778.
18. Mardia, Kanti V. "Measures of multivariate

skewness and kurtosis with applications."
Biometrika (1970): 519-530.
428
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[429 438],2017
2017
The Institution
Lanka
Technical, Economic, Operational and Dispatch

Analysis of Optimum Integration of
Renewable Energy in to the Sri Lankan Power system
Buddhika Samarasekara, Randika Wijekoon and Lathika Attanayake
Abstract: The application of more renewable energy based technologies is intensifying with the
technological advances and global climatic concerns. Renewable energy sources have become
increasingly important in the electricity sector and will continue to grow as a significant source of
electricity generation in future. The Integration of renewable energy sources poses various technical
and economic challenges depending on the characteristic of the resources and the system.
Sri Lanka, being a country with rich potentials of renewable energy sources, it is important to
integrate optimum share of the indigenous renewable energy sources. At present nearly 45% of the
country‟s electricity demand is met by renewable sources. This consists of 34% of major hydro and
11% of wind, solar, small hydro and biomass. In the future it is expected to increase the contribution
from wind and solar. Increased penetration levels of variable renewable energy based generation
sources pose various technical challenges for system planning and operation. Determining the
optimum level of Renewable energy penetration requires the consideration of seasonality and
intermittency of sources, daily electricity demand pattern, power system stability and system
operational flexibility concerns are among the main system specific constraints for the country.
A detailed study was carried out to increase the renewable energy share from 11% in 2015 to an
optimum share of more than 20% beyond 2020. The paper focuses on the analysis on the economic,
operational and dispatch restriction aspects of the above renewable energy integration study. Detailed
system studies that were carried out for the above work is not included in scope of this paper.
Ultimately, it quantifies the results on generation of renewable energy sources and curtailments
required under the given system‟s constraints. Further it highlights the key observation of the results
and prospective mitigation measures to facilitate the integration of more renewable energy sources to
the network.
The paper outlines the renewable energy resource estimation, optimized development plan,
operational analysis and cost estimations to open a pathway to a paradigm shift in Sri Lankan
electricity generation system in near future.
Keywords: Renewable Energy Integration, Intermittency, Operational Constraints, Curtailment
1. Introduction rising economic activities. An optimum strategy

to meet the growing electricity demand of a
1.1 Background developing island nation should mainly cover
Sri Lanka being an island has an isolated power the aspects of supply reliability and security,
system and it was 100% renewable system until economics of supply as well as environmental
mid-nineties. With the growing demand, it has and social considerations. Total renewable and
now evolved to a system with a mix of both thermal installed capacity and the peak
renewable and thermal generation sources and
Eng. Buddhika Samarasekara, B.Sc. Eng. (Hons), C.Eng.
the size of the existing generating system is MIE(SL),MIET, PGDipBM (Col),Chief Engineer (Generation
nearly 4018 MW in 2016. This includes 1903MW Planning)
(47%) of renewable energy sources and
2115MW (53%) of thermal energy sources. Eng. Randika Wijekoon, B.Sc. Eng. (Hons), AMIE(SL),
Electrical Engineer
Sri Lanka electricity requirement was growing
at an average annual rate of around 6% during Eng. Lathika Attanayake, B.Sc. Eng. (Hons), AMIE
(SL),Electrical Engineer, Generation and Transmission
the past 20 years [1], and it is expected to
Planning Branch, Ceylon Electricity Board
continue in the foreseeable future with the
429
demand of the country over the last 30 years renewable energy integration study describing
are shown in Figure 1 and the energy its methodology and results covering the
contribution of the renewable energy sources renewable resource assessment; long term
has been significant in past years. optimized planning, operational analysis and
economic analysis. Detailed analysis conducted
on the system stability as an integral part of the
study carried out by transmission planning unit
is not described in this paper. Renewable
energy (RE) forms except major hydro are
referred to as “Other renewable energy” (ORE)
and intermittent renewable energy sources such
as wind and solar are referred to as variable
renewable energy (VRE) in this work.
2. Study Methodology
Figure 1 – Peak Demand and Installed The main objective of the study is to determine
Capacity of past 30 years the optimum extent of renewable energy
integration subjected to the technical,
Sri Lanka, together with other countries, has operational and economic constraints. A new
expressed its national obligation towards methodology was adopted considering the
mitigating the climate change impacts to United limitations in Sri Lankan Power System
Nations Framework Convention on Climate specifically to investigate the restrictions of
Change (UNFCC). Even though the greenhouse integrating more ORE‟s and that has an
gas (GHG) emission of the country is as small integrated approach of major steps of the
as 0.05% of the global total[2], Sri Lanka has an system planning functions. A simplified
ambitious program towards a reducing the approach of the overall study methodology is
overall GHG emissions. When considering the shown in the figure 2. The study process
greenhouse gases, CO2 is one of the primary started with the assessment of the renewable
gases which contribute towards warming of energy potentials in the country followed by
earths‟ atmosphere. According to International the preparation of the future RE deployment
Energy Agency [2], approximately 41% of CO2 scenarios as necessary. Optimized plant
emission from the electricity sector while major sequences were then determined for each
contributor for CO2 emission is the transport scenario which were then subjected to the
sector which account for 48% approximately. transmission network study and the
Indigenous renewable energy development is a operational study. Additional scenarios were
key area for decarbonizing the electricity sector defined whenever necessary for the network
and that requires special analysis to explore the and operational study to obtain practical
possibilities of renewable energy integration. results.
Renewable Energy Resource Assessment
Technical and economic implications of
renewable energy integration possibilities are Projection of RE Development
different from system to system due to number
of factors. System attributes such as size, plant Optimization of Long term Generation
composition, network features, stability levels, Expansion
operational constraints, interconnectivity and
Transmission Medium to Short term
demand characteristics as well as renewable Network study Operational study
energy resource characteristics such as
seasonality, intermittency and geographical Technical and Economic analysis of
locations are important factors. Therefore, being RE integration
the transmission system operator, Ceylon
Figure 2 – Main Outline of the Methodology
Electricity Board (CEB) conducts a
comprehensive study, to determine the
Similar approaches have been adopted in other
optimum extent of renewable energy
international studies such as Western wind
integration covering its technical, economic,
solar integration [6], Hawaii solar integration
operational and environmental aspects prior to
[7] and have slight differences to suit the study
formulating the long term generation
objective and the system characteristics.
expansion plan. This paper presents the above
430
Figure 3 below illustrates the generation and multiple scenarios. Following Figure 4
transmission planning operation functions by illustrates the probabilistic variation of annual
timescale. The scope of this study covers most energy production of hydro system. It is
of the system constraints that are important important to note that the hydro energy can
from years to sub hourly time steps. vary as much as 2250GWh depending on the
hydrological condition which is a significant
Years Long-Term Planning uncertainty. The average hydro condition
yields 4155GWh of energy for power
Weeks Mid-Term Planning generation.
Days Unit Commitment
As illustrated in figure 5 the Even though the
Hours Economic total capacity of the Major hydro is closely 1384
Dispatch MW, the actual available capacity can vary
Minutes from 85% to 45% seasonally depending on the
Generator Primary month of the year and the hydrological
Seconds Control condition.
Milliseconds Protection
Figure 3 – System Planning functions by

Millisecond
s timescale
The methodology requires use of extensive
amount of data and the use of number of
modelling and simulation tools. Such details
are presented in the relevant sections of this
paper.
3. Renewable Energy Resource

Assessment
Assessment of the performance of the each Figure 4 – Variation of Annual Energy

renewable energy sources is required for the
detail modelling simulation and optimization
work. Major Hydro, Wind, Solar, Mini Hydro
and Biomass are separately modelled
appropriately to assess capacity and energy
contribution of while capturing all the
necessary characteristics such as seasonality,
intermittency and features of geographical
dispersion. Effort is placed to best represent
each resource while harnessing the maximum
potential.
3.1 Assessment of Major Hydro Resources

Major hydro being the largest renewable
energy contributor, accurate determination of Figure 5 – Available monthly average capacity
its hydrological probabilities and total resource of the hydro for hydrological conditions
capability is important. However, this
assessment is complex due to the cascade and 3.1 Assessment of Wind Resources
multipurpose nature of reservoirs with Assessing the performance of wind resources in
numerous operational constraints. The software the country is crucial as it is one of the best
tool Stochastic Dual Dynamic Program (SDDP) wind resource potential in the world. Therefore
which is a hydrothermal dispatch model that capacity and energy contribution, seasonality
determines least-cost stochastic operating intermittency and geographical dispersion have
policy of a hydrothermal system for short, been given importance in the modelling
medium and long term operation studies was activity.
used for the purpose. The simulation tool uses As the wind resources of different quality are
the hydro inflow data on past 35 years to spread across the country, this study
stochastically estimate the hydro potential for considered five main wind regimes for the
431
modelling. 10 minute wind measurement data (GHI) and Diffuse Horizontal Irradiance (DHI)
recorded by Sustainable Energy Authority measurements were available in ten minute
(SEA) and the Ministry of Power and time steps and the Direct Normal Irradiance
Renewable Energy (MOPRE) were used for the (DNI) was estimated with the available GHI
modelling purpose. Using the software tool and DHI using solar zenith angle. Typical
System Advisor Model, wind plants were commercial PV modules and parameters were
modelled for each wind regime and parameters used modelling the solar PV plant. Figure
of each wind plant are set considering multiple below illustrates the estimated output of a
factors while ensuring maximum energy yield 10MW Solar PV plant at Kilinochchi for
[3]. Table 1 provides the main wind plant consecutive five days.
parameters and results obtained for energy for
each wind regime. The figure 6 shows the
power production profiles obtained from
simulation for the Mannar wind resource.
Table 1 – Characteristics of Wind plants

modelled for each wind regime
Capacity (MW)
Annual Energy
capacity (MW)
Annual Plant
Data location
Wind Speed
Parameter
Factor (%)
Turbine
(GWh)
Block
Figure 7 – Output of a simulated 10 MW solar

Nadukuda
Mannar 2.5 25 80 36.7
2015
PV plant at Kilinochchi area
Puttalam 2 20 55 31.4
Udappuwa 3.3 Assessment of Mini Hydro Resources
2009-2010 The Model for Mini-hydro power plant was
Hill Seethaeliya
appropriately prepared based on 10 year
0.5 10.5 17 19.1 historical data to reflect the seasonal energy
Country 2012-2014
and monthly average capacity variation and to
Pooneryn obtain accurate results. The annual plant factor
Northern 2 20 59.7 34.1
2015
is estimated to be 37.4 % in the average hydro
Kokkilai condition. Figure 8 below illustrate the average
Eastern 2 20 47.9 37.3
2015 model used for the mini-hydro resources.
Figure 6- Power output variation of 25MW

wind plant at Mannar
Figure 8 – Monthly average capacity
3.2 Assessment of Solar PV Resources Variation of the mini-hydro Plant in p.u
Performance of solar PV resources was assessed
using the Sam Advisor Model through 4. Optimized long term planning
modelling and simulation. Solar irradiance
measurements were obtained from the Once the renewable energy resources are
Sustainable Energy Authority (SEA) where the assessed, long term planning of the total system
data were available for two locations in is then carried out. Development of renewable
northern and southern areas of the country. energy resources and other thermal generation
Therefore two solar regimes were considered sources is determined through the optimization
for this study. Global Horizontal Irradiance process.
432
4.1 Renewable Energy Development Table 2 – ORE development case -1
Projection
Capacity
Biomass
Hydro
Projection of the development of future
(MW)
(MW)
(MW)
(MW)
Wind
Solar
Total
Mini
Year
renewable energy potentials is based on
number of technical and other driving factors.
Fundamentally, most attractive resources were 2018 344 144 39 210 737
given priority for the development. 2020 374 414 49 410 1246
2025 424 729 74 685 1912
Only few alternatives remain for the major
2030 474 894 99 1009 2476
hydro development as the county‟s potential is
already harnessed to a greater extent. 241MW 2035 524 1184 114 1283 3105
of Major hydro plants already planned [1] will
be added during the period of this study. Table 3 – ORE development Case -2
Capacity additions of other renewable energy
Capacity
Biomass
Hydro
sources have been projected for all five wind
(MW)
(MW)
(MW)
(MW)
Wind
Solar
Total
Mini
Year
regimes, two solar regimes and for Mini-hydro
and biomass resources. Quality of resource,
availability and the development transmission 2018 344 144 39 210 737
infrastructure, maintaining optimum share of 2020 374 414 49 410 1246
other renewable energy have been considered
2025 424 649 74 635 1782
for future capacity additions. Factors such as
past experience in renewable energy project 2030 474 814 99 959 2346
development, reduction of global technology 2035 524 1104 114 1233 2975
costs especially solar PV [4], availability of land
and other infrastructure were given 4.2 Optimized development scenarios
consideration when making the projection. It is Long term expansion planning was carried out
was identified that increasing the total ORE using the WASP IV software tool to identify
share above 20-22% causes system to optimum plant development program. Several
experience operational limitations where the key long term development scenarios with
VRE output have to be curtailed. Such various system flexibilities were analysed to
knowledge has also been considered in making evaluate impact of RE integration.
initial projections.
At this stage focusing on different RE
In this study considered two main ORE integration method is important. Flexibility of
development cases that are defined optimum power plants, storage technologies, demand
ORE share more than 20% by 2020. Basis for side measures and grid support from
defining two cases considered mainly the renewables are among them [5], [9]. In this
curtailment levels due to system operational study, considering the Sri Lankan context, the
flexibilities with different future expansion grid support of pumped storage hydro power
scenarios of major power plants. Many other plant that has been planned mainly for the peak
cases were defined on the basis of regulatory power requirement, was subjected to analysis.
reserve availability under the transmission
network study purpose and they were not Four main scenarios were defined and the
covered under the scope of this paper. But the technical, operational and economic impact of
results of the system study have also been the introducing combined cycle units, coal fired
incorporated in preparing the projections. units and the grid scale storage system such as
Projected ORE capacities for selected milestone pumped storage hydro power plant are key
years are given in the table 2 and the table 3 differences of these scenarios. Each scenario
respectively. represents the least cost plant sequence for the
corresponding renewable energy development
Case (case1 or case 2). Additionally, a Reference
scenario was created with no future ORE
development for the comparison purpose. A
summary and the capacity additions of the
scenarios are given in the table 4 and table 5
below.
433
Table 4 –Optimized Long term planning Main operational constraints in VRE integration
scenarios are limitation of operational flexibilities of
Scenario Description different power plants [11]. Ramping rate of
Only the existing “other renewable different plants that determines how quick the
Reference
energy” by 2016 is included and No future plant can respond to counter the changes in
scenario
ORE development is assumed VRE output, Minimum stable loading level of
New Combined cycle power plants are the power plant which determines how much
Scenario 1 restricted in western region. Coal and room and reserve can be allocated for the
Pumped hydro plants included operation of VRE throughout the day.
Development Coal units are limited with Additionally, plant start-up and shutdown
Scenario 2 pumped hydro and more Combined cycle constraints are also important to a certain
units are permitted in other locations extent. Observing these characteristics requires
Development coal units and pumped the operation to be simulation at least sub-
Scenario 3 hydro units are restricted. All new hourly time step. In this study simulation were
additions are combined cycle units.
conducted in 30 minute time steps.
All new additions are combined cycle
Scenario 4
power plants with pumped hydro The software tool Stochastic Dual Dynamic
development
Program (SDDP) was used for determining the
Table 5 - New Major Capacity Additions in medium term hydro-thermal operation policy
each scenario until 2036 and subsequently the software tool (NCP) was
used for optimized short term dispatch
Reference
Scenario 1
Scenario 2
Scenario 3
Scenario 4
scenario
Type of simulation in 30 minute time steps. Both

plant models requires extensive amount of input data
and computation resources. The input data
requirement ranges from 30 minute VRE output
Major hydro 241 241 241 241 241 data, demand data, plant technical, operational
parameters to system wide operational
Gas
Turbines
105 105 105 105 105 requirements such as reserves and largest
Combined single unit restrictions.
1500 1500 2700 4800 4200
Cycle
The key observation of the dispatch simulation
Coal 3300 2400 1200 0 0
results was the scale of over generation of VRE
Pumped
600 600 600 0 600 resource. In other words some parts of the VRE
Hydro
generation are seen as excessive due to the
ODE Dev. Case Case Case Case Case
Case
operational limitations of the system. In such
1 1 1 2 1
instances VRE generation had to be curtailed.
The requirement of such curtailment depends
5. Operational analysis on many factors such as the VRE production
level, intermittency, demand level and the type
In the power system operation process, of power plants online at the time.
maintaining supply and demand balance at all
times is utmost important to ensure continuous The study specially requires the dispatch
and reliable supply of electricity. Intermittency information for entire year for multiple years
of variable renewable energy sources such as and the short term model is capable of
wind and solar, specially introduce additional simulating limited number of days at a time.
supply side fluctuation to the system. These Therefore considering the scale of the study, a
fluctuations are compensated by the rest of the given year was divided into three periods as
power plants in the system and their flexibility illustrated in Figure 9 and a weekday and a
to that is often finite [8]. Therefore Sri Lankan weekend day was simulated for each period
power system, being a small isolated power that provide simulation of 6 days per each year
system, its ability to integrate more variable to make a reasonable approximation,. This
renewable energy significantly depends on the approximation is mainly to reasonably capture
operational flexibility of the system. The the seasonal variation of renewable energy
objective of the operational study is to examine production and the demand variations.
in detail the operational flexibility and
constraints of the system in integrating high Dry Period High Wind Period Wet Period
levels of variable renewable energy sources. Jan- April May- Aug Sep- Dec
Figure 9 - Approximation of a year
434
Considered plant operational constrains are
crucial in determining the system‟s VRE
absorption capability. Following key
assumptions were used for this study.
•Future VRE Generation profiles are estimated

based on measured data and past performance.
No significant variation was assumed for future
years.
•Maintenance plans of major power plants
(Coal fired and Combined cycle Plants) were
prepared as per the maintenance requirement
and present practice.
•According to the past demand data, relatively Figure 11 –curtailment Requirement of VRE
low off-peak usually occur on Monday early 2025 weekend day in high wind season
mornings and relatively low Day time load is
present in Sundays. To capture such 6. Variable Renewable Energy
bottlenecks a Sunday and a Monday is
Curtailment
consecutively simulated in short term model.
• Wind, biomass, solar and mini hydro were Summary of the scale of curtailment observed
considered as “Must run” power plants and through the operational study results are given
curtailment of VRE is allowed subjected to in the table 6 below. No curtailment was
operational constraints. observed in the year 2018 and Key observations
of the operational analysis are as follows.
In terms of operation reserve, higher level of
reserve is necessary with the increased Table 6 – Observed Curtailment levels
penetration of VRE sources. For this study
purpose 5% of reserve for load variability and Yr High Wind Season
Wet Season (MW)
10% reserve from VRE share are included as (MW)
regulation reserve requirements. Weekday Weekend Weekday Weekend
Simulations were conducted for four key years OP DP OP DP OP DP OP DP

(2020, 2022, 2025 and 2028) covering 10 years
period from 2018 for several scenarios. Sample Scenario 1 (New Combined cycle power plants are
restricted in western region. Coal and Pumped hydro
of the results obtained for a day in high wind plants included )
period weekend of 2025 is given in the figure 10
below. 2020 150 - 80 50 170 - 140 -
2022 220 - 140 100 - - - -
2025 380 330 280 70 - 20 -
2028 70 - 60 185 - - - -
Scenario 2 (Development Coal units are limited with

pumped hydro and more Combined cycle units are
permitted in other locations )
2028 70 - 30 111 - - - -
Scenario 3 (Development coal and pumped hydro units are

restricted. All new additions are combined cycle units)
Figure 10 – 30 minute dispatch simulation of a
2025 weekend day in high wind season 2025 445 - 380 430 - - - -
The figure 11 illustrates the ORE production 2028 80 - 200 276 - - - -

relative to the demand and the red coloured
portion is the VRE over generation that need to *OP- Off peak, DP- Day peak
be curtailed due to operational constraints.
435
 No renewable energy curtailments were Table 7 - Levelized Cost of Other ORE
observed in dry periods, mainly due to the technologies
very low level of wind and mini-hydro LCOE
Renewable Capital Plant
production. (UScts/k
Technology Cost Factor
Wh)
 Curtailments were observed on High wind
Solar PV 1412
period from 2020 onwards in all scenarios (2017-2020) USD/kW
17 % 11.80
and gradually rising until 2015.
Solar PV 1100
 Reduced curtailments can be observed in wet (2021-2025) USD/kW
17 % 9.34
periods compared to high wind period.
Solar PV 900
 Weekend days required higher curtailment 17 % 7.76
(2026 onwards) USD/kW
levels at day time than week days due to the
relatively low demand levels. Wind 7.26 35 % 8.19
 Off peak curtailments were significant in the
lowest off-peak of the week. Mini Hydro 6.22 37 % 7.80
 Both off-peak and daytime curtailments were
observed in high wind periods. Biomass 3.44 80 % 18.19
 Introduction of pump storage units from 2025
to 2028 gradually decreases the curtailment
requirement both in day time and off-peak Table 8 - Levelized Cost of ORE technologies
Difference with
time. Curtailment requirement increases Total PV Cost
Reference
when the pumped hydro development Scenario up to 2036
Scenario
restricted. However, it slightly reduces when (USD million)
(USD million)
the combined cycle plants are operated with
higher flexibility. Reference 12382.0 -
Approximated calculation yields that this level
Scenario 1 12872.0 490.0
of curtailment can result 2-3% reduction in
annual VRE (wind and solar PV) generation Scenario 2 12979.3 597.3
and such curtailment can negatively affect the
economic viability of VRE projects. Scenario 3 13653.3 1271.3
Scenario 4 13618.6 1236.6

7. Economic and Cost Analysis
Economic cost effectiveness of renewable It is observed that the cost of the reference
energy technologies is also a crucial factor in scenario is the lower than other scenarios. The
determining the integration possibilities. There scenario 1 is the lowest among four scenarios.
are number of cost components to be Therefore the most economically optimum
considered when evaluating the RE integration strategy to increase system flexibility to
possibilities and the total Present Value (PV) integrate the level of renewable energy
cost for each scenario over the 25 year period penetration envisage in this study, is to develop
(2017-2036) is considered in evaluating the cost a combination of low cost base load power
effectiveness alternative scenarios. Addition to plants with pumped hydro power plant. When
that the capacity, energy and emission benefits the system flexibility is increased by increasing
of RE can be assessed in comparison with the the number of combined cycle units in the
reference scenario. system, curtailments are reduced but the total
cost increases is significantly.
The PV cost mainly includes the capital cost,
operation costs, relevant infrastructure
development cost and cost of additional system
reserve for VRE integration. Capital costs data,
plant factors and the levelized cost of electricity
(LCOE) of ORE considered for the study is
given in the table below. Most appropriate
technology costs for the Sri Lankan context is
considered and the falling capital cost of solar
PV is incorporated to reflect the present trend.
Table 8 below compares the total PV cost of
each scenario against the reference scenario. Figure 12 – Cost differences of Scenarios
436
8. CO2 Emission Reduction  Designing the future base load power plants
to be de-loaded with increased flexibility in
The renewable energy development order to minimize the VRE curtailment.
significantly supports the country‟s obligation  Completing the planned network
of reducing greenhouse gas emissions. The strengthening projects as scheduled.
level of RE development projected in this study  Prioritizing the development of ORE
successfully contributes for the reduction of locations based on life cycle cost considering
GHG emissions in-line with the nationally plant factors, availability and cost of
determined commitment (NDC) submitted to transmission network.
UNFCC. Latest analysis of power sector  Incorporate the characteristics grid scale
emissions that incorporated the RE projection storage technologies to enhance the system
of this study, compares the CO2 reduction flexibility.
potential in two scenarios with and without  For small scale RE projects, competitive
renewable energy development and it is bidding process should be adopted to reflect
illustrated in the figure 13 below. the economic costs of development.
 If the proposed conventional plants are not
commissioned as scheduled, the VRE
addition in the plan has to be revised
accordingly. Thus it is proposed to review
this planning methodology once in two years.
 The LNG Power Plants may require
minimum plant factors high as more than
70% in order for LNG contracts to be viable
and competitive in the global market. This
would lead to curtailment of more RE sources
in order to dispatch the LNG operated power
plants. Therefore, LNG procurements
contracts should be negotiated to minimize
the „Take or Pay‟ risks.
Figure 13 – CO2 emission comparison of with
and without renewable energy development  This study open a pathway for a paradigm
shift to Sri Lankan Electricity Generation
System in near future by enabling the
10. Conclusions
integration of more renewable energy for
future years as illustrated in the Figure 12.
It is observed that the system flexibility is
limited when increasing in variable energy
share around 20% requires curtailment.
According to the results, inclusion of hydro
units with a combination of low cost base load
plants and combined cycle units increases the
system flexibility as well as the cost
effectiveness for integrating RE share envisaged
in this study.
Sri Lankan power system needs to adopt

following measures for the effective integration
of higher level of Renewable Energy sources in
to the system.
 Establishing a day ahead, hourly basis wind

Figure 14 – Increased renewable energy
and solar PV energy forecasting systems and
integration
a 24 hour (round the clock), Renewable
Energy Desk for the national dispatch centre.
 Proving Variable Renewable Energy (VRE)
curtailment rights to system operator and
introducing a compensation mechanism has
to be studied to future VRE plants.
437
References
1. Long Term Generation Expansion Plan 2015-2034.
Ceylon Electricity Board
2. CO2 Emissions from Fuel Combustion, 2016

Edition, International Energy Agency
3. Wind resource assessment handbook, National

Renewable Energy Laboratory (NREL), U.S.
Department of Energy
4. World Energy Outlook 2016, International Energy

Agency
5. Renewable Energy Road Map 2016 edition,

International Renewable energy Agency
6. Western wind and solar integration study,

National Renewable Energy Laboratory (NREL),
U.S. Department of Energy
7. Hawaii solar integration study: National

Renewable Energy Laboratory (NREL), U.S.
Department of Energy
8. Emerging Issues and Challenges in Integrating

High Levels of Solar into the Electrical
Generation and Transmission System, National
Renewable Energy Laboratory, U.S. Department
of Energy
9. Grid Operations for Efficient Variable Renewable

Integration, Dr. Michael Milligan, National
Renewable Energy Laboratory, U.S. Department
of Energy
10. The Evolution of Wind Power Integration Studies:

Past, Present, and Future, Erik Ela, Michael
Milligan, Brian Parsons, Debra Lew, David
Corbus, 2009 IEEE
11. Integrating Variable Renewable Energy:

Challenges and Solutions, L. Bird, M. Milligan,
and D. Lew, National Renewable Energy
Laboratory
438
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[439 446],2017
2017
The Institution
Lanka
Optimum Use of Solar Inverter by Feeding Reactive

Power at the Night
K. Kanchanee Navoda and Asanka S. Rodrigo
Abstract: Due to the significant increase in the price of Grid Electricity, many countries are
switching to Renewable sources. With the expanded Solar Net Metering system in Sri Lanka, domestic
consumers and commercial consumers are benefited, after installing such a renewable energy system
directly connected to the LV distribution system. However, LV distribution system in Sri Lanka has
many issues related to the under voltages. The PV inverters are not utilized at the night peak.
Therefore, it can be operated in feeding reactive power to eliminate the low voltage occurrence during
the night peak. This paper describes the theoretical background of “How the solar PV inverters can be
operated as a voltage controller in Sri Lankan Electricity Network by feeding reactive power”. The
study was conducted considering the existing PV inverters. An algorithm is developed to calculate
operating power factors for existing PV inverters in the feeders, where the feeders are experiencing
under voltage problems. The proposed algorithm was discussed with a case study, and the results are
verified by modelling the same system in SynerGee. Based on the developed algorithm, a software
tool with a user interface is developed to calculate the operating power factors. In order to promote
this scheme, an economic evaluation was carried out by proposing an incentive scheme for customers.
It is found that by implementing such system, the net benefit to the utility was positive. Furthermore,
by utilizing distributed PV inverters at night peak by feeding reactive power, low voltage issues and
line losses can be reduced.
Keywords: Power Factor, PV Inverter, Reactive Power
1. Introduction improvement of the night grid performance,

this can be promoted by proposing a new tariff
Solar power is only available at the day time for reactive power generation as well.
where the effectiveness of this solar installation
is zero during the night time. However due to 2. Literature Review
the rapid growth of solar PV installations, now
the network is connected with many In the past, several research has been done on
distributed generators [1-3]. It is shown that the the methods for feed reactive power to
solar PV installations with modern grid distribution networks using different methods.
connected inverters can be used to ensure the One of those studies which are described in [10]
voltage stability of the network by feeding by Huajun. Yu, Junmin. Pan, and An. Xiang.
reactive power [4-6]. The theory behind the inverter use to produce
VAr was described by these authors, and that
Most of the utilities are experiencing system was applying appropriate phase shift between
voltage unbalances at many feeders during the reference current and grid.
night time. This situation can be eliminated by
using a net metering inverter as a VAr producer The internal arrangement of the solar grid tie
to the system. That can provide reactive power inverter use to as a VAr injector was described
for maintaining the system voltage within its in the paper [12] by A. Ellis, B. Kirby, C. Barker,
limits. The operating power factor of the E. Seymour, E. Von Engeln, J. MacDowell, J. R.
inverter has to change for feed reactive power Williams, L. Casey, R. Nelson, R. Walling and
to the grid. The efficiency in the use of net W. Peter. Also, the same arrangement further
metering installation is increased and helped to
maintain the grid voltage stability by this
concept. [7][9][12][13] Eng (Ms.) K. Kanchanee Navoda, M.Sc.(Moratuwa),
B.Sc.Eng.(Ruhuna), AMIE(SL), Electrical Engineer,
Ceylon Electricity Board.
This method is not yet popular in Sri Lanka
Eng. (Dr.) Asanka S. Rodrigo, Ph.D.(HKUST),
because the available tariff methodology does M.Sc.(Moratuwa), B.Sc.Eng.(Moratuwa), AMIE(SL),
not address about a rate for reactive power. As Senior Lecturer, Department of Electrical Engineering,
this feed is important as an effective University of Moratuwa.
1
439 ENGINEER
described with simulation results in the paper This research study was conducted using the
[11] by Maknouninejad, M. Godoy Simoes and existing PV inverters. Therefore, the capacities
M. Zolot. The simulation and practical results and the implemented locations were already
were obtained in the paper [8] by A.Cagnano set. All the research works carried out in the
and E.D. Tuglie with varying the parameters literature review was calculated the suitable
and tested with several scenarios. capacities and the pre-defined locations
(selected by another calculation) to minimize
According to the tariff system which was the low voltage issues in the feeder. The
proposing in the reactive power feeding developed algorithm also specified to select
concept in Germany was described in the paper operating power factors for each existing PV
[1] by A. Kozinda, T. Beach and V. Rao, the inverter only.
concept was for the whole day. The cost
comparison was made based on two factors in The economic evaluation mentioned in the
this paper; those were „cost to the customer to paper was described the cost saving factors [1]
produce reactive power‟ and „cost to the utility according to the Sri Lankan tariff system, and
to produce reactive power‟ with the absence of the reactive tariff was proposed to reward such
inverter feeding method. Under that „cost to customers who were expected to feed reactive
customer‟ was again considered with the cost if power to the network through their PV
the customer downgrades power factor and inverters. This total economic evaluation was
cost if the customer over sizes the inverter. On newly discussed in the paper, and it was not
the other hand, the cost to the utility of previously mentioned in other research works
installing capacitor banks to compensate the mentioned in the literature review.
reactive power requirement was instated
feeding it by using an inverter. 3. Mathematical Modelling
Tariff method needs to be addressed a 3.1 Technical Background
rewarding method to the customer who wishes
to feed reactive power at the night peak. Before
proposing such tariff, the tariff for reactive
power consumption also needs to be
introduced. The Maximum Demand Charge for
Bulk Consumers which was the only consent on
reactive consumption, introduced in Sri Lankan Figure 1- Line Diagram of Feeder Points
tariff system. The purpose of introducing The active loads and reactive loads were
Maximum Demand Charge was to encourage obtained in each point, and power flow of
the customer for Demand Side Management throughout the line was represented according
and minimize the system voltage reduction and to the above line diagram. The ith point active
harmonics etc. power load and reactive power load were
represented as PLi and QLi. The active and
Considering the tariff systems in other reactive power at point i was represented as Pi
countries, several methods were recommended and Qi (Figure 1).
to charging for reactive power. Per kVAr rates
in three blocks for reactive power were Active Power at i+1 bus,
introduced Singapore and Spain. From April
2010, a common charging method was forced Pi+1 = Pi - PLi+1 - Ri,i+1 . …. (1)
across all UK Electricity Supply Companies. Reactive Power at i+1 bus,
The way the reactive power charge was
calculated; the customer was „allowed‟ to use Qi+1 = Qi - QLi+1 - Xi,i+1 . …. (2)
reactive power up to a third of the active power
Line Loss between ith and i+1th bus,
consumed for free, „allowed‟ reactive power =
0.33 x total kWh. The charge was only applied
by reducing the allowed reactive power from P Loss (i, i+1) = Ri,i+1 . …. (3)
total consumed reactive power. In German Here the line resistance and line reactance
distribution network operators were charged in between ith and i+1th points were Ri,i+1 and Xi,i+1.
averagely 1.1 c€/kVArh (0.0-2.7 c€/kVArh) for S (kVA) was denoted the installed solar inverter
reactive power consumption only if the power capacity, the operating power factor of the solar
factor was lower than 0.9 (in average). [5] inverter Cosα was needed to be fixed according
to the requirement of the reactive power at the
ENGINEER 2
440
particular location and the distribution line The voltage was improved, and this improved
conditions (Minimum voltage, Power Loss, voltage was maintained between the accepted
etc..). standard tolerances margin is .
Reduced distribution line loss between ith and The current Is was adjusted within the allowed
i+1th points was denoted as, voltage tolerance margin. Is was denoted as,
P Loss new (i, i+1) =Ri,i+1 . …. (4) Q = …. (14)
Here Q was in kVAr which is feeding by the
The current Ii,i+1 in figure 2 was represented as Ir solar inverter.
and Ix where Ir = Ii,i+1 Cosθ, Ix = Ii,i+1 Sinθ
respectively (Cosθ is the power factor). Q = …. (15)
Here S was the apparent power of the solar
inverter and the Sinα was the operating power
factor ( ) of the inverter which was set
Figure 2 - Current between Two Points according to the requirement of voltage
regulation.
Vi = Vi+1 + Ii.i+1 (Ri,i+1 + jXi,i+1) …. (5)
Vi = Vi+1 + (Ir - jIx) (Ri,i+1 + jXi,i+1) 3.2 Objective Function
Vi = Vi+1 + Ir Ri,i+1 + Ix Xi,i+1 …. (6) The objective function was obtained as a
Vi = Vi+1 + Ii,i+1 Cosθ Ri,i+1 + Ii,i+1 Sinθ Xi,i+1 sensitivity factor by considering the power loss
…. (7) function equation no (3) of a point. Line loss
between ith and i+1th sections,
Hence voltage drop between ith and i+1th point, P Loss (i, i+1) = Ri,i+1 .
Vi - Vi+1 = Ii ,i+1 (Ri,i+1 Cosθ + Xi,i+1 Sinθ) …. (8)
Sensitivity factor was defined as,
Considering the scenario with feeding reactive …. (16)
power shown in figure 3,
The locations where the power losses are high
were identified by the Sensitivity Factor. This
was defined as a function of reactive power
since the concept was to feed reactive power by
a solar inverter and reduce line losses and
Figure 3 - Current between Two Points with maintain the system voltage within the
PV Inverter margins.
I1i, i+1 = Ii ,i+1 + jIs …. (9) 3.3 General algorithm for VAr control
Here Is the current flow through the solar with existing PV installations
inverter, The operational data of the selected existing
I1i, i+1 = Ir - jIx+ jIs …. (10) system (LV feeder) with PV installations were
considered as the inputs for the algorithm. The
Hence under above scenario voltage drop load data of each pole and the voltage profile
between ith and i+1th point, on each pole were taken in to account for the
Vi - Vi+1 = I1 i ,i+1 (Ri,i+1 + jXi,i+1) model. Then by considering the data, the
Vi - Vi+1 = Ir Ri,i+1 + Ix Xi,i+1 - Is Xi,i+1 …. (11) locations were identified where the low voltage
profiles existed.
Before feeding of reactive power by solar
inverter, the representing voltage at i+1th point Existing installed solar inverter data such as the
was, inverter capacity and the availability of the
function of Q at night mode was collected.
Vi+1 = Vi Ir Ri,i+1 Ix Xi,i+1 …. (12)
The sensitivity factor was computed to each
pole, and the locations were identified where
After feeding of reactive power by solar
the power losses are high considering as a
inverter, the representing voltage at i+1th point
factor of reactive power.
was,
Vi+1 = Vi Ir Ri,i+1 Ix Xi,i+1 Is Xi,i+1 …. (13) According to the information of installed solar
inverters, the largest capacity of inverter was
3
441 ENGINEER
selected which was the nearest location of the Table 2- Parameters of the Sample Feeder
poles and which was having the highest loss. Length Installed Active Reactive
No of
The operating power factor of the inverter was of the service inverter Power Power
set where it feeds required reactive power, to Feeder connections Capacity Load Load
eliminate low voltage issues and the voltage (m) at the pole (kVA) (kW) (kVAr)
profiles of the poles were increased within the
0
standards.
40 2 0.5 1.20 0.60
80 2 1.30 0.60
The voltage profiles were obtained again, and 120 1 0.56 0.40
the low voltage affected areas were selected 160 2 1 1.50 0.60
and if the issue was there then again the 200 2 0.92 0.51
sensitivity factor for the selected feeder was 240 3 1.36 0.94
computed with the modification, and the next 280 1 2 1.11 0.73
highest loss locations were identified. The 320 2 0.75 0.44
iterations were done by using Genetic 360 2 0.64 0.18
Algorithm. 400 2 0.45 0.25
440 3 3 1.28 0.30
Next, the installed solar inverter was selected 480 2 1.33 0.61
where having a high capacity to the nearest 520 1 0.95 0.70
location among the poles which are having 560 1 0.5 0.64 0.37
600 2 1.20 0.50
with highest losses and operation power factor
640 3 1.5 1.37 0.14
was set as mention in above.
680 1 0.94 0.27
720 3 0.5 2.20 1.20
The function was repeated until eliminate or
760 2 1.06 0.52
reduce the low voltage problems of the selected 800 1 2 1.12 0.60
feeder with existing solar installations. 840 2 0.5 1.23 0.47
880 1 0.86 0.70
4. Case Study 920 1 0.81 0.50
960 2 2 1.80 0.62
A sample model was created for case study 1000 3 2.21 0.72
considering the features shown in Table 1. 1040 2 1.25 0.76
Parameters of the sample feeder were shown in 1080 1 0.5 0.94 0.45
Table 2. 1120 2 1.68 0.49
1160 1 0.85 0.73
The final operating power factors of each PV 1200 1 1 1.20 0.35
inverter installation which are to feed 1240 1 0.76 0.27
1280 2 3.5 0.91 0.50
reactive power to minimize the low voltage
1320 1 0.80 0.20
problem were shown in the Table 3.
1360 1 0.5 0.61 0.37
1400 1 1 0.68 0.34
Table 1- System Features
1440 2 0.80 0.26
Transformer Capacity (kVA) 100 1480 1 1.50 0.31
Feeder Length (m) 1500 1520 1 0.75 0.30
Average Pole Span (m) 40
Fly Conductor Resistance (Ohm/m) 0.000452 The results for voltage profiles were shown
Fly Conductor Reactance (Ohm/m) 0.000321 in Figure 4.
Average Service Wire Length (m) 30
25mm2 ABC Service Wire Resistance 0.00149
(Ohm/m)
25mm2 ABC Service Wire Reactance 0.0001
(Ohm/m)
Total Customers per Feeder (1 64
Phase)
Figure 4 - Graph of Voltage Profiles
ENGINEER 4
442
According to the resultant voltage profiles, it reduced and that balance was fed by the
can be seen that when the absences of distributed solar PV inverters in the feeder.
feeding reactive power, low voltage profile The obtained results for total current
was obtained starting from 640m of feeder variation of the feeder were shown in Figure
length and downwards. Once the reactive 6.
power feeding was enabled the low voltage
profile was started from the 720 m length of The obtained results for power loss variation
the feeder. The point of 640 m and 680m of the feeder were shown in Figure 7.
feeder locations voltage were eliminated
from low voltage issue. Considering all
locations the voltage was increased slightly.
Table 3 - Computed Operating Power Factors

Length of Installed Final
the Feeder inverter Operating
(m) Capacity Power Factor
(kVA)
40 0.5 0.00
160 1.0 0.80
280 2.0 0.93 Figure 6 - Graph of Current Variation
440 3.0 0.99
560 0.5 0.67
640 1.5 0.99
720 0.5 0.00
800 2.0 0.95
840 0.5 0.28
960 2.0 0.95
1080 0.5 0.44
1200 1.0 0.94
1280 3.5 0.99
1360 0.5 0.67 Figure 7- Graph of Power Loss Variation
1400 1.0 0.94
According to the case study total power loss
of all the branches before feeding reactive
The obtained results for reactive power power was 1.95kW, and after feeding
variation of the feeder were shown in Figure reactive power, the total power loss was
5. 1.77kW. Total reduction of power loss was
9.26%.
5. Model Validation
The Case Study results obtained by the

developed algorithm were compared with
the results obtained by the SynerGee 3.5
software for the same Case Study.
The feeder loading at Table 2 have modelled

in SynerGee was shown in Figure 8.
Figure 5 - Graph of Reactive Power
Variation According to the parameter of the feeder and
the connected loads, the low voltage scenario
According to the reactive power variation, it was shown after 640m of the feeder length.
can be seen that the 19.045 kVAr of the
reactive power feeding from transformer was
5
443 ENGINEER
 Loss reduction due to reactive power
feeding by solar inverter [4]
Reduced distribution line loss between ith and

i+1th points was denoted as,
P Loss (i, i+1) = Ri,i+1 I2i,i+1 …. (17)
The cost saving due to the reduction of loss was

represented considering the distribution system
loss.
Saving of Cost (Per Year) = Saving of loss
Figure 8 - SynerGee Model of Case Study Average cost per unit Peak Hours 365
Feeder
Considering the Case Study the saving of cost
The result obtained by the algorithm was due to the reduction of loss was Rs 3979.76.
used, and the inverters were placed in the
feeder with the operating power factors  Increased line capacity due to reactive
according to the Table 3, and Figure 9 was power feeding by solar inverter [4]
shown the model that includes with PV
inverter installations. The line current without compensation was
91.365 A. The line current with compensation
was 87.001 A.
Considering the Case Study the saving of cost
due to the increased line capacity was Rs
45266.88.
 Increased Maximum Transfer Capability [4]
Pmax = Where k = 𝑄𝑄/𝑃𝑃
Figure 9- SynerGee Model of Case Study
Feeder with PV Inverter Installations …. (18)
Considering the starting point of the feeder,
After running the load flow analysis the low
voltage issue was eliminated until 720m The maximum transfer capability without
away the length of the feeder which was
obtained as the same result by the compensation =
implemented algorithm.
The maximum transfer capability with
Compare to the results from the
implemented algorithm and the SynerGee
compensation =
model the percentage of voltages with
feeding reactive power obtained at the same
nodes are almost same in both scenarios. Considering the Case Study the saving of cost
Therefore the validity of proposed algorithm due to the increased line capacity was Rs
for reactive power feeding and its 155593.07.
implemented software version was accepted.
6.2 Proposal for Reactive Tariff in Sri
6. Economic Evaluation Lanka
Proposing the reactive power tariff for Sri
Lankan system was needed only to encourage
6.1 Energy Cost Saving due to Feeding
the Demand Side Management. Therefore the
Reactive Power
rate was defined bit higher than the cost of the
The economic evaluation was conducted
market price of per kVAr when installing the
considering the energy saving at selling end.
capacitor bank to improve the power factor.
The average cost per unit of selling end was
Then the customers were insisted to installed
taken at Rs 15.06 and the feeding time of
such system and improve the power quality
reactive power was limited to peak hours. [2]
ENGINEER 6
444
rather than paying a higher rate for each  Even with the small capacities of
month. distributed PV inverters can be used to
Considering the Case Study results calculation eliminate the low voltage for some extent.
was computed when a capacitor banks fixed at According to the case study, it fixed the
customer end. The assumptions were made in voltage around 80m length of the feeder.
Table 4. [5]  “Sensitivity Factor” was introduced to
identify the critical locations where it is
Table 4 - Assumptions Made for necessary to attend first, under the
Economic Analysis limitations of “over voltage”.
 Since the concept was introduced for the
Life time of the capacitor bank 5 years night peak solar generation module is
Annual maintenance cost (% of 2% mostly utilized, and the efficiency of the
capital cost) installation increased.
Cost of installing 1kVAr Rs 3000
 Due to the reduction of reactive power
Rate of return on equity 20%
feeding from the transformer, it has
reduced the total line current and caused to
Considering the pole 40m away from the
reduce the power loss of the total system.
transformer required reactive power was 0.6
According to the case study, the reduction
kVAr and assuming 1kVAr capacitor bank
of power loss was 9.26%.
was installed operates at the peak time.
 Even after the rewarding back to the
customers, the net saving to the utility was
Total reactive energy required per year
profit, and according to the case study, it
= 876 kVArh
was nearly 185000 LKR per year.
If the rate per kVArh was „R‟ considering the 8. Recommendations
annual cash flow,
R× =3000+ …. (19) As per the research findings, when promoting
R =Rs 1.830 per kVArh feeding reactive power with existing installed
PV inverters following recommendations were
Levelized cost for “Reactive Energy” by each made;
pole for Case Study was calculated, and the  The behaviour of the feeder parameter
computed average cost per kVArh was Rs values at the night peak needs to be
1.809, when installing capacitor bank in analysed such as voltage, power loss along
customer end. the feeder before computing the power
factors. By analyzing the whole feeder
Therefore for encouraging customers to feed under the limitations, the operating power
reactive power at the peak time by using factor for each inverter has to be decided by
installed solar PV inverters the reward was the utility and to be informed to the
proposed by crediting Rs 2.00 per kVArh. customer.
 Solar inverter manufacturers have to
Cost benefit analysis was computed by using manufacture their products with the
the case study results. provision of changing the power factor set
Total saving by reactive power feeding per by the user and with the ability to program
year = Rs 204839.71 the setting on an hourly basis.
Total rewards to the customers per year  General Public, including the customers
= Rs 20032.66 who wish to connect solar power
Net saving to the utility per year generation module, are need to be educated
= Rs 184807.05 on the selection of proper PV inverters
which are capable of operating in reactive
7. Conclusion mode as well.
 The rewarding method for reactive feeding
According to the case study when selecting the by PV inverters needs to be introduced to
optimal operating power factor for the existing customers in order to encourage them for
installed PV inverters, conclusions were maintaining the stability of the system
finalized as follows; voltage.
7
445 ENGINEER
References 12. A. Ellis, B. Kirby, C. Barker, E. Seymour, E. Von
Engeln, J. MacDowell, J. R. Williams, L. Casey,
R. Nelson, R. Walling, W. Peter.” Reactive
1. Kozinda, A., Beach, T., Rao, V., “Latent
Power Performance Requirements for Wind and
Opportunities for Localized Reactive Power
Solar Plants” presented at the IEEE Power and
Compensation," Cal x Clean Coalition Energy
Energy Society General Meeting, San Diego, CA,
C226, 2013”.
2012.
2. CEB Statistical Digest Report -2015.
13. A. Maknouninejad, N. Kutkut, I. Batarseh, Z.
3. Ceylon Electricity Board, “Manual for Qu. “Analysis and Control of PV Inverters
Interconnection of Micro Scale Renewable Operating in VAR Mode at Night” presented at
Energy Based Power Generating Facilities at the IEEE PES Innovative Smart Grid Technologies
Low Voltage Consumer Feeders of National (ISGT), Hilton Anaheim, CA, 2011.
Grid”, August 2016.
14. Inverter Reactive Power Compensation (2014):
4. Kutkut, N., “An AC PV Module with Reactive http://www.blueoakenergy.com/blog/inverter
Power Capability: Need and Benefit," Petra Solar, -reactive-power, Visited, 24th November 2016.
Inc., 2012”.
5. Resource Management Associates (Pvt) Ltd “A

Tariff for Reactive Power in Sri Lanka”, Final
Report, September 2011”.
6. SMA Solar Technology, America “Q at Night,”

Reactive power outside of feed-in operation with
SUNNY CENTRAL 500CP XT / 630CP XT /
720CP XT /760CP XT / 800CP XT / 850CP XT /
900CP XT”.
7. Ellis, A., Kirby, B., Barker, C., Seymour, E., Von

Engeln, E., MacDowell, J., Williams, J. R., Casey,
L., Nelson, R., Walling, R., Peter, W., ”Reactive
Power Interconnection Requirements for PV and
Wind Plants – Recommendations to NERC”
Sandia National Laboratories, USA, 2012.
8. Cagnano, A., Tuglie, E. D. "Online Optimal

Reactive Power Control Strategy of PV
Inverters" Industrial Applications, IEEE
Transactions on, vol.58, no.10, pp.4549,4558,
October 2011”.
9. Rajesh, C. H., Reshma Kiran, S., “Harmonic and

Reactive Power Compensation in a Grid
Connected PV System with Source Side Control
Technique.” The International Journal of Emerging
Technology and Advanced Engineering, vol. 3,
pp.212-220, Sep 2013.
10. Huajun, Yu, Junmin, Pan, and An. Xiang, "A

Multi-Function Grid Connected PV System with
Reactive Power Compensation for the Grid"
Solar Energy, vol. 79, no. 1, pp. 101-106, July.
2005.
11. Maknouninejad, M. Godoy Simoes, M. Zolot,

“Single Phase and Three Phase P+Resonant
Based Grid Connected Inverters with Reactive
Power and Harmonic Compensation
Capabilities” IEEE Tr. IEMDC 09, 3-6 May 2009,
pp. 385-391.
ENGINEER 8
446
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[447 453],2017
2017
The Institution
Lanka
Smart Electronic Analyzer for Determination of

Chicken Meat Spoilage
M.P.U. Isuranga, H. Pasqual and N.S. Weerakkody
Abstract: People in the modern society pay more attention to food safety than the cost, since a
lot of known diseases are transmitted through spoiled food. Freshness is the main factor considered by
a customer when purchasing food items. Microbes can easily spoil perishable food. The traditional
food spoilage detection methods are visual identification, physical detection, chemical detection, and
microbial appraisal methods. However, these methods have many problems like long experiment
cycle, high equipment cost, and necessity of relatively sophisticated laboratory environmental
conditions. Thus, this study presents a spoilage status analyser which can detect evolving gas
concentration of meat products. In this design, concentrations of gases are measured, which are
emitted in the meat decaying process using gas sensors. The classifier operating on support vector
machine based on many parameters gave a quality approach to obtain a final spoilage status of tested
meat samples. The obtained results demonstrated good performance in discriminating meat samples
in one of two quality classes, the accuracy of obtained data by the equipment was further confirmed
by microbiological assay using total plate counts. The developed smart electronic analyser is a quick
and low-cost detector with accuracy to detect freshness of meat.
Keywords: Electronic nose; Gas sensors; Meat spoilage; Support vector machines;
1. Introduction methods in the laboratories. After getting

Meat is easily being spoiled by microbes. As a results from such samples, further
product from animal sources, it can expose to investigations are carried out.
bacteria that cause food spoilage during The early detection of microorganisms in meat
processing, transporting, and selling. Sri is an essential part of any quality control or
Lankan meat industry only large-scale entities food safety plan. Conventional methods for
have any state of the art facilities that can identifying bacterial spoilage have involving
control the above processes following ISO basic steps like pre-enrichment, selective
22000 [15] standard conditions. Presently, enrichment, selective plating, biochemical
butchers openly slaughter the animals without screening [11]. Commonly used methods like
considering the standard environmental SPC (Standard Plate Count), DMC (Direct
conditions. Therefore, the chances of microbial Microscopic Count) are inexpensive and can
contamination are significantly high. give both quantitative and qualitative
The present society gives more attention to information on the number and nature of
meat freshness because of food spoilage can bacteria. Moreover, preparation of the culture
cause many health problems. Therefore, meat medium, inoculation of plates and colony
products are regulated under the Laws which counting make these conventional methods
have specific guidelines to assess the quality of need expert personnel. Also, those methods
meat products. In Srilanka, meat products are require days to complete and to produce
regulated under the Food (Amendment) Act, results.
No 20 of 1991. Food Act states that 106 CFU/g
(colony-forming units per gram) is the limit that
determines whether meat is spoiled or not Eng. M.P.U. Isuranga, B. Tech. Eng. (Hons) (OUSL),
(Microbial Population ≥ 106 CFU/g - Spoiled, Student Member of IESL, Demonstrator, Department of
Electrical & Computer Engineering, The Open University
Microbial Population ≤ 106 CFU/g - Fresh) of Sri Lanka.
[7,16].
Regular monitoring is required to determine Ms. H. Pasqual, B.Sc. (Eng.) (Hons), MEng(Saitama),
whether butchers are selling fresh quality meat Senior Lecturer, Department of Electrical & Computer
Engineering, The Open University of Sri Lanka.
or not. Public health inspectors are the
responsible officers to ensure the Dr. N.S. Weerakkody, BSc Agri (Peradeniya), MSc
implementation of the Food Act. At present, (Peradeniya), Ph.D. (Australia), Senior Lecturer,
Department of Agricultural and Plantation Engineering,
public health inspectors collect samples from The Open University of Sri Lanka.
butcher shops and apply microbial appraisal
1
447
The testing of microbial spoilage needs a spoilage occurs. For early detection,
method that can overcome problems from odour/smell signatures are considered useful
conventional methods. These methods are parametric resources for determining spoiled
validated and evaluated using conventional meat [1]. Apart from that temperature and
culture methods as a reference. Most of these humidity of environment make many
rapid detection devices are expensive. Because variations to the spoiling process. [6,12]
those devices use advanced technologies, which Proposed system can be used to sniff in the
are very costly. For example, FT-IR (Fourier normal environment to identify spoiled items.
transform infrared) spectroscopy uses infrared The proposed system mainly consists two parts,
rays for detection while mass spectrometer and identification and classification. Figure 1 shows
gas chromatography use ionised detectors and the block diagram of the proposed model.
external gases for measuring purposes[11].
Methods like PCR (Polymerase chain reaction)
Collect
uses detecting DNA patterns which require
Sample Gas
expert labour intensive also. Rapid detection
Concen Pre -
methods are needed to operate in a laboratory
tration Processing
environment to get required conditions. An
Data collected
electronic nose introduced as rapid detection
method which can detect and recognise certain Display signal
item using their odour signature. These systems Spoilage Classification
have sensors, pumps, valves, electronics, data Result
extraction methods, data processing methods. Figure 1 - Proposed Model
However, most of the modern devices like
Alabaster UV, FOX 2000, FOX 3000, and food The developed system consists of five key
sniffer provide excellent accuracy in their components: sensor chamber, sampling unit,
results, though they are very expensive. Several data acquisition system, Control unit, and
types of research done to design a low-cost, Display Unit. Details of principal components
effective device [14]. explained in the following sections.
Meat is made up of fibres which are held
together by connective tissue. Most animal 2.1 Sensor chamber
muscle is 75% water, 20% protein, and 5% fat, Sensor Chamber has a gas sensor array which
carbohydrates, and various proteins. The has six different gas sensors. This measures
connective tissue is composed of proteins, concentrations of gases such as ammonia,
collagen and elastin. The fat surrounds the carbon monoxide, natural gas, smoke, alcohol,
fibres in small particles. and volatile organic compounds which
While the content of the fibres is called muscle associated with meat decaying process and
juice.it is composed of water in which are microbial interactions. This sensor array fixed
proteins, salts and extractives. These extractives in the chamber that connects with the sample
contain nitrogen which causes a flow of headspace collection chamber. Also, sensor
digestive juices in the stomach, which aids in chamber temperature and humidity are always
the breakdown of food [4]. Meat decomposes monitored to avoid instability of stable
like any other organic matter. Bacteria, conditions of sensors.
Enzymes, fungi, Oxygen, and moisture all
convert meat back to the core organic
compounds. Protein will decompose into
ammonia, the fat will become aldehydes and
aldehyde acid, and the carbohydrate will
decompose into alcohol, ketone and carboxylic
acid [4]. Food would not decompose if not for
bacteria, enzymes and fungi these microbes are
primary decomposers of all food items. They
decompose meat into volatile gases [8].
2. The Proposed System

Meat spoilage can be detected using, off- Figure 2 - Sensor Chamber
odours, off-flavours and discolouration.
However, mainly this off-flavours, and Perspex plastic container serves as a gas
discolouration are present when most of the chamber. This chamber is airtight and non-
2
448
reactive to chemical or food odours. A tube removed and diluted into 1:10 ratio using 0.85%
connected to the sample section via a saline solution. The meat sample was
diaphragm-type air pump. This air pump is homogenised in a stomacher (IUL masticator,
turned on for 20 seconds to drawing in the torrent del‟Estadella, Barcelona, Spain) for 2
headspace gas from the meat sample via a tube. minutes and serially diluted in 9 ml of diluent
The odours directed by this pump directly (0.85%, Saline) as necessary. Plate count
access the sensors to get measurement obtained by plating out samples on the surface
according to the gas concentration. of nutrient agar with incubation of the plates at
There are no specific rules applied when 37oC for 48 hours. Counts were achieved by
selecting the type of sensors. A considered enumerating typical colonies present and
factor was the detection ability of the required calculated as log10CFU/g of the meat sample.
gas. The type of sensors used and their From this calculation predetermination of meat
sensitivity to a particular gas listed in Table 1. samples quality was obtained before testing
with the device.
Table 1 - Gas Model Sensitivity
4. Analytical Algorithm
Sensor Model Sensitive Gases Data measured with gas sensors react with one
MICS 5524 CO, C2H5OH, H2, NH3, another quickly and may be distracted by the
CH4 thermal noise. Moreover, before applying the
MQ5 LPG, NPG, C4H8, C3H8 classification, in this work we use the “moving
MICS 5914 NH3, C2H5OH, H2, C3H8, average” technique as filtering method to
C4H10 reduce the noise factor from sensor signals.
MQ3 C2H6O, CO, LPG From getting moving averages, final sensor
data can base on the all the sensory data
MQ2 H2, LPG, CH4, CO,
gathered till sensors come to steady state. This
C2H6O
process runs along the 20s to get decent data set
MICS 2714 NO2, H2
from all six sensors.
The two most popular types of moving
MICS (Sensortech) gas sensors and MQ averages are Simple Moving Average (SMA)
(Futurlec) gas sensors both operate with 5V and the Exponential Moving Average (EMA).
supply for both heater section and sensing [5,9] From these two moving average
section. However, to maintain the required algorithms, SMA formed by computing the
voltage level to get a proper heating average sensory data over a specific number of
temperature, a voltage divider circuits periods and old data dropped when the new
implement for each and every sensor according data comes available. Therefore, it is necessary
to datasheets. to determine an exact number of periods from
When compared to the electrochemical gas which we want to get the average. In this work,
sensors, MOS type gas sensors have several twenty periods were used to get the average. In
advantages. Some of them are simple EMA algorithm, recent data get more weight
operations, low manufacturing cost, than old data when getting the final average.
significantly good sensing range, commercial However, in this situation, we want to get the
availability and small size. average of sensory data when it is already in
the steady state. So SMA is the most suitable
2.2 Sensor data acquisition
algorithm for the system.
Voltage captured from the sensor connected to
the LM358N operational amplifier. Op-Amp
circuit with non-inverting configuration used as
5. Classification algorithm
There are three popular algorithms which used
the amplifier model [13]. The gain of the
in the several gas sensors based works
operational amplifier was adjusted on a one to
considered when choosing the correct
one basis, depending on the strength of each
algorithm for the system. These three
sensors‟ response. The amplified voltage
algorithms namely artificial neural networks
signals of the six sensors connected to the ADC
(ANN), support vector machines (SVM), and K-
of the Arduino platform. Then this data was
nearest neighbour (KNN). Training and
processed using the Analytical algorithm.
evaluation for classification of meat spoilage
status use two classes, Spoiled and Fresh.
3. Calibration Method According to the researches done by several
Briefly, a small portion of the meat sample
intuitions most accurate algorithm is KNN
weighing approximately 10g was aseptically
[10,12]. The most underperforming algorithm,
3
449
when used with gas sensors, is ANN algorithm. used. Meat decay rapidly compared to other
KNN and ANN both have a drawback that to items. If rotten meat exposed to the proposed
get a considerably accurate result; these system by placing it in the sampling chamber,
algorithms need a large amount of training data identified the spoiled item using classification
set. However, when using support vector algorithm.
machine, we can get acceptable accuracy with
the much lower amount of training dataset
[3,9]. So for this study, we use SVM as
classification algorithm.
This algorithm is used to classify data based on
features. SVMs hinge on two core mathematical
operations: first, plotting the input data into
high dimensional space using a kernel function
and then splitting them into two different
classes using hyperplane decision surface [5].
When training the SVM algorithm, SVM will
identify the hyperplane that separates two Figure 4 - Prototype implementation
classification class which leaves maximum
margin from data of both classes as possible. The developed prototype system was capable
The maximum margin hyperplane: of measuring gas concentration values which
emit when the spoilage process happens. When
wT x – b = 0 … (1) training experiment conducting temperature
inside the chamber kept almost constant (about
W and b are to be determined to maximise the 30oC) and the humidity value also kept 40-60%
separation between the two parallel level in every experiment. The data measured
hyperplanes as a state in figure 3 where “m” after a regular interval of 15 minutes. The
defines as the distance between two sampling and cleaning periods defined by
hyperplanes. Hence in the training process, can obtaining sensor responses that have been
be definite that (xi, yi) is data points where “xi” tested by the system. For that allocation of the
is the captured data while “yi” is the class label 20s were sufficient to clean and sample the
[yi ∈ {-1,1}] which each sample belongs. headspace using a vacuum pump. Another 20s
was given to get the moving average of sensors.
Total time for the process is the 60s.
In this experiment, 60 randomly selected
samples used from a local hospital, local
butcher shops and supermarkets. The quality of
the meat samples separately identifies with a
microbiological assay using total plate counts.
A support vector supervised training was
carried for 50% of the samples, and decision
boundary hyperplane parameters included in
Figure 3- SVM determine the maximum Atmel Mega 2560 microcontroller. Furthermore,
margin for separation hyperplane other 50% of the samples used in testing the
accuracy of the classifier.
The decision boundary should classify all
Gas concentration of 30 Samples from the
points correctly under the hyperplane which collected meat batch taken to train the support
have a maximum distance to both classes. Then
vector machine algorithm. The relevant
maximisation will reduce with the parameter values received from the six sensors.
minimization of ||w|| under the constraint
However, these sensors did not provide exact
yi (wT xi + b) ≥ 1 or yi (wT xi + b) ≥ -1. gas concentration value of single gas. As an
example, MICS5524 and MICS5914, both
6. Prototype implementation and sensors detect Ammonia gas. So, sensor values
testing are weighted according to the importance of
A prototype system was developed(figure 4) parameter and obtained two different values
for obtaining concentration values and also for Sensortech (MICS) sensors and MQ sensors.
training for classification algorithm which
based on support vector machine. For this MICS Value = (MICS5914*0.5) +(MICS5524*0.3)
study, only one kind of meat samples (Chicken) +(MICS2714*0.2) … (2)
4
450
and we get the support from these data as
MQ Value = (MQ3*0.4) +(MQ2*0.3) +(MQ*0.3) vector form to adjust the boundary well for
… (3) classification process, and here we have five
support vectors to define the optimum
7. Results & Discussion boundary(Figure 6). Below shows the used
The obtained “MICS value” and “MQ Value” support vectors,
2D graph has been plotted in figure 5 to show
freshness status distribution among training
( ) ( )
data set.
( ) ( )
( )
When we use the theory support vector, we

need the bias input 1 to modify existing vectors
into modified support vector form. Modified
support vectors are S1ˊ , S2ˊ , S3ˊ , S4ˊ , and
S5ˊ .Also, we need to find seven parameters α1,
α2, α3, α4, and α5 about the vectors by using
seven equations.
Figure 5- Support Vectors of Training Data
α1 S^1 S^1 +α2 S^2 S^1 +α3 S^3 S^1 + α4
According to this graph, the fresh samples are S^4 S^1 + α5 S^5 S^1 = -1 (-ve Class) … (4)
in blue and the spoiled samples shown in red
colours. When we examine Y, X axis, it clearly α1 S^1 S^2 + α2 S^2 S^2 +α3 S^3 S^2 +α4
demonstrated that there is a separation between S^4 S^2 +α5 S^5 S^2 = -1 (-ve Class) … (5)
these two classes. With this data, the most
suitable method for data classification can α1 S^1 S^3 +α2 S^2 S^3 +α3 S^3 S^3 +α4
found. The proper classification answers are S^4 S^3 +α5 S^5 S^3 = -1 (-ve Class) … (6)
given by the support vector machines. Below
show the steps of hyperplane equation α1 S^1 S^4 +α2 S^2 S^4 +α3 S^3 S^4 + α4
derivation of support vector machine. S^4 S^4 +α5 S^5 S^4 = +1 (+ve Class) … (7)
α1 S^1 S^5 +α2 S^2 S^5 +α3 S^3 S^5 +α4

S^4 S^5 +α5 S^5 S^5 = +1 (+ve Class) … (8)
Let substitute values for above equations S1ˊ ,

S2ˊ , S3ˊ , S4ˊ , and S5ˊ . From these seven
equations, we obtain values for α1, α2, α3, α4,
and α5
α1 = 1.6745 α2 = -2.4937 α3 = 3.6123

α4 = -2.3696 α5 = -4.2355
Using previously obtained values w‟ value

obtained.
∑ … (9)
Figure 6 - Hyperplane support vectors
value of the developed system can obtained
from above equation,
According to these data in feature space, there
is a boundary that can separate these data into ൌ( ) … (10)
two classes as a good class and bad class. This
boundary defined by the help of some data Using above calculations, hyperplane equation
from both classes which is near to the border is,
5
451
When Doing this study, problems like sample
=( ) … (11) odours are not coming to sensor chamber were
occurred. Therefore, two vacuum pumps
b (offset value) = -629.045 introduced for cleaning and sample collection.
Six gas sensors determined the gas
Gradient of a line = ;Ѳ = 2.04 concentration data from the sample headspace
and displayed the status. The support vector
machine; statistical data classification algorithm
From that, Hyperplane 30 samples were tested
did the classification part of the device to get
with spoilage early detection device to
the outcome.
determine the accuracy. Figure 7 shows the test
The smart electronic analyser is a kind of
samples which categorised according to the
electronic nose. As far with all the electronic
above hyperplane.
nose systems, it depends on gas sensor
technology. The next step is integrated more
advanced gas sensors with the device to make it
universal food spoilage detection device which
can detect freshness status of any food item.
Also, can be developed a cloud database using
device test data and rerun the classification
algorithm in cloud platform to get more precise
hyperplane boundary update will do in future
improvements of the project.
References
1. Abdallah, S. A., Al-Shatti, L. A., Alhajraf, A. F.,

Al-Hammad, N. and Al-Awadi, B. (2013) „The
Figure 7- Test Data with SVM Hyperplane Detection of Foodborne Bacteria on Beef: The
Application of the Electronic Nose‟,
the accuracy obtained according to test samples SpringerPlus, 2(1), p. 687. doi: 10.1186/2193-
show below; 1801-2-687.
Accuracy of the selected algorithm
2. Anthony, C. R. M. and Balasuriya, D. N. (2016)
„Electronic Honey Quality Analyser‟, Engineer:

= Journal of the Institution of Engineers, Sri Lanka,

49(3), p. 41. doi: 10.4038/engineer.v49i3.7075.
= 3. Balasubramanian, S., Panigrahi, S., Logue, C. M.,

Gu, H. and Marchello, M. (2009) „Neural
= 86 % Networks-Integrated Metal Oxide-Based
Artificial Olfactory System for Meat Spoilage
Identification‟, Journal of Food Engineering, 91(1),
8. Conclusion and Future Work pp. 91–98. doi: 10.1016/j.jfoodeng.2008.08.008.
Microbial spoilage of food caused major 4. Eom, K.-H., Hyun, K.-H., Lin, S. and Kim, J.-W.
problems such as foodborne illnesses and (2014) „The Meat Freshness Monitoring System
significant food wastage. For controlling those using the Smart RFID tag‟, International Journal of
Distributed Sensor Networks, 2014, pp. 1–9. doi:
diseases, early detection of foods is important.
10.1155/2014/591812.
Existing methodologies for rapid detection like
mass spectrometry, gas spectrometry, Fourier 5. Gutierrez-Osuna, R. (2002) „Pattern Analysis for
Transform Infrared Spectrometry, and Machine Olfaction: A Review‟, IEEE Sensors
electronic nose can identify the foodborne Journal, 2(3), pp. 189–202. doi:
pathogens. Most of these methods operated 10.1109/jsen.2002.800688.
under laboratory environment. Also, most of
those devices are very complicated and very 6. Hasan, N., Ejaz, N., Ejaz, W. and Kim, H. (2012)
expensive. Due to these issues, regular food „Meat and Fish Freshness Inspection System
inspections occupied under traditional methods Based on Odor Sensing‟, Sensors, 12(12), pp.
15542–15557. doi: 10.3390/s121115542.
such as direct microscopic count and standard
plate count which are time-consuming
methods.
6
452
7. ICMSF - International Commission of
Microbiological Specification (2011)
Microorganisms in foods. (8 Vols). New York:
Springer.
8. Kai Zhang, L. Z. (2014) „Volatile Organic

Compounds as Novel Markers for the Detection
of Bacterial Infections‟, Clinical Microbiology:
Open Access, 03(03). doi: 10.4172/2327-
5073.1000151.
9. Kima, E., Lee, S., Lee, T., Shin, B.J., Lee, J., Byun,
Y.T. and Kim, H.S. (2011) „An Intelligent Real-
Time odor Monitoring System using a Pattern
Extraction Algorithm‟, 2011 IEEE Ninth
International Conference on Dependable, Autonomic
and Secure Computing,doi: 10.1109/dasc.2011.92.
10. Mamat, M., Samad, S. A. and Hannan, M. (2011)

„An Electronic Nose for Reliable Measurement
and Correct Classification of Beverages‟, Sensors,
11(12), pp. 6435–6453. doi: 10.3390/s110606435.
11. Mandal, P.K., Biswas, A.K., Choi, K. and Pal,

U.K. (2011) „Methods for Rapid Detection of
Foodborne Pathogens: An Overview‟, American
Journal of Food Technology, 6(2), pp. 87–102. doi:
10.3923/ajft.2011.87.102.
12. Panigrahi, S., Balasubramanian, S., Gu, H.,

Logue, C.M. and Marchello, M. (2006) „Design
and Development of a Metal Oxide Based
Electronic Nose for Spoilage Classification of
beef‟, Sensors and Actuators B: Chemical, 119(1),
pp. 2–14. doi: 10.1016/j.snb.2005.03.120.
13. Pearce, T. C., Schiffman, S. S. and Nagle, H. T.

(eds.) (2006) Handbook of machine Olfaction:
Electronic nose technology. Germany: Wiley-VCH
Verlag GmbH.
14. Schaller, E., Bosset, J. O. and Escher, F. (1998)

„“Electronic Noses” and their Application to
Food‟, LWT - Food Science and Technology, 31(4),
pp. 305–316. doi: 10.1006/fstl.1998.0376.
15. Srilankabusiness.com. (2017). Food & Beverage

Product Standards in Sri Lanka|Food & Beverage
Standards & Regulations. [online] Available at:
http://www.srilankabusiness.com/exporters/st
andards/food-and-beverages-products-
standards.html [Accessed 12 Sep. 2017].
16. Health.gov.lk. (1980). Regulations under the Food

Act. [online] Available at:
http://www.health.gov.lk/enWeb/FOODWEB
/files/regulations.html [Accessed 24 May 2017].
7
453
pp. [1 -- 10],
[455 462],2017
2017
© The
The Institution
Lanka
Power Electronic Interface of a Solar Driven DC

System
W.W.M.D.B. Wijesinghe, H.M.K.C.B. Wijerathne, B.H.T.S. Gunaratne,

P.J. Binduhewa and J.B. Ekanayake
Abstract: The distributed grid connected PV systems are gradually increasing in Sri Lanka and
around the globe, and will continue in the future. With the rapid development of energy storage
techniques and dc loads, a domestic distribution should comprise of dc and ac multi-ports to integrate
loads and sources in the forms of ac and dc. Such an interface would increase the efficiency by
reducing the number of conversion stages. Further an Energy Management System would help to
coordinate the resources effectively. This paper presents the design of a power electronic interface of a
multi-port distribution board which enables to realize the above functionalities. This multiport power
electronic interface combines the grid and a photovoltaic system along with an enhanced storage unit
as power sources. The loads are connected to the dc and ac ports. The simulation results of the
designed power electronic converters and their controllers will be presented in the full paper.
Keywords: boost converter, bidirectional converter, Grid connected voltage source inverter
1. Introduction side will be handled by the PV panel and it can

supply the other two units as well. The
The modern day energy demand and rapid storage unit will ensure that an uninterruptible
improvements of technology has created a supply is maintained when it is required. Even
question as to whether the power and energy though the initial cost will be on the high side,
is being well managed and optimized to match the maintenance and handling cost will not be
the economy as well. The usage of renewable considerable which supports the economy side
energy sources to supply the demand, mainly as well.
in the means of solar power is becoming the 2. Proposed System
future of power management. On the other
hand, the traditional approach of grid supply The proposed multiport power electronic
has not disappointed the customers much, but interface is shown in Figure 1. Power sources
the next step need to be improved. And coupled to the system are utility supply,
uninterruptible power supply in domestic photovoltaic panels and a battery storage unit.
scale is the issue that needs to be addressed The battery bank enables to store energy so
while the efficiency is maintained at a higher that the stored energy can be utilized when the
level. utility is not available or utility tariff is
The power electronic interface of solar driven relatively high. The loads connected to the
dc system combines all the concerns that the system can be either ac or dc.
modern world is seeking answers for. It
enables the combination of a PV panel, the All the sources and loads are connected to a
grid supply and a storage unit to the same bus dc-link through power electronic converters
bar of 48V, which is then expected to be used connected to each source and load. PV panels
as the supply point for different kinds of ac
Mr.W.W.M.D.B. Wijesinghe, B.Sc. Undergraduate, Faculty of
and dc loads. The use of power electronic Engineering, University of Peradeniya.
interface to perform energy management Ms. B.H.T.S. Gunaratne, B.Sc. Undergraduate, Faculty of
results in number of merits to the individual Mr. H.M.K.C.B. Wijerathne, B.Sc. Undergraduate, Faculty of
user as well as to the entire society in a whole. Engineering, University of Peradeniya.
Dr. P.J. Binduhewa, B.Sc.(Peradeniya), Ph.D. (UK), Senior
Since the usage of power is being optimized,
Lecturer at Dept. of Electrical and Electronic Engineering, UOP.
the overall efficiency will increase while Eng.(Prof.) J.B. Ekanayake, B.Sc. Eng.(Peradeniya), Ph.D.,
providing the user with multiple options in FIEE, FIET, FIESL, CEng (SL &UK), Senior Lecturer at Dept. of
Electrical and Electronic Engineering, University of Peradeniya.
energy distribution. The majority of the utility
455
are connected to the dc-link through a dc-dc Thus a simple boost converter, which is
converter. Utility is coupled using a Voltage shown in Figure 2, is used for this function.
Source Converter (VSC) which is capable of bi-
directional power flow. A bi-directional dc-dc
L
D
converter is used to integrate the battery bank 0.00023mH
to the dc-link. In order to facilitate the

interconnection of ac loads, an inverter is used. PV + + Load
A dc-dc converter can be used to match the dc- Cin

Switch
Co
194.6uF
44.98uF
voltage of the load.
Figure 2 - Boost Converter
The component values and the necessary

parameter values were calculated using the
theoretical equations as given below [8].
…(3)
… (4)
… (5)
Figure 1 - System Diagram
…. (6)
This project aims to evaluate the controllers
associated with the power electronic Function of the controller of the converter
converters in the proposed system. Thus a associated with PV panel is to ensure that the
scaled down grid supply is applied. The Grid photovoltaic panel is operating at the
supply is assumed to be 30 V and the DC bus maximum power point. The controller for the
bar is maintained as 48V. converter was designed to control the input
voltage where the MPPT algorithm was used to
3. Design of the Physical Interface maintain the PV output at the optimum level
under varying conditions. Average equations
3.1 PV connected DC-DC converter of the converter operation were used to obtain
The photovoltaic panel produces a dc output the operating model and the equations
obtained for the plant are given below [8], [11].
thus a dc-dc converter is required to interface
PI controller was established with suitable pole
the PV panels to the dc-link of the proposed
placements. The controller is shown in figure 3.
system. The output power of a PV panel varies
with irradiance and temperature. Thus it is
important to know the ranges of voltage and
current. Equation (1) and (2) present the … (7)
variation of short circuit current and open
circuit voltage of the PV panel with irradiance (8)
and temperature. Isc, Ga, T, Voc, ΔIsc, ΔVoc and
..(9)
Vt are short circuit current, irradiance,
temperature, open circuit voltage, temperature
coefficient of Isc, temperature coefficient of V oc Vin(ref)
+ *50 + +
(Kp) -0.0208 DRCl
and thermal voltage. The „s‟ notation - -
-
represents the parameter value at the standard
702.34 *1000
1/s
Vin G/1+sT
conditions. Vin
ǤǤሺͳሻ Figure 3 - Input Voltage Controller Block

Diagram
ǤǤሺʹሻ
456
3.2 Design of Bi-directional DC-DC mode output voltage has been controlled to
Converter connect the dc bus bar, where the stored
power is need to be used by loads. The plant
A bi-directional dc-dc converter is being used equations for the boost mode were obtained as
in this system in order to connect the battery follows [8], [11].
bank with dc bus bar. The main objectives of
connecting a battery bank are to maintain the
stability of the dc link according to the load
requirement and a continuous power flow of …(14)
the dc link.
...(15)
Buck boost half bridge bi-directional converter
topology as referred in figure 4 was selected to …(16)
acquire this objective [10]. Switch Q1 operates
for the buck mode power flow of the converter
and the Q2 operates for the boost mode For the buck mode operation,
conduction. The switches are controlled by
using a PWM signal. In the buck mode of the
converter, power flow facilitates the battery
bank charging and in the boost mode it
converts to DC link power supply. For the ...(17)
boost mode operation, component value ...(18)
calculations were done by using equations 10-
13 [8]. A similar set of equations can be
obtained by considering the buck mode as
..(19)
well.
The controller of this bi-directional converter is
to control the charging and discharging
current. A PI controller based controller was
implemented for this purpose. The battery
storage may need to regulate the dc-link
voltage when the utility is not available. For
such situations, an outer voltage control loop
is required. Figure 5 and 6 presents the
controller of the inner current controller and
the output voltage controller loops.
Figure 4 - Buck Boost Bi-directional Converter V2

*1000
+
+ *3.652 + +
Iind(ref) N/D DRCl
(Kp)
- -
…(10)
…(11) 1162.5 *1000
1/s
Iind G/1+sT
… (12) V1
… (13)
Figure 5 - Inner Current Loop of the
Converter
Voltage and current controlling of the
converter has done in the both modes. In the PI controlling method was used by taking 25
buck mode, output voltage has been controlled kHz as the fundamental frequency level. Poles
to 24V or for an externally given desired value. were selected accordingly making the outer
Output current has been controlled for the voltage loop poles 10 times slower than the
constant current charging purposes. inner current loop poles. Necessary switching
The plant equations for buck mode, with the signals have provided to make changes within
consideration of the operating point model, the control loops according to the mode being
were obtained as follows [8]. In the boost used in the converter.
457
obtained by varying the magnitude and phase
V1(ref)
+ *0.0265 +
-1
angle between the voltages at the ends of the
(Kp)
- - inductor. A DC link capacitor (Cin) is used to
reduce dc side voltage ripple [2], [7], [8].
2.088
1/s
V1 *1000
Selector Switch
+ *0.241 +
V2(ref)
(Kp)
- - }
*75.87
The instantaneous inverter output voltage is
1/s
V2 *1000 (Kp)
given by [8],
Figure 6 - Outer Voltage Loops of the ….(21)

Converter
Where,
3.3 Design of Grid Connected Voltage E(t) – instantaneous output voltage of
Source Converter the inverter
Ma – modulation index
Vdc – DC bus voltage
The voltage source inverter is the link between
f – grid voltage frequency
the utility and the dc-link. The converter is
capable of bi-directional power flow. Power
Using the above equation Ma value was
will be imported from the utility when the
calculated and following theoretical equations
local generation is no enough and power will
were used to calculate the component values
be exported when local generation is excessive
[2].
than generation.
ሺʹʹሻ
Even though there are several common
topologies for a single phase inverter, full
bridge inverter using sinusoidal PWM …(23)
unipolar voltage switching was selected due to
its low harmonic content compared with
E –RMS value of inverter output
bipolar switching [8]. The schematic diagram
voltage V – RMS value of grid voltage
of the full-bridge inverter is shown in Figure 7
Δ - Phase angle difference between E
&
P – Power rating of inverter
ΔV - DC link voltage ripple
Q1 G1 Q3 G3 fs- Switching frequency
L
+ +
Load
48V Cin
4. Implementation of the Physical
Q2 G2 Q4 G4 Interface
4.1 PV Connected DC-DC Converter

For the development of the prototype,
Figure 7 - Voltage Source Inverter Mitsubishi PVUJ225GA6 panel was selected.
The main specifications of the panel are given
The switches are controlled using PWM
signals. A sinusoidal signal (Vref) is compared in Table 1.
with a high frequency triangular waveform.
The frequency of the sinusoidal signal is equal Table 1- PV panel specifications
to the frequency of output voltage waveform. Parameter Value
The operating pattern of switches are Vmmp 30V
explained in Equation X. The output of the Impp 7.50A
inverter is passed through an inductive filter
Voc 36.4V
(L) to filter out undesired harmonic content
providing the desired sinusoidal waveform at Isc 8.30A
50Hz. The rectifier mode of operation can be Pmpp 225W
458
Since the output of the PV panel was expected 4.2 Bi-directional DC-DC Converter
to vary under practical operation, the range of
It was decided to use two 12 V batteries
variation of the input voltage of the converter
connected in series as the storage method. The
was calculated under limits of irradiance and power rating of the bi-directional converter is
temperature. The temperature range was taken made equal to the power rating of the solar
as 100C to 350C and the irradiance level range converter when designing this prototype
was considered to vary from 600W/m2 to interface. Thus the total power generated from
1400W/m2. The effect of the conditions was PV can be absorbed by the battery storage. The
specifications of the bi-directional dc-dc
calculated using the equations 1and 2 [2]. With
converter is given in table 04.
the system requirements the specifications of
the converter are listed in the table below. The Table 4 - Design specifications of the
open circuit voltage values for extreme worst converter
cases were found to be 33.07V and 41.19V. Parameter Value
using the assumption of maximum power Power rating /(W) 250
point voltage to be 80% of the open circuit High side voltage 48
voltage, the variation of PV panel output /(V)
voltage range was calculated. Table 02 Low side voltage 24
presents the specifications of the boost /(V)
converter. Worst case condition analysis was Switching frequency 25
/(kHz)
performed considering both the limits of the
Inductor current 20% of the inductor
input voltage value and the extreme worse ripple current
case was found to 26.4V where the duty ratio Input / output 1% voltage
was 0.313. voltage ripple
Table 2 - Design specifications of the According design specifications listed in table

converter 04, the two instances of converter power flow,
Parameter Value worst case conditions were reviewed through
Power rating / (W) 250 previous literature studies. By analyzing the
Input voltage / (V) 26.4 – 32.9 battery charging and discharging
Output Voltage / (V) 48 characteristics of lead-acid batteries, two worst
Switching frequency / 25 case scenarios were found as the battery bank
(kHz) voltage varies between 27.6V – 23V. Converter
Inductor current 20% of average design was done according to the worst case
ripple inductor current conditions in order to withstand the variations
Input voltage ripple 1% of rated input and the parameter values are given in table 05.
voltage
Table 5 - Calculated parameters of the
Output voltage ripple 1% of rated output
converter
voltage
The obtained parameter values considering the Buck mode Boost Selected
given equations under theory are listed in operation mode value for
table 03. operatio the
The inductor current ripple was selected to be n converte
20% since improving it further would cause r design
the size of the inductor to be large and C1 226.12 μF 225.6 μF 226.12
μF
lowering the ripple would increase the size of
capacitors.
C2 54.17μF 61.5 μF 61.5 μF
Table 3 - Parameters of the converters
L 185.2 μH 184.39 185.2 μH
Parameter Calculated Value μH
Inductance / (mH) 0.23
Input Capacitance / 44.98
(uF)
Output Capacitance / 194.50
(uF)
459
4.3 Grid connected Voltage Source
Inverter
Table 06 present the specifications of the

voltage source converter to be implemented in
the prototype interface.
Figure 8 - Input Voltage and Output Voltage

Table 6 - Design specifications of the voltage
source inverter
Parameter Value
Power rating /(W) 250

AC side rms voltage 30
/(V)
DC side voltage /(V) 48 Figure 9 - Inductor Current at Open Loop
Switching frequency 5 Condition
/(kHz) Closed loop response of the converter is
Grid voltage 50 obtained by connecting the PV panel to the
frequency /(Hz) input of the converter. Step response of the
controller is shown in Figure 10. It is clear
DC side voltage ripple 1% of the DC that the controller follows the reference
output
within 0.000s.
When calculating parameter values, the phase

angle difference (δ) of the voltages at the ends
of the inductor was set to be 10o and voltage
drop across the inductor was assumed to be
10% of grid voltage.
Figure 10 - Response of Input Voltage to a

Table 7- Calculated parameters of the voltage Step Variation in Reference
source inverter
5.2 Bi-Directional DC-DC Converter
Component Calculated Selected Open-loop operation of the bi-directional
value value for converter was first verified for the two
the operating modes. Figure 11 and 12 shows the
design results for the buck mode. A voltage source
L 2.19 mH - was connected to the low-voltage side and load
to the high voltage side when testing the buck
Cin 57 μH 67 μH
mode. For a low-voltage side 24 V is applied
with 0.5 duty ratio. The corresponding output
voltage is 48 V. The inductor current ripple
5. Simulation Results of the corresponds to 20 %.
proposed system
5.1. PV Connected DC-DC converter

The designed dc-dc converter was simulated
in PSCAD. The open loop results of the
converter are shown in Figure 8 and 9. For
input of 30 V, the converter output voltage is
48 V. The current ripple of inductor current is
around 2 A which corresponds to 20 % ripple.
Thus the desired characteristics are shown by
the converter.
Figure 11- Inductor Current and PWM Signal
460
Figure 15 - Buck Mode Output Voltage
Figure 12 - Output Voltage Variation with Given Reference
Boost mode results are shown in Figure 13 and

14. A voltage source of 48 V was connected to
the high-voltage side of the converter while a
load is connected to the low-voltage side. Duty
ratio was set to 0.5. The corresponding low-
voltage side voltage is 24 V and inductor
current ripple is 20 %.
Figure 16 - Buck Mode Output Current
Variation with Given Reference
Figure 17 - Boost Mode Output Voltage

Variation with Reference
Figure 13 - Inductor Current Variation and 5.3 Grid connected Voltage Source Inverter
PWM Signal
First, the open loop operation of inverter mode
was verified using PSCAD simulator for the
calculated Ma value which is 0.97.
Figure 18 shows the inverter output voltage

waveform which is a square waveform with a
varying duty ratio. Unipolar switching makes
the output waveform pattern much similar to
Figure 14 - Output Voltage that of a sinusoidal waveform.
After the verification of open-loop operating,
the closed loop controller was implemented Figure 20 shows the sinusoidal waveform and
for the bi-directional converter. First the its RMS value obtained after filtering via the
current controller was implemented for the inductive filter.
two operating modes. Figure 15 shows the step
response during buck mode operation. Step
response during the boost mode is shown in
Figure 16. In both cases, the current response is
tracked by the controller.
Then the outer voltage control loops were

developed. Figure 17 shows the step response
Figure 18- Inverter Output Voltage Waveform
of the outer voltage loop for buck and boost
modes. The controller is following the
command. Further the results of the inner loop
and outer loop has verified that the inner loop
is faster than the outer loop in both operating
modes.
461
3. Jasim farhood sultani “Modelling, Design and
Implementation of d-q Control in Single-Phase
Grid-Connected Inverters for Photovoltaic
Systems Used in Domestic Dwellings“Faculty
of technology, de montfort university, leicester,
UK
4. Bhim Singh, Sagar Goel,Chinmay Jain and

Umarshankar Subramanium, ” A Sustainable
Solar Photovoltaic Energy System Interfaced
with Grid-tied Voltage Source Converter for
Power Quality Improvement”,[Online] ,
http://www.researchgate.net/publication3117
31886.
5. Erickson, R. W., and Maksimović, D.,

Fundamentals of power electronics. Norwell,
Figure 19 - Desired Output Voltage and its RMS Mass.: Kluwer Academic, 2001.
Value Variation
6. Colonel W.M.T McLyman, “ Transformer and
Inductor Design Handbook” , Idyllwiled
California USA , ch 8, pp 254
7. Mojgan Hojabri and Mehrdad Hojabri,

“Design, Application and Comparison of
Passive Filters for Three-Phase Grid-Connected
Renewable Energy Systems”
Figure 20 - Inductor Current Waveform 8. NED Mohan, Tore M. Undeland & William P.
Robbins, “Power Electronics, converters
applications & design”, second edition.
6. Conclusion
9. K.Suresh & Dr. R. Arulmozhiyal, “Design and
Implementation of Bi-directional DC-DC
The proposed ac-dc multiport power electronic Converter”,[Online],
interface enables to integrate different energy http://www.scirp.org/journal/cs .
sources and load operated by ac and dc.
This paper presented the design methodology 10. M Stadler, G.Cardoso, S.Mashayekh, T Forget,
of the converters and controlled related to this N.De.Forest, A.Agarwal, and A. Schobein,
“Value Streams in Microgrids”, 2015
multiport power electronics interface. A
prototype system was designed based on the 11. Athimulam Kalirasu, Subharensu Sekar Dash,
theory. Then the system was simulated in ”Simulation of Closed Loop Controlled Boost
EMTDC/PSCAD software. The simulation Converter for Solar Installation”,Serbian
results verified the open-loop and closed-loop Journal of Electrical Engineering, vol. 7, No1,
operating of the power electronic converters May 2010, 121-130
associated with the energy sources of the
interface. 12. Nagisetty Sridhar and R. Kanagaraj “Modeling
As future work, the power electronic and Simulation of Controller for Single Phase
and Three Phase PWM Rectifiers”
converters related to the load side will be
developed with controllers. Then the
performance of the systems as a single unit
will be evaluated.
References
1. Parhizi, S., Lotfi, H., Khodaei, A., and
Bahramirad, S., “State of the Art in Research on
Microgrids: A Review,” IEEE Access, vol. 3, pp.
890–925, 2015.
2. Binduhewa, P. J., “Microsource Interface for a

Microgrid,” The University of Manchester,
2010.
462
pp. [1 -- 10],
[463 468],2017
2017
© The
The Institution
Lanka
Indoor Navigation System with RFID Tags for

Virtually Impaired Individuals
J.M.K.V. Jayalath, D.M.G.C. Jayathilake, M.A.C.I.U. Yapa

S.A.H.A. Suraweera and G.M.R.I. Godaliyadda
Abstract: This paper targets to deliver a comprehensive indoor navigation system

especially for blind and elderly people for navigating safely and efficiently. In order to address the
issue of navigating oneself in complex environments such as shopping complexes, hospitals, offices
etc., we propose deploying RFID tags in a network connecting each and every strategic location in the
area. The user is navigated through these RFID tags via the shortest and least obstructive path
available. Fuzzy logic based deviation correction algorithm is developed to provide required voice
commands to the user guiding him safely from the entrance to the destination. User specific step
detection system, which is a major area in inter-waypoint navigation has been developed to track the
user using his step count. Finally, a cloud system is proposed which can keep track of each and every
user who is wearing the device in a particular environment.
Keywords: Indoor Navigation, RFID tags, step counting, fuzzy logic
1. Introduction The discussed paper proposes RFID tags, a

cheap passive element which can be deployed
Navigating in an indoor environment has all over the building for a very low cost. These
become a major issue for visually impaired RFID tags form a local network in the building
individuals as the most popular navigation connecting almost all possible destinations that
system, global positioning system (GPS) a single user is able to visit. A device which is
malfunctions inside buildings [1]. Generally, capable of providing voice commands to the
there might be several floors connecting user and guide him to his desired destination
several building sections where the user will and a white cane which can detect RFID tags
not be localized by GPS. Furthermore, in most will be the proposed user carrying equipment.
of massive structures, people experience less A step recognition and counting system is
internet connectivity than outside due to proposed which can identify a stepping action
attenuated signals. of the user out of his other body movements. A
On the other hand, people who have less eye deviation correction technique is proposed
sight or old enough to get lost inside buildings incorporating fuzzy logic to provide more
would still want to walk around and get their linguistic and accurate results that will
works done by themselves. For an example A smoothly correct the user‟s deviation from the
visually impaired person would like to go to desired path without giving sharp turn
the eye clinic which is located at south wing commands time to time. Further, this device
ground floor which he cannot reach without Eng. J.M.K.V. Jayalath, BSc.Eng(Peradeniya)
getting help from someone around. In order to Automation Engineer, MAS Kreeda.
Eng. D.M.G.C. Jayathilake,
address this general problems, people have
BSc.Eng(Peradeniya).Engineer, IP Core Network
proposed various indoor navigating Operations, Dialog Axiata PLC.
techniques such as Ultrasonic Beacons [2],
Eng. M.A.C.I.U. Yapa, BSc.Eng(Peradeniya).Electrical
Ultra-wideband, and indoor navigation with Engineer, ZPMC Lanka Company(PVT.) Limited.
WIFI and Indoor localization [3] using Dr. S.A.H.A. Suraweera, BSc.Eng.(Peradeniya), Ph.D.
KINECT sensors [4] etc. But all these methods (Monash University), Senior Lecturer, Department of
use active equipment which may affect the Electrical and Electronics Engineering, Faculty of
effectiveness and the practicality of the
navigation system when the power Eng. (Dr.) G.M.R.I. Godaliyadda
BSc.Eng.(Peradeniya), Ph.D. (National university of
consumption factor is considered.
Singapore), Senior Lecturer, Department of Electrical and
Electronics Engineering, Faculty of Engineering,
University of Peradeniya.
463
connects to a common local network where design for this kind of movements of 3 axes
someone can monitor the user throughout [8],[10].
his/her journey.
3. System Description
System initiates at the first point when a user
Our proposed system is similar to the system step into the building from any given entrance.
proposed by [5] But the novel use of deviation This happens when the cane detects a floor
correction method using Fuzzy logic and the mounted RFID tag at the entrance. It is
step recognition technique developed using proposed to use few RFID tags placed in an
standard deviation of foot acceleration signals array so that the user will not miss any
made the human navigation smoothly and waypoint. Passive RFID tags has a maximum
accurately.The limitations of the existing detection range of around 10 mm.
system would be the lack of deviation Then the user receives a voice command
correction which makes it necessary to use a through his earphone to select the destination
large amount of RFID tags.But in the proposed he wants to go. After his selection using up
system, the navigation primarily depends on down and enter buttons, the system calculates
deviation correction using fuzzy logic and the best least distraction path to his
distance estimation with the use of step destination.
recognition algorithm which drastically There are two basic concepts we developed in
decreases the number of RFID tags required. this system, a technique which uses fuzzy logic
Furthermore, in the existing system a network to minimize the deviation of the user while
lag causes a difficulty in Navigation due to walking on the desired path and a specific
usage of GPRS signals, but in the proposed feature which is extracted from the limb
system, this issue is eliminated by Mainly acceleration signals. This feature helps to
relying on the deviation correction and differentiate leg movements with step and
distance estimation for navigation instead of random motion.
relying on data obtained from the server In the implementation, a centralized server
which communicate with the wearable device
Most of the methods have been developed to
via WIFI network in the building is proposed
identify steps using mobile phone‟s inbuilt
to accomplish the following criteria.
gyro sensor and accelerometer [6] to use in
health purposes such as identifying girt
 Keep track on all users in the building.
patterns etc. [7], [8] Since Mobile phones are
 Perform the necessary communication
normally carry in pockets, we could not get a
with the user
better signal enough to count steps in a
 Running Dijkstra‟s algorithm.
reasonable accuracy.
Some of those methods analyzed singles

directly come from the foot by placing an
accelerometer on the foot top [8]. Such
methods can be categorized in to two as
analytical methods and mathematical
methods.
In mathematical prospective, researches were

trying to model the foot dynamics in to a
mathematical model. And trying to generate a
foot stepping signal using the mathematical
model and then match the template with the
original signal [9]. Researches have also tried
such as window overlapping method,
Figure 1 – RFID tag placement at every
Standard deviation based methods [6]
strategic location in each floor. Positions of
Spectrogram analysis and template matching
all the users can also be demonstrated in the
to count the steps accurately. Further, several server room interface
experiments were made to optimize a filter to
464
4. Deviation Correction
Technique Using Fuzzy Logic
Inter waypoint navigation system using
direction and distance would not be robust
enough to assist the user in navigating
between waypoints due to environment
complexities. Therefore, a new algorithm is
proposed to guide the user in between
waypoints using the perpendicular threshold
method.
Figure 4 - Membership function for distance

Moved along the desired path
during the implementation of the proposed

method, turn commands such as turn left,
Turn Right,turn left a lot etc. were issued not
respective to some hard coded parameters but
in a probabilistic way which improve the
accuracy of the system. The commands such as
proceed forward, turn right and turn left were
issued all the time but commands such as turn
Figure 2 – The perpendicular threshold
right a lot and turn left a lot were issued very
method with two parameters
rarely proving the accuracy of the system.
After implementing the perpendicular
Table 1 – Rule base for the fuzzy logic
threshold method, it was found that the
control
accuracy of the system varies drastically with
the walking pattern of the users. To remedy Rule Base
that issue and increase the adaptability of the LL ML Z MR LR
guiding system, fuzzy logic is proposed to be VS TRL TR - TL TLL
incorporated to give more linguistic and S TR - - - TL
probabilistic commands to user rather than M TR - - - -
straightforward turn commands. L TRL TR - TL TLL
For the Fuzzy control system, the two inputs;
Distance travelled along the desired path and TRL – Turn Right a lot TR – Turn
the distance travelled perpendicular path is Right
obtained. Distance travelled along the desired TLL – Turn Left a lot TL – Turn Left
path is obtained as a factor of the distance
between two RFID waypoints which improves 5. Step Identification Using Moving
the system‟s adaptability to user‟s walking standard deviation
pattern.
In statistics, standard deviation is a measure
that is used to quantify the amount of
variation of a set of data from the mean. Low
standard deviation indicates that the data
points tend to be close to the mean while high
standard deviation indicates that the data
points are spread out over a wider range of
values. This is a useful property to
differentiate data from „moving‟ segments to
„not moving‟ segments. Because when the leg
moves, the data points vary in a high
Figure 3 - Membership function for distance frequency. When the standard deviation is
Moved perpendicular to desired path considered, getting a high value can be
465
predicted while low value is predicted for 5.1 Calibration stage
standard deviation when the user is not
From the calibration it is expected to acquire
moving [6].
few details which are unique from person to
First, accelerometer data from low cost gyro person. First the user is asked to walk up and
sensor was collected and filtered against noise down once, first a known number of steps and
[11],[12]. Consider the standard deviation in a then a known distance. Then the program is
non-overlapping moving window, consider executed and in the first turn, the standard
the size of the window 30. The whole data set deviation of the whole walking signal and the
is divided to non -overlapping windows of best value for a „K‟ constant (which will be
size 30 and e standard deviation in each described later) is calculated. In the second
window can be predicted as below. turn the user is asked to walk down a known
distance. It may be 10-15 meters. Program will
calculate the average distance of a step and
will use it to measure the distance.
5.2 Unique constant for each and every user
The K constant is a portion of standard

deviation which will be used to compare with
the real time moving standard deviation
would be the unique constant for a particular
person. As an example, if the user is a low
weight, thin person, the stepping signal will
Figure 5 – Filtered foot acceleration data contain much noisy data where the standard
along z axis (top)and Plot of standard deviation is relatively high. Hence the K value
deviation on Acceleration data set along z will be larger. If the user is an obese person
axis (bottom) who has short steps, a less noisy stepping
signal s expected which lead to a low K value.
As the graph depicts, the blue plot is clustered
properly than the red one (filtered acceleration 5.3 Decision making criteria
signal) proving the expected results. Through
In the proposed work the different standard
the experiment, the two mountain type
deviation can be calculated as below using the
segments gathered together can be identified
data discussed above.
as actual steps. In other words, „moving‟
segments. There is a space between those two ∑
mountains which represents the interval √ … (2)
between steps and is called „not moving‟
segments. First the average standard deviation of 3 sets is
obtained from set one to set 3.
Calculating moving standard deviation:
If the average of standard deviation is greater
Standard deviation of the window I, than , it is considered to be a
candidate step.
Filtered Acc data on X For i=1

axis
Standard Deviation
Step Detection
… (3)
Candidate step if;
… (4)
Figure 6 – Generation of candidate And continue till i = m.

steps
In the proposed work, if it is followed the
∑ same methodology with all the three axes, the
√∑ …. (1) parameters
466
people. Some paths are too dangerous
for blind people to walk. Sometimes,
can be obtained. blind people walking through normal
5.4 Voting people disturb the community. So,
weight between two RFID points is
A voting method is proposed for the determined considering all those facts.
candidate plots generated earlier to get the
final decision to the problem. As a first step  Objects on paths should be avoided.
AND rule as a voting strategy is considered.
After determining weights with above
But it is proposed to use weighted average
mentioned facts, these weights are included to
techniques to vote for a step.
a matrix called "weighted matrix". Then
are the voting of axis weighted matrix is used in Djkstra's algorithm
respectively with source as the input and destination as the
output.
For the demonstration of applicability of

Djkstras' algorithm in RFID based navigation
… (5)
system, we have designed an interface. Source
As an example, if we have given same RFID point and the destination RFID point is
weightage for each axis and the candidate given as the input and the most appropriate
plots vote for a step as, 2 axis vote for a step, path is displayed on the map.
one is not:
6.1 Experiment on Step counting and results
..… (6) This method is tested using randomly selected
7 men and 2 women and they were asked to
Decision- Step is not identified
follow the procedure and collected the results.
It was observed that the user walks fast the
system found it difficult to recognize the steps
Every axis vote for a step: differently because the variation curves
segment together. Furthermore AND voting
mechanism serves its best as three axis(x,y,z)
…(7) are equally informative.
Decision- Step has identified Table 2 – Experiment Results
6. Determining the weights of paths Walking Actual Counted Error

Pattern Steps Steps
According to the above explanation on Normal 10 10 0
Djkstra's algorithm, weight should be Walking 12 12 0
determined between RFID tag points. Weights 14 14 0
of the paths were determined considering 16 16 0
several facts. 18 18 0
20 20 0
 Distance between two RFID tags: 22 22 0
24 21 -3
This is the main factor to be
26 25 -1
considered to determine the weight. 28 27 -1
The blind person should have to be 30 26 -4
navigated through the shortest path as Walking in 20 23 3
possible. Therefore, more weight is a staircase 15 18 3
put on distance. Walking at 20 22 2
human
 Allowable and not allowable paths:
disturbance
In an indoor environment some paths
are not allowed to walk for blind
467
7. Conclusion Applications”,MDPI AG, Basel,
Switzerland,2012,doi:10.3390/s120201437
The proposed work discusses the
implementation of a RFID based navigation 5. Sakmongkon Alessandro Chumkamon et al.,
“A Blind Navigation Systemb Using RFID for
system which can be used by blind people.
Indoor Environments,in ECTI-CON 5th
The system contains RFID tags placed in conference, 2008, DOI:
strategic locations. A shortest path calculating 10.1109/ECTICON.2008.4600543
algorithm and inter waypoint navigation
principle using step counting and deviation 6. Aylar Seyrafi,”Real time Automatic Step
Detection in the Three
techniques. The key novelties of the work are
DimensionalAccelerometer Data ”, M.S. thesis,
the step counting algorithm and the deviation Appl. Signal Pross. Dept., Bleking Institute of
correction algorithm using fuzzy logic and the technology,Karlskrona, SE , 2009
step distance estimation algorithm. The
proposed step counting algorithm uses the 7. Joaquín Torres-Sospedra et al.,“The Smartphone-
acceleration data of the foot in all three Based Offline Indoor Location Competition at IPIN
2016: Analysis and Future Work”,MDPI AG,
perpendicular directions and provides a final
Basel, Switzerland,2017,doi:10.3390/s17030557
result using a voting mechanism. The step
distance measuring algorithm uses this output 8. Akram Bayat et al.,”A Study on Human Activity
and the data obtained from user‟s walking Recognition UsingAccelerometer Data from
pattern to estimate the distance travelled by Smartphone”,International Conference on
Mobile Systems and Pervasive
the user. Once the distance is estimated, the
Computing,Ontario, CA, 2014, pp 3-5.
user‟s location is obtained with the direction
travelled and Turn commands are issued to 9. Christopher M. Bishop, “Pattern Recognition and
correct the deviation of the user from the Machine Learning”, New York, Springer
desired path. The deviation control is done Science+Business Media,2006, ch. 7, sec. 1.
pp.359-370
through a fuzzy control system which adapts
to the human nature of the user which 10. Xiaoping Yun, Eric R. Bachmann,“Self-
improves the accuracy and overall reliability contained Position Tracking ofHumanMovements
of the system. Using these three novel Using Small Magnetic Sensor Modules”, IEEE
concepts, the accuracy of the system was International conference onRobotics and
Automation, Roma, Italy,2007
improved immensely and after actual
implementation of the system at the shopping 11. Yongjian Zhang et al.,(2017 April 28), “A Novel
mall, Kandy City Center. The system proved MEMS Gyro North Finder Design Based on the
to be a complete indoor navigation system for Rotation Modulation Technique”[Online],
visually impaired individuals. Available: http://www.mdpi.com/1424-
8220/17/5/973
Reference
12. Ryo Takeda et al.,(2014 Dec 5), “Drift Removal
1. Babu R and Wang J,”Real-time data analysis of for Improving the Accuracy of Gait Parameters
ultra-tight GPS/INS integration”,Springer- Using Wearable Sensor Systems[Online],
Verlag,2008,doi:10.1007/s10291-008-0097-9 Available: http://www.mdpi.com/1424-
8220/14/12/23230/htm
2. Rohan Kapoor et al.,(2016 Oct 8), “A Novel 3D
Multilateration Sensor Using Distributed
Ultrasonic Beacons for Indoor
Navigation”[Online], Available:
www.mdpi.com/1424-8220/16/10/1637
3. Kai Dong et al.,”Canoe:An Autonomous

Infrastructure-Free Indoor Navigation
System”,MDPI AG, Basel,
Switzerland,2017,doi:10.3390/s17050996
4. Kourosh Khoshelham and Sander Oude

Elberink,“Accuracy and Resolution of Kinect
Depth Data for Indoor Mapping
468
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[469 473],2017
2017
The Institution
Lanka
Energy Efficiency Benchmarking of Pumps

in Water and Wastewater Industries in Sri Lanka
B.G.A. Dinesh
Abstract: Pumping of public portable water in Sri Lanka constitutes the largest power
consumption in the country, 237 million kWh being utilized annually at present levels, but growing at
a rapid rate. The overall efficiency of pumps bears a direct correlation to the efficient use of electric
power. According to the energy audits carried out in the field, there is at least 20% economically
redeemable energy wastage by conservative estimation, while actual loss percentage found was close
to 30%.
To gauge the operating efficiency of pumps, there have to be benchmarks against which to compare
actual efficiency and find out wastages of energy. Benchmarking of pumps poses a challenge due to
the wide-ranging sizes, ranges of flow & head of pumps. Size of pumps ranges up to 850 kW, head up
to 180 m flow rate up to 4,300 m3/h.
The achievable efficiency of pumps is directly related to the pump type, specific speed and similar
parameters. There are widely published data on the empirical relationship among pump types,
specific speed and related pump parameters.
Therefore, an empirical system of determination of efficiency benchmarking was developed and

implemented which is undergoing field verification at the moment. The system is based on empirical
correlation data published by reputed institutions supplemented by methodology devised by us.
Specific Energy Consumption (SEC) was calculated and used as a final extended parameter of
comparison. Benchmark SEC was arrived at, using the empirical method of determination explained
in this paper. From above SEC figures, it can be determined whether any pump is performing below
acceptable benchmark levels.
This exercise needs the ongoing validation of the benchmarks against best efficiencies of pumps
available on the market and results of the actual SEC figures found in the studies.
This paper summarizes the approach, formulation of methodology and findings in the field on the
topic of this paper.
Keywords: Pump Efficiency, Motor Efficiency, Pump Specific Speed, Rotodynamic Pumps
Characteristics, Specific Energy Consumption
1. Introduction the island, analyzing the study data and finding

the causes of inefficiencies. Depending on the
Conveyance of water and wastewater in the nature and the extent of the causes of
National Water Supply & Drainage Board inefficiencies, appropriate remedial actions
(NWSDB), Sri Lanka is mainly carried out by were carried out.
pumping. Pumps used are of a rotodynamic
type. NWSDB annually consumes 237 Million The main cause of the energy wastage in
kWh of electricity for the pumping requirement pumping is the lower efficiencies at which the
of the water and wastewater systems. With the pumps are operated. This is found out in the
new mandatory regulations regarding energy energy audits carried out in the pumping
efficiency coming into effect in Sri Lanka in the stations.
recent past, as well as the awareness created
regarding the efficiency problems that might
exist in the NWSDB pumping systems, NWSDB Eng. B.G.A. Dinesh, AMIE(SL), M.Sc.(Sustainable Energy
undertook a study programme. The Engineering, KTH-Sweden), P.G. Dip. (Energy Technology-
UOM), B.Sc.Eng.(Mechanical-UOM), Chief Engineer
programme mainly consisted of carrying out (M&E), RSC(S), NWSDB.
energy audits at the water supply systems in
1
469 ENGINEER
Pumps in the water and wastewater systems of carried out to find out the causes. Subsequently,
the NWSDB, vary in size from 1.5 kW to 850 the remedial solutions can be determined for
kW. Owing to the inherent characteristics of the implementation and saving energy.
pumps, their achievable highest efficiency
depends on the size and several other factors. 2. Methodology
There is no reliable way to gauge the acceptable The efficiency of a pump is expressed in
efficiency of the pumps to determine whether different forms such as Overall Pump Efficiency
pumps are wasting energy. and Pump Hydraulic Efficiency. Motor
efficiency also comes into the picture when
As a redress to this problem, NWSDB determining these pump efficiencies. Any
commenced establishing an effective pump will have a single operating point where
benchmarking program. An empirical system it exhibits the maximum efficiency of the
was established based on data published by particular pump, in its range of efficiencies
reputed bodies to determine the empirically (Figure 1). An operating point is any point on
possible highest efficiency for the pumps the Head vs Flow rate curve on which the
operating in the field. pump operation range lies.
A comparison was made between the above The maximum efficiency point is termed
efficiency and the operating efficiency of the Guaranteed Duty Point in the industry. At
pumps in the field. This exercise facilitated in operating points away from this point on either
the determination of whether the pumps are side will reduce the efficiency, thus increasing
working at the highest possible efficiency or the wastage of energy.
wasting energy by working at sub-optimal
efficiencies. Therefore in an effort to reduce energy wastage
or to optimize the efficiency, we must make
In case they are working at sub-optimal sure that the pumps are functioning at the
efficiencies, further investigations can be guaranteed duty point or as close to it as
possible.
Figure 1 – Typical manufacturer published characteristics of a pump

ENGINEER 2
470
Overall pump efficiency defines how efficiently Instantaneous pump efficiency is directly
the pump unit uses the electrical power drawn related to the instantaneous flow rate through
from the utility grid in the pumping operation it. This is evident from the manufacturer’s
which is given by, characteristic curves issued for the performance
of any particular pump (Figure 1).
(1)
At the same time, rotodynamic pump exhibits
Where, an inherent design characteristic termed
Specific Speed which is directly related to the
achievable efficiency as well as defining the
configuration of the pump structure within the
rotodynamic category.
When the specific speed of a particular pump is

determined, it is possible to find out the
maximum achievable efficiency for that pump
by using the flow rate through the pump and
empirical data published by a recognized body.
In our study, we used the data published by the
Hydraulic Institute (HI), USA, and McGraw
Motor efficiency deterioration over time is

rather slow compared to the deterioration of
the pump efficiency due to the fact that wear
and tear phenomena associated with pumps are
relatively absent. Similarly, motors are built
with narrow and accurate efficiency tolerances.
However, in this benchmarking exercise, we
also determined the actual operating efficiency
of the motors. For this, we use typical motor
characteristics published by IEC 60034-30 (2008) Figure 2 –Hydraulic efficiency vs Specific speed
(Table 1).
Table 1 – Table with efficiency classes: IE 60034-30 (2008)
3
471 ENGINEER
Hill (Figure 2). …(3)
Equation for the specific speed is,

To make corrections for the actual operating
…(2) efficiency of the motor due to the extent of
instantaneous under loading, Load Factor (LF)
Where, of the motor has to be determined.
Equation for the Load Factor is,
…(4)
Where,
Motor rated power is the rating mentioned on

the manufacturer’s data plate on the motor.
Therefore, the achievable optimal efficiency of a Achievable optimal power consumption (OPC)
pump under particular operating conditions is of the motor was determined by the following
dependent on the actual flow rate through it formula.
and the specific speed of the particular pump.
…(5)
In this programme, we also determined an
extended parameter termed Specific Energy
Consumption (SEC), which is the total amount Benchmark specific energy consumption
of energy consumed per one cubic meter of (BSEC) is determined from the following
water pumped. It is obvious that higher the equation,
efficiency figure for a pump, lower the SEC,
which is the situation that we are striving for.
…(6)
The objective of this study is to benchmark
efficiency of the NWSDB pumps using the
empirical method described and at the same This is the maximum allowable energy
time to find the actual operating efficiencies of consumption by a given pump to deliver 1 m3
the pumps. If the actual operating efficiency is of water under the operating conditions
lower than the benchmark, then it is subjected encountered during the study.
to further investigation action as well as
remedial measures. During the programme, NWSDB has
benchmarked about 500 pumps and compared
As the core of the methodology, the benchmark the actual operating efficiencies of about 200
efficiency or the achievable optimal efficiencies pumps.
were determined.
3. Results
The first step of the process is to determine the
Specific Speed of the pump. Design or rated
values of the Q and H are used in the From the results, it is evident that 127 nos.
calculation. These Q and H values are the pumps show excessive SEC figures over the
values corresponding to the guaranteed duty benchmark SEC’s. These pumps are clearly
point. using excessive power than necessary; are
wasting energy.
Once the specific speed is calculated, the
achievable optimal hydraulic efficiency of a The energy wasting pumps should be further
particular pump is determined from the HI investigated with guaranteed efficiencies of
diagram (Figure 2). similar pumps in the market as an additional
check on the achievable efficiency. Although
Next, the actual power drawn by the pump is this is an essential step of the analysis, it is yet
determined by the following formula, to commence. After its commencement,
changing or improving efficiency data will be
ENGINEER 4
472
used to modify the HI diagram for achievable accurate. Also, the tool must be subjected to
efficiencies. This will allow NWSDB to provide any necessary changes depending on the
detailed analysis on this aspect as well. verification process. This must be an ongoing
process to capture the technological and other
4. Discussion and Constraints changes in the market.
Since this is mainly an empirical exercise, the References

results and deductions from the data must be
compared with the achievable efficiency of 1. Ebara Pump system Engineering Handbook:
similar pumps of the same specific speed range. Ebara Corporation, 2001
This might reveal necessary adjustments to HI
data used in this programme or elsewhere. 2. Garr M. Jones , PE : Pumping Station Design –
Third Edition , 2006
Pumps in the same specific speed range and
3. Hydraulic Institute Standards for Centrifugal,
operating in the same Head/Flow Rate ranges rotary and reciprocating pumps: fourteenth
may show varying efficiencies because the edition -1983
detailed design of the pumps varies from
manufacturer to manufacturer. Some 4. Igor J. Karassik , Joseph P. Messina , Paul Cooper ,
manufacturers are able to achieve comparably Charles C. Heald (2004) : Pump Hand Book –
higher efficiencies due to innovativeness and Fourth Edition , 2008
their R&D achievements. Therefore, the
5. Pumping Station Engineering Handbook: Japan
adjustment process must be a continuous
Association of Agricultural Engineering Enterprises
process whenever necessary. – Tokyo , 1991
A considerable problem in the results is the

inconsistent results are generated with respect
to the SEC value. There are cases when SEC
should show higher values theoretically,
favourable results or lower SEC’s result for
example, when operating well away from the
guaranteed optimal efficiency point, mostly on
the higher flow rate side. This is due to the non-
linear, different behaviour of the characteristics
of the power, H – Q etc. of the pumps, which is
again the result of marginal design
improvements of similar pumps among
different manufacturers. This peculiarity is
interpreted case by case with the actual
characteristics of the particular pumps.
5. Conclusions
The programme is greatly useful as an indicator

to identify the possible energy wasting pumps.
Thus this tool is very useful in screening
problematic pumps. However, once identified,
they must be subjected to further
investigations, including an Investment Grade
Energy Audit to find out the status of the
performance of the pump. Depending on the
results of such studies, remedial action could be
taken, if necessary.
It must also ensure to compare the achievable

efficiency level of compatible pumps from
similar manufacturers in the market, in order to
validate or calibrate the HI Efficiency diagram
or other aspects of this tool to make it more
5
473 ENGINEER
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[475 483],2017
2017
The Institution
Lanka
Feasibility of Hybrid Reactors for Removal of Heavy

Metals in Wastewater
G.S. Nallaperuma and W.M.K.R.T.W. Bandara
Abstract: The research carried out for the removal of heavy metals in wastewater is very much
significant in the field of Environmental Engineering and Science. Although there are a number of
chemical and physical treatment processes to remove heavy metals efficiently from wastewater, the
drawbacks like high operational cost and excess sludge production limit the use of those methods.
However, the biological treatment methods are still economical because the combination of anaerobic
and aerobic treatment methods can be used more efficiently since they have a number of advantages
like low operational cost, less sludge production and energy through gas production. Therefore, an
Up-flow Anaerobic Sludge Blanket Reactor (UASB) and an activated sludge process were combined to
form a hybrid reactor for the removal of heavy metals with less sludge. The hybrid system was
operated with a flow rate of 20 ml/min and an HRT of 7hrs in the UASB reactor. The hybrid reactor
was operated 12 times with the influent having 12 different characteristics as follows: each of four
heavy metals, namely Pb, Cd, Zn and Cu contained wastewater in three different concentrations. The
average COD removal efficiency was 84%. Heavy metal removal efficiencies achieved for Pb and Cu
were above 80%, for Cd is above 75%, and Zn removal efficiency was above 90%. Through this study
it was found that the hybrid reactor is feasible for the removal of heavy metals in wastewater and the
heavy metal removal from the wastewater increases in the order of Cd<Cu<Pb<Zn. The initial heavy
metal concentration is directly proportional to the removal efficiency. The same study can be carried
out to investigate the technical feasibility of the hybrid reactor to treat the actual wastewater
containing heavy metals.
Keywords: Hybrid rector, UASB reactor, activated sludge process, HRT, COD removal, heavy
metal removal, sludge volume
1. Introduction
Nowadays heavy metals are the most popular
It is known that the elements with an atomic environmental pollutants and create most
weight between 63.5 and 200.6 and a specific serious environmental and health [4].
weight of more than 5.0 are heavy metals [1]. It Therefore, the removal of toxic heavy metals is
is well recognized that the industrialization has significant in order to protect both the people
resulted in the collection and discharge of and the environment. The methods available
heavy metals into the environment. The for the removal of heavy metals in wastewater
common heavy metal pollutants in wastewater can be categorized into three sections like
are Cadmium (Cd), Chromium (Cr), Copper biological, chemical and physical. Currently,
(Cu), Iron (Fe), Lead (Pb), Nickel (Ni), Zinc the physical and/or chemical processes are the
(Zn), and Arsenic (As). Zn, Cu, and Ni have major processes of industrial wastewater
toxic effects in both very low and very high treatment.
concentrations [2]. Heavy metals accumulate in
living organisms since they are not
biodegradable as organic compounds. Most of G. S. Nallaperuma, B.Sc.Eng.(Ruhuna), Civil Engineer,
the heavy metals present in the environment Sierra Construction Pvt Ltd, Colombo 6.
are toxic or carcinogenic. As an example zinc
Dr. W. M. K. R. T. W. Bandara, Ph.D.(Hokkaido-Japan),
cause severe health problems like stomach M.Sc (Hokkaido-Japan), B.Sc.Eng. (Peradeniya, Senior
cramps, vomiting, skin irritations and anaemia Lecturer, Department of Civil and Environmental
[3]. Similarly, the U.S. Environmental Engineering, Faculty of Engineering, University of Ruhuna.
Protection Agency says that the Cadmium is a
probable human carcinogen.
475
Chemical precipitation, coagulation, solvent greater demand for alternative energy sources
extraction comes under chemical treatment as a result of increased demand for energy with
methods while ultra-filtration, electrolytic the increasing of cost of available fuels. Due to
processes, and reverse osmosis, oxidation with the rapid industrialization, this resulted large
ozone/hydrogen peroxide, membrane quantity of wastewater with high organic
filtration, ion exchange, photocatalytic content. Therefore, the primary biogas which is
degradation, and adsorption comes under produced as a result of treating this waste
physical methods [5]. Many of these methods water anaerobically will be one of solution for
are not popular nowadays due to the high cost growing interest in alternative energy sources.
and disposal problems. Therefore, these
methods for treating wastewater have not been Anaerobic process does not require aeration.
widely applied at large scale. However, at Thereby the cost associated with equipment,
present, there is no any single process capable maintenance and energy consumption is low in
of sufficient treatment due to complexity in the anaerobic treatment when compare with
effluents. Therefore, a combination of different aerobic treatment [8] due to above benefits,
processes is often used to achieve the best water wastewater treatment process has forwarded
quality in the most economical way [5]. towards the anaerobic treatment. However,
anaerobic processes are not popular among
When compared with physical and chemical treatment plants used for treatment of
treatment processes, Biological treatment is the municipal wastewater. Reasons for the
most economical processes. Biodegradation by reluctance in the implementation of anaerobic
microorganisms such as bacteria, yeasts, algae processes are mainly the spreading of
and fungi degrade various pollutants such as unpleasant odour, unstable behaviour of
hydrocarbons, Pharmaceutical substances and reactors and operation difficulties with the high
metals [5]. Biological treatment mainly divides load rate variations [6]. The effluent quality is
into two groups, aerobic and anaerobic also less in anaerobic treatment. As wastewater
treatment. The main units of most wastewater cannot be treated to low chemical oxygen
treatment plants are aerobic and anaerobic. demand level through anaerobic treatment,
Anaerobic processes produce biogas as a by- quality of the effluent is low. However aerobic
product which can be used to generate energy treatment is capable of producing a good
and produces less sludge other than in aerobic quality effluent without an unpleasant odour.
systems. However, aerobic processes have Therefore, the combination of anaerobic and
disadvantages like high energy requirement for aerobic system was used in this research to
the aeration to maintain the dissolved O₂ level limit the drawbacks in each process in a
which is necessary to maintain aerobic successful way.
condition and at the same time produce excess
sludge which requires handling, treatment and
disposal [6]. 2. Literature Review
In anaerobic treatment, the decomposition of In combined anaerobic-aerobic treatment,
organic and inorganic matter occurs in the anaerobic digester is used to treat the influent
absence of molecular oxygen [7]. They tend to of the aerobic treatment. It means that
produce sludge only 5-10% when compared anaerobic digester acts as a pre-treatment for
with the aerobic process due to many reasons aerobic influent. Therefore, oxygen requirement
[8]. The main reason is the energy available for in the aerobic system for degradation and
the microorganisms to create biomass is less nitrification of remaining organic matter from
due to the energy available in the wastewater is the anaerobic digester effluent is low due to the
preserved in methane gas. This ultimately large fraction of organic matter is degraded
results in less sludge. This saves a significant earlier from the anaerobic digester. Also, this
amount of cost associated with the sludge system can be used to prevent the separate
disposal. Other than that CO₂ and CH4 are sludge stabilization.
produced as a result of the breakdown of
organic molecules by microorganisms. Eighty Less sludge production, low chemical
percent (80%) of biogas which produced consumption and operational simplicity with
through anaerobic treatment is CH4 and minimum equipment. However, anaerobic–
remainder is CO₂. The advantage is that CH4 is aerobic systems have many limitations such as
a source of energy which is the primary biogas high HRT, difficulties in biogas collection and
produced by anaerobic treatment [9]. There is a high space requirement.
476
Anaerobic–aerobic treatments have more sewage treatment plants were monitored for
attention than the other combinations the heavy metal removal efficiency using
throughout the research history since it has activated sludge process. Under this study, the
significant benefits like less energy removal efficiencies of Cd, Cr, Cu, Zn, and Pb
requirement. These limitations can be from each plant were monitored. According to
successfully avoiding by the use of developed the results a high removal efficiency of Cr, Cu,
high rate bioreactors which increase removal of and Pb were identified in secondary treatment
organic compound at a shorter HRT and processes than in primary processes.
produce high methane yields for biogas Chromium was highly removed during
production. In order to limit the problems aeration in the secondary process while copper
arising with respect to space, odour and was strongly adsorbed by the microbial floc,
minimal sludge production, necessary steps and lead was removed more efficiently because
have been implemented by combining the of increased settling time and larger particle
aerobic- anaerobic process in a single size in secondary treatment. Zn removal
bioreactor. It is expected that these bioreactors efficiency was high and equal from all plant
will be able to treat a wide range of municipal types. However, the removal of Cd could not be
and industrial wastewater which contains high made during this study [12].
organic strength with the advantages like
simplicity, effective cost, production of Similarly, Cheng carried out an investigation to
renewable energy and high treatment check whether the sludge is capable of
efficiency. However, large scale removing soluble metal species from
implementation of the integrated bioreactors is wastewater and the attraction of the biological
need for the industry and operation and mass towards the selected heavy metals. There
performance of these reactors should be he found that, at lower metal concentrations,
investigating on a larger scale [10]. metal is taken up by the biofloc through the
formation of metal organic complexes. At
Use of high rate bioreactors in anaerobic– higher metal concentrations, metal ion
aerobic systems tends to achieve more than 70% precipitation from solution may occur in
of COD removal within a short HRT. Hence, addition to sludge uptake.
the hybrid treatment can be used very
effectively and efficiently to treat both The metal uptake by the biomass also depends
industrial and municipal wastewater [10]. on several factors, including pH and the
Treating high strength industrial wastewaters concentration of organic matter and metals
using anaerobic–aerobic processes couples the present in the system. Higher initial
benefit of anaerobic digestion by producing concentrations of metal ions or sludge increase
biogas while removing COD and volatile the overall uptake. In general, the uptake
suspended solid (VSS) through aerobic capacity increases with increasing pH, up to a
digestion. Therefore, the anaerobic–aerobic value at which metal hydroxide precipitation
processes are operationally and economically occurs. Among the metals studied, the
advantageous in the treatment of industrial preferred order of uptake by activated sludge
wastewaters. [11]. was found to be in the sequence of Pb > Cu >
Cd > Ni [13].
The reason for that is when using a single
system some bio degradable organic matter can Metal removal in the activated sludge process
be remained. Due to the additional digestion, was observed by Stephenson and Lester. There,
that residual part was degraded further. It the least soluble metals in the influent such as
caused to destruct VS additionally in hybrid Ag, Bi, Cd, Cr, Cu, Pb, and Zn showed removal
system. Although, sludge reduction was efficiency greater than 67%. The most soluble
achieved using hybrid system, the problem metals in the influent such as Cobalt, Mn, Mo,
associated that was difficulty in maintenance Ni and Tl were poorly removed at
and construction of the anaerobic reactor. approximately 40%. All the other metals except
Therefore, in this research, it is decided to use Mn were more soluble in effluent compared to
UASB reactor, as construction and maintenance influent. In order to remove insoluble metals, in
of UASB is easy compared with other reactors. total, the total was responsible for the total
According to past researches, it is proved that a removal of metals for the metals to be removed
considerable amount of heavy metals can be [14].
removed by the aerobic treatment. As an
example, in 1972 six number of municipal
477
In another study carried out by Chipasa, decreased at the beginning of period but COD
compared the heavy metal removal efficiency and MLSS removal efficiency re-increased due
in wastewater using biological wastewater to micro-organisms adapted to changes.
treatment system [15]. At the same time he However the increase in the later part is less
compared the heavy metal contents in than the best previous stage condition of COD
undigested and digested sludge as a further removal efficiency and it showed a downward
study. At the end, he concluded the study as trend. That concludes the heavy metal removal
follows, the heavy metal content of copper and in the fixed activated sludge is not simply
zinc were highly varying in the influent because of a biological process but also because
wastewater than the cadmium and lead. The of some part of these compounds is removed by
results of the wastewater treatment system adsorption on sludge. Although this part is
investigated by Chipasa indicate that when the negligible in comparison with biological
contents of cadmium and lead at 0.02 and 0.05 removal of these compounds [16].
mg/l their removal efficiency is insignificant
[15]. Therefore he concluded that the heavy According to the past researches, it is found
metal removal from wastewater is influenced that the anaerobic treatment is not much
by the initial contents in the influent. Hence it is popular in heavy metal removal from
certain that the heavy metal removal is directly wastewater since heavy metals are toxic to
proportional to their influent concentrations. micro-organisms in the anaerobic process. Fang
There, he found that the heavy metal removal & Hui in 1994 found that the order of heavy
increased in the order of Cd<Pb<Cu<Zn, with metal inhibition due to anaerobic starch-
respect to their initial influent wastewater degradation in granules is Zn>Ni>Cu>Cr>Cd
contents. [17]. Due to the layered structure of the sludge
granules, it had higher toxicity-resistance than
According to the varying heavy metal contents flocculent sludge. According to many
the digested and undigested sludge were researcher, the methanogenic bacteria are the
characterized. Due to microbial decomposition most sensitive anaerobes to the toxic material in
of organic and inorganic compounds and the waste [18]. However, another two
anaerobic stabilization of sludge, the heavy researches show that, some of the acid-forming
metal content on dry weight basis was bacteria were more sensitive and highly
increased. However, the order of increase in affected due to the heavy metals than the
heavy metal contents in digested sludge was methanogens [19] [20]. In an investigation
opposite to that in the effluent. Therefore, the carried out by Hickey et al. found that some
order of increase is as Zn<Cu<Pb<Cd with the anaerobic micro-organism groups might be
change in positions of copper and lead [15]. severely inhibited with the addition of heavy
metals than the methanogenic groups [20].
Hassani A.H in 2009 studied the performance Also found that acidogens were severely
of activated sludge system in treating affected with the addition of Cu and Zn than
wastewater containing chromium, lead and the methanogens.
nickel [16]. A three section Plexiglas tank with a
downward-flow aerated reactor, an upward- Leighton and Forster investigated the effect of
flow aerated reactor and a gravity sediment Lead and Nickel on biogas production and the
unit was fed with the synthetic wastewater. VFA production [21]. However, Lead and
There the COD removal efficiency is about 96% Nickel was highly influenced on biogas
and the acclimation time for microorganisms is production while they appeared to act quickly
short. In the above activated sludge system at 1 on acidogenesis than on the methanogenic
mg/l concentration the removal efficiency of bacteria. On the other hand, Lin and Chen
chromium, lead and nickel was about 84%, 75% investigated that biogranules have high
and 80% respectively. When the concentration sensitivity to metal toxicity in the
of them is increased up to 5 mg/l, 10 mg/l and methanogenesis process than in the
50 mg/l the removal efficiency of chromium, acidogenesis process [18].
lead and Nickel were increased. Finally by
rising concentration of chromium, lead and According to the above discussed past research
nickel up to 100 mg/l the removal efficiency facts, it is clear that the combination of
became 86%, 97.6% and 97% respectively. In anaerobic and aerobic process will be more
every stage, with the increase of heavy metal effective when removing heavy metals from
concentration of chromium, lead and nickel, wastewater. Therefore, the objectives of this
COD and MLSS removal efficiency was research are to study whether the hybrid
478
reactor is feasible in removing heavy metals Table 1 - Composition of synthetic
from wastewater and the removal efficiencies. wastewater
Concentration
Chemical
3. Materials and Method (mg/L)
3.1 Experimental setup of the hybrid Milk powder 1000
system
Sucrose 300
The experiment was conducted in the
laboratory scale. The hybrid system was Ammonium chloride
30
constructed by combining the aerobic (NH4Cl)
treatment system with the UASB reactor. Magnesium
150
UASB reactor worked as a pre-treatment for sulphate(MgSO4.7H2O)
aerobic system.
Ferric chloride
1.5
(FeCl3.6H2O)
The UASB reactor has a working volume of
8.5L (height: 110 cm with 10 cm diameter). Calcium chloride
25
The reactor body was made using cylindrical (CaCl2.2H2O)
Perspex pipes. The bottom of the reactor was Manganus sulphate
0.1
filled up to 2 L level with anaerobic granular (MnSO4)
sludge obtained from a full-scale UASB Copper sulphate
reactor in Elephant house beverage and ice- 0.1
(CuSO4)
cream company, Ranala, Sri Lanka. Anaerobic Sodium bicarbonate
sludge granules had total and volatile solids 200
(NaHCO3)
concentrations of 26 g/L and 20 g/L
KH2PO4 5
respectively.
Aerobic system was constructed using an

3.3 System operation
aeration tank (20L) and a sedimentation tank
(13L). Aerobic treatment was initiated by
Hybrid reactor was operated over 110 days.
introducing 100 ml of activated sludge into
The methodology followed is in the flow
the aeration tank which was obtained from
chart shown in figure 2. Hybrid reactor was
Hemas southern hospital wastewater plant,
fed with synthetic wastewater and operated
Galle, Sri Lanka.
at 20 ml/min flow rate. Synthetic wastewater
was introduced from the bottom valve of the
A simplified flow sheet of lab scale model of UASB reactor using a peristaltic pump. To
hybrid system was shown in Figure 1. control the fluctuation of pH in the UASB
reactor sodium bicarbonate was added to the
wastewater. First, the reactor was operated
without introducing any heavy metal.
Gas
Bag
According to the literature review it was found
Aeration
that the Cu, Cd, Zn and Pb are the most
Settling
Tank Tank Effluent common heavy metals in municipal
`
UASB
wastewater. Therefore by considering the
Granule maximum and minimum concentrations
Synthetic
available on those wastewaters and the
maximum tolerance limit that can be
`
Wastewater Return Activated

Tank sludge To Sludge
stirrer
P Treatment discharged in to the inland water surfaces,
three concentrations were selected for each
Figure 1 - Experimental set up of the hybrid heavy metal as shown in table 2 and the
system treatment was started. Each heavy metal
concentration was treated for a period of one
week. After treating all the three concentrations
3.2 Synthetic wastewater of one heavy metal, the system was once again
Synthetic wastewater was prepared using fed with synthetic wastewater without any
milk powder as the main substrate. Table 1 heavy metal and operated for one week. This
shows the composition of synthetic was done every time before adding a new
wastewater. heavy metal to the wastewater.
479
Table 2 - Heavy metal concentration used for Table 4 - Measurement methods
synthetic wastewater
Parameter Method/Apparatus
Heavy metal Heavy metal Portable
COD
concentration (mg/l) spectrophotometer
pH pH meter
Lead (Pb) 0.1 0.25 0.5
Temperature Thermometer
Zinc (Zn) 1 2 3
Graphite furnace
Cadmium (Cd) 0.1 0.25 0.5 Heavy metal
atomic absorption
concentration
Copper (Cu) 0.5 1 1.5 spectrometer
Sludge Settleable solid using
Volume imhoff cone

Hybrid system was operated over 110 days at
7 HRT and at ambient temperature. Reactor
performance was observed by measuring
COD, pH, temperature, heavy metal
concentration and sludge volume to check
whether the anaerobic-aerobic treatment was
occurred within the hybrid reactor.
pH and temperature inside the UASB reactor

during the operating period is shown in
Figure 3. pH was maintaining within the
Figure 2 - Operational procedure of the range of 5.8 – 7. The temperature was above
system 300c.
3.4 Analysis
Table 3 and Table 4 shows the measured
parameters and measurement methods,
respectively.
Table 3 - Measured parameter
Parameter
Influent
Effluent
Reactor
Sludge
COD √ √
Figure 3 - pH value inside the UASB
pH √ √ √ reactor
Temperature √
Heavy metal 4.1 Reactor performance
√ √
concentration
The influent T-COD concentration and
Sludge Volume √ effluent T-COD concentration of hybrid
system were measured. The average T-COD
concentration of the influent was 1631 ± 92
mg-COD/L. The average COD removal
efficiency and the sludge volume of the
synthetic wastewater without adding any
480
heavy metal was around 94% and 38 ml above 80% and sludge volume produced has
respectively also decreased with the increase of influent
Cd concentration. The removal efficiency of
4.1.1 Lead (Pb) removal Cd is less compared to other heavy metals
because of its high solubility. However above
The COD removal efficiency of lead 75% removal efficiency was obtained.
introduced wastewater was around 90% and
sludge production also reduced with the
increase of influent Pb concentration.
However above 80% Pb removal efficiency
was obtained from the hybrid reactor. Heavy
metal removal efficiency has increased with
the influent concentration.
Figure 6- COD removal efficiency, heavy

metal removal efficiency and the sludge
volume for the wastewater containing Cd
4.1.4 Copper (Cu) removal
Figure 4 - COD removal, heavy metal removal When Copper (Cu) was introduced to the
efficiency and the sludge volume for the wastewater in 0.5 mg/l, 1 mg/l, and 1.5 mg/l
wastewater containing Pb concentrations, the COD removal efficiency
has been affected with the addition of Copper
to the influent and the efficiency of removal
4.1.2 Zinc (Zn) removal
has decreased. This is because of the
Around 90% COD removal efficiency was reduction of treatment from UASB reactor.
obtained for the Zinc introduced wastewater. Therefore, the sludge volume is also high
However, the sludge volume has increased due compared to other cases. However above 80%
to the reduction of treatment from UASB due to Cu removal efficiency was obtained from the
toxicity of Zinc for anaerobic sludge granules. hybrid reactor.
Above 90% Zn removal efficiency was obtained
from the hybrid reactor.
Figure 7- COD removal efficiency, heavy

Figure 5 - COD removal efficiency, heavy metal removal efficiency and the sludge
metal removal efficiency and the sludge volume for the wastewater containing Cu
volume for the wastewater containing Zn
4.1.3 Cadmium (Cd) removal Although there were no past researches have
conducted on the heavy metal removal from
Cadmium (Cd) was introduced to the UASB reactor, a comparison or a range for
wastewater in 0.1 mg/l, 0.25 mg/l, and 0.5 removal efficiencies cannot be shown. However
mg/l concentrations. The COD removal also the study carried out by Hawari & Mulligan in
481
2006 shows that the feasibility of anaerobic efficiency of Zinc (Zn) from synthetic
granules as a novel type of biosorbent, for lead, wastewater. From the above removal
copper, cadmium, and nickel removal from efficiencies, it can conclude that the reduction
aqueous solutions. Therefore, it can conclude of heavy metal contents has increased in the
that the anaerobic sludge granule used in the order of Cd<Cu<Pb<Zn.
UASB reactor was able to remove heavy metals
by the biosorption process. However more than This study can be carried out further to
60% removal efficiency was obtained from the investigate the technical feasibility of hybrid
UASB reactor for all the heavy metals. reactor for removal of heavy metals in actual
wastewater because in the real case the heavy
According to the literature review done on the metals presence in the wastewater not limited
removal efficiencies of Pb, Zn, Cd and Cu from to Pb, Zn, Cd and Cu. The heavy metal content
activated sludge process it was found that the in sludge is higher than that of the effluent.
high heavy metal removal efficiency for Zinc Although the sludge produced from the
was obtained due to the microbial floc wastewater without heavy metals is much
produced in the activated sludge process lower, with the heavy metal introduced
efficiently adsorbed zinc in both the soluble wastewater the sludge volume has increased.
form and in the finely-divided suspended Therefore a further investigation can be carried
particles. An experiment done by Brown et al out to minimize the sludge production in the
in1973 found that the removal efficiency of zinc wastewater with heavy metals by changing
when the influent concentration within 2.5 reactor conditions, flow rate and the anaerobic
mg/l -20 mg/l is 95% - 74%. Stephenson & sludge granule and activated sludge amount
Lester in 1987 found that the removal efficiency added to the UASB and ASP.
of Zinc from activated sludge process is above
67%. According to Brown et al (1973), the
overall lead removal efficiency varies within References
33%-55%, Cadmium removal is around 25%
and the Copper removal is round 63%. Copper 1. Srivastava, N. K. & Majumder, C. B., 2008.
is mainly adsorbed by the microbial floc, and Novel biofiltration methods for the treatment
lead is removed more efficiently because of of heavy metals. Journal of Hazardous
increased settling time and larger particle size. Materials, Volume 151, pp. 1-8.
These values are also tally with the research
2. Dhokpande, S. R. & Kaware, J. P., 2013.
done by Chipasa in 2003, he found that the
Biological Methods for Heavy Metal Removal-
order of removal efficiency of Pb, Zn, Cd and A Review. International Journal of Engineering
Cu from wastewater is Cd<Pb<Cu<Zn. The Science and Innovative Technology (IJESIT), 2(5),
removal efficiencies obtained from the ASP pp. 304-309.
during this reasearch also tally with the above
values. 3. Oyaro, N., Juddy, O., Murago, E. M. & Gitonga,
E., 2007. The contents of Pb, Cu, Zn and Cd in
Considering above facts the removal meat in Nairobi, Kenya. Journal of Food,
efficiencies of Pb, Zn, Cd and Cu obtained from Agriculture & Environment, 5(3,4), pp. 119-121.
the hybrid reactor can be acceptable.
4. Fu, F. & Wang, Q., 2011. Removal of heavy
metal ions from wastewaters: A review. Journal
5. Conclusions of Environmental Management, Volume 92, pp.
The subject of this paper is the feasibility of 407-418.
hybrid reactor for removal of heavy metals
from wastewater. Hybrid reactor is the 5. Ahmaruzzaman, M., 2011. Industrial wastes as
low-cost potential adsorbents for the treatment
combination of UASB reactor (Anaerobic) and
of wastewater. Advances in Colloid and Interface
the activated sludge process (Aerobic). Science, Volume 166, pp. 36-59.
The reactor was operated over 110 days by
maintain a flow rate of 20 ml/min where HRT 6. Leitao, R. C., Haandel, A. V., Zeeman, G. &
of UASB reactor is 7 hrs at ambient temperature Lettinga, G., 2006. The effects of operational
of 320C. Results showed that the T-COD and environmental variations on anaerobic
removal is above 84% for all the heavy metals wastewater treatment systems: A review.
introduced wastewaters and more than 80% Bioresource Technology, Volume 97, p. 1105–
removal efficiency of Lead (Pb) and Copper 1118.
(Cu). More than 75% removal efficiency of
Cadmium (Cd) and more than 90% removal
482
7. Metcalf & Eddy, 2003, wastewater engineering: methane production and hydrogen and carbon
Treatment and reuse. 4th ed.,MacGraw hill monoxide levels during batch anaerobic sludge
companies, china. digestion. water research, 23(2), pp. 207-218.
8. Castillo, A., Cecchi, F. & Mata-alvarez, J., 1997. 20. Lin, C. Y., 1993. Effect of heavy metals on
A combined anaerobic-areobic system to treat acidogenesis in anaerobic digestion. Water
domestic sewage in coastal areas. water research, research, 27(1), pp. 147-152.
31(12), pp. pp. 305-306.
21. Leighton, I. R. & Forster, C. F., 1998. The effect of
9. Kassam, Z. A., Yerushalmi, L. & Guiot, S. R., heavy metals on a thermophilic methanogenic
2003. A market study on the anaerobic upflow sludge blanket reactor. Bioresource
wastewater treatment systems. Water, Air, and Technology , Volume 63, pp. 131-137.
Soil Pollution 143, p. 179–192.
10. Chan , Y. J., Chong, M. F., Law, C. L. & Hassell,

D. G., 2009. A review on anaerobic–aerobic
treatment of industrial and municipal
wastewater. Chemical Engineering Journal,
Volume 155, pp. 1-18.
11. Ros, M. & Zupancic, G. D., 2004. Two-Stage

Thermophilic AnaerobicÐAerobic Digestion of.
Environmental engineering science, 21(5), pp. 617-
626.
12. Brown, H. G., Hensley, C. P., McKinney, G. L. &

Robinson, J. L., 1973. Efficiency of heavy metals
removal in municipal sewage treatment plants.
Environmental letters, 5(2), pp. 103-114.
13. Cheng, M. H., Patterson, J. W. & Minear, R. A.,

1975. Heavy metals uptake by activated sludge.
Water Pollution Control Federation, 47(2), pp. pp.
362-376.
14. Stephenson, T. & Lester, J. N., 1987. Heavy

metal behaviour during the activated sludge
process- extent of soluble and insoluble metal
removal. The science of the total environment,
Volume 63, pp. 199-214.
15. Chipasa, K., 2003. Accumulation and fate of

selected heavy metals in a biological. Waste
Management, Volume 23, p. 135–143.
16. Hassani, A. H., Hossenzadeh, I. & Torabifar, B.,

2010. Investigation of Using Fixed Activated
Sludge Systemfor Removing Heavy Metals
(Cr,Ni and Pb) From Industrial WasteWater.
Journal of EnvironmentalStudies, 36(53), pp. pp.
22- 24.
17. Fang, H. P. & Hui, H. H., 1994, 'Effect of heavy

metals on the methanogenic activity of starch-
degrading granules', Biotechnology letters, 16(10),
pp. 1091-1096.
18. Lin, C. Y. & Chen, C. C., 1999. Effect of heavy

metals on the methanogenic UASB granule. Wat.
Res. , 33(2), pp. 409-416.
19. Hickey, R. F., Vanderwielen, J. & Switzenbaum,

M. S., 1989. The effect of heavy metals on
483
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[485 491],2017
2017
The Institution
Lanka
Feasibility of UASB Reactor for Removal of Heavy

Metal in Wastewater
U.L.N.C. Jayathileke and W.M.K.R.T.W. Bandara
Abstract: Even though there are many ways to remove heavy metals in wastewater, most of
them are not cost effective and have many drawbacks. The feasibility of the Up-flow Anaerobic Sludge
Blanket (UASB) reactor, for the removal of heavy metals, were studied through this research. A lab-
scale reactor was operated for 110 days with synthetic wastewater influent and HRT of 7 hours. Pb,
Zn, Cd and Cu were added to the influent to 4 different concentrations and one heavy metal at a time.
The parameters analyzed in the treatment was COD, Biogas production rate, pH, temperature, heavy
metal concentrations. The results show that Zn and Cd shows a toxic effect for the wastewater
treatment, and Cu and Pb are not toxic. However heavy metal removal efficiency of more than 60%
was achieved for each heavy metal. Therefore the UASB reactor is feasible for the removing of heavy
metals from the UASB reactor.
Keywords: Biogas, COD removal, Heavy metal removal, HRT, Sludge granules, UASB reactor
1. Introduction and flocculation [3]. However, there are many

Environment contamination with toxic heavy drawbacks from these metals such as high
metals has become one of the major concern for capital cost, high maintenance cost, high
humankind in the current era of globalization operation cost, high usage of chemicals and
and rapid industrialization. Discharge of heavy high sludge production [4].
metal contaminated wastewater causes the
accumulation of heavy metals to the water Therefore new methods with less cost and less
bodies, soil and crops. Heavy metals are sludge production are required. Such as
naturally occurring substances that have a high anaerobic digestion. Anaerobic digesters are
atomic weight and a density at least 5 times commonly used at municipal wastewater
greater than that of water [1]. treatment plants to degrade sludge, which uses
suspended bacterial growth. Most of the
Through the circulation of heavy metals from industrial wastewater treatment plants are
wastewater discharge to water bodies, soil and using anaerobic digesters with the development
crops, finally, it can be consumed by the of anaerobic digesters, having fixed-film
animals and humans. Through that, there are bacterial growth. Anaerobic digesters use a
many toxic effects to the human and animal method called catabolic (destructive) processes,
health through heavy metal consumption like which occur in the absence of free molecular
Skin manifestations, visceral cancers, vascular oxygen (O2). The process of anaerobic digestion
disease Kidney damage, renal disorder etc. [2]. involves biologically destroying a significant
According to above description, heavy metal portion of the volatile solids in sludge, and to
removal from wastewater has become a major minimize the volume of sludge. The main
concern in the present world. Therefore this products from anaerobic digesters are biogas
study focuses on two objectives. The first and digested sludge. Biogas contains methane
objective is to identify whether the Up-flow (CH4) and carbon dioxide (CO2) mostly. This
Anaerobic Sludge Blanket (UASB) reactor is involves four processes called Hydrolysis,
feasible for the removal of heavy metals and if Acidogenesis, Acetogenesis and
feasible the removal efficiencies that can be Methanogenesis [5].
obtained. The second objective is to identify the
effect of heavy metals on UASB reactor
treatment process. U.L.N.C. Jayathileke, B.Sc. Eng. (Ruhuna), Civil
Engineer, Road Development Authority (RDA).
2. Literature Review Dr. W.M.K.R.T.W. Bandara, Ph.D.(Hokkaido-Japan),

M.Sc (Hokkaido-Japan), B.Sc. Eng. (Peradeniya, Senior
In the present era, there are many methods Lecturer, Department of Civil and Environmental
used for heavy metal removal such as Ion Engineering, Faculty of Engineering, University of
exchange, Membrane filtration, Coagulation Ruhuna..
485
UASB reactor is an attractive anaerobic adsorbent. Two types of wastewater were used.
treatment method because of its compactness, One is high strength wastewater with 5mg/l of
low operational cost, low sludge production, heavy metals from each and other is low
and production of biomethane [6]. In the late strength wastewater with 10mg/l of heavy
1970's the UASB reactor was developed by metals from each. For both, the cases good
Dr.Gatze Lettinga and his colleagues [7]. The removal efficiencies of 86% were achieved. The
UASB reactor has been used increasingly in order of heavy metal removal efficiencies was
recent years, to treat a variety of industrial Cu>Pb>Cd>Zn [13]
wastes and municipal wastes. The treatment
occurs with exposure to microorganisms in However, the feasibility of the UASB reactor for
the granule to wastewater- substrates. All the removal of heavy metals has not been taken
biological processes are taken place within the into consideration. Therefore the objective of
sludge bed, and the production of biogas this study is to identify whether the UASB
occurs. The accumulation of incoming reactor is feasible for the removal of heavy
suspended solids and bacterial growth forms metals and to identify the removal efficiencies.
the sludge bed. Retention of active sludge Also, the effect on wastewater treatment
within the UASB reactor facilitates good process by heavy metals are considered
treatment performance at high organic loading through this study.
rates [8].
3. Materials and Method
Although most of the researchers have focused
on the toxic effect of heavy metals on the 3.1 Setup of the UASB reactor
anaerobic digestion [9] [10]. , there are some
researchers who have focused on the removal A lab-scale UASB reactor with a height 110
of heavy metal using the anaerobic treatment cm; diameter 10 cm; working volume 8.5 L
methods . was used for this study. Reactor body is made
with cylindrical Perspex pipes. PVC conical
Haytoglu studied the effectiveness of the shaped funnel with a valve is connected to the
anaerobic biomass in adsorbing heavy metals. bottom of the reactor. The reason to use
This study focused on five heavy metals which conical shape is to make the uniform flow
are Pb, Cr, Cu, Ni and Zn [11]. The biomass throughout the reactor. The top part of the
used for the study were grown under the reactor is designed to collect gases that emit
laboratory conditions using a synthetic during anaerobic reactions and to collect
wastewater composition. The adsorption treated wastewater.
capacity was measured for the anaerobic dead
biomass. The maximum adsorptive capacity for
the biomass for Pb, Zn, Cu, Ni and Cr were
found as 1250, 625, 357, 227, and 384 mg/g dry
biomass.
Packed Bed Biofilm Reactor (PBBR) is an

anaerobic reactor used for wastewater
treatment. Research carried out on this reactor
was focused on the removal of Zn, Cd, Cu and
Ni [12]. The heavy metals were injected to the
influent, and then the removal efficiencies were
checked. The higher removal efficiency was
achieved when the influent heavy metal
concentration was at 8 and 20 mg/l. The
removal efficiency was higher than 71%. At the
28 and 40 mg/l, the removal efficiencies were
reduced. The removal efficiencies were in the
order of Cu>Zn>Ni>Cd. Also, a BOD removal
of 90 % was achieved.
Figure 1 - Experimental setup of UASB
Removal of Cu, Cd, Pb and Zn were studied in reactor
an up-flow anaerobic packed bed reactor which
was filled with porous volcanic rock to act as an
486
Six valves were connected to the reactor at UASB reactor at Ranala Elephant house Ice-
different places. Bottom valve is used to cream factory) blanket. After that system was
introduce wastewater into the reactor. Three commenced by introducing water for few
valves were connected at the body of the days. Then synthetic wastewater was fed
reactor to collect a sample that is taken from from the bottom of the UASB reactor.
inside the reactor. Top two valves are used to without heavy metals for a selected flow
collect treated wastewater and headspace rate of 20ml/min, considering an HRT value
gases. The Synthetic wastewater is fed to the of 7 hrs. As the wastewater flows upward
reactor using a mini pump from a PVC through the granular sludge and with contact
storage tank. A Cross-sectional view of lab of sludge blanket the treatment occurred. Also,
scale UASB reactor is shown in Figure 1. the gas p r o d u c e d f r o m t h e t r e a t m e n t
were collected from the top valve of
A synthetic wastewater composition was t h e r e a c t o r . The heavy metals selected for
used as the influent to the reactor. First, a testing of the feasibility of UASB reactor are as
baseline was established for the performance follows.
of the reactor by running the reactor without
adding any heavy metals. Then the four
heavy metals Pb, Zn, Cd and Cu were added I. Lead (Pb)
to the influent, as one heavy metal at a time in II. Zinc (Pb)
3 different concentrations and the reactor was III. Cadmium (Cd)
operated at 7 HRT. The influent and Effluent IV. Copper (Cu)
wastewater samples were collected to test the
parameters mentioned in Table 2. Then by introducing one heavy metal at a
time, the system was run for treatment, and
the required parameters were checked. For
3.2 Synthetic wastewater different concentrations of heavy metals
For the study, a milk powder based using the HRT as 7hr, the removal of heavy
wastewater composition was prepared using metals and the effect of the reactions of the
the chemicals stated in table 1. UASB were checked. The overall
methodology is shown in Figure 2.
Table 1 - Composition of synthetic
wastewater
Concentratio
Chemical
n (mg/L) Add heavy
metals
Milk powder 1000
Change
Sucrose 300 Run the UASB heavy
Ammonium chloride reactor metal type
30 /concentrat
(NH4Cl) ion
Magnesium sulphate
150 Collect Samples
(MgSO4.7H2O)
Ferric chloride
1.5
(FeCl3.6H2O) Test and analyze
the results
Calcium chloride
25
(CaCl2.2H2O) Figure 2 - Overall Methodology
Mangenus sulphate
0.1
(MnSO4)
Sodium bicarbonate
200
(NaHCO3)
3.4 Experiment
Monopotassium
5 Table 2 and Table 3 shows the measured
phosphate (KH2PO4 ) parameters and measurement methods,
respectively.
3.3 System operation
The reactor body was filled with 2l of

granular sludge (which were taken from the
487
Table 2 - Measured parameter
Influent UASB inside
Gas outlet
Temperature
Parameter
Influent
Effluent
Reactor
10.0 33
pH value
Temperature ͦC
8.0
28
COD √ √ 6.0
4.0
pH √ √ √ 23
2.0
Temperature √
0.0 18
Headspace gas √
0 50 100
Gas Time (days)
√
Composition
Heavy Metal Figure 3 - pH and temperature variation
√ √
Concentration
4.2 Pb removal
Table 3 - Measurement methods
For the treatment of Pb introduced wastewater,
Parameter Method/Apparatus Pb was added to wastewater in 3
concentrations which are 0.1 mg/l, 0.25 mg/l
Portable and 0.5 mg/l. These values were selected after
COD
Spectrophotometer considering the Pb concentration in domestic
pH pH meter wastewater in Asia.
Temperature Thermometer
Headspace gas Water replacement COD removal Pb removal
volume method
Gas Biogas production
Biogas production
Gas chromatograph 100 50
Composition
COD/ Pb removal
rate (ml/h)
efficiency (%)
Heavy metal Atomic absorption

concentration spectrophotometer 50 30
4. Results and Discussion 0 10

0 0.2 0.4 0.6
4.1 pH and temperature Pb concentration (mg/l)
For the treatment in UASB reactor, three types Figure 4 - Pb Introduced wastewater
of bacteria are involved, which were
hydrolysis, acid-producing bacteria, and As for the figure 4, the COD removal
methane-producing bacteria. Therefore, due to efficiencies and the biogas production have
acid-forming bacteria and the methane- increased with the increase of Pb
producing bacteria the pH value should be in a concentrations in the influent wastewater.
range of 6 to 8. Otherwise, there can be less Therefore, the adsorbent of the Pb into the
treatment from the bacteria performances. granules has enhanced the performances of
When considering the pH values inside the the granules. Also, the Pb removal is between
UASB within the operational period, it is at an 63% - 83%.
acceptable level as per the figure 3. In general,
the most of the anaerobic wastewater treatment This matches with the fact that the presence
processes are operated within the mesophilic of heavy metals enhances the biogas
temperature range which is 30°C - 40°C or in production in UASB, as studied by Nkemka
the thermophilic temperature range which is & Murto [14]. Also, the enhanced
45°C - 60°C. The temperature is within the performances have increased the adsorption
mesophilic temperature range which is capacity of the granules as the Pb removal
acceptable. efficiencies also have increased.
488
4.3 Zn removal
COD removal Cd removal
For the treatment of Zn introduced Biogas production
wastewater the Zn was added to wastewater
in 3 concentrations which are 1 mg/l, 2 mg/l, 100 50
Biogas production
COD/Cd removal
and 3 mg/l (These values were selected after
80
efficiency (%)
considering the Zn concentration in domestic 40
rate (ml/h)
60
wastewater in Asia). As for the Figure 5, Zn 30
removal efficiency between 68% -74% were 40
achieved (figure 5). With the increase in the 20 20
Zn concentration, the COD removal efficiency 0 10
and biogas production have reduced. One of 0 0.2 0.4 0.6
the reasons for this might be because of the Cd concentration (mg/l)
high concentrations of Zn than other heavy
Figure 6 - Cd Introduced wastewater
metals. Also, Atlas has found from his
studies, that the methane production ability is The COD removal efficiencies and the biogas
reduced when the Zn is present [9]. production have gained a peak value at
Therefore, this is a reason for the reduction in 0.25mg/l concentration of Cd. Therefore, after
biogas production and the COD removal. that concentration the Cd it is toxic for the
However anaerobic dead biomass has a high anaerobic digestion process. However, the Cd
adsorption capacity of Zn after the Pb removal has increased further with the increase
adsorption [11]. Therefore, it explains the of Cd concentration. This may be due to the
increase in the Zn removal efficiencies inorganic precipitation of Cd with increased Cd
although the COD removal decreased. So, the concentrations [15].
Zn is toxic for the reactions in the sludge
granules.
4.5 Cu removal
COD removal Zn removal For the treatment of Cu introduced wastewater

Biogas production the Cu was added to wastewater in 3
100 50 concentrations which are 0.5 mg/l, 1 mg/l, and
Biogas production
COD / Zn removal
1.5 mg/l (These values were selected after

efficiency (%)
80
rate (ml/h)
40 considering the Cu concentration in domestic

60 wastewater in Asia). A Cu removal efficiency
30
40 between 61% - 68% was achieved. For the COD
20 20 removal (figure 7) and biogas production, there
0 10
is no significant variation with an increase of
0 2 4 Cu concentrations. Therefore, the increase in Cu
concentrations has not affected the anaerobic
Zn concentration (mg/l)
digestion which indicates Cu is not toxic. Also
for the Cu removal the efficiency there is no
Figure 5 - Zn Introduced wastewater significant variations. Therefore’ this Cu
removal efficiency and COD removal efficiency
4.4 Cd removal can be enhanced with the change of the HRT.
Also in the other anaerobic treatment methods,
For the treatment of Cd introduced
Cu has a high removal efficiency [11] [12].
wastewater the Cd was added to wastewater
However, when considering the observations
in 3 concentrations which are 0.1, 0.25 and done during the Cu treatment, the sludge
0.5mg/l (These values were selected after
granules started to float at the final
considering the Cd concentration in domestic concentration without settling at the bottom. As
wastewater in Asia). As for the Figure 6, COD
for Nnaji, large granules tend to possess a
removal efficiencies and the biogas polymeric coat which traps evolved gases and
production have increased and then
thus make the granules fluffy [16]. The larger
decreased with Cd concentrations. A removal
the granules, the more difficult it becomes for
efficiency of 61-71% was achieved.
the substrate to diffuse into the core of the
granule, hence reducing the rate of
bioconversion. Therefore, this might be the
reason for the no change is the COD removal
and the Cu removal efficiencies.
489
COD removal Cu removal 3. Dhokpande, S. R. & Kaware, D. J. P., 2013,
Biogas production ‘Biological Methods for Heavy Metal
Removal’, International Journal of Engineering
100 50
COD/Cu removal
Biogas production
Science and Innovative Technology (IJESIT), 2(5),
efficiency (%)
80 40 pp. 304-309.
rate (ml/h)
60
30 4. Fu, F. & Wang, Q., 2011, ‘Removal of
40
20 Heavy Metal ions from Wastewaters: A
20
R eview’, Journal of Environmental Management,
0 10 92, pp. 407-418.
0 1 2
Cu concentration (mg/l) 5. Gerardi, M. H., 2003, The Microbiology of
Anaerobic Digesters. s.l.: A John Wiley & Sons,
Inc., Publication.
Figure 7 - Cu Introduced wastewater 6. Mudhoo, A. & Kumar, S., 2013, ‘Effects of

Heavy Metals as Stress Factors on
5. Conclusions Anaerobic Digestion Processes and Biogas
Production from Biomass’, Int. J. Environ.
With reference to the main objective of this Sci. Technol., 10, pp. 1383–1398.
study the UASB reactor is feasible for
7. Bandara, W. M. K. R. T. W. et al., 2013.
removing heavy metals. The removal Introduction of a Degassing Membrane
efficiencies are different based on the heavy Technology into Anaerobic Wastewater
metal type and the concentrations. For Pb, the Treatment. Water Environment Research, 85(5),
efficiencies were between 63% - 83%. For Zn, pp.387–390.
the removal efficiencies were between the
68% -74%. On the other hand, for Cd, the 8. Karthikeyan, K. & Kandasamy, J., 2009,
removal efficiencies were between 61-71% for ‘Upflow Anaerobic Sludge Blanket (UASB)
the concentrations of 0.1, 0.25 and 0.5 mg/l. Reactor in Wastewater Treatement’, Water
and wastewater treatment technologies, 2, pp.
Finally, for Cu, the removal efficiencies were
180-198.
between 61% - 68%. The other objective of this
research was to identify whether there is an 9. Altas, L., 2008, 'Inhibitory Effect of Heavy
effect on the wastewater treatment Metals on Methane-Producing Anaerobic
parameters from the heavy metals. Therefore, Granular Sludge', Journal of Hazardous
from the results, it can be concluded that the Materials, pp. 1551-1556.
COD removal has enhanced by the increased
Pb concentrations. However, there is a 10. Azizi, S., Kamika, I. & Tekere, M., 2016,
negative effect on the COD removal by the Cd ‘Evaluation of Heavy Metal Removal from
Wastewater in a Modified Packed Bed Biofilm
and Zn. However, the effect is more
Reactor’, PLos ONE, 11(5).
significant when it comes to Cu. However, in
each case at least 50% removal efficiencies can 11. Haytoglu, B., Demirer, G. & Yetis, U., 2001.,
be achieved. This study can be developed ‘Effectiveness of Anaerobic Biomass un
considering the removal efficiencies of heavy Adsorbing Heavy Metals’, Water Science and
metals when there is more than one heavy Technology, 44(10), pp. 245-252.
metal in influent wastewater and for different
HRT values for better analysis. 12. Azizi, S., Kamika, I. & Tekere, M., 2016,
‘Evaluation of Heavy Metal Removal from
Wastewater in a Modified Packed Bed
References Biofilm Reactor’, PLos ONE, 11(5).
1. Tchounwou, P. B., Yedjou, C. G., Patlolla, K. A. 13. Sekomo, C., Kagisha, V., Rousseau, D. & Lens,
& Sutton, D. J., n.d., ‘Heavy Metal Toxicity and P., 2012, ‘Heavy Metal Removal by Combining
the Environment’, In: A. Luch (ed.), Molecular, Anaerobic upflow Packed Bed Reactors with
Clinical and Environmental Toxicology. Water Hyacinth Ponds’, Environmental
s.l.:Birkhäuser Basel, pp. 133-164. Technology, 33(12), pp. 1455-1464
2. Gunatilake, S., 2015, ‘Methods of Removing 14. Nkemka, V.N. & Murto, M., 2010, ‘Evaluation
Heavy Metals from Industrial Wastewater’, of Biogas Production from Seaweed in Batch
Journal of Multidisciplinary Engineering Science Tests and in UASB Reactors Combined with
Studies (JMESS), I, pp. 12-18.
490
the Removal of Heavy Metals’, Journal of
Environmental Management, 91, pp. 1573-1579.
15. Cheng, H., Patterson, J. & Minear, R., 1975,

‘Heavy Metals Uptake by Activated Sludge’,
Water Pollution Control Federation, 47(2), pp.
362-376.
16. Nnaji, C. C., 2013, ‘A Review of the upflow

Anaerobic Sludge Blanket Reactor’, Desalination
and Water Treatment.
491
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[493 498],2017
2017
The Institution
Lanka
GIS-Based Site Suitability Analysis for Wastewater

Treatment Plants
N.A. Nusky, S.I.S. Wijesingha, W.M.K.R.T.W. Bandara, B.M.L.A. Basanayake and
Terrance M. Rengarasu
Abstract: Identifying a site for domestic wastewater treatment plant involves a

comprehensive assessment to identify the best obtainable location(s) or site(s) that can
simultaneously meet the requirements of minimise economic and regulations, health, environmental
and social costs. While GIS has been a powerful tool to handle spatial data in site suitability analysis,
application of this tool alone could not overcome the issue of unpredictability in expert opinion
when trying to judge and assign relative importance to each of many criteria considered in a site
suitability analysis. To overcome this issue, the Analytical Hierarchy Process method is used in
combination with the GIS tools. This manuscript presents how the integrated tool has used
effectively in the site suitability analysis for Unawatuna tourist area and Puttalam Municipal
Council which considered simultaneously 8 different sub-criteria under 3 different main criteria
such as soil texture, topography, surface water, groundwater, land use, residential area and roads.
Keywords: GIS, Treatment Plant, Site Suitability Analysis, Wastewater
1. Introduction protecting public health from hotel

wastewater and eliminates or reduces the
Wastewater has been defined as the water migration of wastewater pollutants to the
discharged from a community after it has water bodies. Hence, it is necessary to take
been contaminated by various wastes. In the effective steps toward achievement of
wastewater, 99% is water, and only 1% is environmental goals of sustainability through
solid waste. Wastewater is generated from developing treatment plants in suitable sites.
domestic, industrial, agricultural sources and But finding a suitable site for this purpose
surface run-off due to rain. Eventually, it goes involves considering a wide range of criteria
down to the drainage and ends up in which makes decision-making complicated.
wastewater treatment facility. Usually,
surface water needs more filtration and The principles of sustainable development of
treatment than the groundwater because wastewater treatment make site suitability
rivers, streams and lakes contain more analysis more and more complex due to the
pollutants and sediments and are more concern of multiple criteria. It includes
vulnerable to be contaminated than attention not only essential capacity of a land
groundwater. Some water supplies may also unit to support a particular land use for a
contain organic chemicals, disinfection by- long period of time without deterioration, but
products and inorganic chemicals. also environmental and the socio-economic
feasibility.
Municipal domestic wastewater can be
defined as wastewater from residential N.A. Nusky, B.Sc. Eng. (Ruhuna), Planning Engineer,
settlements and services which originates Sierra Construction Pvt Ltd, Colombo.
primarily from the domestic activities the S.I.S.Wijesingha, B.Sc. Eng. (Ruhuna), Civil Engineer,
human metabolism. It provides an essential Nawaloka Construction Company Pvt Ltd.
community service that is vital for the Dr. W. M. K. R. T. W. Bandara, Senior Lecturer,
protection of the environment and public Department of Civil and Environmental Engineering,
Faculty of Engineering, University of Ruhuna.
health. The domestic wastewater treatment
B.M.L.A. Basanayake, Senior Lecturer, Department of
plant undergoes physical, chemical and
Civil and Environmental Engineering, Faculty of
biological processes for cleaning the Engineering, University of Ruhuna.
wastewater and separate the solids. The Dr. Terrance M. Rengarasu, Senior Lecturer, Department
treated wastewater can be used for irrigation of Civil and Environmental Engineering, Faculty of
of food crops. In addition, the nutrients in the Engineering, University of Ruhuna.
wastewater can be recovered and provided to
the water bodies. This gives benefits for
493
Therefore, when selecting appropriate site relatively easy identification of those areas
selection, the technical, social, environmental where all criteria are met. In addition to
aspects need to be considered as main criteria storing, retrieving, displaying spatial data, a
during analysis. The criteria influencing the GIS allows the user to create overlays,
site selection were mainly determined by intersections, buffers, spatial joins, map
temperature, wind direction and speed, algebra, proximity analysis, and other
rainfall soil, topography, groundwater, analytical processes. In the perspective of land
surface waters, land use, residential area and or site suitability, GIS helps the user to
roads [1]. determine what locations/sites are most/least
suitable for development. In this means, the
The site selection studies usually utilize many results of GIS analysis can offer support for
parameters. Thus a systematic methodology decision-making.
is needed to combine the various information
from a wide range of disciplines [2]. Multi- Here the study was carried with the main
criteria decision analysis (MCDA) involves a objective to develop an approach of GIS
set of processes that assign the alternatives suitability analysis to identify an optimal site
values evaluating for a specific purpose. In for domestic wastewater treatment plant. The
spatial MCDA, geographical data are selected study areas are Unawatuna tourist
gathered, processed and transformed into a area in the southern region and Puttalam
decision. For that, we have to implement a Municipal Council of Sri Lanka which is the
technique that allows an estimation of the agriculture-based area. Here we are trying to
weights. Analytic hierarchy process (AHP) is find the feasibility of the application of this
one such technique which was applied to approach in various types of areas. That is the
assign weights for the selected influence reason behind the selection two different
criteria with a combination of GIS analysis types of the study area which is most
tools to provide effective and efficient site common in Sri Lanka.
identification promising results [3].
1.1 Role of GIS in Site Suitability
Analysis
Recent technological advances made in the 2.1 Analytic Hierarchy Process (AHP)
field of spatial technology created significant AHP is a multi-criteria decision method that
impact in planning activities. The purpose of uses hierarchical structures to represent a
using GIS is that maps provide an added problem and then develop priorities for
dimension to data analysis which brings us alternatives based on the judgment of the user
one step closer to visualizing the complex [7]. AHP shows that weighting activities in
patterns and relationships that characterize multi-criteria decision-making method can be
real-world planning and policy problems [4]. effectively dealt with via hierarchical
This field of planning is of prime importance arranging and pair-wise contrasts. According
for the countries which have various to the statements, the AHP is based on three
geographic patterns, cultural activities etc. It steps:
also offers an interpretation of physical • Assemble the decision problem in a
(spatial) data with other socio-economic data, hierarchy
and thereby providing an important linkage • Comparative judgment / Pair-wise
in the total planning process and making it Comparison of the elements
more effective and meaningful [5]. • Synthesis of the priorities
When we consider the site suitability analysis, Advantages:

every factor required for the suitability • The AHP method supports group
analysis can be represented in a digital layer decision-making through consensus by
which can be queried individually or overlaid calculating the geometric mean of the
with other layers. In contrast to manually individual pairwise comparisons [8].
querying and overlaying the different • The advantages of AHP over other multi-
information sources, a GIS can prove to be criteria methods are its flexibility, intuitive
very efficient, accurate and time-saving [6]. appeal to the decision-makers and its
This is particularly true for large areas or for a ability to check inconsistencies [9].
great number of variables, as is the case for • AHP is uniquely positioned to help model
this research. Therefore, using a GIS allows a situations of uncertainty and risk since it is
494
capable of deriving scales where measures 60m distance from any public stream or water
ordinarily do not exist [10]. body to the WWTP.
• AHP method has the distinct advantage
that it decomposes a decision problem into Ground Water: The groundwater protection
its constituent parts and builds hierarchies responses are concerned with the site
of criteria. Here, the importance of each selection process for disposal effluent site and
criterion becomes clear [11]. the associated design, operation and
monitoring of disposal sites. Therefore,
drinking water quality of groundwater
2.2 Factors Influence on Site Selection
should be protected from migrating
 Technical Aspects pollutants and contaminants. The minimum
distance to the groundwater table will be
Soil texture: Physical characteristics
1.5m [12]. Generally, greater depths are
influence the permeability of the soil. The
usually preferred.
permeability must be within a range in which
adequate residence time for renovation is
given and prevent rapid downward  Social and Cultural Aspects
movement of effluent. The bearing capacity of Land Use and Building Area: The sewage
the soil also an important factor in site must be applied to land in a way that does
selection process for treatment. The allowable not cause a nuisance and does not endanger
bearing capacity of the soils and the allowable health. Nuisance relates particularly to odour
differential settlement will control the need to which is allocated with by buffer zones to be
support tanks and structures on piles. kept around building/residential areas. As
According to the permeability condition and our study area is a densely populated area,
bearing capacity of the soil, the red-yellow we cannot go for big buffer limit. Mainly to
podzolic is preferable to the WWTP in the minimize odour, noise, aerosols and air
study area. particles, it is preferable to keep a 20m buffer.
Regarding the land use, our study has a land-
Slope (Topography): A low-lying site use type such as cultivation area, rock area,
facilitates the wastewater flow from the built-up area, sandy area, bare land, etc. Here
network service area by a gravitational flow we should avoid the area like built-up area,
which minimizes the number of pumping rock area, cultivation area and forest area.
stations in the collection system. If the slope is Bare lands are more suitable for the site.
a too high surface runoff and soil erosion will
be very high. And, such sites require flood Roads: Distance from the roads increases the
protection. In addition, the steep slope will cost of wastewater treatment plant
also affect cultivation. According to the construction and maintenance. However, the
wastewater treatment plant construction presence of the wastewater treatment plant
guideline, the appropriate slope is less than close to the roads affects the landscape,
35% which provides suitable irrigation of climate, and the public health. According to
treated wastewater for agricultural purposes. the wastewater treatment plant construction
guideline, it has to be a buffer zone of 500m
from the highway and railways.
 Environmental Aspects
Surface Water: The wastewater will be
2.3 Methodology
discharged to the surface water bodies or sea
All research procedures can be mainly
directly or indirectly. It pollutes the inland
categorized as defining site suitability criteria,
and coastal surface water when the pollutant
AHP analysis and weighted overlay. Defining
content increases. And also, the natural water
site suitability criteria was done by referring
bodies, wetlands, oceans quality can be
previous studies and laws on conservation.
deteriorated by wastewater nitrogen,
And relevant GIS maps were created using
phosphorus, chloride, faecal coliforms and
ArcGIS. For AHP analysis questionnaire
salinity removals. The intentional and
survey was conducted among experts such as
accidental discharges, leakages, seepage will
Galle irrigation chief engineer, engineers,
cause this more. Therefore, an excluded area
doctors, undergraduates and the public in the
should be created between surface waters and
study area. In the survey, the main and sub-
WWTP site. According to the crown land
criteria were compared pair-wise among
ordinance (Chapter 454), there should be the
them. The weights derived from the AHP
analysis was used for weighted overlay
495
process using ArcGIS. The weighted overlay The criteria and the suitability classification
was done in two phases, with the sub-criteria with scores were developed which is a
and main criteria. Finally, field verification primary procedure to identify an optimal site
was carried out. The design flow chart is for municipal WWTP for Unawatuna is
shown in Figure 1. shown in table 2.
Table 2 - Suitability Criteria for Unawatuna
Criteria
Criteria
Scores
Main
Class
Sub
Red-
yellow 3
Technical Aspects
Soil Podzolic
Soil
Texture Regosol 2
Bog and
1
Figure 1 - Design Flow Chart half dog
0 – 15 2
Slope
Topography 15 – 30 3
[%]
The digital layers used for spatial analysis 30 < 1
were 1: 10000 map obtained from the 0 – 1.5 1
Ground Depth
Environmental
department of surveying, Sri Lanka. The 1.5 – 2.0 2

water [m]
Aspects
groundwater map was created by collecting 2.0 < 3

the data around the study area using ArcGIS 0 – 60 1
tools. Ultimately, based on the above Surface Buffer
60 – 100 2
procedures the final suitability map was Waters [m]
100 < 3
created.
Building/ 0 – 15 1
Buffer
3. Results and Discussion Residential 15 – 30 2
[m]
Areas 30 < 3
Criteria which have highest weights influence Built-up
more in the output overlaid map. The area
weighting criteria for Unawatuna area is Boggy area
shown in table 1. Here, among main criteria Forest area
Social and Cultural
social and cultural aspect influence around Cloud area 1

61%. Land Use Type Cultivation
Aspects
area
Rock area
Table 1 - Weightage of criteria for Unawatuna Water area
Sand area 2
Sub-
Main criteria Weight Weight Bare area 3
criteria
0 – 400 1
Soil Buffer
Technical 0.750 Major Roads 400 – 600 3
0.090 Texture [m]
Aspects 600 < 2
Slope 0.250 0 – 50 1
Ground Buffer
0.250 Minor Roads 50 – 100 3
Environmental Water [m]
0.303 100 < 2
Aspects Surface 3 - suitable 2 - Moderate 1 - Unsuitable
0.750
Waters
Land use 0.303
Social and The criteria and the suitability classification
Residential
Cultural 0.607 0.607 with scores for Puttalam Municipal Council
Area
Aspects are shown in table 3.
Roads 0.090
496
Table 3 - Suitability Criteria for Puttalam Table 5 - Weightage of criteria for Puttalam
Municipal Council Municipal Council
Sub
Main criteria Weight Weight
criteria
criteria
Scores
Main
Class
criteria
Sub
Technical Soil Texture 0.750
0.090
Aspects Slope 0.250
Building/ 0 - 50 1 Ground
Buffer 0.250
Residential 50 - 100 2 Environmental Water
[m] 0.303
Areas 100 < 3 Aspects Surface
0.750
Waters
Build up area
Land use 0.303
Cultivation Social and
Residential
area 1 Cultural 0.607 0.607
Area
Social and cultural Aspects
Water area Aspects

Roads 0.090
Rock area
Land Use Type Cloud area The final suitability maps for Unawatuna and
Puttalam area (see Figure 2 and Figure 3)
Sand
2 were created from weighted overlaid maps
Forest and majority filtering processes. The
Boggy area particular sites were identified from field
verification and land area requirement.
Barren land 3
0 - 400 1
Buffer
Major Road 400 - 600 3
[m]
600 < 2
0 - 50 1
Buffer
Minor Road 50 - 100 3
[m]
> 100 2
Alluvial
Red-Yellow
Lotosois;
3
gently
undulating
Technical Aspect
terrain
Soil Texture Type
Reddish Figure 2- Final Suitability Map for
Brown Earths Unawatuna
and Low 1
Humic Gley
Soils
Regosols 2
0 - 0.15 2
Slope Slope [%] 0.15 - 0.3 3
0.3 < 1
0 - 1.5 1
Ground Depth
Environmental
Water [m] 1.5 - 2 2

Aspect
2< 3
0 - 60 1
Surface Buffer Figure 3 - Final Suitability Map for Puttalam
60 - 100 2
Waters [m] Municipal Council
100 < 3
3 - suitable 2 - Moderate 1 – Unsuitable
In the Puttalam Municipal Council,
approximately 1200 hectare was identified as
The weighting criteria for Unawatuna area is
a suitable area. For Unawatuna final suitable
shown in table 4.
areas A, B, C, D were identified according to
497
the land area requirement and field Nutrients Can Be Recycled’, GeoWorld, Apr
verification. The area requirement from the 1999, 12(4): 52-54.
client (NWSDB Galle) was 16000m2. 7 Saaty, T.L., 1980,’ the Analytic Hierarchy
Process’, New York.
4. Conclusions 8 Zahir, S., 1999, ‘Clusters in a Group: Decision

Making in the Vector Space Formulation of the
The defined criteria were more reliable for the Analytic Hierarchy Process’, European Journal of
selected study area. AHP method is superior Operational Research, 112(3), pp.620-634.
for multi-criteria decision-making technique.
GIS is greater flexibility and accurate tool for 9 Ramanathan, R., 2001.’ A Note on the Use of
domestic wastewater site suitability analysis. the Analytic Hierarchy Process for
Hence, Wastewater treatment plant site Environmental impact assessment’, Journal of
suitability has been very effective due to the Environmental Management, 63, pp.27–35
application of spatial analysis tools and
10 Millet, I., Wedley, W.C., 2002. ‘Modelling Risk
decision-making results in the GIS. Using and Uncertainty with the Analytic Hierarchy
AHP technique for multi-criteria weight Process’, Journal of Multi−Criteria Decision
assigning process is quantitatively and Analysis, 11: pp.97−107.
qualitatively reliable, and a priority order for
each criterion selected in the study area had 11 Macharis, C., Springael, J., De Brucker, K. and
been easy to be obtained using this method. Verbeke, A., 2004, ‘PROMETHEE and AHP:
The integration of AHP technique and GIS The Design of Operational Synergies in
analysis provided the efficient result of Multicriteria Analysis.: Strengthening
PROMETHEE with ideas of AHP’, European
suitability analysis of wastewater treatment
Journal of Operational Research, 153(2), pp.307-
plant. Authorities can easily apply this new 317.
approach as it is fast and high accuracy. The
approach applied in this thesis can be 12 USEPA (U.S. Environmental Protection
adopted in another part of Sri Lanka for Agency), 1988, ‘Quality criteria for water. EPA
sustainable future development. 440/5-86-001. USEPA, Office of Water
Regulations and Standards’, U.S. Government
Printing Office (PB81-226759), Washington,
References D.C., USA.
1 Meinzinge, F., 2003, GIS Based Site

Identification for the Land Application of
Wastewater, Linclon University.
2 Siddiqui, M., Everett, J. & Vieux, B., 1996,

‘Landfill Siting Using Geographic Information
Systems: A Demonstration’, Journal of
environmental engineering, 122(6), pp.515–523.
3 Duc, T. T., 2006, ‘Using GIS and AHP

Technique for Land-Use Suitability Analysis’,
International Symposium on Geoinformatics for
Spatial Infrastructure Development in Earth and
Allied Sciences, pp.1–6.
4 Malczewski, J., 1996. ‘A GIS-Based Approach to

Multiplecriteria group Decision making.
International’, Journal of Geographical Information
Systems 10(8), 955-971.
5 Verma, R., Kumari, K. & Tiwary, R., 2009,

‘Application of Remote Sensing and GIS
Technique for Efficient Urban Planning in
India’, Geomatrix Conference Proceedings, pp.1–
23.
6 Fancek, M., Klopcic, J. & Klopcic, R., 1999, ‘GIS

Helps Return Sewage Sludge to Soils Where
498
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[499 506],2017
2017
The Institution
Lanka
Air Pollution Monitoring Through Crowdsourcing

M.I.N.P. Munasinghe, G.I.U.S. Perera, J.K.W.D.B. Karunathilaka, B.C.S. Cooray
and K.G.D. Manupriyal
Abstract: Air pollution, a silent killer in the modern world, is responsible for 7 million deaths in
2012 according to WHO. Sri Lanka, as a developing country, does not possess a real-time air quality
monitoring system but only has a mobile unit, which collects air quality data periodically according to
the Central Environment Authority (CEA). The solution discussed in this paper addresses several
critical aspects of the problem such as collecting, transmitting, storing, analyzing and finally giving
active feedback to individual users as well as general public using a crowdsourcing solution. In our
approach, we have developed a miniature micro-controller based handheld device, which collects
hazardous gas levels (CO, SO2, NO2) using semiconductor sensors and sends data to an android
device using a serial interface. An Android app provides real time warnings to users while it syncs
data to a cloud based web application. This would provide analytical data from all devices using
multiple visual dimensions giving air quality information to the general public through an Open API.
The device, which we have developed is handheld, cheaper and is easily extensible to detect multiple
gases compared to other devices currently in the market and provide rich information to the public
regarding air quality in the country as well as anywhere in the world where the device is used.
Keywords: Air quality monitoring, Mobile data measurement, Air pollution, Crowdsourcing,
Map Reduce
1. Introduction 2. Problem
Air pollution is a staggering worldwide
Air pollution has become one of the most problem. The impacts of air pollution rank it
perilous problem in modern world, which among the most serious and widespread
causes harmful illnesses, dangerous smog and human health hazards in the world. Air
gravely affects to the people’s everyday life. pollution occurs when any harmful gases, dust,
Unfortunately, despite the very real impacts air smoke enters the atmosphere and makes it
pollution cause on people’s lives, it often goes difficult for plants, animals and humans to
unnoticed, unless the pollution is critically survive as the air becomes dirty. Current
high, because of its invisible nature. That is to atmosphere is quite different from the natural
say, most developed countries have atmosphere that existed before the Industrial
implemented effective processes to regulate air Revolution in terms of chemical composition
pollution. Nonetheless, air quality monitoring [1].
and relaying that information to public is at
poor level because government maintained air Breathing polluted air causes chronic illnesses
quality monitoring networks are sparse and do such as asthma and bronchitis and contributes
not possess a proper method to deliver to terminal illnesses such as cancer and heart
gathered information to public. disease. The World Health Organization
(WHO) estimates that in 2012 around 7 million
―Breathe Free‖ is a new generation of hand held premature deaths resulted from air pollution,
device, paired with the android smart phone, more than double the previous estimates [2].
which reveals various hazardous gas levels to
users in real time. Moreover, air quality details In Sri Lanka, presently it is impossible to get
are gathered through crowdsourcing and real-time air quality details of any cities
visualized in a web application. according to Central Environment Authority
Aforementioned web application visualizes gas (CEA). CEA is the only institute, which
levels in a three-dimensional geographical map
Eng. M.I.N.P. Munasinge, Dept. of CSE, University of
with time axis. This paper presents a detailed
Moratuwa.
description regarding development of ―Breath Eng. (Dr.)G.I.U.S. Perera, Dept. of CSE, University of
free‖ device and related applications. Moratuwa.
Mr. J.K.W.D.B. Karunathilaka, Dept. of CSE, University
of Moratuwa.
Mr. B.C.S. Cooray, Dept. of CSE, University of Moratuwa.
Eng. K. G. D. Manupriyal, Dept. of CSE, University of
Moratuwa.
1
499 ENGINEER
measures air quality in Sri Lanka and it follows 3.1 Crowdsourcing
a procedure to gather data in whole country. Crowdsourcing means act of a company or
CEA possess a mobile unit which is taken to institution taking a function once performed by
main town of every district and collects employees and outsourcing it to an undefined
environmental data throughout a week by network of people in the form of an open call.
using instruments in this mobile unit. There is This can take the form of peer-production, but
no real-time air quality monitoring and is also often undertaken by sole individuals.
furthermore, general public cannot access to The crucial prerequisite is the use of the open
this collected data. call format and the large network of potential
laborers [4].When applying crowdsourcing to
Similar products and their limitations this application, function of collecting data is
There are few devices that has been developed transferred from CEA to public. This is a
to monitor air pollution but clearly there are strategic model to attract an interested,
major limitations which have been addressed motivated crowd of individuals capable of
by proposed solution. providing solutions superior in quality and
 AirCasting quantity to those that even traditional forms of
This device aims to collect air pollution methods can.
data using a device and android app.
[3]However, its main limitations are the Furthermore, it would be better, if the people
higher cost, being battery powered have the knowledge of what it might do to the
makes it bulky, and difficult to use crowd and how it may reinstitute long-standing
since people have to charge it. mechanisms of oppression through new
Nowadays people even don’t prefer to discourses. Following reasons are the key
charge their mobile phone. Therefore, it reasons to use crowdsourcing in this project.
would be highly unlike they will be
pleased to use a device, which require • Low cost: Remunerations of Crowdsourcing
charging. Moreover, its data can be relatively low cost [5].
visualization is in very poor state since • Higher accuracy: Large number of people
it uses 2D map with colored squares. contribute for same gas in particular area
reduce the error percentage.
 Mob Geo Sens • Cover more area: Large number of people
This device use phones internal and from diverse locations can contribute, system
other connected external sensors to can cover more area than conventional systems.
collect data which use external power
source [4]. Main limitation is it only Following components are the major
works with old Nokia phones because components of the system.
it depends on Symbian operating
system, which is obsolete now. 3.2 Hand Held Device
This is an Arduino driven small device which
3. Solution consists of three main sense modules. Those are
CO sensor module (MQ-7) [6], SO2 sensor
―Breathe Free‖ is a solution which is not only module (MQ-136) [7] and NOx sensor module
limited to monitoring real time air quality, but (MiCS-2714) [8]. This device is capable of
also a complete system to transmit, store, collecting ambient gas level details and sending
process and present data to public in simple them using serial communication method to an
easily understandable interface. It is composed Android smartphone. In order to connect the
with three major components. A hand-held device and the phone, mini USB to micro USB
device, android application and visualization cable is used and moreover this device is
platform. powered by the smartphone itself hence no
need of external battery or Bluetooth/Wi-Fi
In the problem mentioned above, one critical transferring module for operating the device.
component is data collection method. At the Since this device is aimed to be a low-cost
moment, it was done by the CEA and it has device, this method is more beneficial because
some limitations as mentioned before. battery cost and data transfer module costs are
Therefore, in this new approach data collection reduced from the manufacturing cost.
is done via crowdsourcing.
This device further has an ability to reduce the
power consumption by sleeping the Arduino
ENGINEER 2
500
module and other sense modules. ―Sleep‖ is a Finaldevice with its housing is shown in the
special mode of the processor, which halts following figure.
normal program execution and shuts down its
internal components to reduce power
consumption. Once a processor is asleep it can
only be awakened by specific events such as an
external interrupt from a button press, or a
timer counting a period of time. Different sleep
modes disable various internal clocks and
components to reduce power consumption. In
this product, to save the maximum amount of
power, ―SLEEP_MODE_PWR_DOWN‖ has
been used. In this mode, all clocks and
oscillators are turned off and this is the
maximum power down mode that can be
achieved in Arduino[9].
Detecting ranges of aforementioned gas Figure 2 – Device with housing

modules are shown in the following table.
3.3 Android Application
Table 1 – Sensor details An Android application is developed in order
6HQVRU 'HWHFWLRQ 'HWHFWLQJ to provide an interface to handheld device by
0RGXOH *DV 5DQJHSSP allowing user to control the device and
04 &2 providing readings from the device.
04 62
Functionalities of the Android application are,
0L&6 12
1.Displaying current gas levels.
2.Warn user when gas levels are above a
These sensors should be operated within their
threshold value.
operational temperature levels in order to have
3.Send data to online server.
good accuracy. Therefore, the circuit was
4.Change settings of the device.
designed to minimize heat exchange between
sensors by keeping them apart while making Primary requirement of the Android
the circuit footprint minimal in order to application is to sync gathered data to online
provide easy usage. The design of the circuit is database which is also the crowdsourcing
shown in the figure below. component. In order to achieve this, android
application uses a NoSQL database called
Couchbase as the medium to store data
received by the device. Couchbase provides
automatic syncing capability which satisfy the
main requirement. Application syncs data to
online server periodically when an internet
connection is available. This sync is a one-way
NO2 Module CO2 Module sync from the mobile phone to the online
database server. Moreover, nature of the
gathered data is not structured. Same app
should be able to use with multiple type of
devices since the device can be easily extensible
by adding multiple sensors. Therefore, NoSQL
type database is most suitable for requirement.
Arduino Nano
SO2 Module Measured gas level data are fed to the mobile
phone via USB interface and Android
application collects received data and saves to
local Couchbase database. Real-time gas levels
are displayed on the main screen of the
Figure 1 – Device Circuit application. User can easily view what are the
pollution gas levels in the area the user is at.
3
501 ENGINEER
Furthermore, application warns to the user of communication to make efficient use of the
any increased reading of a pollution gas. The network and disks [1]. MapReduce allows
application produces an alert if it senses that a programmers with no experience with parallel
pollution gas level is above some threshold. and distributed systems to easily utilize the
These threshold values are determined resources of a large distributed system. A
according to standard maximum gas levels a typical MapReduce computation processes
human can withstand without a problem. many terabytes of data on hundreds or
. thousands of machines [11].
Another key feature in the application is it
provides the user to control the device through In this application, large amount of data is
few important settings. Application allows user gathered from large number of devices across
to configure the time interval, which the device various locations and data is distributed in
should collects data from the sensors. This is multiple nodes. Because of this large volume
important since this interval becomes smaller, and distributed manner, calculations include
power consumption increases but provide more collected data is a challenging task. Therefore,
real-time data. Another configuration it MapReduce support in Couchbase DB is used
provides is the device sleep settings which lets do calculation which are required to produce
user to set a time which put device in to sleep maps. For example, in average map view it is
mode in order to preserve battery power. required to calculate gas level averages in each
Another setting is to select type of internet the city. In order to do that map function will emit
connection which should be used to sync data key as city and gas values as value parameter.
like Wi-Fi or mobile data. Corresponding reduce function will reduce that
values and calculate average of each value and
3.4 Visualization and Data Processing return a document with key as city and value as
Platform average values of each gas.
In this project this is the most important part
that distinguishes this product from other 3.4.2 Outlier Detection and Removal
similar products. Key thing to mention is that In this crowdsourcing approach, there is a
this visualization platform can be easily higher probability of getting incorrect values to
extended to add more maps to visualize data in the online sever because of various reasons like
rich formats. malfunctioning sensors because of humidity
and temperature differences and even
Visualization platform consists of web server intentional incorrect data inputs. Therefore,
and Couchbase NoSQL database. Reason to use data cleaning or outlier removal is a very
NoSQL database is that data which are crucial aspect in order to filter accurate data set
gathered are schema less, as mentioned before. to visualize.
And the other advantage is that Couchbase
server uses cashing to optimize the data access. An outlier can be defined as an observation that
In the case of data growth Couchbase server lies an abnormal distance from other values in a
can be easily extended by adding more nodes random sample from a population. In a sense,
to the database. this definition leaves it up to the analyst (or a
consensus process) to decide what will be
3.4.1 Data Processing Using MapReduce considered abnormal. Before abnormal
Major challenge in this solution is handling observations can be singled out, it is necessary
large amounts of data and process them in to characterize normal observations [12]. In [13]
order to generate useful information to end the authors define an outlying observation, or
user. To overcome this MapReduce has been outlier, is one, that appears to deviate
used. considerably from other members of the sample
in which it occurs. Another definition as
MapReduce is a programming model and an observed in [14] is, it is an observation that
associated implementation for processing and deviates so much from other observations so as
generating large datasets that is amenable to a to arouse suspicion that it was generated by a
broad variety of real-world tasks. Users specify different mechanism.
the computation in terms of a map and a reduce
function, and the underlying runtime system One of the major tool in descriptive analysis is
automatically parallelizes the computation quartiles whichdivides the range of data in four
across large-scale clusters of machines, handles parts each having 25% ofthe data. Q1 is the 25%
machine failures, and schedules inter-machine point, Q2 which is also same asmedian, is the
ENGINEER 4
502
50% point and Q3 is the 75% point [15]. In order Earth was its web plug-in was set to
to remove outliers in this application, data of discontinue after December 2015 due to a
each gas has been grouped by city and security issue. With that problem, the only
quantities are calculated for each city. Location option was to use next best library. Cesiumjs
data is gathered using mobile phone location has client side rendering of data which was
information. Following method is used to much suitable for this use case since it leverages
identify lower and upper bound in each city. rendering the map to the client which reduce
the load on server. Cesiumjs is also built on
1. Arrange data in order. plugin free technology which gives assurance
2. Calculate first quartile (Q1) that there will not be problems like with Google
3. Calculate third quartile (Q3) Earth. For the past few years,progressively
4. Calculate inter quartile range (IQR)=Q3-Q1 more users were moving to Cesiumjs from
5. Calculate lower boundary= Q1-(1.5*IQR) Google Earth. There were two main reasons for
6. Calculate upper boundary= Q3+(1.5*IQR) this migration [19]. The first reason is the lack
7. Anything outside the lower and upper of high resolution image content. Google lost its
boundary is an outlier [16]. source of high quality imagery in few years
back. Since then Google did not have a scalable
Figure 3 shows the graphical representation of high resolution image supply. The other reason
this calculation. was it is required to use a plugin to view
Google Earth in web applications. This became
a major issue when Google released a new 64-
bit Chrome version which does not support
Google Earth plug-in. On the other
hand,Cesiumjs uses webGL [21] to render 3D
maps. Therefore, it does not have problems
associated with Google Earth. As of now
Cesiumjs does not have much high-quality
imagery. But Cesiumjs is holding discussions
with early imagery providers for Google Earth
and future seemsoptimistic.
Google Maps for 2D visualization

In order to provide 2D visualization of city wise
pollution data, several visualization libraries
were considered as before. One of them was
Google Maps [22] and other alternatives
Figure 3 – Identifying outliers
included Bing maps and open street maps.
3.4.3 Data Visualization Google Maps has better and more features
Visualization component of this project is considering other visualization alternatives.
intended to provide simple and informative Support for heatmaps is a one major concern
representation of collected, cleaned and and additionally Google maps consists with
processed data to public. Following are the easy to use API (Application Programming
types of maps which are currently added in Interface) and better documentation.
order to showcase the possible multiple modes
of visualization and their related technologies. Following maps are the different visualization
modes that has been implemented.
Cesiumjs for 3D map visualization
With environmental pollution data, 3D  3D Map
visualization can provide users a better and In this map, each bar represents a reading of a
more artistic picture. After considering multiple certain hazardous gas. Importance of this map
chartinglibraries, Cesiumjs [17] and Google compared to other maps is that it is more
Earth [18] were selected because they match convenient to get an idea about gas level
with most of the requirements. Google earth changes from the relative height of each bar.
provides better visualization options and From other 2D maps that has been developed, it
proved to be better with a higher existing user is difficult to get idea about relative levels from
base. Cesiumjs was the other library which the just color and shown gas value. Moreover,
comes after Google Earth in both feature wise the color of this map is related to the intensity
and user base wise.Main issue with Google
5
503 ENGINEER
level of the gas. Higher intensity gives intense average, map reduce technique which was
color. discussed before has been used.
Figure 6 – Gas level average map

Figure 4–3D map  Heat Map
This map represents the amount of data
 2D Dot Map available for a given gas. If higher number of
Visualization of each data point as a dot with a users are available, color becomes red.
color. Here each data point represents a single Importance of this map is, if there are large
reading observed by a user. Advantage of this number of users in particular area, that means
map it gives clear top down view of data and the accuracy of data in that area is high.
gives straight forward color representation of
different gas levels rather than using intensity
in 3D map. Which gives direct idea about gas
level in particular location.
Figure 7 – Heat map
3.4.4 Public API (Application Programming

Interface) to retrieve air pollution data
It was able to develop an API for public which
can be used to retrieve air pollution data in
Figure 5– 2D dot map particular area. By providing latitude and
longitude to this API, it returns average
 Average Map pollution gas levels of that location.
Aim of this map is to visualize city wise data to Development of this API is very useful because
provide overall notion. Here each cylinder it facilitates large number of other types of
represents a city. Height of the cylinder applications in air quality monitoring domain
represents the average value of a certain which depends on data collected by this
pollution gas. Radius of the cylinder represents system. For an example, billboards in cities that
the radius covered by city. Importance of this shows real time air pollution data, desktop or
map is it represents the average pollution of mobile widgets that show air pollution like
particular city. In order to calculate this
ENGINEER 6
504
weather apps which currently have or small Acknowledgement
devices which shows air pollution data can be Authors thanks to central environment
developed. authority for helping us by giving access to
their laboratories to calibrate sensors and
4. Discussion providing access to mobile air pollution
General use case of this device would be the monitoring vehicle to get more information
ability to view air pollution details to the public regarding current air quality data gathering.
using visualization platform, which is exposed
via web application. In order to test that References
scenario, testing is carried out by using the
device while traveling the train; Figure 5, 1. Daly, A., and Zannetti, P., An Introduction to
Figure 6 and Figure 7 shows the actual data Air Pollution, California: The Arab School for
gathered from the device while traveling from Science and Technology, 2007.
Colombo Fort railway station to Gampaha
railway station and near the University of 2. U. N. E. Programme, "Air Pollution: World’s
Moratuwa premises.These are the maps which Worst Environmental," UNEP Year Book 2014
emerging issues update, December 2014, pp. 43-
are generated by the visualization platform and
47.
exposed via web application. Test results shows
bright red dot in heat map (Figure 7) indicating 3. http://www.takingspace.org/airbeam-
large number of data points near university technical-specifications-operation-
area and this is due to higher amount of testing performance/. Visited 9th Sept 2015.
near the university premises. Moreover,
average map (Figure 6) shows there are higher 4. Kanjo, E., Benford, S., Paxton, M., Chamberlain,
concentration of CO near university area and A., Fraser, D. S., Woodgate, D., Crellin, D., and
Gampaha area. But near Colombo Fort area, it Woolard, A., "MobGeoSen: Facilitating Personal
is considerably low. This may be caused by the Geosensor Data Collection," Pers Ubiquit
sea wind near Colombo area. Since only one Comput, vol. 12, 2008, p. 599–607.
device was used in small amount of time in one
5. http://crowdsourcing.typepad.com/cs/2006/0
particular area, collected data is not sufficient to
6/crowdsourcing_a.html,Visited August 2015.
support these claims.
6. Schenk, E., and Guittard, C., "Crowdsourcing:
Major limitation of this project was lack of more What can be Outsourced to the Crowd, and
devices and more data to support these claims Why?", 2009.
with less error percentage. However, with the
resources available to the project, only one 7. https://www.sparkfun.com/datasheets/Sensor
device could be developed, and was used as a s/Biometric/MQ-7.pdf, Visited 20th Jan 2016.
proof of concept.
8. http://www.china-total.com/Product/meter/
gas-sensor/MQ136.pdf, Visited 20th Jan 2016.
5. Conclusion
9. http://sgx.cdistore.com/datasheets/e2v/1107-
It can be concluded that the authors were able Datasheet-MiCS-2714.pdf, 20th Jan 2016.
to bring forward a vital solution to help
support for improving human health, through 10. http://playground.arduino.cc/Learning/Ardui
real-time air pollution monitoring. The noSleepCode, 20th Jan 2017.
discussed work here presents a complete
solution which is composed of parts for 11. Jeffrey, D., and Sanjay, G., "MapReduce:
collecting, transmitting, processing, warning simplified data processing on large clusters",
and providing information to public. The Communications of the ACM - 50th anniversary
issue: 1958 - 2008, vol. 51, no. 1, 2008, pp. 107-
system fills the gap of awareness not only in Sri
113.
Lanka but also in any country in the world.
Above all, data collected by the system will be 12. Jeffrey, D., and Sanjay, G., "MapReduce: A
crucial element for researches to study adverse Flexible Data Processing Tool," Communications
effects and predict trends in air pollution. The of the ACM - Amir Pnueli: Ahead of His Time,
device can be a cost-effective tool for research vol. 53, no. 1, 2010, pp. 72-77.
and field studies relevant to air-quality
monitoring as well 13. http://www.itl.nist.gov/div898/handbook/ind
ex.htm, Visited 23rd Jan 2016.
7
505 ENGINEER
14. Mehmed Kantardzic, ―Data Mining—Concepts,
Models, Methods, and Algorithms ―.
15. Hawkins D,‖ Identification of Outliers‖,

Chapman and Hall, 1980.
16. Goswami, S., and Chakrabarti, D. A., "Quartile

Clustering: A quartile based," JOURNAL OF
COMPUTING, vol. 4, no. 2, 2012.
17. D. M. D. J., E. R. L. Sunitha, "Automatic Outlier

Identification in Data Mining," International
Journal of Advanced Research in Computer and
Communication Engineering, vol. 3, no. 6, 2014.
18. https://cesiumjs.org/ , Visited 25th Jan 2016.
19. https://www.google.com/earth/, Visited

25thJan 2016.
20. https://medium.com/@CyberCity3D/globetrot
ting-the-fall-of-google-earth-the-rise-of-cesium-
and-what-it-all-means-for-the-virtual-
1d261fdc25c9#.bg0olksk3, Visited 25th Jan 2016.
21. https://www.chromeexperiments.com/webgl ,
Visited 25th Jan 2016.
22. https://developers.google.com/maps/?hl=en,
Visited 25th Jan 2016.
ENGINEER 8
506
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[507 515],2017
2017
The Institution
Lanka
A Conceptual Tool for Assessing Building Life Cycle

Carbon Emissions in the Context of Sri Lanka
R.P. Kumanayake, H.B. Luo and R.M.P.S. Bandara
Abstract: At present, the world is facing an enormous challenge due to the adverse
environmental impacts of various human activities including construction. Construction is one of the
major industrial sectors of Sri Lanka, within which buildings sub-sector has the highest contribution to
the value of work and raw material consumption. Consequently, buildings are responsible for a
significant share of the total energy use and carbon emissions. Therefore, the development of suitable
mechanisms for assessing building environmental performance with regard to energy use and carbon
emissions is a timely requirement. Most of the existing systems are focused on the developed and
temperate climate countries and these have a limited use in Sri Lanka, due to a range of climatic,
geographical, technological, economic, social, political and cultural diversifications unique to the
country. This paper presents the initial design for developing an automated tool for assessing life
cycle carbon emissions of buildings in the specific context of Sri Lanka. Within the proposed tool,a
building information database will be developed using MySQL database management system, by
making use of existing databases such as Inventory of Carbon and Energy (ICE), relevant data from
recent research literature and field data from building contractors, material manufacturers and
equipment suppliers. The database is integrated with a computerized programme for estimating life
cycle carbon emissions which is developed using Visual Studio C# Object Oriented Programming
language in the .NET framework. Using the proposed tool, the life cycle carbon emission assessment
of buildings from raw material extraction to the deconstruction can be estimated. The tool will
facilitate the quantification of the impacts of life cycle carbon emissions of buildings and provide a
much needed environmental decision making mechanism to the construction professionals, hence
contributing to environmentally sustainable construction.
Keywords: Carbon emission, building life cycle, environmental assessment, sustainable

construction
1. Introduction emissions over long term. For any

improvements to take place, assessment of the
Buildings have a significant environmental current building performance is essential. Life
footprint which includes 40% of global energy cycle assessment (LCA), that assesses a range of
consumption and one third of greenhouse gas environmental impacts throughout the lifetime
(GHG) emissions worldwide [1].On the other of a product is considered to be the best
hand, it was found that the building sector has technique to evaluate the full extent of the life
a considerable potential for energy efficient cycle carbon emissions of a building [4].
improvements and GHG reduction without
significant increase in the investment cost [2]. Life cycle carbon emissions assessment is a
specialized form of LCA, which evaluate the
The Kyoto protocol was targeted at reducing carbon emissions over the building life cycle,
5% of global GHG emissions within the time facilitating the selection of low-carbon emitting
period of 2008–2012[3]. Carbon dioxide (CO2) is
considered as the predominant GHG and the Eng. R.P. Kumanayake, B.Sc.Eng. (Hons.) (Moratuwa),
M.B.A. (Colombo), AMIE(SL), Ph.D. Student, School of
CO2equivalent emission values of the other Civil Engineering and Mechanics, Huazhong University of
greenhouse gases are usually given in terms of Science and Technology, China.
global warming potential (GWP)[1]. Many Senior Lecturer, Department of Civil Engineering,
General Sir John Kotelawala Defence University, Sri Lanka.
studies related to the assessment of carbon
emissions of buildings were initiated as a result Prof. H.B. Luo, B.A. (Civil Engineering), Ph.D.,
of introducing the Kyoto Protocol. The energy Deputy Dean, School of Civil Engineering and Mechanics,
Huazhong University of Science and Technology, China.
consumption is identified as the main cause of
carbon emissions in buildings. As buildings Eng. R.M.P.S. Bandara, BSc.Eng. (Hons.) (Moratuwa),
have long life spans, the decisions taken at M.Eng. (Moratuwa), AMIE(SL), Senior Lecturer,
present have the ability to affect the carbon Department of Mechanical Engineering, General Sir John
Kotelawala Defence University.
1
507 ENGINEER
materials, systems, and processes for buildings. well as a carbon emission assessment tool for
In recent years, carbon emission assessment has the green building certification in Sri Lanka.
gained attention worldwide due to its direct
association with global warming. In 2. Literature Review
conventional buildings, over 80% of carbon
emissions take place in the operational phase Based on a detailed study of literature, it was
due to energy used for heating, cooling, found that the majority of the existing energy
ventilation, lighting and appliance usage[5]–[7]. and carbon related research and databases
The other life cycle phases such as focus on the developed and temperate climate
manufacturing of building materials, building countries. Only a few examples of developing,
construction, maintenance, renovation and tropical countries exist[13]–[16]. Ruuska[17]
demolition are responsible for the rest of the investigated how the energy and carbon
carbon emissions. profiles of these two categories of countries
differ. The specific areas which should be
Although a variety of building environmental focused when applying the existing research
assessment systems are being used globally, findings to developing and tropical countries
most of these were established by the were identified as “ level of operational energy
developed countries. Despite their effectiveness use, transition from traditional to modern
in assessing building environmental building materials, role of insulation, role of
performance in their countries of origin, they advanced building systems, technology of
are less successful in other regions of the world material production, energy production
such as South Asia, due to a wide range of methods and energy carriers”[17]. In their
climatic, geographical, technological, economic, studies, Abeysundara et al. [13], Chau et al. [18]
social, political and cultural diversifications and Gonzalez and Navarro [19] made suitable
unique to these regions[8].In Sri Lanka, adjustments to make the existing data
Construction is one of the major industrial compatible with the specific context of a
sectors, within which the buildings sub-sector particular country.
has the highest share of value of work and raw
material usage[9].Sri Lanka has already A significant lack of country specific research
identified the importance of developing and related data on building energy and carbon
building environmental assessment systems, emissions in Sri Lanka can be observed at
codes and standards which take the regional present. Dias and Pooliyadda[20] developed a
and national differences into account. The database management system to present and
Energy Efficient Building Code developed by compute the embodied energy and carbon
Sri Lanka Sustainable Energy Authority[10] and emission of a number of building materials
GREENSL Rating System developed by Sri commonly used in Sri Lanka. Although some
Lanka Green Building Council [11] are two Sri Lankan researchers have been using the
such initiatives. But the assessment tools existing building environmental assessment
necessary to monitor the compliance to the tools and life cycle inventory databases for their
codes and to assist the green building studies, these do not reflect the true conditions
certification are yet to be developed. of the country as they were originally intended
for a different country or region. Thus, the
The objective of the present research is to necessity of developing building environmental
develop the initial framework and the assessment programs to match the unique
associated conceptual model for a conditions of the country has emerged.
computerized building life cycle carbon
emission estimation tool, which is integrated With the current level of integration of the
with a carbon emission database. The construction industry with information
insufficiency of the relevant regional and technology, there are numerous software and
country specific data was found to be a huge other computerized tools developed globally
challenge. Therefore, the data available in the for energy and carbon emission assessment of
recent literature and globally accepted buildings. Most of these were developed by the
databases such as Inventory of Carbon and researchers and organizations in developed
Energy (ICE) [12]were referred and adjusted to countries with the use of databases and
match the specific context of Sri Lanka. The information readily available. These include
proposed program can be used as a decision SUSB-LCA (Korea), AIJ-LCA (Japan), GEM-21P
making tool in low-carbon building design as (Japan), Carbon Navigator (Japan), BECOST
(Finland), LISA (Australia), ENVEST2 (UK),
ENGINEER 2
508
LCA-MCDM (U.S.) and Eco-Quantum (The Kang et al.[34]suggested a systematic model for
Netherlands)[21]. developing sustainable building assessment
tools by establishing a logical performance
Apart from the above, researchers from around assessment framework. It can also be used as a
the world have been developing frameworks, guideline in developing a building life cycle
models and related computerized programs carbon emission assessment framework and the
based on the existing green building related model in the context of Sri Lanka.
certification systems, life cycle databases,
standards and codes of the respective countries. For developing these systems, the existing
BEGAS 2.0, a framework to support the Green carbon emission databases, standards and
Building Index (GBI) certification system of codes of the country concerned were used. In
Korea was introduced by Roh et al.[22]. In the existing literature a significant gap can be
order to assess life cycle carbon emissions and observed in relation to the availability of
identify environmentally-friendly construction related research, data sources and carbon
techniques for a specific CO2 emission target, an emission programs in the context of
optimum design program (SUSB-OPTIMUM) developing, tropical countries like Sri Lanka.
for apartment houses was developed[23]. A The present study is expected to bridge this gap
LCA model to estimate the energy use and to some degree by establishing a framework
carbon emission of buildings in Beijing, China and the associated conceptual model as an
was developed by Aden et al. [24].A life cycle initial step in developing a life cycle carbon
carbon emission program to be used in the emission estimation program for buildings in
schematic design stage was introduced by Baek the specific context of Sri Lanka.
et al. [25] which was based on the global
environmental management system GEM-21P. 3. Methodology
A simplified life cycle carbon emission There are a number of LCA techniques
assessment model and the associated tool (B- available today. ISO 14040 standard series
SCAT) was developed especially focusing on introduced by International Organization for
the early design stages, which is expected to Standardization (ISO) presents a general
support low carbon building design [26]. Fu et methodological framework, which make the
al. [27] developed a computer based carbon comparisons of different LCAs easier. In recent
emissions calculation system for the years, it has emerged as a preferred framework
construction phase of buildings. A computer- and followed by many researchers [27], [35]–
based model for assessing building life cycle [37].ISO 14040 Environmental Management -
carbon emission was developed by Kwok Life cycle assessment standard [38]identifies
[28]and an assessment framework to evaluate four interactive phases of a LCA;goal and scope
the embodied carbon impacts of residential definition, inventory analysis, impact
buildings in China was modelled by Li et assessment and interpretation. The framework
al.[29].In a subsequent research, an automated for the proposed LCA tool which is illustrated
tool for estimating life cycle carbon emission of in Figure 1 was developed according to it.
residential buildings in China, was
developed[30]. Some researchers have 3.1 Goal and scope definition
integrated carbon emission estimation
programs with Building Information Modelling The proposed tool is intended for assessing the
to enhance the effectiveness of carbon emission life cycle carbon emissions of buildings in Sri
estimation at early design stages [31], [32]. Lanka. The system boundary includes both
Moncaster and Symons spatial boundary and life cycle process
[33]introduced“Embodied Carbon and Energy boundary for the calculation of carbon
of Buildings (ECEB)” tool, designed as an emissions. The spatial boundary is defined as
approach for early design stage of UK the closed 3-dimensionsl space of the building.
buildings. Life cycle process boundary includes all phases
of cradle-to-grave life cycle process and 1m2 of
Reviewing of the processes, assumptions, gross floor area of the building per year was
specific features, highlights and limitations considered as the functional unit of the study.
related to these existing frameworks and
systems provide a good starting point for
developing a similar system for Sri Lanka.
3
509 ENGINEER
Figure 1- LCA framework for developing the building life cycle carbon emission estimation tool.
3.2 Inventory analysis the life cycle carbon emission of the building
will be produced as the final output. For this,
The life cycle inventory includes the inputs of the proposed carbon emission estimator
the building life cycle process in the forms of program will be developed using Visual Studio
energy and materials. By analysing a sample of C# programming language in .NET framework.
buildings, the major construction materials that Given the basic project information and
make up more than 80% of carbon emissions in material quantities, it will estimate the carbon
typical Sri Lankan buildings were identified. A emissions using the programmed formulae for
similar approach was followed in previous each life cycle stage incorporating material and
studies [21], [22], [29], [32], [39]. The material energy related data stored in the MySQL
quantities and building information are database. In order to make the LCA results
obtained from design drawings, bill of more representative of Sri Lanka, the energy
quantities and technical specifications. The coefficient values obtained from the existing
energy data are obtained from recent literature databases and research literature will be
as well as local and international reports. Due converted to carbon emission coefficients by
to the lack of national databases in relation to considering the specific energy generation mix
construction machinery, transportation, of the country at present.
demolition and waste disposal in Sri Lanka,
suitable assumptions are made by consulting 3.2.2 Interpretation
experienced construction professionals and
referring to records of previous construction The results will be interpreted to give the total
projects. Due to the common challenge of data life cycle carbon emissions of the building in
unavailability in building life cycle studies, kgCO2/m2/year.
especially in developing countries, many
researchers have followed a similar approach 3.3 Estimation of carbon emissions
[14], [16], [40]. The computerized life cycle
inventory is developed using MySQL database Based on the carbon emission coefficient
management system. method and the LCA approach, total life cycle
carbon emissions of a building can be
3.2.1 Impact assessment calculated as given in equation (1)[41].
The proposed tool is designed to compute the …(1)

carbon emissions of each life cycle phase and
ENGINEER 4
510
Where CLC represents the total life cycle carbon 3.3.3 Carbon emissions ofon-site construction
emissions of a building (kgCO2) and CM, CT, CC,
CO&M and CD represent the carbon emissions at Carbon emission during on-site construction,
material production, transportation, CCis due to the operation of construction
construction, operation and maintenance and machinery and equipment during various
deconstruction phases (kgCO2) respectively. construction activities, which is given as:
The carbon emissions at each life cycle stage are
presented by equations (2)-(7). ∑ …. (5)
3.3.1 Carbon emissions at material production Where j is the total number of on-site
construction activities,QiandRi are the quantity
The carbon emissions at the material (m3 or kg)and fuel/electricity use rate (l/m3,
production stageCM, which includes the raw kwh/kg etc.) of the construction activity
material extraction and building material irespectively.fc,i(kgCO2/l or kgCO2/kwh) is the
production can be calculated as in equation carbon emission coefficient of the type of
(2)[30]: energy used for the construction activity i.
∑ ….(2) 3.3.4 Carbon emissions at building operation

and maintenance (2)
Where n is the total number of material types,
miand fm,iare the quantity (kg) and the carbon Carbon emission at the operation and
emission coefficient (kgCO2/kg) of type i maintenance stage, CO&Mis due to the
material respectively. The value of fm,ican be fuel/electricity consumption of the building
calculated using the equation (3)[41]. operation and the materials replaced in
maintenance, which is estimated as[30]:
….(3) (3)
∑ ... (6)
Where eiand βi are the embodied energy
coefficient (MJ/kg) and the average carbon Where Qe is the average annual electricity
emission factor (kgCO2/MJ) of the energy consumption of the building (kWh/year), fe the
source used for manufacturing type i material carbon emission coefficient of electricity
respectively. As multiple sources are used for (kgCO2/kWh) and Y the assumed lifespan of
energy production, an average carbon emission the building (years). Electricity is considered as
factor (βi) is generally used for a specific the energy source for building operation and it
country. For example, the value of βi for the is calculated based on the monthly electricity
power generation in Sri Lanka can be calculated bills. In the case of new buildings, the energy
by considering the primary energy mix of the consumption can be computed based on the
country at present that comprises of biomass expected functions, power demands and
(38.8%), petroleum (38.9%), coal (9.9%) major working time of the building[44]. For the
hydro (9.5%) and new renewable energy maintenance stage, k is the total number of
(2.9%)[42]. In several previous studies, a similar material types required for repairs and
process for computing an average carbon replacement, ri the amount of building material
emission factor was carried out [19], [43]. i that needed to be replaced or used for repairs
(kg), fm,i is the carbon emission coefficient of
3.3.2 Carbon emissions at material type i material (kgCO2/kg) , Y is the assumed
transportation lifespan of the building (years) and R the
renewal or repair interval (years) of material i.
The equation (4) is used to calculate the carbon The value (Y/R) is the replacement factor of the
emitted during the delivery of materials to the material considered[18].
construction site (CT)[30].
3.3.5 Carbon emissions at deconstruction
∑ ) …. (4)
The carbon emissions at the deconstruction
WhereDi is the average transportation distance stage, CDcan be regarded as the summation of
(km) of type imaterial and ft,iis the carbon carbon emissions for demolition, transportation
emission coefficient for unit material to landfill and recycling which is given as[30]:
transported over unit distance of type i material
(kgCO2/kg.km).(1)
5
511 ENGINEER
∑ ∑ site or recycling facility (km) and ft,iis the
∑ …. (7) carbon emission coefficient for unit material
transported over unit distance of type i material
The carbon emissions due to building (kgCO2/kg.km).
demolition is given by the component
∑ where m is the total number of The carbon emissions due to recycling of
demolition activities, Qd,iand fd,iare the quantity materials is given by the component∑
and carbon emission coefficient of demolition , where r is the total number of
activity respectively. recyclable material, mr is the weight of
recyclable material i(kg), wr the proportion of
The carbon emissions due to the transportation recycling and fm,i the carbon emission coefficient
of demolished material is given by the of the type i recyclable material (kgCO2/kg).
component∑ , where p is the The data related to building demolition,
total number of demolished materials, md the disposal and recycling are taken from the
weight of demolished material to be literature on similar studies and, whenever
transported for landfill or recycling (kg), Dd the available, the existing practices in Sri Lanka are
distance from the demolition site to the landfill considered.
Figure 2 - The conceptual model for the building life cycle carbon emission estimation tool.
4. Development of the carbon coefficients for the relevant energy sources are
emission estimator tool referred from Sri Lankan and international data
sources [1], [45], [46]. The embodied energy
The development of the building carbon coefficients for building materials are taken
emission estimation tool will be carried out in from recent research literature and
two stages; development of carbon emission international databases [12], [47] and converted
database and development of the estimator to the carbon emission coefficients by using the
program including the user interface. average carbon emission factor for Sri Lanka
which is described in section 3.5.1.A similar
4.1 Development of carbon emission database approach in „localization of data‟ has been
followed in previous research [19], [43].
The data needed for computing carbon Transportation and machinery data are
emission of each life cycle phase according to collected from research literature and by
equations (2) to (7)are of four types; energy, consulting construction professionals in Sri
materials, transportation and construction Lanka. In selecting most appropriate data,
activities. The lack of data availability for recent data from Sri Lanka and Asian countries
energy and carbon emission related studies has with similar construction practices are given
been a major challenge to the researchers of the priority.
developing countries. The carbon emission
ENGINEER 6
512
MySQL database management software is used the unique conditions of the country. After the
to create the carbon emission database, which development of the proposed tool, it will be
can be easily integrated with the estimator verified by using several case study buildings
programme. in Sri Lanka. The database will need periodical
revision due to the changes in energy and
4.2 Development of carbon emission carbon emission values in future. This tool can
estimator program be used in different life cycle stages for multiple
purposes including low-carbon building design
The carbon emission estimator programme is and green building certification, hence
developed using Visual Studio C# promoting sustainable building practices in Sri
programming language in the .NET Lanka.
Framework. The model for developing the
proposed tool is illustrated in Figure 2.The user References
interface of the program is developed using
Visual Studio C# forms application. Upon 1. United Nations Environment Programme,
entering the program, a user has to provide two “Common Carbon Metric: Protocol for
types of input manually;(1)basic information of Measuring Energy Use and Reporting
the project such as project name, location, type Greenhouse Gas Emissions from Building
of building, type of structure, number of floors, Operations”, 2010.
gross floor area and building life span (2)
2. Intergovernmental Panel on Climate Change,
building life cycle information such as major
“Climate Change 2007: Synthesis
material quanities and annual energy Report”,Contribution of Working Groups I, II
consumption. and III to the Forth Assessment Report of the
Intergovernmental Panel on Climate Change,
Using this data the building life cycle carbon Core Writing Team, Pachauri, R.K and
emission estimator program calculates the Reisinger, A (eds.).IPCC, Geneva, Switzerland.
carbon emissions of the each life cycle stage by 2007.
using the programmed formula and the carbon
3. Kim, S., Lee, S., Na, Y. J., Kim, J. T.
emission database. As the output the total
“Conceptual Model for LCC-based LCCO2
building life cycle carbon emission is given in Analysis of Apartment Buildings”,Energy
kgCO2/m2/year. The carbon emissions of Build., Vol. 64, 2013, pp. 285–291.
different life cycle stages and material types can
be viewed both in tabular and graphical forms, 4. UNEP SBCI (Sustainable Buildings and
facilitating easy comparison of the results. The Climate Initiative), Buildings and Climate
proposed tool can be used for multiple Change. United Nations Environment
purposes. Low-carbon building design, Programme, 2009.
evaluation of alternative project solutions and
5. Junnila, S., “The Environmental Impact of an
as an assessment tool in green building
Office Building throughout its Life Cycle”,
certification are a few among these. PhD Thesis -Helsinki University of
Technology,2004.
5. Conclusion
6. Suzuki, M., Oka,T., “Estimation of Life Cycle
This study aims at developing the initial Energy Consumption and CO2 Emission of
framework and the conceptual model for a Office Buildings in Japan”,Energy Build., Vol.
28, No. 1, 1998, pp. 33–41.
computerized building life cycle carbon
emission estimation program in the context of 7. Adalberth, K., Almgren, A., Petersen, E. H.,
Sri Lanka. The framework is based on a cradle- “Life-Cycle Assessment of Four Multi-
to-grave building life cycle assessment. The Family Buildings”,Int. J. Low Energy Sustain.
proposed tool integrates a computerized Build., Vol. 2, 2001, pp. 1–21.
database and an automated estimator program
with a user-friendly interface. The concept of 8. Ding, G. K. C., “Sustainable Construction-
the major building materials, which are The Role of Environmental Assessment
responsible for 80% of carbon emission in a Tools”, J. Environ. Manage., Vol. 86, No. 3,
2008, pp. 451–464.
typical Sri Lankan building, is used for
developing the tool. Due to the lack of country- 9. Survey of Construction Industries
specific energy and carbon emission data, the 2013,Department of Census and Statistics Sri
data taken from recent literature and Lanka, 2015.
international databases are adjusted to match
7
513 ENGINEER
10. “Innovative, Energy Efficient Building Cycle CO2 Simple Assessment System using
Concepts”, [Online]. Available: Standard Apartment Houses of South
http://www.energy.gov.lk/your- Korea”, Renew. Sustain. Energy Rev., Vol. 15,
business/building-energy/energy-efficient- No. 3, 2011, pp. 1454–1467.
building-design, Accessed : May 7, 2017.
22. Roh, S., Tae, S., Suk, S. J., Ford, G., Shin, S.,
11. Green Sl ® Rating System for Built “Development of a Building Life Cycle
Environment, Green Building Council of Sri Carbon Emissions Assessment Program
Lanka,2015. (BEGAS 2.0) for Korea‟s Green Building
Index Certification System”, Renew. Sustain.
12. “Inventory of Carbon and Energy (ICE) Energy Rev., Vol. 53,2016, pp. 954–965.
Version 2.0”, University of Bath, UK, 2011.
23. Roh, S., Tae, S., Shin, S., Woo, J.,
13. Abeysundara, U. G. Y., Babel, S., Gheewala, “Development of An Optimum Design
S. H., “A Matrix in Life Cycle Perspective for Program (SUSB-OPTIMUM) for the Life
Selecting Sustainable Materials for Buildings Cycle CO2 Assessment of an Apartment
in Sri Lanka”,Build. Environ., Vol. 44, No. 5, House in Korea”,Build. Environ., Vol. 73,
2009, pp. 997–1004. 2014, pp. 40–54.
14. Kofoworola, O. F., Gheewala, S. H., “Life 24. Aden, N., Qin, y., Fridley, D., “Lifecycle
Cycle Energy Assessment of a Typical Office Assessment of Beijing- Area Building Energy
Building in Thailand”,Energy Build., Vol. 41, Use and Emissions  : Summary Findings and
No. 10, 2009, pp. 1076–1083. Policy Applications”, China Energy Group,
Energy Analysis Department, Environmental
15. Ramesh, T., Prakash,R., Shukla, K. K., “Life Energy Technologies Division, Lawrence
Cycle Energy Analysis of a Residential Berkeley National Laboratory, September,
Building with different Envelopes and 2010.
Climates in Indian Context”,Appl. Energy,
Vol. 89, No. 1, 2012, pp. 193–202. 25. Baek, C., Park, S. H., Suzuki, M., Lee, S. H.,
“Life Cycle Carbon Dioxide Assessment Tool
16. Pinky Devi, L., Palaniappan, S., “A Case for Buildings in the Schematic Design
Study on Life Cycle Energy Use of Phase”, Energy Build., Vol. 61, 2013, pp. 275–
Residential Building in Southern India”, 287.
Energy Build., Vol. 80, 2014, pp. 247–259.
26. Roh, S., Tae, S., “Building Simplified Life
17. Ruuska, A., “Role of Embodied Energy, Cycle CO2 Emissions Assessment Tool
Operational Energy and Related Greenhouse (B‐ SCAT) to Support Low‐ Carbon Building
Ggas Emission of Buildings in the Context of Design in South Korea”, Sustainability, Vol. 8,
Developing Tropical Countries”, in SB13 No. 6, 2016.
Singapore-Realising Sustainability in Tropics,
2013, pp. 205–211. 27. Fu, F., Luo, H., Zhong, H., Hill, A.,
“Development of a Carbon Emission
18. Chau, C. K., Yik, F. W. H., Hui, W. K., Liu, H. Calculations System for Optimizing Building
C., Yu, H. K., “Environmental Impacts of Plan Based on the LCA Framework”,Math.
Building Materials and Building Services Probl. Eng., 2014.
Components for Commercial Buildings in
Hong Kong”, J. Clean. Prod., Vol. 15, No. 18, 28. Kwok, K., “Assessment of Building Lifecyle
2007, pp. 1840–1851. Carbon Emissions”, PhD Thesis - Univ.
Kansas, 2013.
19. González, M. J., García Navarro, J.,
“Assessment of the Decrease of CO2 29. Li, X., Yang, F., Zhu, Y., Gao, Y., “An
Emissions in the Construction Field through Assessment Framework for Analyzing the
the Selection of Materials: Practical Case Embodied Carbon Impacts of Residential
Study of Three Houses of Low Buildings in China”, Energy Build., Vol. 85,
Environmental Impact”,Build. Environ., Vol. 2014, pp. 400–409.
41, No. 7, 2006, pp. 902–909.
30. Li, D., Cui, P., Lu, Y., “Development of An
20. Dias, W. P. S., Pooliyadda, S. P., “Quality Automated Estimator of Life-Cycle Carbon
Based Energy Contents and Carbon Emissions for Residential Buildings: A Case
Coefficients for Building Materials: A Study in Nanjing , China”, Habitat Int., Vol.
Systems approach”,Energy, Vol. 29, No. 4, 57, 2016, pp. 154–163.
2004, pp. 561–580.
31. Peng, C., “Calculation of a Building‟s life
21. Tae, S., Shin, S., Woo, J., Roh, S,. “The Cycle Carbon Emissions Based on Ecotect
Development of Apartment House Life and Building Information Modeling”, J.
ENGINEER 8
514
Clean. Prod., Vol. 112, 2014, pp. 453–465. Analysis of Energy Sector Performance”,Sri
Lanka Sustainable Energy Authority, 2015.
32. Knight, D., AddisB., “Embodied Carbon
Dioxide as a Design Tool – A Case Study”, in 43. Aye, L., Ngo, T., Crawford, R. H.,
Proceedings of ICE - Civil Enginering, Vol. 164, Gammampila, R., Mendis, P., “Life Cycle
Nov. 2011. Greenhouse Gas Emissions and Energy
Analysis of Prefabricated Reusable Building
33. Moncaster, A. M., Symons, K. E., “A Method Modules”,Energy Build., Vol. 47, 2012, pp.
and Tool for „Cradle to Grave‟ Embodied 159–168.
Carbon and Energy Impacts of UK buildings
in Compliance with the New TC350 44. Zhang, X., Wang, F., “Life-Cycle Assessment
standards”, Energy Build., Vol. 66, 2013, pp. and Control Measures for Carbon Emissions
514–523. of Typical Buildings in China”,Build.
Environ., Vol. 86, 2015, pp. 89–97.
34. Kang, H. J., “Development of a Systematic
Model for An Assessment Tool for 45. "IEA Statistics: CO2 Emissions from Fuel
Sustainable Buildings Based on a Structural Combustion", International Energy Agency,
Framework”,Energy Build., Vol. 104, 2015, 2015.
pp. 287–301.
46. “Emission Factors for Greenhouse Gas
35. Scheuer, C., Keoleian, G. A., Reppe, P., “Life Inventories”,US EPA Center for Corporate
Cycle Energy and Environmental Climate Leadership, April, 2014.
Performance of a New University Building:
Modeling Challenges and Design 47. “Embodied Energy Coefficients”,Centre for
Implications”, Energy Build., Vol. 35, No. 10, Building Performance Research (CBPR),
2003, pp. 1049–1064. School of Architecture , Victoria University
of Welliington, Newzealand. 2007.
36. Biswas, W. K., “Carbon Footprint and
Embodied Energy Assessment of Building
Construction Works in Western Australia”,
Int. J. Sustain. Built Environ., Vol. 3, 2014, pp.
179–186.
37. Dong, Y. H., Jaillon, L., Chu,P., Poon, C. S.,

“Comparing Carbon Emissions of Precast
and Cast-in-situ Construction Methods - A
Case Study of High-Rise Private Building,”
Constr. Build. Mater., Vol. 99, 2015, pp. 39–53.
38. “ISO 14040-Environmental Management -

Life Cycle Assessment - Principles and
Framework”, International Organization for
Standardization,1997.
39. Roh, S., Tae, S., Shin, S., “Development of

Building Materials Embodied Greenhouse
Gases Assessment Criteria and System
(BEGAS) in the Newly Revised Korea Green
Building Certification System (G-
SEED)”,Renew. Sustain. Energy Rev., Vol. 35,
2014, pp. 410–421.
40. Atmaca, A., Atmaca, N., “Life Cycle Energy

(LCEA) and Carbon Dioxide Emissions
(LCCO2A) Assessment of Two Residential
Buildings in Gaziantep, Turkey”,Energy
Build., Vol. 102, 2015, pp. 417–431.
41. Chau, C. K., Leung, T. M., Ng, W. Y., “A

Review on Life Cycle Assessment, Life Cycle
Energy Assessment and Life Cycle Carbon
Emissions Assessment on Buildings”, Appl.
Energy, Vol. 143, No. 1, 2015, pp. 395–413.
42. “Sri Lanka Energy Balance 2015-- An
9
515 ENGINEER
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[517 524],2017
2017
The Institution
Lanka
Upgrading of the Existing Sedimentation Tank at

Hemmathagama Water Supply Scheme
L. Wijesinghe and D.R.I.B. Werellagama
Abstract: Demand of a water supply scheme may exceed designed production capacity due to
elapsing of design period, population growth, changing consumption patterns and water service
extensions to adjoining areas. With increasing demand, existing water supply schemes need
upgrading. Financial and time constraints should be resolved for best augmentation strategy.
Efficiency of existing unit processes need increasing. With long settling times, large sedimentation
tanks require large investment and longer construction period. This study researched capacity
upgrading of plain sedimentation tank at Hemmathagama water supply scheme (HWSS) in 2005.
Sediment removal efficiencies were compared for clarifier design discharge of 1000 m 3/day and
augmented discharge of 4000 m3/day. A conceptual design was proposed to improve the solid
particles removal efficiency for increased flow rate in the same plain sedimentation tank by fixing
plate settlers. Plate settler modules were designed and constructed. Removal efficiency was evaluated.
Influent and effluent turbidity were measured weekly for six months; for three scenarios, namely
plain sedimentation tank with 1000 & 4000 m3/day flow rates and same tank with plate settler for
4000 m3/day flow rate. Results showed that the plain sedimentation tank at HWSS could achieve 86%
reduction of turbidity, when operated with the design flow rate of 1000 m 3/day. Only 27% reduction
of turbidity could be achieved when the design flow rate was increased by four times up to 4000
m3/day. When a properly designed plate settler module was fixed to the existing plain sedimentation
tank, it achieved 82% reduction of turbidity with the increased flow rate of 4000 m3 /day. It was
demonstrated that using a plate settler; the existing plain sedimentation tank at HWSS could be
upgraded to cater for an increased flow rate, which is four times of its original design flow.
Performance in 2017 (after 10 years in operation) is also reported.
Keywords: water treatment, high rate sedimentation, plate settlers
1. Introduction Due to the long settling time needed,

rectangular or circular sedimentation tanks are
The production capacity of a water supply very large, thereby occupying large areas
scheme is decided considering the consumers compared to the treatment plant area. Since the
within a defined service boundary and a settling efficiency largely depends on the
specified design horizon [1]. The demand of the surface area of the sedimentation tank, further
water supply scheme often exceeds the increasing settling area to upgrade the
production capacity due to elapsing of design treatment plant capacity is a very difficult
period, unpredicted population growth, option due to financial and space constraints.
changes in consumption pattern and requests Using tube or plate settlers, an internationally
for extensions to the adjoining areas. To cater accepted option (relatively new to Sri Lanka in
for the increasing demand the existing water 2005) to increase the sedimentation efficiency
supply scheme can be augmented. When was investigated in this study, to increase the
selecting the best augmentation strategy settling efficiency in a working plant, within
financial and time constraints must be resolved. few weeks, with a limited budget.
Often there is no money available and the plant
cannot be taken offline as it is already serving
the consumers. Therefore, the engineer should Eng. (Dr.) D.R.I.B. Werellagama Int. PEng. (SL), C. Eng.,
plan to increase the efficiency of the existing MIE (SL), CMEngNZ, B.Sc. Eng. (Peradeniya), M.Eng.
(AIT), Dr. Eng. (Nagoya), Principal Lecturer in Civil
unit processes as far as possible. This study Engineering, School of Engineering, Wellington Institute of
showed the feasibility to carry out applied Technology, New Zealand.
research on the full-scale model, under Sri Eng. Lalith Wijesinghe, C.Eng., MIE(SL), B.Sc. Eng.
Lankan (tropical) conditions, without (Peradeniya), M.Sc.Eng. M.Eng. (AIT), MBA,
District Manager (World Bank Funded Water Supply and
disrupting the water supply. Sanitation Improvement Project), National Water Supply and
Drainage Board, Kegalle, Sri Lanka.
517
Plate settler technique has been successful in weir loading rate to minimize turbulence to
developed countries (with different feed waters avoid re-suspension of settled particles [2].
and temperatures). As plate/tube settlers do Settling basins require large land areas. Plant
not require any mechanical devices, except for a upgrades increasing water demand are
mechanical sludge scraper (required in any constrained by land availability. High-rate
large clarifier), authors attempted a full-scale settler modules can be installed underwater in
trial in 2005 as a cost effective option for water flow channels, increasing the tank loading rate
plant upgrading under local conditions. without adding basin volume [3].
Presently (2017) Plate settlers are well accepted
in Sri Lankan water treatment plants. Shallow basins provide the same settling as
deep basins. But horizontal shallow basins are
The objective of this 2005 study was studying subject to scouring action of the flow-through
possible capacity increase at Hemmathagama velocity and wind action, lowering removal
Water Supply Scheme (HWSS) after fixing plate efficiency. If horizontal shallow basins are
settlers designed by authors, to the existing stacked to reduce plant surface area, sludge
plain sedimentation tank. Quality parameter removal becomes difficult. Therefore, plate
monitored was turbidity removal efficiency. settlers are designed to be vertically inclined.
Chemical coagulation and flocculation was not Settled solids can slide down inclined surfaces
introduced to the HWSS, saving capital and and drop into the basin below. Distance
maintenance costs. between plates is designed to provide an up-
flow velocity lower than the setting velocity of
The scope of this study was to: the particles, allowing particles to settle on the
* Study the sediment removal efficiency of the inclined plate surfaces. The effective settling
existing plain sedimentation tank at HWSS for area is the horizontal projected area of the plate,
its design flow rate of 1000 m³/day and for an calculated by multiplying the inclined area; by
increased flow rate of 4000 m³/day. cosine of the angle of the plate to horizontal.
* If increased flow deteriorated water quality, Total settling area is the sum of the effective
then to propose a conceptual design to improve areas of each plate [4]. Depending on the type
the solid particles removal efficiency in the of plate setting modules, water enters either
same plain sedimentation tank, by fixing plate horizontally parallel to the plates or upward
settlers (for the increased flow rate). along the 60˚ tilted plates [5].
* Carry out actual design, identify construction
materials and locally construct an actual (Full) Main design criterion for plate settlers is the
scale model. surface loading rate for each plate. Typical
* Evaluate the removal efficiency for flow rate loading rates for plates range from 0.7 to 1.7
4000m³/day after fixing the locally m/h, depending on the settling characteristics
manufactured plate settlers in existing tanks. of the solids, water temperature, and desired
* Evaluate the turbidity removal efficiency for effluent quality. These loading rates allow for
flow rate 4000m³/day in 2016 (after 10 years of overall basin loading from 5 to 15 m/h, which
fixing the plate settlers in existing tanks), to is a multiple of conventional basin loading rate
assess the reliability of introduced low-cost [4]. Commercially available Plate settlers are
technology. typically manufactured in modules. Dimension
* Evaluate the durability of the designed plate of modules and plates vary by manufacturer
settlers after 10 years of operation. and are proprietary (and therefore expensive).
Basin dimensions are controlled by sludge
2. Literature Survey removal method. In this 2005 research work,
design was done in-house and plate modules
2.1. Sedimentation Theory were manufactured locally.
Sedimentation uses gravitational settling
principle. Efficiency of sedimentation process Materials selection for fabrication of settler
relates to inlet water quality, particle/floc size modules is important. Plates can be made of
and weight, tank geometry, etc. In a plastic, wood, galvanized steel, and asbestos –
conventional sedimentation tank, the flow is cement. Each of them has its advantages and
usually horizontal. The main design criteria for limitations. Asbestos – cement plates must be
a sedimentation tank with horizontal flow are coated with plastic or a similar type of coating
the surface loading rate (discharge/surface area material to prevent corrosion from alum treated
= Q/A), adequate depth and detention time for water. Galvanized steel needs anti-corrosive
settling, suitable horizontal flow velocity and coatings. Wood needs preservatives and should
518
be cleaned often as sludge does not slide easily full treatment within the limited budget
on wood [2]. Modules fabricated with paper- allocation. Also, acquiring new lands to
thin PVC plates were said to be deteriorate construct new civil structures within allocated
within 10 to 15 years due to weathering and time was impossible, and no usable land was
wave actions in the basins. Settler modules available within the HWSS. Considering the
made with thick plastic plates are more durable inadequacy of funds to acquire lands & the
and may have a module height over 2m [5]. In limited time available to complete the
this study 10 mm thick PVC sheets were augmentation, a plate settler module was
selected, inclined by 60˚. considered for existing plain sedimentation
tank to increase plant capacity to 4000 m3/day.
3. Project Description
The plate settlers were designed by authors,
The Hemmathagama Water Supply Scheme fabricated and fixed within the month of
(HWSS) is located 16km away from the January 2006 and samples were collected
Mawanella town, in Kegalle district. The weekly for turbidity measurements from
existing water supply scheme constructed in February 2006 to July 2006. As shown in table
year 2000 comprised an intake at Asupini Ella, 1.3, it was observed that the performance of the
a plain sedimentation tank without pre- sedimentation tank with plate settlers for a
treatment by coagulation/flocculation and a discharge of 4000m3/day met the required
distribution network providing 1000 m3/day to standard (of Turbidity less than 2 NTU).
Hemmathagama town and suburbs. By 2004
about 1000 service connections were provided 4. Methodology
by HWSS and the NWS&DB had to stop
providing service connections as the 4.1 Sampling
production capacity was barely sufficient to Samples were collected from influent & effluent
supply the demand of existing customers. of the sedimentation tank at HWSS (without
There were many requests for new service coagulation and flocculation) for three different
connections and pipe line extensions. A scenarios and the turbidity was measured.
demand survey revealed that capacity of the Samples were collected in 250ml sterilized glass
HWSS needed to be increased to 4000 m3/day. bottles. One sample each from influent and
effluent of the plain sedimentation tank were
In 2005 funds were allocated to construct a new collected (weekly) from first week of January
intake at a higher elevation at Asupini Ella and 2005 to final week of June 2005 with volume
replace the existing 150 mm diameter GI raw flow rate at 1000m3/day. After discharge was
water gravity main with 225 mm diameter increased up to 4000 m3/day, without
uPVC pipes to increase inflow to the treatment modifying the plain sedimentation tank, one
plant. Sadly, no funds were allocated for sample each from influent and effluent of the
capacity improvement of the treatment works. plain sedimentation tank were collected weekly
from first week of July 2005 to final week of
Data collection for this study started from December 2005. Plate settlers were designed,
January 2005. Then discharge through the plant assembled in to modules and fixed to the
was 1000 m3/day (full design capacity). existing plain sedimentation tank in January
Samples were collected weekly for turbidity 2006. One sample each from influent and
measurement, for six months. The WTP was effluent of the sedimentation tank were
performing well during this period (Table 1.1). collected weekly from February 2006 to July
In July 2005 construction of intake and new raw 2006 with discharge at 4000 m3/day. Samples
water main was completed. The discharge were transported to the district laboratory and
through the WTP increased to 4000 m3/day. the turbidity was measured. These (2005)
From July 2005 NWS&DB started distributing results were used to assess the viability of
the increased volume but without augmenting installing plate settlers as plant capacity
the treatment plant capacity. Samples collected augmentation option.
weekly for turbidity measurement from July to
December 2005 (Table 1.2) show water quality To assess performance of plate settlers after 10
deteriorating with quadrupled discharge years, one sample each from influent and
(existing plain sedimentation tank was effluent of the sedimentation tank were
overloaded). While augmenting the existing collected monthly from January 2017 to June
treatment plant was desirable, it was 2017 with discharge kept constant at 4000
impossible to design and construct a plant with m3/day (Table 1.4).
519
4.2 Turbidity measurement 4.3 Flow measurements
The turbidity of the collected samples was The flow rate to the sedimentation tank at
measured using HACH Model 2100P HWSS was measured using the Arkon-iMag
turbidimeter. These results are tabulated in magnetic type flow meter fixed to the inlet line.
Tables 1.1, 1.2, 1.3 and 1.4. The HACH Model The flowmeter shows Flow Rate (L/s or m3/hr)
2100P Portable Turbidimeter measures with Accuracy ±0.5% of reading [7].
turbidity (in manual mode) in three ranges;
0.001 to 9.99, 10 to 99 and 100 to 1000 NTU [6]. 4.4 Physical observations
Physical observations were made on PVC
plates and fixing arrangement for durability.
Table 1 - Turbidity data for 3 Experimental Scenarios (2005-6) and 10 years later (2017)
Table 1.1 Table 1.2 Table 1.3
Turbidity in NTU in Turbidity in NTU in Turbidity in NTU with
existing Plain Settling existing Settling tank improved
Tank for 1000m3/day Plain Settling Tank for with
design flow 4000m3/day augmented flow Plate Settlers for
4000m3/day increased
design flow
Date Inflow Outflow % Date Influent Effluent % Date Influent Effluent %
2005 (NTU) (NTU) Removal 2005 (NTU) (NTU) Removal 2006 (NTU) (NTU) Removal
05/1 25.0 1.9 92 05/7 6.0 4.0 27 03/2 6.5 1.3 80
13/1 30.0 2.0 93 12/7 8.0 6.0 25 12/2 7.3 1.9 74
18/1 15.5 1.8 88 17/7 7.8 5.8 26 24/2 6.4 1.2 81
25/1 17.0 1.9 89 24/7 4.8 2.5 48 07/3 8.0 1.6 80
09/2 0.7 0.6 14 04/8 7.6 5.6 26 15/3 8.8 1.8 80
18/2 6.5 1.5 77 12/8 8.0 5.7 29 27/3 6.0 1.1 82
25/2 5.5 1.3 76 17/8 8.3 6.3 24 02/4 8.0 1.7 79
13/3 8.8 2.0 77 27/8 2.8 2.0 29 10/4 6.1 1.0 84
19/3 8.0 1.9 76 06/9 6.2 4.5 27 18/4 4.0 1.0 75
27/3 8.5 1.9 78 13/9 7.4 5.9 20 29/4 7.5 1.2 84
05/4 8.0 1.9 76 21/9 4.2 2.1 50 05/5 3.8 0.9 76
10/4 6.1 1.3 79 26/9 7.5 6.0 20 10/5 8.3 1.9 77
19/4 6.1 1.0 84 06/10 8.1 5.0 28 17/5 6.0 1.0 83
25/4 8.8 2.0 77 14/10 7.6 5.7 25 26/5 7.6 1.3 83
04/5 3.5 1.0 71 19/10 5.3 3.1 42 06/6 25.0 4.0 84
09/5 8.2 1.5 82 25/10 9.7 5.9 39 11/6 20.0 3.1 84
19/5 8.1 1.2 85 05/11 5.4 3.1 43 18/6 13.5 2.0 85
25/5 8.0 1.8 77 18/11 8.0 5.0 38 26/6 12.0 1.9 84
07/6 16.3 1.6 90 17/11 17.0 12.0 26 07/7 15.8 2.3 85
13/6 25.0 1.9 92 26/11 20.0 15.4 23 09/7 20.0 3.2 84
16/6 7.8 1.4 82 07/12 9.3 4.8 48 15/7 6.0 1.0 83
27/6 7.4 1.0 86 15/12 8.5 6.4 25 25/7 7.8 1.3 83
23/12 30.0 22.0 27
28/12 25.0 19.0 24
All Effluent turbidity values All Effluent turbidity values All Effluent turbidity values
satisfied SLS614(1983) except on 27/8/2005 failed satisfied SLS614(1983)
Maximum desirable water SLS614(1983) Maximum Maximum desirable water
quality value of 2 NTU desirable water quality limit of quality value of 2 NTU
2 NTU
520
sedimentation tank could not reduce the
turbidity to meet required standard (<2NTU).
As the remedial measure, Plate Settlers were
Table 1.4 designed and fixed to the existing plain
Turbidity in NTU after 10 years sedimentation tank. The design was carried out
operation with Plate Settlers for 4000 following design recommendations by
m3/day flow Vigneswaran & Viswanathan (1995) [2]; ASCE
Date Influent Effluent % & AWWA (2005) [4], Kawamura (2000) [5] and
2017 (NTU) (NTU) Removal NWS&DB Design Manual D3 (1989) [8]. Pre-
05/1 2.0 1.5 25 treatment was not introduced.
06/2 6.3 1.6 75 • Surface loadings of the tank and plate
04/3 13.0 1.9 85 settlers for 4000 m3/day flow rate were
calculated and verified against above design
04/4 12.5 1.8 85
recommendations. The design surface loading
06/5 2.4 1.9 21 in the plate settler is 9.64 mm/min. It is less
03/6 2.4 1.8 25 than 28 mm/min which is the recommended
All Effluent turbidity values satisfied SLS614 value by ASCE & AWWA [4].
(1983) Maximum desirable water quality • The upward velocity through the plates
value of 2 NTU was calculated and checked against the design
recommendations to verify it was less than the
settling velocity.
5. Results and Discussion • Material for the plates was selected
considering the durability, non-corrosive
The observations are shown in the Tables 1.1, nature & smooth texture of the surface,
1.2, 1.3 & 1.4. The relevant data of final design availability and economy.
are given (and compared with literature values) • Angle of inclination of plates were
in Table 2. Table 1.1 gives effluent turbidity selected as 60˚ satisfying the design standards.
when discharge was 1000 m3/day. The • Detention time of plate settlers was
percentage reduction of turbidity was more calculated and checked against the design
than 70% always (Figure 1) and the effluent recommendations.
turbidity was within the maximum desirable • The capacity of the existing inlet
level of SLS 614 (1983). channel was checked to verify whether it could
When the discharge increased to 4000 m3/day accommodate the increased flow rate. The
(without modifying the existing sedimentation existing inlet channel could accommodate a
tank), the values of the major design flow rate of 2.78 m3/s. However the design
parameters shifted out of the recommended flow rate was 0.05 m3/s only.
ranges. The percentage reduction of turbidity • Inlet ports were designed to provide
(for this case) is tabulated in Table 1.2. Except laminar flow in to the sedimentation basin.
one sample with the minimum raw water Recommended port velocity was 0.10 - 0.20
turbidity (2.8 NTU) all other samples failed to m/s [1]. Design flow velocity was 0.13 m/s.
achieve the desirable treatment level. The • The clear height (of water) below plate
effluent water quality deteriorated. The settlers was provided as per the design
percentage reduction of turbidity ranged from recommendations to minimize turbulences due
20% to 50% with varying raw water turbidities, to localized velocity effects (Figure 4).
the average percentage reduction being 27.3 % • The capacity of the existing outlet
only. channel was checked to verify whether it could
The above findings showed that the turbidity of accommodate the increased flow rate.
raw water can be reduced to maximum • The scour chamber was checked
desirable level {SLS 614 (1983) standard} by against design recommendations.
sending the raw water at designed flow rate
through a properly designed plain After above calculations, Plate Settlers with
sedimentation tank. When discharge increased, following details were fixed (Figures 2, 3 and 4)
shifting the surface loading beyond to the existing plain sedimentation tank.
recommended range, the same plain
521
Design Capacity 4000 m3/day Slope length of plates 1000 mm
Number of tanks 02 Nos. Slope to horizontal plane 60˚
Plate Thickness 10 mm # Distance between plates 63 mm
Width of plates 2490 mm # Length of the tank 9.50 m
Number of plates 101 per tank
Figure 1 - Graph of Effluent Turbidity Vs Influent Turbidity for Three Different Scenarios
Table 2 - Comparison of selected values of design parameters with “recommended range for the
plate settlers” for volume flow rate of 4000 m3/day
Design Criterion Design Recommended Range of Values

Value D3, NWS&DB, Kawamura ASCE & AWWA
(This (1989) (with no (2000) (2005)
study) pre-treatment)
Surface loading 3.9 < 9.0 < 12.5 5.0 - 15.0
rate/(m/hr)
Surface loading in 9.6 - 7.5 - 8.8 11.6 - 28.3
plate settler
rate/(mm/min)
Tank depth / (m) 2.5 2.5 - 4.0 3.0 - 5.0 2.1 - 4.3
Length /width ratio 4:1 3:1 to 8: 1 4:1 to 5: 1 3:1 to 5: 1
Detention time/(min) 11 - - 5 - 20
Weir overflow rate/ 34 < 18 9.0 – 13.0 < 10
( m3/m.hr)
Sludge storage/ (m) 0.30 0.15 - 0.30 > 0.15 > 0.15
Upward velocity in 75.9 - 150 - 200 -
basin / (m/s)
Velocity in inlet 0.13 < 0.2 < 0.01 < 0.3
pipes/ (m/s)
Detention time / (min) 11 - - 5 - 20
522
Figure 2 - Plate settler sedimentation tank after Figure 3 - Fixing details of plate settlers
completion (2007)
Figure 4 - Detail of Plate Settler (Section); (width is 5300mm with two channels)
Figure 5 - Plate settler sedimentation tank after 10 Figure 6 - Fixing details of plate settlers after 10
years years
523
The performance of sedimentation tank with and physical observations made after 10 years
plate settler is tabulated in Table 1.3 and shown showed that the installed plate settlers are still
in Figure 1 too. The maximum influent raw reliable and durable.
water turbidity was 25 NTU and it was reduced
up to 4 NTU, which is not within maximum It was practically demonstrated (in a working
desirable level but within the maximum full scale plant) that plate settlers can be used to
permissible level (Turbidity reduction of 84%). upgrade the plain sedimentation tank within a
The plate settlers could reduce the raw water short time frame without completely stopping
turbidity up to the desirable level when the the water supply. The upgrade produced 4
influent turbidity is ≤ 13.5 NTU. Turbidity times more water in the same settling tank
reduction of more than 75% has been achieved (with four times original design flow rate) with
for all the samples. Compared to the inferior local design and fabrication (a limited capital
performance of the plain sedimentation tank input). Not having to introduce coagulation
with increased flow rate (27% only), the and flocculation to cater for an increased flow
turbidity removal significantly improved (to rate is an advantage both capital-cost wise and
82% removal) after fixing plate settlers. The operations (O&M) cost wise.
whole augmentation process was achieved
while the tanks were on-line too. Acknowledgement
The performance status is further confirmed in Authors wish to acknowledge the assistance
Table 1.4 where the data of the % removal after given by the National Water Supply &
10 years is presented. These effluent turbidity Drainage Board in preparation of this paper.
data confirm that the plate settler continues to
be a reliable technical option to remove high
References
(and low) raw water turbidity inputs, even after
a long service life.
1. National Water Supply & Drainage Board
(NWS&DB), (1989), “Design Manual D2 – Urban
The physical status of the plate settler (after 10 Water Supply & Sanitation”, Sri Lanka Water
years of operation) is shown in Figures 5 and 6. Supply and Sanitation Sector project.
The plates and the fixing arrangement are still 2. Vigneswaran, S. & Viswanathan C., (1995), “Water
intact except for minor discolouration of plates Treatment Processes – Simple Options”, CRC
at few locations. This confirms that PVC plate press, Florida, USA.
settlers (with design values used) is a durable 3. MWH (Montgomery Watson Harza, Inc), (2005),
“Water Treatment: Principles and Design”
technical option in tropical conditions too.
4. ASCE (American Society of Civil Engineers) &
AWWA (American Water Works Association),
6. Conclusions (2005), “Water Treatment Plant Design, 4th Ed.”,
McGraw Hill Publishing Company.
The plain sedimentation tank at 5. Kawamura S., (2000) “Integrated Design and
Hemmathagama WSS could reduce the Operation of Water Treatment Facilities”, Wiley-
turbidity up to the desirable level (2 NTU; SLS Interscience Publishers.
614) for its design flow rate of 1000 m3/day, 6. HACH, (2005), “Model 2100P portable
Turbidimeter. Instrument & Procedure Manual”
when the influent turbidity was less than 30
7. Arkon Flow Systems, (2005), “i-Mag Magnetic
NTU, with average percentage reduction of Flow Meter Manual”
turbidity of 86%. However, when the flow rate 8. National Water Supply & Drainage Board
was increased by four times up to 4000 m3/day (NWS&DB), (1989), “Design Manual D3 – Water
the quality deteriorated to 27% average Quality and Treatment”, Sri Lanka Water Supply
percentage reduction of turbidity. and Sanitation Sector Project.
9. American Public Health Association, American
Therefore, plate settlers were designed, Water Works Association & Water Pollution
modules fabricated and fixed to the existing Contract Federation, (1995), “Standard Methods
for the Examination of Water and Waste Water,
plain sedimentation tank. These plate settlers
16th Ed.”
removed turbidity up to the desirable level of 10. Wijesinghe, L. (2007), Upgrading the existing
SLS 614 (1983) standard (for the augmented sedimentation tank at Hemmathagama Water
design flow 4000 m3/day), with average Supply Scheme, Unpublished Master of
percentage turbidity reduction of 82%. The Engineering Thesis, Department of Civil
plate material selected was 10mm thick PVC, Engineering, University of Peradeniya. (Research
for availability, durability, smoothness, Supervisor Dr. D.R.I.B. Werellagama).
lightweight & low cost. Water quality testing
524
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[525 530],2017
2017
The Institution
Lanka
Risk Assessment of the Active Ingredient, Glyphosate

of the Herbicide, ROUNDUP® Present in Water
Sources of CKDu Prevalent Areas for Humans
S. Gunarathna, B. Gunawardana, M. Jayaweera and K. Zoysa
Abstract: Glyphosate, which is commercially available as Roundup®, was the most widely used
herbicide in Chronic Kidney Disease of unknown etiology (CKDu) prevalent areas in Sri Lanka until
recently; hence, it has still been persistent in the soil and water environment. Ingestion of water
contaminated with Glyphosate is believed to create adverse impacts on human health. Previous
studies provide evidence of the presence of Glyphosate in trace levels in urine samples of CKDu
subjects. Hence, Glyphosate is suspected to be a causal factor for CKDu. Therefore, this study focuses
on (1) investigating the presence of Glyphosate in water sources of CKDu prevalent areas and, (2) risk
assessment based on exposure levels of Glyphosate present in water for humans to determine whether
daily consumption of water contaminated with Glyphosate would create any adverse health impacts.
Water samples were collected from nine shallow wells and seven surface water sources in the CKDu
prevalent areas, in Anuradhapura District. Glyphosate in water was quantified using LC/MS and
GC/MS. Average daily dose and hazard quotient corresponding to Glyphosate was obtained
according to USEPA guidelines on risk assessment. Glyphosate was detected both in well water and
surface water samples tested. Concentrations of Glyphosate measured were in the range of 3-4 µg/l
and 30-45 µg/l in well water and surface water samples respectively. Hazard quotient was calculated
for Glyphosate concentrations measured in all water samples. With respect to well water, a maximum
hazard quotient of 0.0001 was obtained for both male and female for body weight 71 kg and 59 kg,
respectively. When considering surface water, maximum hazard quotients of 0.0010 and 0.0009 were
obtained for male and female for the same group of body weight respectively. In conclusion, all
hazard quotients determined in this study were profoundly less than 1 indicating no adverse health
impacts from Glyphosate when present in the detected levels in the water sources tested.
Nevertheless, it is advisable to conduct comprehensive studies on other possible pathways of human
exposure to Glyphosate, such as direct ingestion of Glyphosate due to careless handling and
consumption of food contaminated with Glyphosate and/or dermal exposure etc. before Glyphosate
being ruled out as a triggering factor for CKDu.
Keywords: CKDu, Exposure assessment, Glyphosate, Hazard Quotient
1. Introduction 2. Literature Review
Consumption of water contaminated with 2.1. Chronic Kidney Disease of unknown

agrochemicals is implicated to be one of the etiology (CKDu)
triggering causes of the Chronic Kidney Disease
of unknown etiology (CKDu). Glyphosate, CKDu is one of the critical health issues
which is commercially available as Roundup®, prevailing in Sri Lanka and found to be widely
was the widely used herbicide in Sri Lanka spreading in the North Central Province. To
until recently; hence, it is still persistent in the date, CKDu prevalent area is approximately
environment. Ingestion of water contaminated Eng. Ms. S. Gunarathna, AMIE(SL),
with Glyphosate is believed to create adverse B.Sc.Eng.(Moratuwa), Research Assistant, Department of
impacts on human health[1]. Previous studies Civil Engineering, University of Moratuwa.
Eng. (Dr) (Ms.)B. Gunawardana, AMIESL, B.Sc.Eng.
provide evidence of the presence of Glyphosate (Moratuwa), M.Sc. (Auckland), Ph.D.(Auckland), Senior
in trace levels in urine samples of CKDu Lecturer, Department of Civil Engineering, University of
subjects[2]. Hence, Glyphosate is suspected to Moratuwa.
be a triggering factor causing CKDu. Eng. (Prof.) M. Jayaweera, C.Eng., FIE(SL), B.Sc.Eng.
(Moratuwa), Ph.D.(Saitama), Professor of Civil
Engineering, , Department of Civil Engineering, University
of Moratuwa.
Mr. K. Zoysa, B.Sc. (Banglore), M.Sc. (Colombo),
Analytical Chemist, Department of Civil Engineering,
1
525 ENGINEER
17,000 km2[3]with reported morbidities and impact on the degradation of Glyphosate,
mortalities over 400, 000 and approximately 20, where decreased levels of degradation of
000respectively[1]. Majority of the reported Glyphosate is expected when sorption affinity
CKDu victims is males above 40 years, whose of Glyphosate onto soil matrix is strong [7].
occupation is farming [4].
Glyphosate can be mobilized into surface water
The exact causal factors for the CKDu have not sources as solutes or free compounds and as
yet been determined. However, direct ingestion combined with the colloidal particles present in
of toxins due to careless handling of the top soil layers which get carried away with
agrochemicals, ingestion of toxins that are being the surface runoff during the rainfall events[9].
absorbed/accumulated by various edible However, due to the strong adsorption affinity
products and prolonged exposure to toxins of Glyphosate onto soil particles, percolation of
through drinking water consumption is Glyphosate into groundwater through the
considered as potential pathways that are likely vertical soil profile is considered to be
to cause CKDu[5]. Since the majority of the significantly low compared to the Glyphosate
reported CKDu victims is from the farming getting carried away with the surface runoff [7].
communities and also above the age of 40 years, Hence, Glyphosate concentration in
it is suspected that drinking water groundwater could be profoundly low
contamination due to agrochemicals (e.g. compared to the detectable levels of Glyphosate
pesticides, herbicides) could be a leading cause in surface waters [9].
of the disease[4].
Glyphosate is recognized as a low acute toxic
2.2. Glyphosate compound [10]. Nonetheless, due to the
potential health risks that have been recognised
Glyphosate [N-(phosphonomethyl)glycine] is a with exposure to Glyphosate [10], the USEPA
broad-spectrum systemic herbicide and crop has imposed a maximum contamination level of
desiccant that has been widely used in the 0.7 mg/L for Glyphosate when present in water
CKDu prevalent areas. Glyphosate is [11]. In addition, the reference dose or the
commercially available as the herbicide average daily dosage that will not create
“Roundup®” which is a very powerful weed adverse effects throughout the life time is set as
killer used worldwide for the past few decades 2 mg/kg/day[10].
due to its high effectiveness in weed killing [6].
According to the WHO study conducted in the
Glyphosate is known to be strongly sorbed by CKDu prevalent areas in Sri Lanka, the
soil minerals [7] showing a greater affinity to proportion of CKDu victims with a level above
sorb onto variable-charge surfaces and not onto reference value for Glyphosate residue is
permanent-charge sites present on the soil 3.5%[2] Thus, Glyphosate seems to play an
surfaces [7].The sorption of Glyphosate important role for CKDu. Furthermore, in
depends on various factors such as soil organic animal studies, it was found that Glyphosate
matter, cation exchange capacity of the soil, alone could cause adverse effects on Kidney
minerals present in the soil matrix, soil type, pH function. Changes in proximal tubular cells in
and various cations that are present in the soil Kidney were observed in Nile Tilapia due to
matrix[7].In addition, Glyphosate has shown a exposure to Glyphosate [12], and Ayoolaet al.
greater affinity to create bonds with trivalent (2008) have noted changes in the proximal
cations such as Al3+ and Fe3+thus, the main tubule, Bowman space and degenerated tubules
sorption sites in the soil matrix for Glyphosate in African catfish exposed to Glyphosate [13].
retention were reported as the surfaces of Furthermore, glutathioneperoxidase dependent
amorphous aluminium and iron oxides [7]. reduction of cumenehydroperoxide in kidneys
of rats exposed to Glyphosatehas also been
Glyphosate degradation occurs mainly due to observed [14]. Therefore, Glyphosate could be a
the microbial activity of the soil triggering factor causing CKDu. However, there
microorganisms[8]. Microbial transformation of are no comprehensive studies reported in the
Glyphosate degrades the Glyphosate molecules literature to investigate the probable presence of
to Aminomethylphosphonic acid (AMPA), Glyphosate in the water sources and the
which is the major derivative of Glyphosate due associated human health risk levels due to the
to direct cleavage of the C-N bond [8]. consumption of waters contaminated with
However, strong sorption characteristics of Glyphosate in the CKDu prevalent areas in Sri
Glyphosate onto soil could have a negative Lanka.
ENGINEER 2
526
3. Objectives (PESTANAL®, 98% purity) was used for the
preparation of standards for the calibration and
This study focuses on (1) investigating the quantification.
presence of Glyphosate in water sources of
CKDu prevalent areas and, (2) risk assessment 4.3 Exposure Assessment
based on exposure levels of Glyphosate present
in water for humans to determine whether Exposure Assessment test was performed
daily consumption of water contaminated with according to the USEPA guidelines on risk
Glyphosate would create any adverse health assessment[15].Average daily dose (ADD) and
impacts. hazard quotient (HQ) corresponding to
Glyphosate levels detected were obtained
according to equations (1) and (2)[15].
ADD= (CxIRxED)/(BWxAT) ....(1)
4.1 Selection of test sites and sample
collection
Where:
ADD -Average Daily Dose
Nine shallow wells and seven surface water
C -Exposure Concentration (mg/L)
sources were selected in the CKDu prevalent
IR -Ingestion Rate (L/hr)
areas, in Anuradhapura District. All selected
ED -Exposure Duration (hr)
sites of water sources were located near the
BW -Body Weight (kg)
Chena cultivation sites, which have been
AT -Average Time period (day)
subjected to intense application of Roundup®
over long periods of time. Such sites have also
HQ = ADD/Rfd ...(2)
been subjected to series of rainfall events,
tilling, followed by cultivation of a variety of
Where: Rfd- Reference dose (mg/kg/day)
crops and application of fertilizers, especially
Triple Super Phosphate. The locations of
Glyphosate concentrations measured in shallow
shallow wells were selected after identifying
wells and surface water sources were
the greater possibility of seepage of percolated
considered as the exposure concentration
water from Roundup®-applied Chena
during calculations. Males and females in the
cultivation sites, that would be occurring
age group of 20-50 were considered for the
followed by the rainfall events, to the selected
exposure assessment test due to high risk of
wells. Also, the surface water sources were
exposure [4]. It was considered that daily
selected on the basis where such surface water
consumption would be 13 and 9 cups of water
bodies are located in places exposed to the path
by males and females, respectively, per cup
of surface runoff occurring from Roundup®-
volume to be 250 mL and water would be
applied Chena cultivation areas followed by
sustained in the human body for 2 hours.
rainfall events, and such runoff would
Average body weight for a male and female for
eventually end up in the selected surface water
the selected age group was considered as 71 kg
bodies.
and 59 kg, respectively [16]. These assumptions
were used during determination of the ADD
Water samples were collected from the selected
value for Glyphosate concentration present in
wells and surface water sources. All water
wells and surface water sources.
samples were properly sealed, carefully
transported and stored below 4˚ C in the
laboratory to ensure the preservation of 5. Results and Discussion
chemical properties.
Table 1 presents the detected levels of
4.2 Glyphosate and AMPA analysis Glyphosate in surface water and well water
samples. As shown in the results, Glyphosate
Glyphosate and AMPA in water were was detected both in well water and surface
quantified using Liquid Chromatography/ water samples tested. Concentrations of
Mass Spectrometer (LC/MS)Agilent 1200 Glyphosate measured were in the range of 3-4
Infinity series LC with G1600 series Mass µg/l and 30-45 µg/l in well water and surface
Selective Detector coupled with Zorbax Eclipse water samples respectively (Table 1). Surface
XDB- C18 column was used for the water sources near Chena cultivation sites 5
quantification. Samples were derivatized using and 6 have been dried due to the absence of
9-flurenylmethylchloroformate.Glyphosate rain for long periods of time,and thus water
3
527 ENGINEER
samples could not be collected from these two Glyphosate can be mobilized into surface water
sites at the time of sampling. sources along with the surface runoff followed
by a rainfall and Glyphosate can be present as
Table 1 – Glyphosate concentration in wells solutes and/or attached to colloidal particles
and surface water sources [9]. Therefore, the probability of Glyphosate
Chena Glyphosate Glyphosate mobilization into surface water sources is
Cultivation concentration in concentration in greater compared to that of groundwater as
site wells (µg/L) surface water was observed with the presence of higher
sources (µg/L)
concentration of Glyphosate in surface water
1 3.94 36.00 sources compared to that in well waters.
2 ND* 38.00
3 2.80 33.00 ADD values for Glyphosate concentrations
4 ND* 45.00 present in well water and surface water were
5 3.90 - calculated for both male and female groups.
6 ND* - Based on the ADD values obtained, hazard
7 3.60 28.00 quotient (HQ) was calculated for Glyphosate
8 ND* 34.00 concentration measured in all water samples.
9 3.70 43.00 ADD and HQ values obtained are presented in
*ND: Not detected Table 2. HQ values of Glyphosate concentration
present in well water were extremely lower
It is apparent that Glyphosate concentration
than the HQ values for surface waters. Lower
present in well water is profoundly lower than
Glyphosate concentrations in well water have
Glyphosate concentrations detected in surface
resulted in lower HQ values for well water. In
water sources (Table 1). Glyphosate is known to
Table 2 - ADD and HQ values obtained for well water and surface water samples
Well water Surface water

Chena
Male Female Male Female
Cultivation
site ADD ADD ADD ADD
(mg/kg/ HQ (mg/kg HQ (mg/kg HQ (mg/kg/ HQ
day) /day) /day) day)
1 0.00018 0.00009 0.00015 0.00008 0.00165 0.00083 0.00137 0.00069
2 - - - - 0.00174 0.00087 0.00145 0.00073
3 0.00013 0.00007 0.00011 0.00006 0.00151 0.00076 0.00126 0.00063

4 - - - - 0.00206 0.00103 0.00171 0.00086
5 0.00018 0.00009 0.00015 0.00008 - - - -
6 - - - - - - - -
7 0.00016 0.00008 0.00014 0.00007 0.00128 0.00064 0.00107 0.00054

8 - - - - 0.00156 0.00078 0.00130 0.00065
9 0.00017 0.00009 0.00014 0.00007 0.00212 0.00106 0.00164 0.00082
summary, when considering both well water

demonstrate a high affinity to adsorb onto soil
and surface water, all the HQ values obtained
particles [7]. Characterization of soil in the
were profoundly less than 1 indicating no
selected Chena cultivation sites showed the
adverse health impacts from Glyphosate when
presence of amorphous Al and Fe oxides
present in the detected levels in the water
providing evidence of greater retention of
sources tested. With respect to well water,
Glyphosate onto these elements (data not
maximum HQ of 0.0001 was obtained for both
shown).Thus, the strong adsorption behaviour
male and female groups respectively. When
of Glyphosate with soils in the selected sites
considering surface water, maximum HQs of
would not permit Glyphosate to mobilizeand
0.0010 and 0.0009 were obtained for male and
percolate through the vertical profile of the soil
female groups respectively (Table 2).
to meet the groundwater sources.
ENGINEER 4
528
HQ values for the male population were nephrotoxic metals: Are they the culprits behind the
slightly higher in surface water samples (Table epidemic of chronic kidney disease of unknown
2), which could be attributed to the higher etiology in Sri Lanka? Int J Environ Res Public
Health, 2014. 11.
amounts of water consumptions by males
compared to the females. Further, the majority 2. WHO, Investigation and evaluation of Chronic
of the people in the selected study areas were Kidney Disease of uncertain aetiology in Sri Lanka,
farmers. Hence there is a possibility that the 2013, WHO.
actual amounts of water consumption would be
greater than the values considered during the 3. Noble, A., et al., Review of literature on chronic
calculations. Therefore, HQ values for the male kidney disease of unknown etiology (CKDu) in Sri
population could be higher than the average Lanka. , 2014, Colombo, Sri Lanka: International
value considered. Water Management Institute (IWMI). p. 41.
4. Wanigasuriya, K., Aetiological factors of Chronic

6. Conclusions Kidney Disease in the North Central Province of Sri
Lanka: A review of evidence to-date. Journal of the
In conclusion, all HQ values were profoundly College of Community Physicians of Sri Lanka,
lower than 1, which indicates no adverse health 2012. 17(1): p. 21-42.
impact would be imposed from Glyphosate by
consuming the water from the water sources 5. Dharma-wardana, M.W.C., et al., Chronic kidney
disease of unknown aetiology and ground-water
tested. However, water consumption could be
ionicity: study based on Sri Lanka. Environmental
greater than the amounts taken during Geochemistry and Health, 2014. 37(2): p. 221-
calculations due to the factors such as the 231.
prevailing dry climatic conditions and intense
farming activities exposing to outdoor 6. Schuette, J., Environmental fate of glyphosate.,
conditions. Hence, under such conditions, the E.m.a.p. management., Editor 1998,
HQ values could be greater than the obtained Department of Pesticide Regulation:
values in this study. Also, the reference value Sacramento, . p. 95824-5624.
considered for Glyphosate is established for a
7. Borggaard, O.K. and A.L. Gimsing, Fate of
healthy person, and the reference value for glyphosate in soil and the possibility of leaching to
Glyphosate for a CKDu subject could be lower ground and surface waters: a review. Pest
than that of the reference value. Hence, the HQ Management Science, 2008. 64(4): p. 441-456.
value for a CKDu subject could be greater than
the values obtained. Furthermore, there are 8. Sviridov, A.V., et al., Microbial degradation of
other pathways of human exposure to glyphosate herbicides (Review). Applied
Glyphosate such as direct ingestion of Biochemistry and Microbiology, 2015. 51(2): p.
Glyphosate due to careless handling and 188-195.
consumption of food contaminated with
9. Vereecken, H., Mobility and leaching of
Glyphosate and/or dermal exposure etc. glyphosate: a review. Pest Management Science,
Therefore, it is advisable to conduct 2005. 61(12): p. 1139-1151.
comprehensive studies on other possible
pathways of human exposure to Glyphosate 10. EPA, U., R.E.D. Glyphosate, 1993, September.
before Glyphosate being ruled out as a
triggering factor for CKDu. 11. EPA, U., National Primary Drinking Water
Regulations, in Glyphosate1995.
12. Jiraungkoorskul, W., et al., Biochemical and

Acknowledgement histopathological effects of glyphosate herbicide
onNile tilapia (Oreochromis niloticus). Environ.
Authors wish to acknowledge the financial Toxicol, 2003. 18: p. 260-267.
assistance given by the National Research
Council, Sri Lanka and the Senate Research 13. Ayoola, S.O., Histopathological effect of glyphosate
Committee Grant of the University of on Juvenile African Catfish(Clarias gariepinus).
Moratuwa for providing funding for Am. Eurasian J. Agric. Environ. Sci, 2008. 4: p.
362-367.
conducting the present study.
14. Larsen, K., et al., Effects of sub-lethal exposure of
References rats to the herbicide glyphosate in drinking water:
Glutathione transferase enzyme activities, levels of
1. Jayasumana, C., S. Gunatilake, and P. reduced glutathione and lipid peroxidation in liver,
Senanayake, Glyphosate, hard water and
5
529 ENGINEER
kidneys and small intestine. . Toxicol. Pharmacol.
, 2012. 34: p. 811-818.
15. EPA, U., Guidelines for Exposure Assessment,

1992. p. 104.
16. WHO, Measuring change in nutritional status;

Guidelines for Assessing the Nutritional Impact of
Supplementray Feeding programmes for Vulnerable
Groups, 1983.
ENGINEER 6
530
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[531 536],2017
2017
The Institution
Lanka
Field-Scale Study on Treating Landfill Leachate Using

Anaerobic Filters Packed with Low-Cost Filter Media
W.K.C.N. Dayanthi, B.S.R. Nanayakkara, K. Kawamoto, M.D.B.A. Premathilaka
and O.V.N. Dulvin
Abstract: Though the anaerobic filters (AFs) are effective in treating landfill-leachate, it is an
expensive treatment method mainly due to expensive filter materials. If low-cost materials can
effectively be utilized as filter media, AFs can be made suitable for the day-to-day practice. This study
included a treatment of leachate generated from a sanitary landfill model, using two types of AFs. One
AF contained expensive plastic pall rings as filter media, while the other was filled with washed sea
sand (WSS), dewatered alum sludge (DAS), saw dust (SD) and bio char (BC) in layers of equal
thickness. The former acted as the control experiment while the latter was the experimental filter.
BOD5, COD, Nitrogenous compounds, pH, ORP, TSS, TDS, conductivity, turbidity, cations and heavy
metals were analyzed for influent and effluent samples. Overall, the treatment efficiencies of TN,
cations and heavy metals, especially Cu and Fe, of the experimental filter column were higher than
those of the control column. Removal efficiencies of BOD and COD in the control and experimental
filter columns were more or less similar. Therefore, it can be concluded that the identified low cost
and locally available materials could potentially be used as efficient packing media in AFs to treat
landfill-leachate.
Keywords: Field-scale experiments, Filter media, Heavy metals, Nitrogenous compounds,

Organic matter
1. Introduction in treating landfill-leachate, with different

Landfill-leachate is highly contaminated with media emerging as the best for different
carbonaceous materials, nitrogenous contaminants. Nanayakkara et al. [2] found that
compounds, inorganic compounds, soluble and the same organic materials as layers in one
suspended particulate materials, heavy metals media bed and inorganic waste materials,
etc. It has a great potential to pollute the water namely Dewatered Alum sludge (DAS), Sea
bodies, streams and even ground water. Sand (SS), Fire wood Charcoal (FC) filled as
Toxicity, turbidity, odor, bad taste and deposits single medium in separate experimental
are resulted due to the presence of heavy columns showed different efficiencies in
metals such as Zn, Pb, Ni, Cu, Cd and Fe. removing different contaminants. Nanayakkara
Therefore, treating landfill-leachate before et. al. [3] concluded that the filter media with a
releasing to the environment is a compulsory mixture of FC, SS and DAS showed high
need. potential in removing Total-N and BOD,
while it had a low potential in removing COD
As municipal solid waste is high in organic for both up and down flow loading conditions.
matters, biochemical treatments with attached
growth anaerobic filters (AFs) are well suited Eng.(Dr.) W.K.C.N. Dayanthi, Ph.D.(Kyoto),
for the leachate treatment. This process is MEng(AIT), MEng-Moratuwa, B.Eng.(Peradeniya),
C.Eng., Senior Lecturer, Department of Civil and
widely used because it reduces volumes and Environmental Engineering, Faculty of Engineering,
masses of the input materials and generates University of Ruhuna, Hapugala, Galle.
useful byproducts as well. Eng. Mrs. B.S.R. Nanayakkara, BSc.Eng. (Moratuwa),
C.Eng., Civil Engineer, Sri Lanka Ports Authority, MPhil
However, the conventional packing media used Candidate, Department of Civil and Environmental
Engineering, Faculty of Engineering, University of Ruhuna,
in AFs are very expensive, therefore finding Hapugala, Galle.
suitable low-cost and locally available efficient Prof. K. Kawamoto, Ph.D.(Tokyo), Professor, Graduate
packing materials is a crucial and timely need. School of Science and Engineering, Saitama University,
Dayanthi et al. [1] stated that the organic Saitama, Japan.
materials, namely coconut coir fiber (CCF), rice Eng. Mrs. M.D.B.A. Premathilaka and O.V.N. Dulvin,
husk (RH) and rice straw (RS) and saw chips B.Sc. Eng.(Ruhuna), Civil Engineers, Maga Engineering
Pvt. Ltd.
(SC) packed in separate up-flow and down-
flow anaerobic filter columns were satisfactory
1
531 ENGINEER
Therefore, this study focused on investigating layers of equal thickness in the experimental
the suitability of the materials, namely, washed filter column (Figure 3). Influent was
sea sand (WSS), DAS, Bio Char (BC) and Saw distributed in downward direction to the
Dust (SD) as filter media in field-scale experimental and control filter columns under
anaerobic filters to treat the landfill-leachate. gravity. The approximate hydraulic retention
time inside anaerobic filters were 2 days. The
2. Methodology influent and effluent samples of both the filter
columns were collected periodically, and they
were analyzed in terms of several wastewater
2.1 Experimental Set-Up and Experimental
parameters as per the Standard Methods for the
Run
Examination of Water and Wastewater [4]
Figure 1 and 2 show the isometric view and the
image of the field-scale experimental set-up,
2.2 Characterization of the Filter Materials
respectively.The field-scale experimental set-up
Figure 4 shows the images of the selected filter
consisted of a lysimeter modelling a sanitary
materials. Material characterization was done
landfill; a receiving tank; two dilution tanks the
mainly for the experimental low-cost materials.
inlet of each was connected to the receiving
Physical properties such as specific gravity,
tank and the outlet of each was connected to an
porosity, surface area and packing density, and
overhead tank; a control anaerobic filter column
mechanical properties such as hydraulic
receiving influent from one overhead tank and
conductivity, shear strength parameters such as
an experimental anaerobic filter column
friction angle and cohesion of the materials
receiving influent from the other overhead
were measured using core samples extracted
tank. The effluent of both the anaerobic filters
from the packed bed of the experimental
flowed into a collection tank, water from which
column. Particle size distribution was obtained
was recirculated back onto the lysimeter via a
using the sieve analysis test. Table 1 to 4 and
pump. The recirculated water together with
Figure 6 indicate the values of the above-
rain water generated the leachate while passing
mentioned properties of the selected filter
through the municipal solid waste inside the
materials
lysimeter.
Overhead
Lysimeter tanks
Experimental
unit
Collection
tank
Control
unit
Receiving
tanks
Figure 2 - Image of the Field-Scale
Experimental Set-Up
Figure 1 - Isometric View of the Field-Scale

Experimental Set-Up
The control filter column was filled with

‗plastic pall rings‘ (PPR), which is a
conventional expensive filter media. BC, SD,
DAS and washed sea sand (WSS) were filled in
ENGINEER 2
532
Table 2 - Packing Density, Surface Area and
Particle Size Distribution of the Filter
Materials
Filter Packing Surface Particle
Materials Density Area Size
(kg/m )
3 m /g
2 (mm)
WSS 1279.1 1.0 0.3-5

DAS 670.5 32.9 1-20
SD 201.4 2.3 -
BC 386.9 281.6 4-20
PPR 105.7 - -
100
90
80
70
Percentage Passing (%)

60
50
Figure 3 - Sectional View of the Experimental
40
Filter Column
30
20
10
0
0.01 0.1 1 10 100
WSS
-10
Sieve Size (mm)
FWC
BC DAS SS
PPR BC
Figure 5 - Particle Size Distribution of the
Filter Materials
Table 3 – Hydraulic Conductivity of Filter

Materials
Filter Hydraulic Conductivity (cm/s)
Materials
SD WSS DAS WSS 0.01
DAS 0.01
Figure 4 - Selected Filter Materials SD 0.01
Table 1- Specific Gravity and Porosity of the BC 0.01
Filter Materials
Filter Specific Porosity (n) Table 4 - Shear Strength Parameters of Filter
Materials Gravity Materials
WSS 3.116 0.589 Filter
DAS 2.459 0.761 Material Friction
SD 0.790 0.801 s Cohesion (kN/m2) angle ( )◦
BC 0.962 0.695
WSS 1.1 0.09
DAS 2.9 0.09
3
533 ENGINEER
3. Results & Discussion in the experimental set-up was 35.3%, which is
higher than PPR in the control column i.e.
Table 5 shows the mean values of the treatment 26.4%. The removal efficiencies of cations such
efficiencies and their standard deviations. as Ca2+ and Mg2+ (Figure 11 and 12) and heavy
Figures 7 and 8 show the variation of removal metals such as Cu2+ and Fe2+ (Figure 13 and 14)
efficiencies of BOD5 and COD, respectively. in the experimental column were considerably
Bodik et. al. [5] reported that the BOD removal greater than those of the control column. The
is around 68% with the plastic tube media. mean removal efficiencies of Fe2+ and Mg2 were
However, in this study, PPR used as the filter 59.5, 26% respectively in the experimental
material in the control unit gave a mean column whereas those for PPR were 22.47,
removal efficiency of 42.8%, which is slightly 7.1%, respectively. This could be due to the
higher than the result of the experimental unit promotion of treatment mechanisms such as
(40.3%). COD removal efficiencies of the ion exchange, precipitation and adsorption in
experimental and control units were 40% and the experimental system owing to the different
44.4%, respectively. Attached growth processes properties possessed by these materials.
in waste water treatment are very effective for
BOD removal and denitrification. Young and Figure 15 shows the pH variation during the
McCarty [6] stated the potential of the experimental run. The variation has an increase
anaerobic filter process for such treatments by at the beginning and a decrease toward the end
treating a medium strength wastewater within a range of 6 to 10.
successfully.
The ORP (oxidation reduction potential)

variation (Figure 9) indicates a reducing
environment in both the filters.
Table 5- Mean and Standard Deviations of the

Removal Efficiencies
Experimental
Control System
System
Parameter
Mean (%)
Mean (%)
deviation
deviation
Standard
Standard
BOD5 42.9 9.0 40.3 7.3

COD 44.4 8.6 40.0 10.3 Figure 6 - Removal Efficiency of BOD5
TSS 42.1 24.6 53.1 21.8
TDS 37.6 13.6 28.7 9.9
Turbidity 42.9 18.2 40.4 9.4 70
Cu2+ 60.7 25.6 69.5 19.8 60
% Removal of COD
Pb2+ 15.6 8.3 11.6 5.9 50

Fe2+ 22.5 10.2 59.5 30.8
40
TOC 25.0 14.5 5.4 2.4
TC 29.3 12.9 27.2 19.6 30
IC 31.2 17.5 36.0 30.8 20
TN 26.4 13.0 35.4 21.9 10
TP 16.0 5.2 11.3 1.4 0
PO43- 10.1 5.4 6.4 6.1 0 10 20 30
Na+ 31.6 22.9 13.7 8.5 No of Days
Ca2+ 16.7 15.3 31.9 18.8
K+ 30.8 30.2 10.3 11.0 Control Unit Experimental Unit
Mg2+ 7.1 4.9 26.0 7.6 Figure 7 - Removal Efficiency of COD
It could cause the removal of biodegradable

organic matter and TN due to anaerobic
decomposition and denitrification, respectively.
The mean removal efficiency of TN (Figure 10)
ENGINEER 4
534
0
-50 0 10 20 30
ORP variations
-100
-150
-200
-250
-300
-350
No. of Days
Control- Influent Control- effluent
Experimental- Influent Experimental-effluent
Figure 8 - Variation of ORP Figure 11 - Removal Efficiency of Mg2+
80 120
70
% Removal of TN
60 100
% Removal of Cu+2
50
80
40
30 60
20
40
10
0 20
0 10 20 30
No of Days 0
0 10 20 30
Control Unit Experimental Unit
No of Days
Figure 9 - Removal Efficiency of TN
Control Unit Experimental Unit
Figure 12 - Removal Efficiency of Cu2+
Figure 10 - Removal Efficiency of Ca+2

Figure 13 - Removal Efficiency of Fe+2
5
535 ENGINEER
10 3. Nanayakkara, B.S.R, Dayanthi, W.K.C.N,
―Landfill-Leachate Treatment in Up and Down
8
pH variations
Flow Anaerobic Filters with Combined Packing

6 Media of Locally Available and Low-Cost
Materials‖ 14th Academic Sessions, University of
4 Ruhuna, 2017
2
4. Standard Methods for the Examination of Water
0 and Wastewater, American Public Health
0 10 20 30 Association, 1015 Fifteenth Street, NW,
No of Days Washington, DC, USA, pp.20005-2605, 1988.
Control- Influent Control- effluent 5. Bodik, I., Herdova, B., Krotochvil, K., ‖The
Experimental- Influent Experimental-effluent Application of Anaerobic filter for Municipal
Waste Water Treatment‖, 26th International
Figure 14 - Variation of pH value Conference of the Slovak Society of Chemical
Engineering, Jasná – Demänovská dolina, 24—
28, May 1999.
4. Conclusions
6. Young, J. C, McCarty P. L, 1968, ―The Anaerobic
Filter for Waste Treatment‖, Stanford University
The performances of the selected filter materials
Technical Report No 87
are on a par with conventional filter materials
for organic compounds removal. The selected
materials surpassed the conventional materials
in treating several cations and heavy metals
owing to the promotion of treatment
mechanisms such as ion exchange, precipitation
and adsorption due to different properties
possessed by these materials. Therefore, these
materials can effectively be applied to treat
landfill leachate.
Acknowledgement
The authors would acknowledge the financial

support given by the research grant from the
JST/JICA Science and Technology Research
Partnership for Sustainable Development
(SATREPS). They would also like to offer their
gratitude to Mr. Ganaila Paranawithana, a PhD
student of the Saitama University for
measuring surface areas of the packing media.
References
1. Dayanthi, W.K.C.N., Ranga, U.K.S., Sanjeevani,
H.K.M., ―Applicability of Low-Cost Organic
Materials as Packing Media in Anaerobic Filters
to Treat Landfill Leachate‖ Engineer Journal,
Annual transaction of IESL, pp. [72-79], 2013
2 Nanayakkara, B.S.R., Dayanthi, W.K.C.N,

Vidanapathirana, V.P.H.S, Siriwardana, M.P, ―Up
and Down- Flow Anaerobic Filters Using
Firewood Charcoal, Dewatered Alum Sludge and
Silica Sand for Landfill-Leachate Treatment‖
Engineer Journal, Annual transaction of IESL,
2014
ENGINEER 6
536
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[537 543],2017
2017
The Institution
Lanka
Variations to Key Chemical Parameters during Dried

Faecal Sludge and Municipal Solid Waste
Co-Composting
H.M.N. Jayawardena, S.C. Fernando, S.H.P. Gunawardena and N. Jayathilake
Abstract: Co-composting of Municipal Solid Waste (MSW) together with one or more feedstock
has been proven to be addressing typical drawbacks such as low nutrient density, nutrient imbalance,
elemental deficiency, etc. when it is compared with compost that has been produced solely by MSW.
Meantime the disposal of septage from onsite sanitation systems (such as pit latrines, septic tanks, etc.)
has emerged as one of the major problems for local authorities because of rapid urbanization and land
scarcity. This paper focuses on the co-composting of the dried faecal sludge (DFS) and organic fraction
of MSW and its influence on pH, electrical conductivity, nutrient parameters, heavy metals and
biological parameters.
Co-compost trials were conducted using turning windrow composting method consisting of 1000 kg
of total mass; where one windrow was made solely with MSW as a control and other was made with
90% MSW and 10% DFS by mass. Each windrow was allowed to undergo a ten-week active
composting period and a two-week curing period. Results show that addition of faecal sludge assist to
improve pH of co-compost pile within the favourable range during the overall time span enhancing
the microbial activity. Further, pile conditions were set up, so that Nitrogen loss could be expected to
be at minimal level and hence higher Total Nitrogen (TN) content within the compost pile could be
observed together with the high and favourable improvement of electrical conductivity. Improvement
of Total Phosphorus availability and water-soluble Phosphorus availability could be observed. Heavy
metal availability and pathogenic species availability were under allowable ranges under stipulated
standards as well.
Keywords: Composting, Dried Faecal Sludge, Municipal Solid Waste.
1. Introduction techniques has to be revisited due to significant

changes in characteristics and quantity of MSW
With the advanced technology development that have taken place within last few years.
and the growth of urban population at an Therefore, the interest for reusing and recycling
alarming rate change of consumption patterns of MSW based on novel technological aspects is
of the modern world, which in turn has developing currently in the world [3].
significantly increased the quantity of
Municipal Solid Waste (MSW) generated. This MSW consists of diverse types of raw materials,
MSW is not properly treated, disposed and which can be classified into two main groups:
managed in many parts of both developed and agro-industrial and those derived from civil
developing countries [1]. MSW contributes to establishments [4]. These wastes can be further
the production of various types of gases such as categorized into groups such as animal wastes,
methane, ammonia together with leachate, crop residuals, fruit and vegetable wastes,
which could cause various problems ranging aquatic biomass and bio fertilizers, fish and
from simple odours to impacts on climate marine wastes, industrial wastes and human
change. Environmental scientists adopt various habitation wastes [4]. The resource potential of
solid waste management measures with mixed municipal solid waste deviates as it is
appropriate modifications to match both
Eng. (Dr.) S.H.P. Gunawardena, B.Sc. Eng.(Moratuwa),
quantity and characteristics of MSW [2].Source
Ph.D. (Birmingham), Senior Lecturer, Dept. of Chemical &
reduction, recycling, composting, controlled Process Engineering, University of Moratuwa.
burning where emissions are managed Eng. (Dr) S.C. Fernando, B.Sc. Eng.(Moratuwa), M.Sc.
properly with well stipulated mitigation Eng. (Moratuwa), Ph.D.(Soton).
measures or landfilling are proven to be Eng. H.M.N. Jayawardena, B.Sc.Eng.(Moratuwa)
effective management measures [2]. However,
Eng. A.M. Jayathilake, M.Sc.Eng(Moratuwa), B.Sc.Eng.
the effectiveness of these conventional (Moratuwa).
1
537 ENGINEER
dependent upon the waste consumption, drawbacks associated with MSW composting
which also differs considerably from country to and is known as Co-composting [10].
country and in a particular country from a city
to city due to income levels and consumer Several abundant and cheap organic materials
habits [5]. can be utilized for co-composting such as
animal manure, saw dust, wood chips, barks,
It was suggested that average MSW generation slaughterhouse waste, sludge or solid residue
(mixed) between 0.4-0.6 kg per capita per day from food and beverage industries. A
in low-income countries whereas 0.7-1.8 kg per considerable number of past studies that have
capita per day in high-income countries. been carried out related to MSW co-composting
Typically, in low-income countries, the with animal manure, poultry manure [11] [12]
biodegradable fraction is significantly higher and several other co-composting agents [13]
(40-85 %) than in high-income countries (20-50 [14] suggest that the finished co-compost has
%) [5]. ensured higher stability than compost derived
from MSW only. MSW co-composting with
Most commonly executed strategies such as cow manure, poultry manure, and yard
recycling, reusing and resource recovery have trimmings has found to maintain soil fertility
proved to yield several considerable while improving moisture holding capacity and
advantages; such as reducing the acceptable release of plant nutrients [15].
environmental impact, extending the existing
landfill capacity and replenishing the soil The aim of this study was to evaluate the
humus layer. Reducing the organic fraction in effectiveness of faecal sludge (FS) as low-cost
the waste, allows reducing the collection, nutrient enhancement agent while
transportation and disposal costs as well as the understanding the solubility of phosphorous
health and environmental risks associated with containing constituents in the presence of
appropriate waste handling and management organic acids that are likely to be formed
[6]. Usage of organic fraction of MSW as a during the process of decomposition.
source of food for urban animal livestock, direct
untreated application onto soils and production
of fuel pellets as an energy source are other 2. Methodology
several novel applications, which are under
discussion [7]. Compost piles with base dimensions of 1 m × 1
m were constructed. Each pile contained 1
Controlled biological treatment of the organic tonne of total mass. The first pile was
fraction of MSW into compost for organic constructed solely with sorted market waste as
farming is getting wide attention these days the source of MSW in order to keep it as the
due to its cost factor and health benefits in long control. The other pile was constructed with the
term utilization Composting is a process of addition of dried faecal sludge (DFS) as a
controlled biological decomposition of raw fortifying agent.
organic material into stable humic substances
with the presence of oxygen [8]. Once the Drying bed for FS was prepared by applying a
compostable organic portion of solid waste single layer of medium size stone pellets
streams is converted into compost, it can be together with three layers of gravel particle
used for landscaping, enhance crop growth, screened by 20 mm and 10 mm mesh. A metal
enrich topsoil etc. [2]. Apart from being a net with appropriate mesh size was laid on top
source of nitrogen, phosphorous, potassium of all four layers in order to avoid mixing of
and other plant nutrients, compost is also gravel particles with faecal sludge. Black
believed to suppress soil-derived diseases in waters were fed onto the drying bed directly
plants [7]. Incomplete decomposition of through gully trucks that are currently being
complex constituents, the existence of utilized by the municipal council. DFS was
pathogenic microorganisms, the existence of obtained after 21 days of drying of black water
heavy metals in undesirable concentrations, in the drying bed.
and nutrient imbalance are on the top of
discussions of MSW composting drawbacks [9]. 100 kg of DFS was added to a pile in order to
The addition of several other bio-solids such as keep FS composition at 10% by mass so that
faecal sludge and inorganic salts to MSW is initial C/N ratio is maintained around 30 [16]
found to be addressing most of pre-mentioned based on available values of carbon and
nitrogen content of the market waste.
ENGINEER 2
538
The addition of 100 kg of DFS with 900 kg of parameter lab testing in order to determine
MSW was done by a series of constituent Electrical Conductivity (SLS 1219 / 1:6 V/V),
layering method. Where 90 kg of MSW was pH Value (ISO – 10390:2005) and C/N Ratio
spread at the beginning to cover 1 m × 1 m base (ISO – 11261/14235), together with Total
dimension. 10 kg of faecal sludge was then Nitrogen (as N) (ISO – 11261: 1995), Total
distributed on top of MSW layer and a hand Phosphorous (ISO 11263: 1994) and Water
rake was used to mix the content so that Soluble Phosphorous (SLS 645).
maximum possible homogeneousness can be
expected. Once raked mixture was properly 3. Results and Discussion
arranged on 1 m × 1 m base spread, another 90
kg of MSW and 10 kg of DFS were added and Co-composting of FS and MSW is expected to
mixed with the rake. This process was followed be beneficial since they complement with each
until all the constituent quantities were other. Nitrogen and moisture content in
introduced to the pile. DFS contains pathogen human waste is relatively high whereas organic
loads and should be handled with care. The co- carbon content in MSW is relatively high.
compost can only be considered safe after the Meantime, high temperatures generated in
active composting period and appropriate composting may effectively inactivate
personal protection equipment should be worn pathogens present in both FS and MSW,
to avoid any possible health risks to the eventually generate hygienically safe soil
workers. conditioner and a fertilizer.
Each pile was first turned when the Electrical Conductivity

temperature exceeded 550C of the windrow
(usually 2 to 3 days after piling of windrow). Electrical conductivity is an indicator of soluble
Piles were turned weekly after the initial salts and ions prevailing in the compost. From
turning until 10 weeks were completed from the results presented in figure 01 it is clearly
the day of first turning. The moisture level of visible that the sample only with MSW has very
each pile was maintained in the range of 45 - low conductivity when compared with the
60% with the addition of water based on MSW and DFS co-compost pile. This may be
squeeze test [17] performed on site. The due to the fact that organic components are the
squeeze test is a simple field methodology prime element in MSW and it lacks inorganic
where it is believed that water can be squeezed ions and salts. Free ion availability will
out if compost has moisture content above 60% increase if more inorganic ions are added to the
and neither water is released nor remained compost.
when compacted.
Samples collected at each turning were 6000

subjected to testing in order to determine MSW : FS Pile
MSW only Pile
Electrical Conductivity (SLS 1219 / 1:6 V/V), 5000
Electrical Conductivity ((µS/cm)
pH Value (ISO – 10390:2005) and C/N Ratio

4000
(ISO – 11261/14235), which were pre-decided
as characterization parameters. 3000
Each pile was allowed to cure for two weeks 2000

after 10th turning. The organic municipal solid
1000
waste material used for this was market waste
(vegetable and fruit waste). In general, the 0
minimum curing time is considered 21 days but W1 W3 W5 W7 W10 W12
can be extended to many months. Bioassay can Time (Weeks)
help to predict the success of the compost
curing. By considering the input material, a
Figure 1- Electrical Conductivity Variation
reduced curing time was used. In this case the
with composting time
bioassay test was conducted successfully and
confirmed the quality of the product. Any
obstruction or water addition was not Increased electrical conductivity in DFS added
performed during this curing phase. Sample samples indicates the presence of inorganic
drawn after two weeks of curing (i.e. 12 weeks matter in sludge. Salts and other water-soluble
after first turning) were subjected to full- inorganic ions that have been consumed by
3
539 ENGINEER
humans may have been transferred to the DFS. At this pH, it could expect to enhance the
DFS added MSW pile shows gradual decrease kinetics of composting process due to
in conductivity with time. This could be due to prevalence of favourable condition for
the irregularities in decomposing of constituent microbial existence as reported by (Chen, et al.,)
components available in the feedstock. It is [19], Further, this favourable pH environment
reported that unnecessarily high conductivity is will improve the nutrient value of co-compost
due to sodium chloride in soil and is not by reducing nitrogen loss as ammonia which
beneficial to plants. Further, soil salinity greater happens under alkaline pHconditions [9].
than 4000 µS/cm could lead to inhibition of
plant growth and germination [1]. The During the period from W7 to W12 a drop in
conductivity of final compost obtained in 12 pH could be observed in faecal sludge added
weeks from this study has an acceptable pile when compared with the control, and this
conductivity value of 4300 µS/cm. could be due to the neutralization of alkaline
constituents by acidic waste generated during
Throughout the time span of 12 weeks the decomposition.
electrical conductivity does not show a
significant improvement in MSW only sample;
it remained constant around 1000 µS/cm Carbon Nitrogen (C/N) Ratio
throughout the monitoring period. According
to reported work, less EC value tends to cause 20
adverse effects compared to higher EC values 18
MSW: FS pile
MSW only Pile
in compost since composts are used as a soil
16
conditioner in agriculture, which will be
14
incorporated into the soil and thus diluted [18].
12
Therefore, an addition of DFS as a co-
C/N Ratio
10
composting agent has positively affected the
quality of compost. 8
6
pH 4
2
with reference to figure02, it could be observed
0
that the addition of faecal sludge, a significant
W1 W3 W5 W7 W10 W12
increase in pH value of the contents in the pile
Time (Weeks)
can be seen from the week 1 onwards and
remained around 8.0 (7.5 – 8.5). The pH
increases from week 1 in DFS added pile Figure 3 - C/N Ratio Variation with
should be due to the electron accepting ability Composting Time
of ions released with the process of
decomposition as reflected in high electrical
conductivity in figure 1. When analyzing results represented in figure
03, it can be clearly seen that addition of faecal
10 sludge has decreased the C/N ratio initially.
9
MSW : FS Pile
MSW only Pile
This is a clear indication of rich in Nitrogen
8 components when compared with constituents
7 of MSW. Higher water-soluble nitrate
6 concentrations lead to faster discharging of
5 nutrients enhancing the fertility value of
pH
4 compost since all nitrates are highly water-

3 soluble. These results showcase the correlation
2 of pHvariation plot in figure 02, since
1
significantly high pH conditions supposed to
0
W1 W3 W5 W7 W10 W12
release Nitrogen in the form of ammonia and
Time (Weeks) result in high C/N ratio. When it comes to end
user, co-composting has added a considerable
value in terms of Carbon to Nitrogen ratio. But
it is better to have the proper C/N ratio so as it
Figure 2 - PH Variation with Composting will maintain a favourable environment for
Time microbes to proceed with degradation. It is
accepted to have a C/N ratio in the
ENGINEER 4
540
specifications of 20% to 30% for proper total phosphorous is taken as characterization
microbial action [20]. Results have shown parameter, it could be seen that MSW based
acceptable C/N ratio below 20 at the maturity control pile has got about 20% of its total
according to (Wong, et al.) [21] and even phosphorous in water-soluble forms while the
recommended level at maturation below 16 ratio is about 30% in MSW: DFS co-composting
suggested by [22]. The rapid decrease in value pile. This is due to low level of alkalinity
of nitrogen for MSW: DFS with time may be maintained with the addition of faecal sludge
due to the loss of ammonia with time. as reflected in pHvariation curve i.e. figure 2.
According to (Kauwenbergh) [24] acidic nature
Nutrient Content Parameters of faecal sludge has assisted the task of the
acidic constituent formation during
Figure 4 shows the total Nitrogen (TN), total
decomposition process and hence has been
Phosphorus and water-soluble Phosphorus
promoted, phosphorous dissolution and
content in both compost samples after 12
ultimately shown the higher total and water-
weeks. According to figure 4, MSW based
soluble phosphorus content. This would
control pile has low TN content when
enhance further investigations for evaluating
compared with MSW: DFS co-compost pile.
the effectiveness of the addition of faecal sludge
This is a reflection of figure 2 i.e. end of the line
as a phosphorous enhancement agent.
acidic condition is generated through an
introduction of DFS has enhanced nitrogen
availability in nitrite and nitrate forms by
Heavy Metal Availability
demoting nitrogen loss in the form of ammonia
tendency in alkaline conditions.
The presence of acidic constituents that 120
Raw MSW
maintains pH within favourable range Raw FS
100
Heavy Metal Availability (mg/kg)
increases nitrogen availability in the forms of MSW: FS Compost
nitrites and nitrates where alkaline constituents 80

decrease the nitrogen availability in the forms
of nitrites and nitrates [23]. 60
40
20
1.6
MSW: FS Pile
1.4 MSW only Pile 0
Cr (mg/kg) Cu (mg/kg) Cd (mg/kg)
Nutrient Content (mass %)
1.2
1.0
0.8
Figure 5 - Heavy Metal availability in raw
0.6 materials and finished compost
0.4
0.2
Heavy metals availability in raw MSW, DFS
0.0 and co-compost is shown in Figure 05. MSW
Total Nitrogen Total Phosphorous Water Soluble
(as N) (as P) Phosphorous and raw dried faecal sludge (DFS) contained a
Cr concentration of 14 and 23 mg/kg on dry
basis respectively while MSW: DFS co-compost
Figure 4 - Total Nitrogen, Total Phosphorus sample contained 58 mg/kg Cr concentration at
and Water Soluble Phosphorus content in the maturity. This should be due to the mass
finished compost loss occurred during the decomposition process
[25]. However, Cr concentration is still
satisfactory and has a value less than 60 mg/kg
Variations observed in total phosphorous and comply the strictest environmental regulation
water-soluble phosphorous content follow the stipulated by European Compost Network
characteristic low proportional relationship (ECN) [26]. Cu availability and Cd availability
between total phosphorous and water-soluble also represent allowable values of 46.3 and 0.4
phosphorous in both pile configurations. If the mg/kg, which are satisfactory when compared
ratio between water-soluble phosphorous and
5
541 ENGINEER
with 70 mg/kg and 0.7 mg/kg guideline values nutrient content especially Total Phosphorous
[26]. (TP) and Water-Soluble Phosphorous (WSP)
content availability within the derived co-
compost, where Phosphorous is a non-
Biological Parameters renewable source
Faecal coliform availability is believed to be one
 The addition of 10% of the mass FS has not
of the parameters of concern in evaluating FS
caused any serious safety consideration due
application due to hygienic aspects and has
to the availability of permissible
been subjected to discussion in the recent times.
concentrations of Heavy metals
It is reported that if faecal Coliformis below
1000 MPN/g then that particular DFS mixed
compost is suitable for ground application [27].  High temperatures were achieved during the
co-composting process and has been
confirmed biologically safe (sanitized final
120
Raw MSW
product.
Raw FS
100
Biological Parameters (MPN / g)
MSW: FS Compost
Acknowledgement
80
60 Authors would like to acknowledge the

support extended by CGIAR research program
40
on water, land and eco systems in funding this
20
study.
0 Authors would like to acknowledge the

Total Coliform (MPN/g) Faecal Coliform (MPN/g)
support extended by Municipal Council -
Kurunegala, North Western Province in
conducting this study
Figure 6 - Total and Faecal Coliform
Availability in raw materials and Finished
compost References
1. Cofie, O., Kone, D. & Rothenberger, S., 2009.

According to figure 6, MSW: DFS co-compost Co-composting of Faecal Sludge and Organic
produced in this study has a satisfactory level Solid Waste for Agriculture: Process Dynamics.
of biologically available pathogens with 13.3 Water research, 43(1), pp. 4665-4675.
MPN/g of faecal coliform. Therefore, the
2. Sherman, R., 1999. Large scale organic materials
addition of 10% by weight DFS to MSW has not
composting, Carolina: NC State university.
affected the hygienic parameters of the co-
compost. 3. Mollazadeh, N., 2014. Composting: A New
Method for Reduction of Solid Waste and
Waste Water. International Research Journal of
4. Conclusions Applied and Basic Sciences, 8(3), pp. 311-317.
4. Sharma, V. K., Canditelli, M., Fortuna, F. &

The following conclusions can be drawn from
Cornacchia, G., 1997. Processing of Urban and
the study. Agro-Industrial Residues by Aerobic
Composting: Review. Energy conservtion and
 The addition of DFS as the co-composting management, 38(05), pp. 453-478.
agent has improved Electrical Conductivity
within the derived co-compost by increasing 5. Strauss, M., Drescher, S., Zurbrügg, C. &
free ion availability and hence it favourably Montangero, A., 2003. Co-composting of Faecal
maintains soil characteristics. Sludge and Organic Municipal Solid Waste,
Ghana: International Water Management
 The addition of DFS leads to favourable pH Institute.
maintenance within the derived co-compost 6. Tweib, S. A., Rahman , R. A. & Kalil, M. S.,
compatible with the SLS standard. 2011. A literature review on the Composting.
Singapoore, International Conference on
 DFS acts as a compatible co-composting agent Environmental and Industrial Innovation.
for Municipal Solid Waste and improves the
ENGINEER 6
542
7. Hirrel, S. s. & Riley, T., 2005. Understanding the 20. Kutzner, H. J., 2008. Microbiology of
Composting process, Fayetteville: University of Composting. In: H. Rehm & G. Reed, eds.
Arkansas. Biotechnology set. Ober-Ramstadt: Wiley library,
pp. 34-47.
8. Kathryn, P., 2003. Chemistry and Compost,
Washington: American Chemical Society. 21. Wong, J. et al., 2001. Co-composting of Soybean
Residues and Leaves in Hong Kong. Bioresource
9. Kokhila, M. S., 2015. Composting: Advantages Technological Journal, 76(02), pp. 99-106.
and Disadvantages. Tbilisi-Georgia, Ilia state
university. 22. Bernal, M., Navarro, A., Roig, A. & Cegarra, J.,
1996. Cabon and Nitrogen Transformation
10. Dinis, M., 2009. Co-composting: A Brief Review, during Composting of Sweet Sourghm Bagasse.
Porto-Portugal: University Fernando Pessoa-
Porto-Portugal. 23. Ameen, A. & Ahmad, J., 2016. Effect of pH and
Moisture Content on Composting of.
11. Varma, S. V. & Kalamdhad, A. S., 2013. International Journal of Scientific and Research
Composting of Municipal Solid Waste (MSW) Publications, Volume 6(Issue 5), pp. 35-37.
mixed with cattle manure. International Journal
of Environmental Sciences, 6(3). 24. Kauwenbergh, V. S., 2010. World Phosphate rock
reserves and resources, Alabhama-USA: IFDC
12. Ferreira, N. L. & Oliveira, R., 2013. Co-
composting cow manure with food waste: The 25. Brinton, F. W., 2000. Compost Quality Standards
influence of lipids content. dubai, World and Guidelines, New York: New York State
Academy of Science, Engineering and Association of Recyclers.
Technology, pp. 986-991.
26. Cofie, O. et al., 2016. Co-composting of Solid
13. Hasanimehr, M., Rad, H. A., Babaee, V. & baei, Waste and Faecal Sludge for Nutrient and Organic
M. S., 2011. Use of Municipal Solid Waste matter recovery. 3rd ed. Germany: CGIAR
Compost and Waste Water Biosolids with Co- Research Program on Water, Land and
Composting Process. World Applied Sciences Ecosystems (WLE).
Journal, 11(Special Issue of Food and
Environment), pp. 60-66. 27. Miller, C. M., 2015. Microbiological Safety of
Organic Fertilizers Used to Produce Production.
14. Ogunwande, G. A., Osunade, J. A. & Ogunjimi, Carolina: Clemson University.
L. A. O., 2008. Effects of Carbon to Nitrogen
Ratio and Turning Frequency on Composting
of Chicken Litter in Turned-Windrow Piles.
Agricultural Engineering International: the CIGR
Ejournal, x(07).
15. Davidson, G., 2011. Waste management practices,

Halifax, Norva Scotia: Dalhousie university.
16. Cooperband, L., 2002. The Art and Science of

Composting.[Online]
Available at: https://www.cias.wisc.edu/wp-
content/uploads/2008/07/artofcompost.pdf[A
ccessed 14 November 2016].
17. David, A. & Gamble, S., 2013. Municipal Solid

waste Organics Processing. s.l.:ISBN: 978-1-100-
21707-9.
18. Dimambro, M., Lillywhite, R. & Rahn, C., 2006.

Biodegradable municipal waste composts: analysis
and application to agriculture, United Kingdom:
the university of Warwick.
19. Chen, L., Morti, M. d. H., Moore, A. & Falen,

2011. The Composting Process. [Online]
Available at:
http://www.extension.uidaho.edu/nutrient/
wastemanagement/PDF/dairycomposting.pdf
[Accessed 17 january 2017].
7
543 ENGINEER
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[545 552],2017
2017
The Institution
Lanka
Life Cycle Assessment Based Tool to Visualize

Environmental Impact of Construction Materials &
Products
H.M.M.M. Jayawickrama, K.G.N.H. Weerasinghe, R.L. Peiris and A.K. Kulatunga
Abstract: The construction industry has boomed significantly in Sri Lanka after the 30 years of
war. This has created unprecedented demand for some construction materials, and some of them are
extracted or mined from virgin sources. Even several building materials such as cement, roofing
sheets, etc. need to manufacture before they are being used which contribute substantially to the
environment during the manufacturing phase. Conversely, due to the growing interest towards
sustainable built environment or greener buildings has made curiosity within the construction
industry to move towards sourcing of greener materials or to secure greener supply chains for the
construction industry. However, it is difficult to identify the environmental impacts of construction
material and specially to assess or compare in a quantifiable manner. Therefore, this research focuses
on developing a tool to indicate environmental impacts taken into account the phases of the supply
chain which are rarely come to the surface. This tool has been developed based on life cycle
assessment concept. Life Cycle Assessment is a well-recognized methodology used for quantifying the
magnitude of the potential environmental impact of a particular system during a selected period of its
lifetime. Commercial software’s have been developed to conduct these studies, but they are
dependent on the primary databases which have been prepared considering regional geographical
representations. Therefore, it is difficult to use these software due to non-availability of country and
region-specific data/information in the commonly used databases. However, this research overcomes
this barrier with some alternative means to develop the final tool. Based on the developed tool, it is
easier to compare environmental hotspots for the selected impact categories specifically highlighting
the particular phase which has a significant impact in the way of different colour coding system which
can be easily compared by the construction industry professionals without deep knowledge in
Sustainability or Greenness of the supply chains.
Keywords: Life Cycle Assessment, Environmental performance, Construction Material
1. Introduction eco systems. Conversely, developing countries

have been unable to control illegal material
Construction industry has evolved enormously extraction processes [3, 4]. Sri Lanka also as a
during the last two centuries all over the world developing country, has been contributing to
while consuming gigantic amount of resources this damage for a while [5]. After the 30 year of
and covering free lands available [1]. Since it civil war, the development of construction
directly links with the economic development industry is rapidly increasing (Figure. 1). This
of a country [2], all over the world this industry has increased the construction material
continued to be developed rapidly in consumption. For instance, LKR 38 billion used
developing countries and operate at stable state for construction raw materials during 2010 [6]
in developed countries alike. Due to the rapid and it has been increased up to LKR 221 billion
development of the technologies and for 2015 [7].
population increase, size of buildings also Mr. H.M.M.M. Jayawickrama, B.Sc. Eng. Hons
keeps on increasing continuously in to larger (Peradeniya),
scale and the impact of buildings and this MPhill Candidate, Department of Manufacturing and
Industrial Engineering, University of Peradeniya
industry on environment also increasing in
parallel. This has created severe damage to Mrs. K.G.N.H. Weerasinghe, B.Sc. in Export Agriculture
(Uva Wellassa University, MSc.in Environmental Science,
many eco systems as most of the materials use University of Peradeniya
for constructions or to manufacture
Eng. R.L. Peiris, BSc. Eng. Hons (Ruhuna), MPhill
construction materials are virgin and extracted Candidate, Department of Manufacturing and Industrial
from earth. (Ex: Sand, Clay and Lime). Due to Engineering, University of Peradeniya
stringent regulations available in developed Eng. (Dr.) A.K. Kulatunga, BSc.Eng. Hons (Peradeniya),
countries material extraction process happening Ph.D. (Australia), Senior Lecturer, Department of
in controlled manner with minimal effect to the Manufacturing and Industrial Engineering, University of
Peradeniya.
1
545 ENGINEER
voluntary in nature. Type I is done by
stakeholders and advisory committees which
considers product life as a whole. Type II is a
self-declaration claim without pre-clearance in
criteria development and it leads less consumer
trust. Type III is a LCA approached criterion
which the setting and verification is done by a
third-party. Though the standards discussed
above have addressed the total product life
Figure 1 - Sri Lanka GDP - construction [8] cycle, the detailed quantification of
environmental impacts generated by each life
When considering above statistics it is a fact cycle phase targeting the eye open on
that the construction material consumption is responsible party is not addressed [15, 16, 17,
escalating at an alarming rate resulting a great 18, 19]. However, regional divergences can be
loss of the ecosystem and associated bio identified in such methods as LCA base line
diversity. The availability of low quality data are not applicable to the geographical
materials at low cost, and availability of representation of developing countries and
materials without having safe end use therefore, wide range of assumptions are used
mechanisms are the facts identified for the in analysis which cause to reduce the credibility
above issue [5]. Consequently, the necessity of of LCA results. The limitation of a baseline/
consideration of all four phases of the life cycle benchmark to compare is the main weakness
that are Pre Manufacturing, Manufacturing, identified. Providing a roadmap to above issue,
Use and Post –Use when selecting of a decision making methodology by evaluating
environmental friendly construction materials environmental performance of products against
has been raised by recent studies [9, 10, 11] as a pre-defined benchmark specifically calculated
well. to each country /region is suggested by
Finkbeiner in his research [19].
Meantime the attention of local industrialists
has gone towards environmental friendly When look at in to locally developed
buildings or green buildings with global trend environmental performance monitoring tools or
towards sustainable Built environment. Some of methodologies developed aiming construction
the land mark buildings such as Kandalama materials, Green Label Scheme introduced by
Hotel, MAS Thuruliya apparel plant etc. has GBCSL [20] focus on issuing a Green Label to
gone into LEED [12] certification and many construction material manufactures. This
other buildings have been accredited by the methodology specified for providing a green
GREENSL® [13] of Green Building Council of Sri label through evaluating greenness of
Lanka (GBCSL). In both LEED and GREENSL® manufacturing phase of the construction
accreditation schemes, considerable percentage materials by observing the company’s
has been given to construction material and in adherence to the country regulations where the
GREENSL® to disposal. Though selection of final rating is based on the evaluators without
construction materials is a paramount factor for having an overall representation of the
environmental friendly buildings, available environmental impacts. Moreover, the major
tools and methods are not sufficient in gap of this methodology is lack of influence in
visualizing the environmental impact in all four Pre-manufacturing, Manufacturing, Use and
life cycle phases. Mainly the lack of awareness Post-use phases for the evaluation process.
of the decision makers in environmental impact Therefore, the final green label does not
calculation methods has to be addressed via visualize the overall impact of the labeled
such a methodology. Therefore, it is better to material. Other than GBCSL Green Label there
develop a country specified tool to visualize the are no any Eco-Labeling systems which are
environmental impacts of construction focus on Sri Lankan requirement. However,
materials and products considering whole life there are tools [21, 22, 23] developed based on
cycle. LCA which focused on sustainable construction
material selection. Though these can be used as
International Organization for Standardization decision making tools in sustainable material
(ISO) has introduced ISO14020 series of selection it is unable to visualize the
standards for environmental labelling [14] with environment performance in different life cycle
three types of environmental labels such as phases. When consider the local requirement,
Type I, Type II and Type III which are still there is no any tool or method which is able
ENGINEER 2
546
to visualize the environmental performance of extraction, pre-processing and transportation to
the different phases of a product life cycle. the manufacturing plant are considered under
Since there is a big requirement for such a pre-manufacturing phase. Gate to gate
system specially for construction industry, the processing inside the manufacturing facility is
proposed study focuses on develop a tool to considered under manufacturing phase.
evaluate the environmental performance of Distribution of finished products and impact
construction materials throughout the life cycle generating during consumption of the product
and visualize them via Eco-Labeling system. are considered at use phase. Finally reuse,
This tool will be elaborated in research recycle, recover or disposal of the product after
methodology and then results interpretation its use is considered at post-use phase. This has
and decision making strategies will be been elaborated in Figure 3. It is expected to
described through a case study, discussion and develop a data collection format to collect
conclusion. environmental impact related data of
construction materials use in Sri Lanka and to
2. Methodology develop a methodology to issue a graphically
represented Eco- Label for construction
The methodology of the study mainly consists materials.
of six phases and the structural overview of the
methodology is given in Figure 2. Phases of
methodology are, Goal and scope definition of
the study, Identification of assessment
indicators, Inventory preparation for data
collection, Development of analytical
framework, Introduction of a technique for
results interpretation and validating the
proposed methodology through a case study.
Figure 3 - Total Life Cycle representation of a

product
2.2 Impact indicator identification
There are 14 impact categories defined for LCA.

However, to calculate the impact under each
category will require software assistant and
database support. Commercially available
software such as SimaPro, Gabi and Athena or
freely available Open LCA can be used with the
support of a database such as Eco-invent.
Apparently the available data sets in currently
available databases does not represent Sri
Lankan data. Other than that conducting a LCA
Figure 2 - Methodology for total life cycle is time taken effort which
requires expertise knowledge. Therefore, for a
2.1 Goal and Scope definition convenient and user-friendly tool, number of
impact indicators were limited to four.
The goal of the study is to develop a tool to Otherwise decision makers will face
compare environmental impact of construction complications. Following impact categories for
materials throughout the life cycle phases each phase of the product life cycle are selected
under selected environmental impact categories for the proposed tool. Selection of impact
specifically highlighting the particular phases categories was subjected to the data availability
which has a significant impact which can be and calculation simplicity for the better
easily compared by the construction industry understanding of consumers and users.
professionals without deep knowledge in 1. GHG emission = I GHG
Sustainability or Greenness of the supply 2. Material Consumption = I Material
chains. The life cycle phases considered here 3. Energy Consumption = I Energy
are Pre-manufacturing, Manufacturing, Use 4. Solid Waste = I Waste
and Post-use. Activities such as material
3
547 ENGINEER
GHG emission and Energy consumption  Absolute value of GHG emission in
directly considered in each phase. Solid waste manufacturing phase = I mfg., GHG
amount considered under two categories such  Maximum value of the GHG emission in
as hazardous waste and non-hazardous wastes. manufacturing phase among all
Since the material consumption occurs in construction materials = Max I mfg., GHG
different ways along life cycle phases, virgin  Minimum value of the GHG emission in
raw material percentage in pre-manufacturing, manufacturing phase among all
packing and indirect material in construction materials = Min I mfg., GHG
manufacturing, supportive materials  Indicator score of GHG emission in
consumption in use and total of the ratio of manufacturing phase = S mfg., GHG
reusability and recyclability under post use are
considered. Then,
2.3 Inventory Preparation

…. (1)}
Considering above indicators, a data collection
Calculation of weighted impact under each
sheet was prepared by covering all four life
phase of GHG emission
cycle phases. It is expected to collect data for all
the construction materials manufacturing in Sri
After calculating the performance scores they
Lanka. And also, the data from published
were normalized along the phases by
research papers and recommended web sites
considering the weighted average of maximum
also need to be collected for the imported
indictor value under each phase.
construction materials. The sample inventory
Let,
table which was completed for the case study is
Maximum values of GHG emissions at four
given in Table 1. 
phases for all construction materials are;
Max I pre,GHG, Max I mfg,GHG, Max I use,GHG
2.4 Development of Analytical
and Max I post,GHG
Framework
 Weight assign for GHG emission in
manufacturing phase = W mfg., GHG
Analytical frame work developed in an Excel
work book which calculates the resulted values  Weighted impact of GHG emissions at
for above four measuring indicators in each life manufacturing phase = WI mfg,GHG (X12 in
cycle stage for defined functional unit. Table 2)
Calculation process consists three steps such as:
calculation of performance scores, calculation of Then,
weighted impact under each phase for
indicators, calculation of overall impact. The
calculation methodology has been presented
bellow for the GHG emission indicator by
deriving equations for each step. The same
methodology can be applied for other three … (3)}
indicators. Within this analytical frame work
the problem oriented approach (mid-point) was Calculation of overall impact
used for the simplification and final results also
presented directly for impact categories. After calculating the weighted impact for all
indictor values in each phase, weighted impact
Calculation of performance scores matrix (Table 2) can be developed. Total life
cycle impact of each indicator can be calculated
Performance score is the 0-1 scaled value for by taking the summation of values in each
each indicator. Scaling has done by considering phase. In this study it has considered that all
minimum and maximum values identified for four indicators are equally important for a
Sri Lankan construction materials under each simplified calculation methodology. However,
phase. However, still in Sri Lanka there is no if it is not applicable the Analytical Hierarchical
such database available. In this proposed tool, it Process is recommended to calculate weights
is assumed that these data are available for the for each indicator. Overall impact can be
decision making and all calculation steps of obtained by taking the average of total life cycle
analytical framework was developed based on impact values which is a value between 0 - 1.
that. Let,
ENGINEER 4
548
Table 1 - Completed Inventory for S1 clay roof tile
Parameter Unit Pre-mfg Mfg Use Post-use

Virgin Raw Material kg per Product 4.52
Recycled Raw material kg per Product 0.502 2.62
Packing material kg per Product 14.59
Other material required for use kg per Product 0.015 2.33 0.72
Water l per Product 4.2
Electricity (National grid) kWh per product 0.28 0.79
Electricity (Generator) kWh per product 0.24
Process fuel oil consumption kg per Product 0.9 38.5
Process Gas consumption kg per Product 0.44
Process wood consumption Kg per Product
Transport distance (air) tkm per product 0.049
Transport distance (land) tkm per product 0.291 0.010 0.003 0.16
Transport distance (water) tkm per product 0.140
Solid waste amount (Hazardous) g per product
Solid waste amount (Non- kg per product 0.021
hazardous)
Recycle Amount % from product 20
Landfilling Amount g per product 336
Reuse Amount % from product 65
Figure 4 - Graphical representation of environmental impact
5
549 ENGINEER
Table 2 – Weighted impact matrix obtained through the case study were given in
the next chapter.
Indicator Pre Mfg Use Post Total life
mfg use cycle
GHG X11 X12 X13 X14 ∑ X1
3. Case Study
emission
Material X21 X212 X23 X24 ∑ X2 Selected manufacturing industry process more
consumption than 2000MT of clay per month and produces
Energy X31 X32 X33 X34 ∑ X3
consumption over 300,000 clay tiles per month. Cradle to
Solid waste X41 X42 X413 X44 ∑ X4 Gate data were collected through monitoring
Overall - - - - ∑ X/4 their available data and taking measurement
impact during the process. Data for use phase and post
use phase were collected from expertise and
2.5 Results Interpretation through calculations. Calculated results for
each indicator are presented herewith Table 2.
The results obtained through the weighted
impact matrix are to be represented graphically Table 3 – Calculated results of each indicator
using a pre-defined criterion (Figure 4). The for S1 clay roof tile
radius of the outer most circle represent the Indicator Unit Pre Mfg Use Post
mfg use
value range from 0 to 1. And the radius of the
GHG kg CO2 Eq 185.4 2848.5 42 34.5
inner circle represents the total impact of the emission
particular product which is to be considered in Material kg 75.6 0.27 10.8 -
the environmental labelling. consumption
Energy MJ 22.05 79.2 4.95 3.3
consumption
The four phases of production that are, pre- Solid waste kg 0 0.315 0.45 7.2
manufacturing, manufacturing, use and post
use phases were named separately in the
diagram against the area govern by each phase. Material consumption under post use phase has
Then each phase was sub divided in to four measured as the total of reusability and
areas that represent the relative proportions of recyclability ratios. That value has been
impact categories naming, GHG emission, estimated 0.85. The results above founded were
Material Consumption, Energy Consumption scaled to the performance score and converted
and Solid Waste generation. There are to the weighted impact values following the
occasions, where you can find the equations 1,2,3. Calculated weighted impact
manufacturing phases only with three sub values are given in Table 4. Further the total
divisions where the value related to the impact for the entire life cycle of S1 clay roof
category is zero. To find the area to be covered tile according to the proposed tool was
by each indicator under each phase can be calculated as 0.51. The resulted Eco-Label
found by normalizing the values of weighted representation is illustrated at Figure 4.
impact matrix.
Table 4 – Weighted impact of each indicator
for S1 clay roof tile
Since such a graphical representation interprets Indicator Pre Mfg Use Post Total life
the environmental impact through entire life mfg use cycle
cycle of a particular product, this can be GHG 0.056 0.863 0.013 0.010 0.942
recommended as an environmental label which emission
Material 0.214 0.001 0.013 0.025 0.253
will be a solution to the identified research gap.
consumption
Energy 0.090 0.345 0.014 0.012 0.461
2.6 Validating through a case study consumption
Solid waste 0 0.008 0.011 0.38 0.399
The case study was conducted for validating Overall - - - - 0.514
the developed tool. The study was done for S1 impact
clay roof tiles manufactures in Anuradhapura,
Sri Lanka. Further data available [24, 25] for
Asbestos roofing sheets and Calicut roof tile 4. Discussion
were referred for selecting minimum and According to the results, the overall
maximum values for each indicator. The environmental impact of S1 clay roof tile is
functional unit selected was the amount of each 0.514 out of 1. Since it is at the middle of the
roofing options required to cover one square impact scale, it can be recommended as an
metre (1 m2) area of the roof. The results environmental friendly product comparatively.
ENGINEER 6
550
However, in order to have a fair result it is a strong analytical method and it represents a
required to expand the data set so that deterministic approach.
benchmark values cover a large distribution. In
this case study, it has only considered the data
References
for 3 products which cause to deliver a more
relative outcome. 1. Krausmann, F., Gingrich, S., Eisenmenger,
N., Erb, K. H., Haberl, H. and Fischer-
As shown in Figure 4, the highest impact occurs Kowalski, M., 2009. Growth in Global
at manufacturing phase and next at pre- Materials Use, GDP and population during
manufacturing. Most critical indicator for S1 tile the 20th century. Ecological Economics,
is GHG emission and it contributes more at 68(10), pp.2696-2705.
manufacturing phase. Material consumption at
pre-manufacturing, and energy consumption at 2. Crosthwaite, D., 2000. The Global
both manufacturing and pre-manufacturing are Construction Market: A Cross-Sectional
Analysis. Construction Management &
the other significant indicators. By reducing
Economics, 18(5), pp.619-627.
these high impacts, the radius of the impact
circle can be got closer to the centre and it will 3. Ofori, G., 2000, November. Challenges of
be a more environmental friendly material. Construction Industries in Developing
Countries: Lessons from Various Countries.
5. Conclusions In 2nd International Conference on
Construction in Developing Countries:
The tool developed as the result of this research Challenges Facing the Construction Industry in
can be recommended as an initial approach of Developing Countries, Gaborone (pp. 15-17).
achieving the product environmental impact
representation for Sri Lanka. The unique 4. Othman, E. and Ahmed, A., 2013.
features which can be highlighted in this tool Challenges of Mega Construction Projects in
Developing Countries. Organization,
out of the other tools developed are;
Technology & Management in Construction:
 Simple representation of the An International Journal, 5(1), pp.730-746.
environmental impacts using a
graphical representation which can be 5. Pereira, K. and Ratnayake, R., 2013. Water
Integrity in Action: Curbing.
understand by any person without
deeper knowledge on manufacturing, 6. http://www.tradingeconomics.com/sri-
LCA, Environmental impacts etc. lanka/gdp-from-construction, Visited, 19th
 Representation of the four phases of May 2017.
product the life cycle in one pictorial
diagram 7. http://www.statistics.gov.lk/industry/SCI
 The easiness of comparing two or more _Final_Report_2010.pdf, Visited, 19th May
products together as a better decision 2017.
making support system
8. http://www.statistics.gov.lk/industry/SCI
However, the developed tool is to be validated _Final_Report_2015.pdf, Visited, 19th May
by applying it to different construction 2017.
materials available in Sri Lanka to overcome the
9. Reiter, S., July, 2010. Life Cycle Assessment
pros and cons before introducing it to the
of Buildings–A review. In Proceedings of
interested parties including manufactures,
the Sustainability Workshop and Third
consumers, contractors, certification bodies, Plenary Meeting, Brussels, Belgium (pp. 1-
and other third party. 19).
The main disadvantage of this proposed tool is
that it only developed based on the 10. Borg, M., Paulsen, J., and Trinius, W, 2001,
environmental perspective of a product. Since Proposal of a Method for Allocation in
the economic and social aspects also important Building-Related Environmental LCA
in making purchasing decisions the Based on Economic Parameters,
International journal of Life Cycle
representation of other two aspects is important
Assessment, pp. 219 – 230.
and that makes the entire overview of a
product. The developed tool is consisting with 11. Frischknecht, R., Wyss, F., Knöpfel, S.B. and
a graphical representation that elaborates the Stolz, P., 2015. Life Cycle Assessment in the
magnitude of each impact as per each phase. Building Sector: Analytical Tools,
And also, this tool development is not based on Environmental Information and Labels. The
7
551 ENGINEER
International Journal of Life Cycle 22. Abeysundara, U. Y., Babel, S. and
Assessment, pp .421-425. Gheewala, S., 2007. A Decision Making
Matrix with Life Cycle Perspective of
12. U.S. Green Building Council (USGBC). Materials for Roofs in Sri Lanka. Materials
LEED® for New Construction & Major & design, 28(9), pp.2478-2487.
Renovations (LEED-NC), 2005, Version 2.2”.
23. Abeysundara, U.Y., Babel, S. and Gheewala,
13. Green Building Council Sri Lanka. S., 2009. A Matrix in Life Cycle Perspective
GREENSL® Rating System for Built for Selecting Sustainable Materials for
Environment. Version 1. Buildings in Sri Lanka. Building and
Environment, 44(5), pp.997-1004.
14. ISO 14020, Environmental labels and
declarations — General principles. 24. Kuruppuarachchi, K. A. B. N., Life Cycle
Assessment of Two Different Clay Roofing
15. Mungkung, R., De Haes, H.U. and Clift, R., Tiles KABN Kuruppuarachchi, RK
2006. Potentials and limitations of Life Ihalawatta, AK Kulatunga Department of
Cycle Assessment in Setting Ecolabelling Production Engineering, Faculty of
Criteria: A Case Study of Thai Shrimp Engineering, University of Peradeniya, Sri
Aquaculture Product (5 pp). The Lanka.
International Journal of Life Cycle
Assessment, 11(1), pp.55-59. 25. Charles, C. V., Jeyamathan, S. J. and
Rameezdeen, R., 2008. Sustainability Index
16. Fet, A. M. and Skaar, C., 2006. Eco-labeling, for Roof Covering Materials. Building
Product Category Rules and Certification Resilience, p.140.
Procedures Based on ISO 14025
Requirements (6 pp). The International
Journal of Life Cycle Assessment, 11(1),
pp.49-54.
17. Thrane, M., Ziegler, F. and Sonesson, U.,

2009. Eco-labelling of Wild-Caught Seafood
Products. Journal of Cleaner Production,
17(3), pp.416-423.
18. Canals, L. M., Domènèch, X., Rieradevall, J.,

Puig, R. and Fullana, P., 2002. Use of Life
Cycle Assessment in the Procedure for the
Establishment of Environmental Criteria in
the Catalan eco-label of leather. The
Assessment, 7(1), p.39.G. L. Baldo, S.
Rollino, G. Stimmeder and M. Fiesch, The
Use of LCA to Develop Eco-Label Criteria
for Hard Floor Coverings on Behalf of the
European Flower, International Journal of
LCA (2002) 269-275.
19. Finkbeiner, M. and Hoffmann, R., 2006.

Application of Life Cycle Assessment for
the Environmental Certificate of the
Mercedes-Benz S-Class (7 pp). The
Assessment, 11(4), pp.240-246.GBCSL
labelling scheme.
20. Green Building Council Sri Lanka. GREEN

Labeling System for Construction Materials.
Version 1.
21. Abeysundara, U. Y., Babel, S., Gheewala, S.

and Sharp, A., 2007. Environmental,
Economic and Social Analysis of Materials
for Doors and Windows in Sri Lanka.
Building and Environment, 42(5), pp.2141-
2149.
ENGINEER 8
552
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[553 563],2017
2017
The Institution
Lanka
Safety Analysis of Anaerobic Digestion Plant

Operation
Deshai Botheju and P. G. Rathnasiri
Abstract: Biogas generation by anaerobic digestion is rapidly gaining ground as a popular

option for sustainable energy generation in Sri Lanka (SL). More importantly, italso contributes to
formulating a successful waste management strategy SL is desperately in need for. However, despite
the great attention put on research and development related to efficiency enhancement, process
control, degradation of resilient organic matter, and gas utilisation techniques, only a marginal focus
is so far being made on the essential safety technologies that must be incorporated into industrial and
pilot scale biogas plants. This technical paper aims to provide a general overview of various safety
dimensions related to biogas plants and offer guidance to plant designers and operators to incorporate
essential technical, procedural, and organizational barriers with respect to process safety, occupational
safety, and environmental care. A case study is also presented where it relates to the context of the
paper.
Keywords: Anaerobic digestion, Biogas, Environment, Explosion, Process Safety.
1. Introduction 1.1 Biogas safety – current perspective

In the early days of biogas generation, the
The anaerobic digestion (AD) is accepted as a processes were rudimentary, the generated
sustainable process in the context of waste volumes were quite small, and the large scale
management and carbon-neutral energy plants were rare, thereby safety was not
generation in the form of biogas. Mostly free perceived as a challenge. However, during last
from the issues that infested old generation AD two decades, biogas generation has been
reactors like unstable operation, poor methane steadily rising throughout the globe, as a result
yield, and low reliability, the new era of biogas of the focus on climate neutral energy
energy is promising and is rapidly becoming generation. Many governments supported this
adopted in Sri Lanka as a workable technology trend by heftily subsidising the biofuel and
for organic waste treatment and energy renewable energy industry. The increased
generation. The Sri Lankan experience at this number of biogas plants, as well as the
point has shown that, with proper use of significant increase in the size (scale) of biogas
process technologies, biogas generated from installations during recent times, demand high
organic waste streams, such as the organic attention on associated safety risks. In
fraction of MSW, can effectively be utilized in addition, dramatic shifts in the social
multiple platforms such as electricity acceptability and perception of industrial risks
generation, transport fuel for vehicles, or as a [1] have led to the adoption of stricter safety
heating/cooking fuel. However, only a standards and regulatory controls all over the
peripheral attention has so far being made on globe.
the important process safety and HSE
considerations of AD plant operation and
biogas handling.
1.2 Scope
We emphasize here that process safety aspects
are critically important as biogas contains
Eng. (Dr.) Deshai Botheju, Ph.D.(NTNU), M.Sc.Tech
highly flammable methane that poses a (Norway), M.Sc. & B.Sc.Eng. (Moratuwa), AIChE,
significant safety risk when mixed with air AMIE(SL). Currently working for Norwegian Petroleum
within its flammability range (≈ 5 -15 vol % for industry, as an expert consultant in Safety Engineering
and HSE Managemnt.
methane in air). AD reactors can also lead to
significant HSE risks such as exposure to the Eng. (Dr.) P. G. Rathnasiri, Ph.D.(NTNU, Norway),
toxic H2S byproduct gas, asphyxiation danger, M.Sc. (UK), B.Sc.Eng (Moratuwa), AMIE(SL).
and biological hazards associated with wastes Senior Lecturer & the former Head of the Dept. of
Chemical and Process Engineering, University of
feeds and reactor sludge. Accelerated corrosion Moratuwa.
resulting from H2S is another risk factor.
1
553 ENGINEER
This technical paper presents a comprehensive biogas stream. In most cases, it is sufficient to
safety analysis covering AD plant operation. consider biogas primarily as a mixture of
The work is founded on long term experience methane (CH4) and carbon dioxide (CO2) with
of the authors related to the fields of Anaerobic some minor amounts of Hydrogen sulphide
Digestion and Safety Engineering. Further, a (H2S), ammonia (NH3), and water vapour.
case study is presented based on the pilot AD Nevertheless, biogas generated (naturally) from
plant treating canteen food wastes at the highly complex sources such as landfills might
University of Moratuwa contain a large number of other trace
compounds[3].
In this paper, we intend to analyse related
safety considerations under three (3) main The composition of the main flammable agent
categories. They are: (i) Process safety, (ii) of biogas, i.e. methane, may also widely vary
Occupational health and working environment, (from about 10 %to 80 % vol.) depending on the
and (iii) Environmental care. Note that, “safety” biochemical state of the process. Methane and
can be briefly defined as “freedom from carbon dioxide compositions close to 50 - 60 %
unacceptable risk” [2].Accordingly, the concept and 38 - 34 %, respectively, are quite common
of industrial safety encompasses many for the digestion of municipal organic
discretely recognized disciplines such as wastes[20].
Process Safety and Loss Prevention,
Occupational Health & Safety, Environmental Table 1 - Typical composition and flammable
Management, etc. properties of biogas constituents [4, 5, 3, 22].
Components Typical LEL UEL
composition (% in air at (% in air
It is emphasized that only through a systematic
STP) at STP)
and holistic approach to process safety and CH4 50 - 70 % 5 15
HSE can avert catastrophic plant accidents like CO2 30 - 50 % Non-flammable
explosions and process fires, tragic incidents of H 2S 1-2% 4.3 46
personal injuries, as well as unwarranted NH3 <1% 15 28
H2 <1% 4 75
damages to plant equipment and client utilities.
VOC <1% vary vary
H2O vapour <1% Non-flammable
Biogas As a 10 30
2. Process safety considerations mixture (of
60 % CH4)
The loosely defined feed and gas compositions, 2.1 Flammability / Explosivity
and varying production rates are among the Methane, the main energy generating
key challenges related to biogas safety. Biogas component of biogas, with 36 MJ/m3 of gross
is generated as a result of a multitude of calorific value [6], creates explosive mixtures
biochemical reactions; thereby the product with atmospheric air within the concentration
composition and the volumetric gas generation range of ≈ 5 - 15 % [4]; i.e. the limits of
rate depend on a number of less-controllable explosion (Lower explosion limit –LEL, and
parameters such as feed material composition, Upper explosion limit - UEL). Therefore, it is
ambient temperature, the presence of inhibitory critically important to avoid biogas being
agents, etc. In many cases, the digester feed is a allowed to mix with air either within the
mixture of organic waste material (such as digester or inside any gas storage vessel. Biogas
municipal solid waste, waste water sludge, must only be allowed to mix with combustion
agricultural residue, food wastes, etc.) of which air within a suitable burner device at a final
the composition and mixing proportions can energy releasing step (e.g. gas burners, internal
vary throughout the year. Therefore, the gas combustion engines, orgasflares). Table 1
composition and the volumetric generation further shows the flammable and explosive
rates to be used at the plant design stage are properties of the other constituents of biogas.
often the best possible estimates available at H2S, NH3, and H2 can, in standalone mode, also
that time. All process safety measures and other act as flammable agents, although their typical
safety procedures must, therefore, be concentrations in biogas mixtures are so low
determined after careful consideration of the that they do not normally pose a significant
foreseeable worst-case scenarios. explosion hazard.
Table 1 illustrates some of the key

physicochemical parameters, important for
safety considerations, applicable for a “typical”
ENGINEER 2
554
2.2 Hazardous zone classification (explosion protected) for the respective zone.
Hazardous zone classification (HZC) is a safety Zone 0 environments must usually be devoid of
engineering technique used in process plants any potential ignition sources (e.g. inside of the
handling potentially flammable and /or digester), as only a handful of devices exists if
explosive substances (i.e. flammable liquids, at all, that can be used within a zone 0
gases, and combustible dust particulates). The environment. Explosion protected electric
intention of this technique is to tag the areas equipment utilise different techniques to avoid
surrounding hazardous installations (e.g. or minimise the likelihood of spark generation
reactor vessels, process areas, storage areas, (such as low energy non-sparking equipment,
engine and boiler rooms, etc.) based on the encapsulated equipment, oil submerged
likelihood of presence (or release) of flammable equipment, etc.).
materials. Table 2 indicates a generally accepted
criterion for the four (4) different types of zones 2.4 Gas detection
often defined for process plants.
The only reliable way to identify gas hazards
Table 2 - Hazardous zone during the early phase of a leakage is to use gas
classification(abstract definitions are based on sensors which could warn the impending
IEC 60079 Part 10-1 standard; [7]) danger by audio-visual alarms (i.e. sounds and
Zone Abstract definition Example light flashes). For biogas plants, two types of
classification gas detectors, i.e. flammable gas detectors, and
Zone 0 Explosive gas Inside the toxic gas detectors, are relevant.
atmosphere is digester
present continuously
2.4.1 Flammable gas detection
Zone 1 Explosive gas The adjacent
atmosphere is likely area near Fixed gas detection devices that can identify
to occur periodically pressure flammable methane at about 20 – 40 % of the
during normal relief valve. LEL must be installed at strategic locations
operation where a flammable biogas cloud is likely to be
Zone 2 Explosive gas Process area present during a gas leak event (e.g. near
atmosphere might /process digester top, gas storage tanks, near gas
be present for short buildings utilization devices such as electric generators or
durations near the boilers, etc.).
occasionally digester.
Non- Explosive gas is not Office
2.4.2 Toxic gas detection
hazardous perceived as a buildings of
zone danger in this area. the plant.
Continuous detection of H2S is recommended
near medium to large scale biogas plants. Being
a highly toxic gas, it is necessary to identify the
2.3 Ignition source control presence of H2S even at low concentrations (See
Table 3). If a permanently mounted H2S
If a flammable gas cloud (e.g. methane from detector is not available, it is vital to use a
biogas mixed in air) is formed within the plant handheld H2S detector when working near
(e.g. due to a gas leak or failure of process active digesters with low ventilation conditions
containment), the last barrier preventing a (e.g. a digester located inside a plant house).
catastrophic fire or an explosion is to keep
ignition sources (such as electrical sparks or hot In addition, carbon monoxide (CO) detectors
surfaces) from igniting the flammable gas must be installed at enclosed machine rooms
cloud. Therefore it is important to use explosion containing power generators and fired boilers
proof (EX rated) equipment, tools, and electric where combustion might generate deadly CO
& electronic equipment within areas where gas.
flammable gases are likely to be present.
Accordingly, proper hazardous zone 2.4.3 Portable gas detectors
classification (HZC) is the first step towards a Portable gas detectors are often used as a part
sound ignition source control (ISC) philosophy. of occupational safety procedures whenever
operational or maintenance tasks are to be
All electrical and electronic devices (e.g. conducted near or within digester tanks and
pumps, motors, light fixtures, control circuits, pits. These are handheld devices that can
instrumentation and sensors, fire and gas indicate both the presence and concentration of
detectors, etc.) that are to be used within a Zone dangerous gases (flammable and toxic).They
1 or Zone 2 environment must be EX certified are also useful for recognizing
3
555 ENGINEER
potentiallyasphyxiating environments if they 2.7 Ventilation
could indicate the available oxygen level in the Plant buildings near digesters must have
environment. There are many handheld gas suitable ventilation arrangements, in order to
detectors which could simultaneously detect avoid accidental accumulation of flammable
multiple gases and indicate their and toxic gases. A ventilation air change rate of
concentrations, and also emit alarm sounds if 5 air exchanges per hour [4] is recommended. If
dangerous levels are detected. It is, however, natural ventilation is the only available
important to keep such devices properly ventilation mean, then large openings must be
calibrated all time. used with ample air circulation access to ensure
adequate ventilation could be realized even
2.5 Fire detection and fire fighting under unfavourable stagnantatmospheric
conditions.
The biogas plants must be protected with fire
detection devices (e.g. infrared flame detectors, For mechanical (forced draft) ventilation
smoke detectors, etc.) and firefighting means systems, it is recommended to install at least 2
such as deluge nozzles, sprinklers, fire hoses, gas detectors (one for flammable methane and
portable fire extinguishers etc. the other one for toxic H2S) at the ventilation air
inlet. Upon detection of gases at this location,
2.5.1 Fire and gas detector actions the mechanical ventilation system must
automatically stop.
Medium to large scale biogas plants need
automatic safety actions based on input from Special ventilation conditions required for
gas detectors and fire detectors. Confirmed maintenance and repair activities within
detection of gas (based on multiple detectors) digesters is briefly described in section 3.3.2.
must preferably deactivate /shutdown
potential ignition sources such as electrical 2.8 Pressure safety
equipment, electronic circuitry, hot surfaces
(e.g. electric heaters), flames (e.g. gas burners, An anaerobic digester is a biochemical reactor
boilers), and internal combustion engines which gradually builds up pressure due to the
(biogas powered electricity generators). Fire or gas generation. If gas is not removed from the
gas detection must preferably shut the main system, the pressure will continue to build up
gate valve of the gas supply connected to the until it leads to a material failure (e.g. pipe
digester or the gas storage vessel. In addition, rupture, or vessel blow up). Therefore, medium
various other safety actions such as activation to large scale biogas plants, require dedicated
of deluge or sprinkler systems, change of pressure safety designs to avoid such
ventilation conditions, etc. may be required dangerous failures which could then lead to
according to the plant design and the Fire & process fires and explosions. If normal process
Gas control logic defined based on actual risk control system fails to maintain the operating
levels. pressure conditions (e.g. by controlling digester
feeding rate and controlling gas removal flow
2.6 Corrosivity rate), an emergency safety system must kick in
to bring the digester into a “safe status”. The
Anaerobic digesters and the surrounding plant intended safe status, in this case, can be
equipment can be subjected to highly corrosive achieved by depressurizing the digester (i.e.
conditions due to the presence of corrosive lowering the energy content of the system).
digester liquors and the gaseous components This could be achieved by using emergency
such as H2S, CO2 (through carbonic acid pressure safety devices such as the opening of a
formation), and also H2. Therefore, the plant pressure safety valve (PSV) and venting or
design must incorporate such considerations allowing to burst a rupture disk. In more
and the material selection must be done with advanced designs, various instrumented safety
due care. High corrosion, if not properly taken functions shall be used to bring the entire plant
care of, is a safety threat to the plant’s to a safe shutdown (together with ignition
functional integrity as well as to the structural source control and depressurization)when
strength. Corrosion could lead to dangerous detecting such dangerous developments.
gas leaks also. Various process equipment such
as pipes, valves, vessels must have Continuous monitoring of digester pressure
inbuiltresilience against the highly corrosive using process instrumentation is a sound safety
environment. strategy for any biogas plant irrespective of the
ENGINEER 4
556
scale. For larger plants, it is advisable to vessel containment or gas piping, it can quickly
incorporate two independent pressure safety develop into a fire and explosion scenario.
barriers (e.g. a PSV and a rupture disk; or a PSV Various lightening protection means,[8] such as
and an instrumented pressure safety function). air termination & down conductor systems,
may be utilized for this purpose.
2.8.1 Negative pressure development
Plant electric supply and the electronic control
An anaerobic digester must always run with a systems (e.g. process control and emergency
positive pressure inside. In case, a negative control systems) must also have suitable surge
pressure development is allowed, then air protection devices installed so that critical
could draw into the digester and mixed with instrumented safety functions remain
biogas to create an explosive methane-air operational even if the plant is struck with a
mixture. A negative pressure development lightning induced surge current.
could occur if a gas treatment and utilization
systems such as a gas blower or a compressor
draw more biogas from the digester than the 2.11 HAZID, HAZOP, and Risk
system is capable of producing. This situation Assessments
must be avoided by incorporating necessary
process control strategies (such as gas flow Hazard identification (HAZID), hazards and
control), and also by designing emergency operability studies (HAZOP), and various risk
safety barriers such as instrumented safety assessment techniques must diligently be used
functions to shut down gas blowers or pumps based on the plant capacity and other relevant
upon the detection of low digester pressure. risk factors affecting the particular plant. For
small to medium scale plants, a HAZOP study
2.9 Flame arresters (based on plant P&IDs) and a qualitative risk
assessment study must suffice. Large scale
Installation of flame arresters on biogas supply plants and the plants with significant risks to
lines is recommended in order to avoid a flame surrounding third parties must be based on
travelling back through a gas pipe and igniting more rigorous risk assessment and safety
the gas vessel (i.e. digester or a gas storage techniques such as quantitative risk
vessel). A flame arrester is simply a short pipe assessments, gas dispersion studies, working
segment filled with a mesh of fine metal wires environment and health risk studies, etc.
(see Figure 1). The wire mesh must be corrosion
resistant against H2S (e.g. copper is a suitable 2.12 Dispersion analysis
material). The metal wire mesh absorbs heat For large biogas plants, it is advisable to
from a propagating flame and quenches it conduct a numerical simulation of gas
within the flame arrester. dispersion together with actual plant geometry
and surroundings. This will be used to generate
Slightly larger a realistic picture of a gas leak scenario and to
Wire mesh
pipe segment acquire more accurate details on potential
Gas line impacts. Hazardous zone classification and
locating of gas detectors must then be based on
this information. Often, such numerical studies
are a part of a wider quantitative risk
assessment.
Figure 1 - A simple flame arrester
3. Occupational safety, health, and
A flame arrester must be located upstream the
working environment
gas burner (or a flare). In addition, it is
recommended to include a flame arrester
considerations
downstream the main gas valve of a digester or
Apart from critical process safety concerns and
gas storage vessel.
safety barriers introduced at the design stage,
day-to-day operation of biogas plants requires
2.10 Lightening protection.
attention to various occupational hazards,
Taller digesters and the digesters in open health impacts, and working environment
considerations.
landscapes must be lightening protected. If
lightning strikes a digester and damages its
5
557 ENGINEER
3.1 Asphyxiation challenging at the design stage of a biogas
plant. Therefore, it is always advisable to use
Asphyxiation is the greatest working H2S detectors near digesters.
environment hazard for those who perform Table 3 illustrates toxicity information on
work near digesters, especially the inside of various biogas components.
anaerobic digesters, during occasional
maintenance works or breakdowns. Table 3 - Toxicity and Health hazards of gases
Asphyxiation happens when rapidly generating found in biogas plants [10, 11, 12, 21].
biogas pushed out air from a confined space, or Gas Toxicity Olfactory Impact
when biogas is accumulated within an enclosed limit sense
area such as a pit or a digester vessel. When the (ppm)
CH4 Non Odourless Asphyxiating
oxygen level in theair is diluted below 10 %[9],
toxic
a person may quickly succumb. However,
CO2 1000 Odourless Drowsiness
already at 19 - 16 % oxygen levels, a person and headache,
would be subjected to impaired judgement [5, Asphyxiating
9]. Therefore it is not generally recommended H2S 20 Rotten egg Highly Toxic
to enter any area with less than 19 % smell at
oxygenunless a self-contained breathing device low
is worn. concentrati
ons
Wet waste storage pits or a manure tanks can NH3 50 “ammonia Eye, throat,
” smell and lung
inadvertently act as anaerobic digesters,
irritation. Fatal
depending on the environmental conditions
CO 50 Odourless Highly Toxic
(temperature, humidity, etc.). A worker used to
enter such area without harm in previous
CO2, though not considered as a toxic gas, can
occasions may suddenly succumb to
have negative health impacts at elevated
asphyxiation if the waste material is now
concentrations. If breathing CO2 concentration
undergoing digestion process generating
increases beyond 1000 ppmlevel [12],impaired
biogas. Also, even an open top pit (or a tank)
thinking and difficulty of concentration may
that could look like safe enough due to its open
occur. At further increasing concentrations,
nature might be a hazardous location to enter
symptoms like dizziness increases respiration
as the rate of gas generation may be so high to
rate, and heart rate will occur. Eventually
drive out oxygen (air) from the tank /pit
passing out would occur as CO2 displaces
interior.
oxygen supply in the blood stream. OSHA
recommends 5000 ppm [11] as the permissible
3.1.1 Gas layering
limit of CO2 in theair for an 8 hour work day
Layering refers to the stratification of various
and a 40 hour work week for a lifetime of
biogas components based on the relative
exposure.
density of various gases. Methane has a lower
density than air; therefore it floats upwards and
NH3 also acts similar to CO2 in displacing blood
usually accumulates at higher levels of a pit or
oxygen levels [13].
a tank. If biogas ingress happens into a closed
building (e.g. plant house), the methane will be
Carbon monoxide (CO), even though not a
accumulating close to the ceiling level. On the
component of biogas itself, can be a safety
other hand, CO2 has a higher density than air;
hazard when internal combustion engines (such
therefore it accumulates close to the bottom of a
as electric power generators running on
tank or a pit (or a closed building). Considering
biogas), are operated within enclosed buildings.
the minor components; H2S is heavier than air
Exhaust gas from internal combustion engines
and accumulates at bottom layers, and NH3 is
contains CO in deadly amounts. In addition,
lighter than air and rises to upper layers.
some burners (e.g. biogas operated boilers) can
generate CO if complete combustion does not
3.2 Toxicity
occur. Use of CO detectors at such locations is
recommended.
Even thoughit is a minor component in biogas,
H2S could pose a formidable threat to personnel
3.3 Confined area entry
as it is a highly toxic gas. The possibility of
varying biogas composition (as explained
Entering a confined area (like a manure pit or
before) makes judging the H2S level more
an anaerobic digester) must only be carried out
ENGINEER 6
558
if it is really needed, and with a range of safety 3.4 Biological hazards
precautions.
Biogas is generated in a digester due to the
3.3.1 Eliminate entry requirements activity of a variety of microorganisms acting
The “safety in design” concept must always be on, often, waste materials such as manure,
adopted. Digesters must be designed with sewage, wastewater sludge, organic fraction of
minimum need of entry for maintenance; like municipal solid wastes, food wastes, etc. Such
installing mechanical devices such as mixers feed materials can also easily carry various
that can be replaced or repair from outside. pathogenic microorganism and parasites (e.g.
warms). Workers looking after digester
3.3.2 Ventilation operations usually exposed to these biohazards.
Whenever possible, excessive forced ventilation Therefore it is critical to use proper safety
must be used before and during any confined procedures and use good housekeeping
area entry. Either dedicated ventilation air practice within the plant facility. Digester
blowers may be installed as a part of the design, residence time and the operating temperature
or else, if such arrangements are not available, play a key role in determining the sanitary
external temporary fans or blowers must be condition of the digested sludge which must be
used to expel gases from the confined area handled after exiting the reactor. For example;
before and during the entry. Oxygen level in one study related to a plug flow digester
measurements and readings for toxic gases feeding dairy manure and operating under
(H2S) must be taken using portable gas mesophilic temperature conditions (i.e. ≈ 37 oC),
detectors before entry. with a retention time of about 36 days, a Fecal
coliform destruction of 99.9 % was achieved
3.3.3 Rescue [14]. This is considered to be a safe level of
If a person is found unconscious within an pathogen destruction. However certain
enclosed environment, one should never parasitic worms such as roundworms can
attempt a rescue by directly entering, without survive at a much greater percentage.
safety precautions. It is all more common that Therefore digester sludge, especially the sludge
multiple victims succumb to asphyxiation (or to resulting from digesting human and animal
toxic gases) by trying to rescue previous excretes, must be handled with care, and must
victims[5]. Such chain casualties can be avoided be considered as a biohazardous material for
by only letting professional or trained rescuers plant workers.
attempt rescuing victims. An enclosed area
with suspected asphyxiation or toxic gas 3.4.1 Vaccination
presence must only be entered by wearing a Exposed personnel such as plant workers,
self-contained breathing apparatus. Further, the researchers, and research students, must get
enterer must have a rescue rope attached to his appropriate vaccination shots at regular
body and a second person must be present with intervals as recommended by immunologists.
full attention on the enterer. The second person
must have some sort of pulling arrangement 3.4.2 Personnel protective equipment
(like a pulley) to withdraw the enterer by Proper use of personal protective equipment is
pulling on the rescue rope while standing essential. Gloves, overalls, safety goggles,
outside the danger zone (e.g. out of the pit or masks, and boots are required for the personnel
tank). The use of the self-contained breathing who are exposed to hazardous materials.
apparatus requires special training, and often
such use is restricted to fire fighters and 3.4.3 Plant house keeping
professional rescuers. Therefore, whenever, an Standard cleaning and sorting procedures
enclosed area entry with suspected gas must be followed throughout the plant facility.
presence is attempted during normal The use of 5S concept [15]is a proper approach
maintenance or repair work in a biogas plant, to this. Workers must also pay attention to
the other aforementioned safety precautions proper personal hygiene practice.
must be followed (i.e. rope attachment, asecond
person outside the area, and pulling 3.5 Safety signs
arrangement).
Safety signs are vital for delivering critical
safety related information to the plant workers
as well as to any occasional /first-time visitors
to the plant, in a simple and effective way.
7
559 ENGINEER
Prohibition of open flames, attention on alarms investigated, and necessary mitigation
and light signals, escape directions, assembly measures must be taken.
stations, the location of first-aid and firefighting
equipment etc. are among many things that 4.1 Emissions
must be indicated by proper signage. A mix of Emissions can primarily occur when the
illustrations and wording should be used in generated biogas is combusted to utilise its
such a way that the maximum comprehension useful energy (e.g. burning to generate heat or
is possible within the shortest possible time. combustion within an internal combustion
engine to produce electricity, or to achieve
3.6 Heat radiation vehicle mobility). Also,emissions occur when
unused or extra biogas is burned in a gas flare
Potential heat radiation impact from biogas (e.g. in a flare stack, or in an open flare burner).
flaring should be evaluated. Intense heat Even though the emission of air pollutants and
radiation can be hazardous for plant workers. climate gases such as CO2, SOx, NOx, unburnt
Expected heat radiation levels at exposed plant and molecularly rearranged hydrocarbon
locations must be estimated /calculated. For a components, and particulates are not expected
person without special protection suits (but to be a serious problem in relation to biogas
with ordinary clothing), maximum bearable (unlike liquid and solid fuels), particular
heat radiation exposure is given as 6.31 kW/m2 attention must still be paid to the air emission
for 30 seconds period[16]. This factor should standards within the relevant regulative
especially be evaluated for very large scale regime. The content of CO and unburnt
biogas plants. The height of the flare stack must hydrocarbons can most often avoided by
be designed to minimize this heat radiation realizing > 850 oC flame temperatures and > 0.3
impact on the plant, and also to properly sec. residence times [6]. At higher flame
disperse air pollutant emissions (section 4.1). temperatures (>1200 oC) increasing amounts of
atmospheric N2 oxidizes to form thermal NOx ;
Further, installation of a knock out drum before often explained by a mechanism called
the biogas flare is recommended in order to extended Zeldovich mechanism that comprised
avoid any liquids (such as reactor liquor or any of a set of highly temperature dependent
condensate from the gas stream) entering the chemical reactions [23] . The cleanliness of the
gas burner /flare stack. Liquids could not only generated biogas is an important factor here. If
increase the level of air emissions (section 4.1) there is a gas cleaning stage within the process,
but also, at worst cases, could result in then the level of CO2, H2S , NH3, and various
adangerous “burning rain” type scenario where other trace components (such as various
fiery liquid droplets / mists raining down from volatile organic components) will be
the flare stack. significantly low, thereby leading to a cleaner
combustion flue gas stream.
3.7 Noise
Exposure to high noise levels near machinery 4.2 Odour
like electric power generators, gas compressors,
biogas flares can be an occupational safety Due to the handling of various types of organic
concern in biogas plants. Procedural controls waste material, theodour might become a
must be in place to avoid workers being problem if not properly managed. Odours often
exposed to high noise levels for extended spread well beyond plant perimeters and may
period of times. Maximum allowable work impact neighbourhood areas.
duration shall be determined based on the
expected noise levels. For short exposure Following key steps must be considered in
intervals, ear protection devices such as ear order to minimize odour issues:
plugs and earmuffs must be used.
- Waste transportation, unloading, and
feeding the plant must be done with
4. Environmental considerations minimum exposure to open air.
- Digester residence time must be long
As described elsewhere [17] various enough for the waste to undergo complete
environmental care perspectives are also a part degradation. Only partially digested waste
of the concept of “safety”. Therefore any would emit significant odours.
environmental impacts related to the building - Reactor sludge stabilization and suitable
and operation of a biogas plant must be disposal means must be adopted.
ENGINEER 8
560
4.3 Contaminated sludge design and operation phases of this pilot
facility.
The sludge resulting from biogas reactor
operation has to be disposed of using Pilot plant building
sustainable means. It usually needs further By taking into account, safety aspects and
stabilization or treatment before release to the proper natural ventilation, the plant building
environment. Use of anaerobically digested was located in a remote part of the university
sludge as a soil conditioner or fertilizer is quite land, well away from the building clusters with
common. However, attention must be paid to student lecture rooms and laboratories. All
the fact that such sludge quite often contains sides of the building were kept half way open
high concentrations of heavy metals [18]. enabling proper natural ventilation and
Therefore, it is perceived as unsafe to use them dispersion of any potential gas leakage.
on food crops without careful analysis of the
content of heavy metals. Metals like Pb, Ni, Cd, Feed preparation system
etc. can gradually accumulate in living A crusher was installed to reduce the particle
organisms and could lead to chronical health size of the feed stock (food waste). A mesh
problems [19]. guard was installed around the crusher by
taking into account of the working environment
5. Case study considerations related to the personal safety of
the operator.
A pilot scale demonstrative anaerobic digester
(high rate reactor) was constructed on the Feed transfer system
premises of University of Moratuwa (UoM) A submersible sewage pump was installed to
with the objective of utilizing food wastes pump the food waste slurry into the anaerobic
generated in university canteens as the feed reactor. A submersible pump does not act as an
material and then using the generated biogas as ignition source even if a flammable gas cloud is
a transport fuel (for three wheel vehicles) after present.
upgrading. The project was aimed at
technology demonstration related to waste Material selection
management, process, biochemical, reactor, and Based on the project budget, the selected
safety technologies. Figure 2 illustrates the construction material was mild steel. However,
schematic arrangement of this pilot facility. with proper attention to the corrosive
atmosphere, inside of the reactor vessel was
The digester was designed to operate on 20 surfaced coated with high-grademarine paints
days retention time and at a maximum pressure and the outside with anti-corrosive paints.
of 1.5 bar. The bulk liquid capacity was 4.9 m3,
Pressure safety instrumentation
and the total reactor volume including head
A pressure gauge and safety valve were fixed
space was 5.2 m3. This designed gas
on the top of the vessel so that a continuous
production rate is 11.4 m3/d with an average
pressure reading can be obtained by the
CH4 composition of 55 %.
operators.
Gas cleaning and pressure safety

A water scrubber was designed and fabricated
to remove CO2and H2S from the biogas
mixture. This should greatly reduce the
corrosive nature of the product biogas stream.
The raw gas enters the scrubber at 10 bar
pressure from the bottom and water is sprayed
from top of the column. Operating pressure is
also 10 bars inside the vessel. Safety valve for 10
bars has been fixed on the top of the scrubber
which is fabricated from stainless steel.
Figure 2 - Pilot scale anaerobic digester at
UoM aimed at technology demonstration.
Intermediate gas storage and compression
Biogas is required to storage at two stages. i.e.
The following is a brief description of the key
between the reactor and scrubber and before
safety considerations evaluated during the
pumping to the vehicle. The compressor used
was a scuba type compressor which is operated
9
561 ENGINEER
by gasoline and has a maximum capacity of 250 2. International Electrotechnical Commission (IEC).
bars. Upgraded biogas was stored in high- 2004. Functional safety and IEC 61508; A basic
pressure cylinders at 20 bar pressure. The guide. Available online:
http://www.ida.liu.se/~simna73/teaching/SCRTS/IEC
cylinders were of proper safety design and
61508_Guide.pdf
approved for the application and the pressure
condition. 3. Walsh, J. L., Ross, C. C., Smith, M. S., Harper, S. R.,
Wilkins, W. A. 1988. Handbook on biogas
Inspection for gas leaks utilization. Published by U. S. Department of
Before starting up of the pilot plant, the piping Energy.
and anaerobic reactor were tested for possible
gas leaks. To conduct this test, thesystem was 4. CH4Biogas. 2017. Hazardous zones in CH4 biogas
over pressurized and then kept one hour at test systems. Technical article. Available online:
http://www.chfourbiogas.com/uploads/1/4/1/9/141994
pressure condition.
62/hazardous_zones_in_our_biogas_systems.pdf
Waste handling and sludge disposal 5. NIOSH (The national institute for occupational
Canteen food waste is transported into thepilot safety and health). 1990. Preventing deaths of
plant using closed containers in order to avoid farm workers in manure pits. NIOSH publication
odour and contamination due to spilling. number 90-103. Available online:
Sludge generated from the digester is https://www.cdc.gov/niosh/docs/90-103/
transferred to a nearby isolation pit and then
subsequently utilized as a fertilizer for 6. International Energy Agency (IEA). 2000. Biogas
flares: State of the art and market review. Topic
gardening (non-food crops).
report of the IEA Bioenergy Agreement Task 24 –
Biological conversion of municipal solid waste.
6. Conclusions 7. International Electrotechnical Commission (IEC).

2015. Explosive atmospheres – Part 10-1:
This technical paper presents a detailed outlook Classification of areas – Explosive gas
on various safety dimensions that have to be atmospheres. International Standard IEC 60079-
10-1. Edition 2.
properly assessed in the design phase as well as
in the operation phase of anaerobic digestion 8. DEHN International. 2015. Lightning and surge
processes. The presented process safety, protection of biogas plants. White paper. Article
occupational safety & working environment, online:https://www.dehn-
and environmental care considerations are international.com/sites/default/files/uploads/dehn/pdf/
critical to the sustainable and safe operation of white-papers/ab-juli15/wp003-e-biogas.pdf
biogas plants. It is emphasized that safety and
sustainability are the most vital aspects that 9. Air Products and Chemicals. 2014. Dangers of
oxygen-deficient atmospheres. Safetygram 17.
could determine the social acceptance and the
Article available online:
economic viability of any industrial technology.
http://www.airproducts.com/~/media/files/pdf/compan
y/safetygram-17.pdf
A case study is presented where a pilot scale
anaerobic digester was designed, constructed 10. Heezen, P. A. M., Gunnarsdottir, S., Gooijer, L.,
and operated as a technology demonstration Mahesh, S. 2013. Hazardous classification of
project. A wide attention is given to the biogas and risks of large scale biogas production.
relevant safety considerations during all life Chemical Engineering Transactions, Vol. 31.
cycle phases of this pilot scale digester.
11. Aldrich, B. S., Brown, N. J., Hallman, E. M. 2005.
Innovations in manure management: Manure gas
can be deadly. Cornell Manure Management
References Program.
Article online: http://extension.psu.edu/natural-
1. Botheju, D., and Abeysinghe, K. 2014. Public risk resources/energy/waste-to-
perception towards chemical process industry: energy/resources/biogas/links/biogas-and-anaerobic-
Comprehension and response planning. Safety & digestion-safety/manure-gas-can-be-deadly.pdf
Reliability: Methodology and Applications,
Nowakowski et al. (Eds.) (Book), pp. 453-60, 12. Renewable Fuels Association (RFA). 2015.
Taylor & Francis Group, London, ISBN 978-1-138- Carbon dioxide safety program. Report;
02681-0. (Also presented at European Safety & Available online:
Reliability Conference, ESREL 2014, Wroclav, http://ethanolrfa.3cdn.net/2d1bea9178e0e876bd_idm6i
Poland. Organized by European Safety & 8enq.pdf
Reliability Association ( ESRA).
ENGINEER 10
562
13. Aria, A. 2006. Ammonia emissions and safety. Report – Research Programme of the Research
Agri-Facts, June 2006. Available online: Fund for Coal and Steel. Tecnico Lisboa.
http://www1.agric.gov.ab.ca/$department/deptdocs.nsf Available online:
/all/agdex8271/$file/086-6.pdf?OpenElement http://projects.ctn.tecnico.ulisboa.pt/Dioxinas/1
8.pdf
14. P. E. Wright, P. E., Inglis, S. F., Stehman, S. M.,
Bonhotal, J. 2003. Reduction of selected
pathogens in anerobic digestion. In Proceedings
of the 9th international symposium of ani mal,
agricultural and food processing wastes. Article
available online:
http://www.manuremanagement.cornell.edu/Pages/Ge
neral_Docs/Papers/Reduction_Selected_Pathogens_A
D.pdf
15. Michalska, J. and Szewieczek, D. 2007. The 5S

methodology as a tool for improving the
organization. J. of Achievements in Materials and
Manufacturing Engineering. Vol. 24, No. 2., pp.
211- 214. Available online:
http://www.academia.edu/10065153/5S_methodology_
as_a_tool_for_improving_the_organization
16. American Petroleum Institute (API). 2014.

Pressure relieving and depressurizing systems.
API standard 521. Ed. 6, pp. 106.
17. Botheju, D., Abeysingha, K. and Rathnasiri, P. G.

2014. Safety and emergency preparedness in oil
and gas industry: Experience based on North Sea
offshore platforms. Annual proceedings of the
Institution of Engineers of Sri Lanka (IESL) 2014,
pp. 163-170.
18. Botheju, D., Svalheim, Ø., Rydtun, B., Johansen,

J., Haraldsen T. K. and Bake R. 2007. Fertilizer
production by digestate nitrification. In
proceedings of the 12th European Biosolids and
Organic Resources Conference, Manchester, UK,
November 2007.
19. Stasinos, S., Nasppoulou, C., Tsikrika, C.,

Zabetakis, I. 2014. The bioaccumulation and
physiological effects of heavy metals in carrots,
onions, and Potatoes and dietary implications for
Cr and Ni: A review.
20. Biogas Renewable Energy. 2009. Biogas

composition. Information page available online.
http://www.biogas-renewable-
energy.info/biogas_composition.html
21. NIOSH (The national institute for occupational

safety and health). 1994. Table of IDLH values –
Carbon monoxide. Available online.
https://www.cdc.gov/niosh/idlh/630080.html
22. Coward, H. F., Jones, G. W. 1952. Limits of

flammability of gases and vapors. Bulletin 503,
US Bureau of Mines.
23. Niska, J., Kaufmann, B., Mortberg, M., Almeida,

S. M., Zanusso, E., M., Gitzinger, H-P.,
Fernandez, J. M., Mori, M. F. C. 2007. Minimizing
NOx emissions from reheating furnaces. Final
11
563 ENGINEER
564
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[565 570],2017
2017
The Institution
Lanka
Simulating the Effect of Modifying Reversed A2O

Process on Combined Removal of Nitrogen,
Phosphorous and Organic Matter in Municipal
Wastewater
W.B. Roshen P.S. Fernando and P.G. Rathnasiri
Abstract: In evolution of biological wastewater treatment processes, extended aeration, A2O

(combination of anaerobic, anoxic and oxic (aerobic) processes) and Reversed-A2O (RA2O)
(combination of anoxic, anaerobic and oxic processes) processes are significant milestones. While
other methods (than RA2O) have been subject for improvements, improvements for RA2O process
have been minimal. Thus, the effect of modifying Reversed-A2O (RA2O) process via mixed liquor
recirculation, activated sludge recycling and optimizing the percentage recycled on efficiency of
combined removal of nitrogen, phosphorous and organic matter in municipal wastewater has been
studied via mathematical modelling and simulation. The results suggest that mixed liquor
recirculation from aerobic to anoxic reactor reduces nutrient removal efficiency and that RA2O
process can be improved to achieve significant nutrient and organic matter removal efficiencies by
recycling activated sludge only to anaerobic reactor without recycling activated sludge to anoxic
reactor.
Keywords: Reversed-A2O process, Nutrient and organic matter removal, Municipal wastewater,
Activated Sludge Modelling
1. Introduction Kang et al. in 2011 [7], proved that the RA2O

process was also more economically
When wastewater treatment practices competitive with respect to installation,
improved its focus on nutrient removal, construction and operation than the A2O
biological nutrient removal (BNR) methods process.
were more attractive compared to chemical
nutrient removal methods due to savings in However, Zhang and Gao (2000) [8] did not
investment in machinery, less energy observe significant improvements in removal of
requirement (Lettinga, 1996 [1]) and lower organic matter in the RA2O process compared
environmental impacts (Arbuckle and Griggs, to the A2O process. Furthermore, in 2014, Xu et
1982 [2]). Following the various BNR models, al. [9] argued that the RA2O process to be
such as University of Cape Town model, significant only in phosphorous removal, in a
Anaerobic/Anoxic/Aerobic (A2O) model comparative study among the two processes,
(Spector, 1993 [3]), in 1997, Zhang and Gao [4] under same operating conditions, including
proposed the ability to further improve A2O mixed liquor recirculation from aerobic zone to
process, by changing the sequence of anaerobic anoxic zone. This had also been supported by
and anoxic reactors, and this process was Zhou et al. in 2011 [10] via a simulation study.
named Reversed-A2O (RA2O). Nevertheless, via the projection pursuit
method, Fang et al. in 2016 [11], showed that
The main arguments supporting this still the A2O process to be predominant,
improvement were the preference given to compared to the RA2O process, in both organic
denitrification by the carbon source, from matter and nutrient removal.
influent and returned activated sludge, and the
greater biochemical efficiency in aerobic Eng. W. B. Roshen P. S. Fernando, B.Sc. Eng. (Hons)
reactor, due to deeper anaerobic conditions in (Moratuwa).
the anaerobic reactor with the exhaustion of Eng. (Dr.) P. G. Rathnasiri, B.Sc.Eng.(Hons)
nitrates in the anoxic reactor. This innovation (Moratuwa), M.Sc. (UMIST-UK), PhD (NTNU-Norway),
was supported to be higher in efficiency with Senior Lecturer, Department of Chemical and Process
respect to nutrient removal by Xuejun and Bo Engineering, University of Moratuwa.
in 2006 [5] and Bi et al. in 2009 [6]. Furthermore,
1
565 ENGINEER
Accordingly, it is observed that there exists, were within the operational parameters of the
still, wide spread opinions on the efficiency and actual plant, too were input as Formula
effectiveness of the RA2O process, even though Variables, and four more Formula Variables
it is currently, widely applied all over the were defined for COD (Chemical Oxygen
world. Moreover, application of mathematical Demand), TN (Total Nitrogen), TP (Total
modelling for the RA2O is at a fairly young Phosphorous) and TSS (Total Suspended Solids)
stage. Thus, it was decided to investigate the removal efficiencies from effluent. The twenty-
possibility to improve the RA2O process with one processes, specified in ASM 2d, were
respect to combined removal of nutrients and developed as Dynamic Processes, and their rate
organic matter by changing intermediate equations and stoichiometric coefficients were,
recycling streams and percentage quantities thereafter, fed to the software. Four mixed
recycled, and thereby, suggest modifications to reactor compartments were developed for the
the RA2O process, via a modelling approach. anaerobic, anoxic and aerobic reactors and
secondary clarifier, using the actual volumes of
2. Materials and Methods the reactors and average inflow and outflow
rates. After that, three links among the four
With the intention of extending the findings compartments and three bifurcations from the
from the study to real world practices, a clarifier for effluent, waste activated sludge and
municipal wastewater treatment plant, already recycled activated sludge were developed.
employing RA2O, was selected for the study. It
was one of the two RA2O municipal After that, two samples taken from the inflow
wastewater treatment plants operated by to the actual plant were characterized for the
National Water Supply & Drainage Board nineteen model components, according to the
(NWSDB) at Ratmalana, handling a daily flow experimental procedures stated in Standard
of 17,009 m3, on average. Even though, the Methods for Water and Wastewater
plant currently meets the discharge standards, Characterization by APHA. The obtained
stipulated by Central Environment Authority, values were, then, input to the model as the
Sri Lanka, in order to identify further room for initial values of the model components.
improvement, the performance of the plant for
a period of 18 months, from January, 2015 to Following that, the links among the
June, 2016 was analyzed. compartments were modified to reflect the
standard RA2O process (Fig. 1) and three
Thereafter, in order to build the simulation modified configurations of the RA2O process
platform for modelling, using Aquasim 2.1 (Fig. 2-4). Then, the results from simulation for
(Reichert, 1998), the guidelines given in them, for four Formula Variables, removal
Activated Sludge Modelling (ASM) 2d efficiencies of COD, TN, TP and TSS from the
framework (Henze et al., 2000), were followed. effluent were evaluated. After identifying the
Nine soluble components and ten particulate optimum placement of recycling streams from
components, given in ASM 2d, were developed the simulation, the recycling percentages were
as Dynamic Volume Variables, and their initial changed, and the resulting removal efficiencies
values were given as Formula Variables. The for the same four Formula Variables were
empirical constants, given in ASM 2d, as they evaluated.
Figure 1 - Reversed A2O Process
ENGINEER 2
566
Figure 2 - Modified RA2O Configuration 01
3. Results
3.1 Influent Characterization
Table 1 - Results from Influent Characterization
Characterizati Characterizati Characterizati Characterizati

Component Component
on I (mg/l) on II (mg/l) on I (mg/l) on II (mg/l)
SO2 0.86 0.11 XI 157.036 102
SF 105.636 195.84 XS 166.67 173.8293
SA 95.87 122.4 XH 0.65 0.01
SNH4 60 65.88 XPAO 0.03 0.01
SNO3 1.95 2.95 XPP 0.01 2.626
SPO4 1.975 6.69 XPHA 0.02 27.2451
SI 40.914 307.2629 XAUT 0.32 97.68
SALK 5 5 XTSS 425 342
SN2 0.1 0.01 XMeOH 89.12 13.865
XMeP 92.7642 10.504
3
567 ENGINEER
3.2 Removal Efficiencies
Table 2 - Results for the Removal Efficiencies
Configuration COD TN TP TSS

Standard RA2O Process 29.47% 39.76% 36.55% 8.16%
Modified RA2O Process 01 29.34% 39.59% 36.38% 8.13%
As per Table 2, it is observed that the highest 8

removal efficiencies for all the parameters
Removal Percentage
considered have been obtained in Modified 8
RA2O Process 02. As this involved recycling of 8
activated sludge both to anoxic and anaerobic
8
reactors, thereafter, the percentage recycled to
each reactor from the total recycled flow was 8
optimized. There, the percentage recycled from 0% 50% 100%
the total recycled flow, from the secondary Recycled Percentage of Activated Sludge
clarifier, to anaerobic reactor was changed from
0% to 100%, and the variations in removal Figure 8 - TSS Removal Efficiency
efficiencies were observed (Fig. 5-8).
Thus, it was observed that the removal
31 efficiencies of all the parameters improved,
with the highest improvement being in
Removal Percentage
phosphorous removal, when the activated

30 sludge was completely recycled to the
anaerobic reactor, without recycling anything
to the anoxic reactor.
29
0% 20% 40% 60% 80% 100% 4. Discussion
Recycled Percentage of Activated Sludge
The main reason for changing the sequence of
Figure 5 - COD Removal Efficiency anaerobic and anoxic reactors in the previously
used A2O process was to give preferential
42 treatment to denitrification by the carbon
Removal Efficiency
42 source for nitrogen removal, thereby, improve

41 achievement of deeper anaerobic conditions, in
41 the anaerobic reactor, due to exhaust of nitrates
40 (Zhang and Gao, 1997 [4]) and to enhance the
efficiency of nitrification in the aerobic zone,
40
due to higher non-aerobic retention time and
0% 50% 100%
greater energy generation for PAOs in the
previous reactors (Bi et al., 2009 [6]).
Figure 6 - TN Removal Efficiency
One of the major modifications introduced in
39 previous studies to improve the RA2O process
Removl Percentage
39 has been in recirculation of mixed liquor from

38 aerobic zone to anoxic zone. However, the
38 results, via modified configuration 01, portray
37
that recirculation of mixed liquor from aeration
reactor decreases both the organic matter and
37
nutrient removal efficiencies, compared to the
0% 50% 100%
standard RA2O process. This reduction is
observed in modified configuration 03,
compared to modified configuration 02 as well,
Figure 7 - TP Removal Efficiency
where the difference among the two
ENGINEER 4
568
configurations are in recirculation of mixed heterotrophs, resulting from optimized
liquor. Thus, the results are in disagreement favourable conditions, the organic matter
with the findings of Zhou et al. (2011) [10] and removal efficiency too has got improved in
Xu et al. (2014) [9], who had observed increased modified configuration 02, compared to the
efficiency in TP removal, at an expense of TN standard RA2O process.
removal, with mixed liquor recirculation. This
deviation from previous findings is possibly Thus, it can be claimed that in order to further
due to the higher concentration of dissolved improve the RA2O process, from its standard,
oxygen in the recirculation stream from aerobic there should be no recirculation of mixed liquor
reactor, which prevents achieving anoxic from aerobic reactor to anoxic reactor and
conditions in the anoxic reactor, which is in line activated sludge from secondary clarifier
with the findings of Fang et al. (2016) [11]. should only be recycled to the anaerobic
reactor.
Furthermore, although previous researchers
(Zhang and Gao, 2000 [8] and Bi et al., 2009 [6]) 5. Conclusions
have claimed that recycling activated sludge to
the anoxic zone from the secondary clarifier to RA2O process could achieve a higher efficiency
improve the nutrient removal efficiency, due to in removal of organic matter and nutrients,
remaining generated nitrates from aerobic zone highest being in phosphorous removal, from
to be exhausted in the anoxic zone prior to municipal wastewater, when activated sludge
transfer to the anaerobic zone, the results, from is recycled only to the anaerobic reactor.
modified configuration 02 and optimization of
recycling percentages, suggest that organic Recirculation of mixed liquor from aerobic
matter and nutrient removal efficiencies to be reactor to anoxic reactor inhibits achievement of
greatest when activated sludge is solely anoxic conditions in the anoxic reactor, and
recycled to the anaerobic reactor. This is partly hence, deteriorates the efficiency of RA2O
due to the increased sludge concentration in the process.
anaerobic reactor, which helps it achieving
deeper anaerobic conditions faster. Moreover, Acknowledgement
the sludge is expected to be richer in oxygen
compared to nitrate, due to a highly efficient The support received in this study from Eng.
aeration process. As PAOs can generate energy (Mr.) Dhanesh Gunatillake, Eng. (Mr.) Lasantha
at a higher efficiency using oxygen as electron Rupasinghe, Eng. (Mr.) A. Ranatunge, Chemist
acceptors compared to nitrate, this would lead (Ms.) Lakmini Dhanawardhane, staff at the
to a higher biochemical efficiency in the wastewater treatment plant of NWSDB at
phosphorous uptake in aeration process. Thus, Ratmalana and Eng. (Dr.) Sanjeewa
although remaining nitrates in activated sludge Wickramaratne from Lanka Hydraulics
would slightly inhibit the anaerobic conditions Institute is much appreciated.
(Bi et al., 2009 [6]), due to the higher aggregate
favourable impact, the overall efficiency of the
References
anaerobic reactor would improve.
Consequently, this would result in increased 1. Lettinga, G., “Sustainable Integrated Biological
phosphorous removal. Wastewater Treatment”, Water Science and
Technology, 1996, Vol. 33, No. 3, pp. 85-98.
Furthermore, prevention of contamination of
anoxic reactor with oxygen in the activated 2. Arbuckle, W. B. & Griggs, A. A., “Determination
sludge has improved nitrogen removal, via of Biomass MLVSS in PACT Sludges”, Journal of
denitrification, as well. However, still, the Water Pollution Control Federation, 1982, pp. 1553-
1557.
highest improvement in nutrient removal is
observed in phosphorous removal, indicating
3. Spector, M., “Biological Process for Enhanced
the higher potential for RA2O to improve removal of Ammonia, Nitrite, Nitrate, and
phosphorous removal efficiency. Phosphate from Wastewater”, US5182021A, 1993.
Moreover, although attempts to improve RA2O 4. Zhang, B. & Gao, T., “Enhancement of Nitrogen
process largely were focused in spheres of and Phosphorus Removal by Reversal of
nutrient removal, it is observed that organic Anaerobic and Anoxic Zones Sequence”, China
matter removal too could be improved. Due to Water & Wastewater, Vol. 13, No. 3, 1997, pp. 7-9.
improved activity of autotrophs and
5
569 ENGINEER
5. Xuejun, B. & Bo, Z., “The Principle and Full-Scale
Application of Reversed A2O Process for
Removing Nitrogen and Phosphorous”,
Environmental Engineering, Vol. 3, 2006, pp. 008.
6. Bi, X., Liu, C.Q., Cheng, L., Zhang, F. & Zhang, B.,
“The Principal and Full-Scale Application of
Reversed A²/O Process for Removing Nitrogen
and Phosphorus from Municipal Wastewater”,
3rd International Conference on Bioinformatics and
Biomedical Engineering, June, 2009, pp. 1-4.
7. Kang, X. S., Liu, C. Q., Zhang, B., Bi, X. J., Zhang,

F. & Cheng, L. H., “Application of Reversed
A2/O Process on Removing Nitrogen and
Phosphorus from Municipal Wastewater in
China”, Water Science and Technology, 2011, Vol.
63, No. 10, pp. 2138-2142.
8. Zhang, B. & Gao, T., “An

anoxic/anaerobic/aerobic Process for the
Removal of Nitrogen and Phosphorus from
Wastewater”, Journal of Environmental Science &
Health Part A, 2000, Vol. 35, No. 10, pp. 1797-1801.
9. Xu, S., Bernards, M. and Hu, Z., “Evaluation of

Anaerobic/anoxic/oxic (A2/O) and Reverse
A2/O Processes in Biological Nutrient Removal”,
Water Environment Research, 2014, Vol. 86, No. 11,
pp. 2186-2193.
10. Zhou, Z., Wu, Z., Wang, Z., Tang, S., Gu, G.,
Wang, L., Wang, Y. & Xin, Z., “Simulation and
Performance Evaluation of the
anoxic/anaerobic/aerobic Process for Biological
Nutrient Removal”, Korean Journal of Chemical
Engineering, 2011, Vol. 28, No. 5, pp. 1233-1240.
11. Fang, F., Qiao, L. L., Cao, J. S., Li, Y., Xie, W. M.,
Sheng, G. P. & Yu, H. Q., “Quantitative
Evaluation of A 2 O and reversed A 2 O Processes
for Biological Municipal Wastewater Treatment
using a Projection Pursuit Method”, Separation
and Purification Technology, 2016, Vol. 166, pp.164-
170.
12. Reichert P., AQUASIM 2.0 - User Manual

Computer Program for the Identification and
Simulation of Aquatic Systems, Switzerland, 1998.
13. Henze M., Gujer W., Mino T., Matsuo T., Wentzel
M. C., Marais G. v. R., Van Loosdrecht M.C.M.,
Activated Sludge Model No. 2d, The
Netherlands, 2000.
ENGINEER 6
570
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[571 578],2017
2017
The Institution
Lanka
Extraction of Cellulose from Sri Lankan Agricultural

Waste for Engineering Applications
M.P.A. Nanayakkara, W.G.A. Pabasara, A.M.P.B. Samarasekara,
D.A.S. Amarasinghe and L. Karunanayake
Abstract: Polymers play a critical role in engineering applications today. Cellulose is the most
abundant renewable natural biopolymer on earth. It is present in a wide variety of living species
including plants and some marine animals. Rice is the main food of the inhabitants of Sri Lanka. The
total land devoted for paddy is estimated to be about 708,000 Hectares at present in Sri Lanka. Rice
straw is a rice by-product produced after harvesting paddy. Rice straw is a major agricultural waste
product in Sri Lanka. Farmers are not economically using rice straw today. It is noted that rice straw
contains a considerable amount of cellulose with hemicelluloses and lignin. This research is based on
investigating the possibility of extraction of cellulose from most frequently used traditional rice
varieties (Suwandel and Raththal) and technically modified rice varieties (BG300 and BG352) in Sri
Lanka. The dried and cleaned rice straw was milled using a grinder to produce a fine powder of rice
straw. Sieve analysis method was used to analyze the rice straw powder, and equal or less than 150
µm particle size was selected for the cellulose extraction. Pure cellulose was extracted from BG 352 rice
straw variety after following de-waxing, delignification, hemicellulose and silica removal processes.
The Same extraction procedure was followed for BG 300, Suwandel and Raththal rice varieties. The
complete removal of non-cellulosic materials from rice straw was confirmed by FTIR spectroscopy
after each chemical purification step. According to the experimental results, highest cellulose yield
was obtained from rice straw of Suwandel (35.2%). Other types of rice varieties showed cellulose yield
of BG300 (27.8%), BG352 (29.3%) and Raththal (26.8%). This research finding proves that this
developed environmental friendly chemical purification process can be successfully utilized to isolate
pure cellulose from rice straws. This extracted cellulose material can be used for different engineering
applications.
Keywords: Agro waste, Cellulose, Rice straw
1. Introduction necessary to explore the capabilities and

options regarding maximization of value
Paddy is the leading cereal crop grown in Sri generation in rice straw.
Lanka. The total land devoted for paddy is Rice straw, being a lignocellulosic biomass,
estimated to be about 708,000 Hectares [1]. But consists of cell wall polymers: lignin,
regrettably, it has resulted in a large amount of hemicellulose and cellulose [5]. Among them,
agro waste. Among them, the highest cellulose is known as the most abundant
percentage is occupied by rice straw. Each 1 kg natural biopolymer on earth which satisfies the
of milled rice produced results in roughly 0.7– criteria for being an excellent engineering
1.4 kg of rice straw depending on varieties, material including as reinforcing fibres in bio
cutting-height of the stubbles, and moisture composites [6].
content during harvest [2]. It is estimated that Ms. M.P. A. Nanayakkara, B.Sc. Eng. (Moratuwa), Dept.
800 GT of rice straw is produced in Sri Lanka of Materials Science & Engineering, University of
Moratuwa.
per year [3]. Generally, this rice straw produced
is surplus and is either left in the field as Ms. W.G.A Pabasara, B.Sc. Eng. (Moratuwa), Dept. of
uncollected or to a large extent open-field burnt
which is neither an environmental friendly nor Eng. A.M.P.B. Samarasekara, B.Sc. Eng. (Moratuwa),
a productive method. Several researchers have Dept. of Materials Science & Engineering, University of
studied the adverse effects of rice straw Moratuwa.
burning on the environment and reported that Dr. D.A.S. Amarasinghe, BSc. (Kelaniya), MSc.
(Moratuwa), M.Phil. (CUNY, USA), Ph.D. (CUNY, USA),
toxic emissions from rice straw result in air Senior Lecturer, Dept. of Materials Science & Engineering,
pollution, water pollution, soil degradation and University of Moratuwa.
overheating which ultimately generate risk Prof. L. Karunanayake, B.Sc. (USJP), Ph.D. (North
factor towards human health [4]. Therefore, it is London), Professor, Department of Chemistry University of
Sri Jayewardenepura.
1
571 ENGINEER
Rice straw can be considered as the most cultivation area, Padaviya, Anuradhapura, Sri
abundant source of cellulose from all Lanka. Rice straws of Suwandel were collected
agricultural residues with 32 - 47% cellulose [7]. at maturity after harvesting in 2016 Yala season
By considering above facts, developing an from a traditional paddy cultivation area,
environmentally friendly method to isolate Kurunegala, Sri Lanka. Toluene (C6H5CH3,
cellulose from Sri Lankan rice straws would be ≥99.7%, ACS, Sigma-Aldrich), ethanol
most effective way of value addition to rice (H₃CCH₂OH, Min. 99.85 %, ACS, VWR
straw. Since, there have been hardly any Chemicals), acetic acid (CH3COOH, ≥99.8%,
researches on evaluating and analyzing the ACS, Sigma-Aldrich), hydrogen peroxide
yield of cellulose from rice straws of various (H2O2, 50% (w/v), AR, Glorchem), sulfuric acid
rice varieties, this research is significant not (H2SO4, 95.0 - 97.0%, ACS, Sigma-Aldrich),
only because as this focuses on a unique area of potassium hydroxide (KOH, ≥ 85.0 %, GR,
research but also the rice straws were used in MERCK Chemicals) were used for separation of
this research consist of both frequently used cellulose from rice straw. Hydrochloric acid
traditional and technically improved rice (HCl, ≥37.0%, ACS, Sigma-Aldrich) was used
varieties in Sri Lanka. for determination of silica content and used as
Under the most abundant traditional rice received without further purification. All water
varieties, Suwandel and Raththal were chosen used was purified by Barnstead™
and under the most abundant technically Smart2Pure™ Water Purification System
modified varieties BG300 and BG352 were (Thermo Fisher Scientific, Waltham, MA).
chosen. These technically modified rice
varieties are created by a process known as 2.2 Determination of Inorganic and Silica
hybridization. They have been modified to Content in Rice Straw Varieties
acquire some characteristics that the traditional 1 g of rice straw powder sample of each rice
rice varieties lack such as increased yield variety was transferred to a crucible and
potential, increased resistant to pest and disease carbonized gently in a chamber furnace
etc. The traditional rice varieties are cultivated (Nabertherm, Germany) at 550°C for 6 h. The
by using completely traditional methods while weight of the ash obtained was measured. The
the technically modified rice varieties are ash was treated with concentrated HCl. The
cultivated using conventional methods [8]. acid-insoluble residue was filtered, washed
Hence, these rice varieties are cultivated in with hot water until no chlorides were
completely different agronomic, soil and detectable, ignited, and finally weighed as
climatic conditions besides their genetic silicon dioxide. All the tests were carried out in
variations. triplicate.
The main objectives of this research are to
establish an environmental friendly method to 2.3 Isolation of Cellulose from Rice Straw
isolate cellulose from rice straws of frequently Varieties
used traditional (Suwandel, Raththal) and Pure cellulose was isolated from rice straw by
technically modified (BG300, BG352) rice step wise removal of non-cellulosic materials.
varieties and analyzing the cellulose yield from The quantity utilized was 200g for each variety.
above varieties in order to identify the most Initially, foreign material such as dried grass,
suitable rice straw variety for commercial rice husk and other plant residues were
cellulose production. Up to date, high-quality removed from rice straw. Then rice straw was
cellulose was separately isolated from above thoroughly washed with warm water followed
rice straws varieties by streamlining and by washing with normal tap water twice to
optimizing extraction and dissolution remove dirt and aqueous soluble substances.
conditions to remove wax, lignin, hemicellulose Then cleaned rice straw was subjected to air
and silica in a step-wise manner. drying at room temperature for 48 h followed
by oven drying (Memmert GmbH, Germany) at
2. Experimental 60 °C for 5h. The dried and cleaned rice straw
was milled using SINGER Easy Blend grinder
2.1 Material (Model No - KAEASYBLEN, SINGER, Sri
Rice straws of BG300, BG352 were collected at Lanka). Sieve analysis of rice straw powder was
maturity after harvesting in 2016 Yala season done with the use of 250µm, 150 µm, 106 µm, 75
from Rice Research Center, Batalegoda, µm and, 53 µm aperture sized screens. Then the
Kurunegala, Sri Lanka. Rice straws of Raththal rice straw powder retained in each screen
were collected at maturity after harvesting in (except for 250 µm screen), and the pan was
2016 Yala season from Hela Suwaya paddy mixed together in order to obtain a
ENGINEER 2
572
(µm2) (µm2)
(µm2) (µm2)
Figure 1 - Log Normal Particle Area Distributions of Rice Straw Varieties
homogeneous mixture. Rice straw powder (10 tests were carried out in triplicate separately for
g) was first extracted in Soxhlet apparatus with four rice straw varieties.
2:1, v/v toluene/ethanol (450 mL) mixture at
400°C for 15 h to remove wax, pigments and 2.4 Characterization
oils, followed by oven-drying at 55°C for 24 h.
Then the dry weight of the dewaxed rice straw 2.4.1 Optical Microscopy
was measured. The dewaxed powder was then Unbiasedly spread distributions of rice straw
immersed in 3:10, v/v H2O2/CH3COOH powder of above four rice varieties were
solution with solid: liquid ratio 1: 60, w/v at imaged under an optical microscope (MT, Meiji
70°C for 3h to dissolve lignin. H2SO4 was used Techno Co. Ltd, Japan) at 50 magnification. The
as a catalyst. Extraction mixture was constantly area distributions of rice straw powder from
stirred at 125 rpm. The insulated flask above four varieties were analysed using
containing extraction mixture was heated with ImageJ software (ImageJ 1.46, National Institute
water from the thermostatically-controlled of Health (NIH), USA) by analysis of
water bath. Extraction was conducted in a flask micrographs. The statistical analysis was
with volumes two to three times those of performed using Origin Pro 2016 software.
extraction mixtures to prevent oxygen
evolution and substantial frothing during initial 2.4.2 Fourier Transform InfraRed (FTIR)
stages. The insoluble residue was collected by Spectroscopy
vacuum filtration (Whatman 1, Maidstone, FTIR Spectra were obtained using a Bruker
Kent, England), washed with deionized water ALPHA spectrometer (Bruker Corporation,
until the PH of the filtrate was neutral, and then Billerica, MA) at ambient conditions. Samples
dried at 60°C. Then the dry weight of the de- were prepared by grinding with KBr (1:100,
lignified rice straw was measured. w/w) and pressing into transparent pellets. The
Hemicellulose and silica in the delignified rice spectra were obtained in the transmittance
straw were leached with 110 mL 5% KOH at mode from an accumulation of 128 scans at a 4
room temperature for 24 h and then at 90°C for cm−1 resolution over 4000–600 cm−1 range.
2 h. The insoluble residue was collected by
vacuum filtration (Whatman 42, Maidstone, 2.4.3 Scanning Electron Microscopy (SEM)
Kent, England), washed with deionized water The microstructures and surface morphologies
until the PH of the filtrate was neutral, and then of cellulose from above four rice straw varieties
oven dried to remove water. Then the dry were examined by a scanning electron
weight of the cellulose was measured. All the microscope (EVO 18, Carl Zeiss AG, Germany).
The samples were mounted on aluminum stubs
3
573 ENGINEER
Figure 4 - Cellulose Extracted from Rice Straw
Figure 2 - Inorganic Material Percentage of Varieties
Rice Straw Varieties
3.2 Determination of Inorganic and Silica
with conductive carbon tape and sputtered Percentage of Rice Straw Varieties
with gold under vacuum at 10 mA for 90 secs Inorganic (Ash) content of rice straws of BG300
(Sc7620 sputter coating system). The samples (15.4%) and BG352 (16.8%) was higher than that
were observed and imaged at 14.5 mm working of rice straws of Suwandel (13.5%), Raththal
distance and 20 kV accelerating voltage. (13.1%) (Figure2). These values were similar to
the inorganic content reported for Asian and
3. Results and Discussion Californian rice straw varieties [9]. Since
around 75% of the inorganic material present in
3.1 Particle Size Distribution of Rice rice straw is silica [10], variation in inorganic
Straw Varieties content of rice straws should be directly
Prior to cellulose isolation process, particle size influenced by the variation in the silica content.
distributions of rice straw powder of above- Rice straw is unique compared to other cereal
mentioned rice varieties were studied using straws due to its high silica content. It is
microscopy analysis. This method involved absorbed as monosilicic acid from soil and
spreading rice straw powder on a glass slide deposited as amorphous silica in the plant.
unbiasedly to obtain a very thin layer of rice Silica content is affected by its concentration in
straw powder followed by microscopy soil and on the uptake capacity of the plants
imaging. These microscopy images were [11]. Highest Silica content was showed by rice
analysed using ImageJ software to obtain straws of BG300 (11.5%), BG352 (12.3%) while
particle area distributions of each rice straw lowest was showed by rice straws of Suwandel
varieties assuming that particle size is (10.8%), Raththal (9.9%) (Figure 3). Hence, this
proportional to the particle area. Particle area variation in silica content may be due to the
distributions were fitted to a log normal curve genetic variation in rice varieties as well as
using Origin Pro 2016 software. The variation in monosilicic acid concentration in
observations showed that most particles of the soil in straw retrieved locations.
these varieties have particle area between 100-
200 µm2 range (Figure 1). 3.3 Isolation of Cellulose from Rice Straw
Varieties
Cellulose was successfully isolated from
cleaned and dried rice straw powder of above
mentioned four rice varieties through three-
step de-waxing, delignification and
hemicellulose (and silica) removal process. The
faded light brown colour of rice straw reduced
with each step of delignification and
hemicellulose (and silica) removal to become
pure white (Figure4).
The surface of the rice straw is covered with a
protective epicuticular wax layer which is a
complex mixture of different hydrocarbons
[12]. Hence its presence prevents the
permeation of water and organic solvents into
Figure 3 - Silica Percentage of Rice Straw
rice straw, preventing it from further treatment
Varieties
ENGINEER 4
574
Rice Straw Variety
Raththal
Suwandel
BG352
BG300
0 5 10
Hemicellulose Percentage (%)
Figure 5 - Wax Percentage of Rice Straw Figure 7 - Hemicellulose Percentage of Rice

Varieties Straw Varieties
to isolate cellulose. Hence, removal of wax was Peracetic acid treatment was generally used for
achieved using conventional Soxhlet extraction wood de-lignification and also it was a time-
method. It was observed that rice straws of consuming treatment since it was done under
BG300 (15.9%), BG352 (16.8%) varieties had static conditions. Regardless of the above
more wax content than rice straws of Suwandel difficulties, when adopted this treatment to the
(9.5%), Raththal (9.6%) (Figure5). research, it did not return the expected
Removal of lignin from dewaxed rice straw was bleaching level. Due to that reason, it was
essential to proceed forward with the research. necessary to modify the conditions in the
Acidified NaClO2 treatment was the most peracetic acid treatment to get the expected
conventional de-lignification method found in results.
the literature [13]. However, the use of In order to do so, concentrations of acetic acid
chlorinated compounds has been restricted due and H2O2 and liquid: solid ratio used were
to its adverse impact on the environment. modified. Also, time, temperature and agitation
Therefore, this research focuses on using an conditions were modified accordingly. In order
environmental friendly de-lignification method to accommodate these changes to the original
which satisfies Totally Chlorine Free (TCF) method, insulated reaction vessel (by
conditions. Therefore, next option was to concerning the jacketed reaction vessel design)
proceed de-lignification with the hydrogen was created. Due to these process optimizations
peroxide treatment in an alkaline medium [14]. complete de-lignification and bleaching with
Even so, there were two constraints associated minimum deterioration of the cellulose
with this method. First one was that this structure was achieved.
treatment does not provide sufficient level of Highest lignin content was observed from rice
de-lignification and bleaching and the second straws of Suwandel (43.4%), Raththal (46.4%)
one was the unavailability of the jacketed while rice straws of BG300 (39.2%), BG352
reaction vessel which was required to host the (38.2%) showed the lowest (Figure6). These
reaction of this method. Due to these short observations were significantly higher
comings, it was decided that a new method was compared the lignin contents (<7%) reported
required to get the expected level of results in for the international rice straw varieties [16].
this research. Therefore, peracetic acid Nevertheless, these observations suggest that
treatment was chosen which possess high besides genetic variations, environmental
bleaching and de-lignifying action with higher conditions may also greatly influence the lignin
selectivity compared to other TCF oxidizing content.
agents [15].
Rice Straw Variety
Raththal
Rice Straw Variety
Raththal Suwandel
Suwandel BG352
BG352 BG300
BG300 0 10 20 30 40
0 20 40 60 Cellulose Percentage (%)

Lignin Percentage (%) Figure 8 - Cellulose Yield from Rice Straw
Figure 6 - Lignin Percentage of Rice Straw Varieties
Varieties
5
575 ENGINEER
[7]. Rice straws of Raththal had the lowest yield
of cellulose (25.1%) while cellulose yields from
rice straws of BG300 (27.9%), BG352 (29.3%)
were intermediate to above results (Figure8).
The genetic and agronomic diversity between
the rice varieties might be the reason behind
this variation in chemical composition.
3.4 Fourier Transform Infrared

Spectroscopy
The resultant white products obtained after
chemical purification process of each rice straw
variety were confirmed to be cellulose by FTIR
spectra. The spectra also clearly verified the
sequential and complete removal of lignin (1516
cm−1, aromatic skeletal vibrations) during
peracetic acid treatment (Figure. 9) and
leaching of hemicellulose (1725 cm−1, carbonyl
stretching) and silica (796 cm−1, Si–O–Si
Figure 9 - FTIR Spectrum of Delignified Rice Straw - Suwandel
Removal of hemicellulose and silica was stretching) during the alkaline treatment
achieved with the conventional alkaline (Figure. 10).
treatments mentioned in the literature. Rice
straws of Raththal (9.1%) showed highest 3.4 Morphological Analysis of Cellulose
hemicellulose content while rice straws of Extracted from Rice Straw Varieties
Suwandel (1.2%) showed the lowest (Figure 7). SEM micrographs (Figure 11) showed that
However, hemicellulose percentages (19.8%- cellulose isolated from rice straw was mostly in
31.6%) observed for international rice straw the form of fibres with diameters ranging from
varieties [17] were considerably lower than 2-8 µm. Moreover, large aggregates of cellulose
these results. consisted of bundles of cellulose fibres with
The highest yield of cellulose was observed clear boundaries, which indicated the potential
from rice straws of Suwandel variety (35.3%). to break down these bundles to individual
Cellulose yields reported for international rice fibers further. occurred during the oven drying
straw varieties were similar to this observation step.
Figure 10 - FTIR Spectrum of Chemically Purified Cellulose - Suwandel
ENGINEER 6
576
5. Kadam, K. L., Forrest, L. H. and Jacobson,
W.A., ”Rice Straw as a Lignocellulosic
Resource: Collection, Processing,
Transportation, and Environmental
Aspects”. Biomass and Bioenergy, vol. 18(5), 2000,
pp.369-389.
6. Bledzki, A. K. and Gassan, J., “Composites

Reinforced with Cellulose Based
Fibres”, Progress in polymer science, vol. 24(2),
1999, pp.221-274.
7. Sarkar, N., Ghosh, S.K., Bannerjee, S. and Aikat,

K., “Bioethanol Production from Agricultural
Wastes: An Overview”, Renewable Energy,
2012, vol. 37(1), pp.19-27.
Figure 11 - SEM Micrograph of Cellulose from
8. http://dh-
BG300 Rice Straw Variety
web.org/botany/Paddycultivation.pdf, Visited
10th August 2016.
4. Conclusions
The established environmental friendly 9. Balasta, M. L. F., Perez, C. M., Juliano, B. O.,
chemical purification process can be Vlllareal, C. P., Lott, J. N. and Roxas, D. B.,
successfully utilized to isolate pure cellulose “Effects of Silica Level on Some Properties of
from rice straws of frequently used traditional Oryza Sativa Straw and Hull”, Canadian Journal
(Suwandel, Raththal) and technically modified of Botany, vol. 67(8), 1989, pp.2356-2363,
(BG300, BG352) rice varieties. This process Guzmán, A., Delvasto, A. and Sánchez, V.,
“Valorization of Rice Straw Waste: An
consisted of three steps: dewaxing, de-
Alternative Ceramic Raw Material”, Cerâmica,
lignification, and removal of hemicellulose and
vol. 61(357), pp.126-136, 2015.
silica treatments. Highest cellulose yield was
observed from rice straws of Suwandel variety 10. Guzmán, A., Delvasto, A. and Sánchez, V.,
(35.2%). Modified peracetic acid treatment can Valorization of Rice Straw Waste: An
be used successfully for de-lignification of rice Alternative Ceramic Raw
straw as it was able to provide the required Material. Cerâmica, 61(357), Guzmán, A.,
level of bleaching and de-lignification with less Delvasto, A. and Sánchez, V., 2015. Valorization
time consumption. This study shows that of Rice Straw Waste: An Alternative Ceramic
Raw Material. Cerâmica, vol. 61(357), 2015,
promisingly effective results for extraction of
pp.126-136.
cellulose from rice straw can be given by this
chemical purification method. 11. Raven, J. A., “The Transport and Function of
Silicon in Plants”, Biological Reviews, vol.58 (2),
References 1983. pp. 179-207.
1. http://www.cbsl.gov.lk/pics_n_docs/10_pub/ 12. Buschhaus, C. and Jetter, R., “Composition

_docs/statistics/other/econ_&_ss_2011.pdf, differences between Epicuticular and
Visited 26th August 2016. Intracuticular Wax Substructures: How do
Plants Seal their Epidermal Surfaces?, Journal of
2. http://irri.org/networks/rice-straw-project, experimental botany, vol. vol.62(3),2011, pp.841-
Visited 25th October 2016. 853.
3. http://nerdc.lk/en/sub_pgs/1_dept_6_renew. 13. Lu, P. and Hsieh, Y.L., “Preparation and

html, Visited 20th November 2016. Characterization of Eellulose Nanocrystals from
Rice Straw”, Carbohydrate Polymers, vol. 87(1),
4. Romasanta, R. R., Sander, B. O., Gaihre, Y. K., 2012, pp.564-573.
Alberto, M. C., Gummert, M., Quilty, J.,
Castalone, A. G., Balingbing, C., Sandro, J., 14. Lewis, S. M., Holzgraefe, D. P., Berger, L. L.,
Correa, T. and Wassmann, R., “How does Fahey, G. C., Gould, J.M. and Fanta, G.F.,
Burning of Rice Straw Affect CH4 and N2O “Alkaline Hydrogen Peroxide Treatments of
Emissions? A Comparative Experiment of Crop Residues to Increase Ruminal Dry Matter
different on-Field Straw Management Disappearance in Sacco”, Animal Feed Science
Practices”. Agriculture, Ecosystems & and Technology, vol.17 (3), 1987, pp.179-199.
Environment, 239, 2017, pp.143-153.
15. Jääskeläinen, A. S., Poppius-Levlin, K. and
Sundquist, J., Kraft Pulp Delignification with
7
577 ENGINEER
peroxy Compounds. Paperi ja puu, 82(4), 2000,
pp.257-263.
16. Abou-El-Enin, O. H., Fadel, J. G. and Mackill,

D. J.,” Differences in Chemical Composition
and Fibre Digestion of Rice Straw with, and
without, Anhydrous Ammonia from 53 Rice
Varieties”, Animal feed science and
technology, vol.79(1), 1999, pp.129-136.
17. Jin, S. and Chen, H., “Near-Infrared Analysis of

the Chemical Composition of Rice
Straw”, Industrial Crops and Products, vol.26 (2),
2007, pp.207-211.
ENGINEER 8
578
pp. [1 -- 10],
[579 586],2017
2017
© The
The Institution
Lanka
Environmental Impacts of Organic Fertilizer Produced

from Industrial Zone Central Effluent Treatment Plant
Sludge using Delta-D Technology
M.F.H.M. Aadhil and S.A.S. Perera
Abstract: Conversion of Central Effluent Treatment Plant Sludge (CETPS) from industrial
zones, consisting textile and garment processing plants, into a nutrient rich organic fertiliser (OF)
using Delta-D Technology has been explored by Perera et al. CETP primary and
tertiary sludge is not acceptable to be sent to landfills. As a result, CETPS is transported and
incinerated in the rotary kiln of a cement plant at high temperatures. This solution is expensive and
non-sustainable as it has many negative environmental impacts. A comparative study reveals that the
high cost and the negative environmental impacts of incineration can be greatly reduced by
converting CETPS to produce OF. An acute toxicology test on aquatic life revealed that this fertilizer is
non-threatening to marine and aquatic life.
A deeper analysis of the compounds in the OF by conducting FTIR and GC-MS tests have been carried
out and discussed in detail in the paper affirming scientifically that the production of OF is
ecofriendly and is a sustainable solution to CETPS.
Keywords: Solid Waste Disposal, Waste Management, Delta-D Technology, Rapid Digestion,
Organic Fertiliser Production, CETP Sludge Sludge
1. Introduction could be sold and the cost recovered.

Delta-D Technology (DDTech) is a patented Moreover it was found that the objectionable
technology (Sri Lanka Patent No: 12544 and smell emanating from the CETP Sludge was
Patent No: 12947) developed by a Professor in completely removed resulting in an odorless
the Department of Chemical and Process environment in the vicinity. At the
Engineering of the University of Moratuwa experimental stage the manufacturing cost of
[7,8,9,10,11]. In DDTech, a digestive fluid OF produced was approximately Rs. 18.75 per
called Delta-D rapidly digests organic waste kg. [1] However, in depth analysis revealed
(OW), by a combination of enzymetic and that the OF manufactured on a commercial
acidic action and finally converts OW into a scale, DDTech based OF manufacture and sale
nutrient rich organic fertiliser (OF). DDTech could result in a lower cost of production and
removes bad odour and destroys pathogenic an income of over Rs. 40 million per annum.
microorganisms as well [7,8,9,10,11]. This solution was found to have major
economic benefits, as well as, improved
A leading apparel manufacturer in Sri Lanka environmental conditions due to removal of
approached the research team to develop a bad odour of CETPS [12,13,14,15].
low cost solution to dispose 800 tons/year of
sludge generated at the Central Effluent The environmental sustainability of the
Treatment Plant (CETPS) of a major industrial proposed solution was analyzed and is
park. CETPS had around 70% moisture and presented in this paper. The composition of
emanated a bad smell. The disposal method CETPS and OF produced were analysed and
practiced was to transport CETPS to Puttalam impacts of components at the given
and hand over to incinerate in the rotary kiln concentrations were studied. Further an acute
of a cement manufacturing plant. The annual toxicology test was carried out at high
cost of CETPS disposal was around Rs. 24 concentration level to study the toxicity to
million. marine life due leachate produced by
application of OF on agricultural land.
The team selected DDTech to rapidly digest Eng. (Prof.) S.A.S. Perera, DSc, BASc(ChEng&FSc),CEng,
CETP Sludge and to convert it into organic PE(SL), IntPE(SL), FIE(SL), MAIChE (USA), Professor,,
Department of Chemical and Process Engineering, University
fertilizer with the addition of saw dust, of Moratuwa.
Eppawela Rock Phosphate (ERP) and
Eng. M.F.H.M. Aadhil, B.Sc. Engineering (UoM), Post
Dolomite. The main objective was to produce a Graduate Research Student, Department of Chemical and
nutrient rich organic fertilizer (OF) which Process Engineering, University of Moratuwa.
579
2. Literature Review (b)After 1 day, 10 kg of sawdust.and 5L of
water were added and to the digested CETPS
slurry and well mixed.
Sanitary land filling, composting, bio gas
production incineration, artificial drying and (c) Two days later, 5kg’s of ERP was added
combustion to recover thermal energy, and well mixed.
disposal of combustion ash by mixing with
cement mortar and concrete are some of the (d) Two days later 10 kg of Dolomite was
methods of CETPS disposal in the world. Some added and the mixture was left closed for 5
of the problems associated with the above days to obtain the final nutrient rich organic
techniques are; landfilling and composting– fertilizer (OF), of which the total weight was
slow OW digestion rate, foul smell, high land approximately 44.35 kg.
consumption and leachate production with
pathogens that could contaminate soil and (e) A well-mixed sample of OF was taken and
ground water table.; incineration and was sent to Bureau Veritas for GC-MS, FTIR,
combustions – air pollution, volatile organic Microbial Testing.
compounds, chlorinated orgamic compounds
and ash containing heavy metals. A recently (f) A representative sample of OF was applied
developed method issu per critical water to plants and the plant growth was compared
oxidation based sludge processing[17]. All with several other types of fertilizers available
these methods are expensive in terms of in the market.
capital cost and operational cost. In literature,
there is no rapid digestion process to produce
The applicability of any fertilizer into soil
OF, to recover the cost and to earn an income.
directly depends on the constituents of the
References pertaining to this section are
fertilizer and the form each constituent is
[17,18,19,20,21,22,23].
present in. It also helps to understand the
complementary behavior certain chemical
3. Objectives compounds display in the presence of other
chemical compounds. However though the
The main objectives were to provide a low above is necessary it may not be sufficient to
cost, environment friendly, economically prove the harmfulness or otherwise of a
viable, sustainable disposal method by using fertilizer. While the above may result in
DDTech to produce a nutrient rich organic predictions being drawn the reality check of
fertilizer (OF) which could be sold and the cost how the environment will respond to
recovered and to remove the objectionable foul fertilizers of unknown composition is revealed
smell emanating from the CETPS. The only by studying the changes it exerts on the
objectives of testing the OF produced from environment in the short term and the long
CETPS using DDTech, by, Chemical term.
Composition, GC-MS, FTIR and acute Therefore the paramount test to establish that
toxicology tests, were to determine the a fertilizer is safe for application in the short
elements present, main chemicals and their term is to study the soil fertility and carryout
structures and the toxicity plant and marine acute toxicology tests for aquatic and marine
life. life due to the potential leachate the fertilizer
will produce when applied on land. To
4. Experimental material and methods establish the long term suitability chronic
diseases have to be evaluated. However it
The CETPS sample received was well mixed requires long term studies of the fertilizer. As
and a representative sample was used for in the case of any fertilizer many of the chronic
laboratory analysis. The stepwise experimental conditions can be prevented by strictly
procedure, as recommended by the inventor of adhering to the fertilizer guideline especially
DDTech is given below. in managing the heavy metal content in
fertilizer.
(a)A quantity of 15.059 kg of de-watered
ETPSwith 20% moisture content was well 4.1 Acute toxicology test on marine and
mixed with 2L of Delta-D.solution in a 50L aquatic life
plastic barrel. After a few hours the thick Aquatic toxicology is the study of the effects of
CETPS became a uniform slurry. 1 day was manufactured chemicals and other
allowed for further digestion. anthropogenic and natural materials and
580
activities on aquatic organisms at various (for analytical methods see AOAC, 1990 and
levels of organization, from subcellular USDA, 1989).
through individual organisms to communities For the test results to be acceptable, control
and ecosystems. survival must equal or exceed 90%. Therefore
There are different types of toxicity tests that if 10 fishes are used in the control sample and
can be performed on various test species. one die due to the conditions in the control
Different species differ in their susceptibility to sample the experiment is terminated and the
chemicals, most likely due to differences in number of living species are counted. Further
accessibility, metabolic rate, excretion rate, the accuracy of the test can be improved by
genetic factors, dietary factors, age, sex, health using a facility that is well ventilated and free
and stress level of the organism. Common of toxic fumes. Sample preparation, culturing,
standard test species are the fathead minnow and toxicity testing areas should be separated
(Pimephalespromelas), daphnids (Daphnia to avoid cross contamination of cultures or
magna, D. pulex, D. pulicaria, toxicity test solutions with toxic fumes.
Ceriodaphniadubia), midge Laboratory ventilation systems should be
(Chironomustentans, C. ruparius), rainbow checked to ensure that return air from
trout (Oncorhynchus mykiss), sheepshead chemistry laboratories and/or sample
minnow (Cyprinodonvariegatu), rosy barbs handling areas is not circulated to test
(Pethiaconchonius), scud (Hyalalla Azteca), organism culture rooms or toxicity test rooms,
grass shrimp (Palaemonetespugio), goldfish or that air from toxicity test rooms does not
(Carassiusauratus). As defined by ASTM, contaminate culture areas. When the toxicity
these species are routinely selected on the basis test is concluded, all test organisms (including
of availability, commercial, recreational, and controls) should be humanely destroyed and
ecological importance, past successful use, and disposed of in an appropriate manner. [3]
regulatory use. [3] For the purpose of the acute toxicology test 10
A static non-renewal test is appropriate to test fish each was used in the fish tank used as the
the acute toxicology of the sludge by control set up and the fish tank which was
dissolving known weights of the sludge supplied with the fertilizer. The ten fish
sample in incremental dosage(10g per day) to selected included 8 Pethiaconchonius (rosy
the same static water tank with barbs) and 2 Carassiusauratus (gold fish) for
Carassiusauratus (gold fish(GF)) and each tank. All the fish were from the same
Pethiaconchonius (rosy barbs (RB)) to ensure aquarium and were of same age. The tanks
the validation is done precisely. The test were suitable to accommodate 8L of water
samples should be monitored for a minimum each and was supplied oxygen at a constant
of 1 week prior to renewal of water and rate. Nontoxic readymade fish food was
change of dosing rate if required. However supplied as per daily recommendations
sample there is a higher possibility of sufficient for the fish in the tanks. One of the
dissolved oxygen (DO) depletion from high tanks were supplied with the fertilizer 10g
chemical oxygen demand (COD) and/or every day up to 7 days. The daily fish count
biological oxygen demand (BOD), and ill was obtained in both the tanks.
effects from metabolic wastes from organisms
in the test solutions. Therefore it is important 4.2 Determining the composition of the
to provide nutrients and oxygen to the test and produced fertiliser
control sample to ensure these shortcomings It is tedious to find the exact composition of a
are eliminated. It is important to select a fish fertilizer derived from various waste raw
food that does not contain toxic organics and
materials of relatively unknown composition.
metals or whenever difficulty is encountered
in meeting minimum test acceptability criteria As often is the case most of the types of
for control survival and reproduction or compost fertilizers do not measure an exact
growth. If the concentration of total composition. However it is important to
organochlorine pesticides exceeds 0.l5 µg/g minimize the environmental and human
wet weight, or the concentration of the total health risk by identifying the nature of the
organochlorine pesticides plus Polychlorinated
fertilizer. The following parameters were
biphenyl’s (PCBs) exceeds 0.30 µg/g wet
anaysed using Inductive coupled plasma –
weight, or toxic metals (Al, As, Cr, Co, Cu, Pb,
Ni, Zn, expressed as total metal) exceed 20 Optical Emission
µg/g wet weight, the food should not be used Spectroscopy (ICP-OES) according to standard
Environmental Protection Agency (EPA) 3051
581
A. Total Potassium (K), Sodium (Na), of water was validated prior to the toxicology
Magnesium(Mg), Lead (Pb), Nickel (Ni), experiment to ensure the components
Copper (Cu), Iron (Fe), Chromium (Cr), Zinc dissolved or remain as a suspension in the fish
tank.
(Zn), Mercury (Hg) Cadmium (Cd) and
The number of remaining fish species was
Arsenic (As). Nitrogen (N) was determined recorded up to 7 days on a daily basis in both
using standar SLS 645: Part 1: Section C: 2009, the tanks. The results proved to be positive in
and Phosphorous (P) was determined using the environmental toxicology test after 70 g of
SLS 645: Part 5:1985. Organic Carbon (C) was fertilizer was added 10g a day and the living
determined using SLS 1246: Appendix D: 2003. number of fish were counted. All fish species
survived during the 7 day period indicating no
acute toxicology risk. The results are presented
A separate Gas chromatography and Mas
in Table 2 below.
spectroscopy (GC-MS) test was carried out to However it has to be noted that there are
determine organic compounds present in the several vulnerabilities of this examination as
fertilizer sample. Further a Fourier Transform some times fish may harm each other and
– Infrared Spectroscopy (FT-IR) was carried some fish may die due to these kind of
out to identify organic classes such as O-H behavioral aspects. This factor is
uncontrollable. Therefore the test data would
groups, C=O groups, C=N groups, C≡N
be more accurate if both the control and test
groups and C-N groups. Through carryout
samples contained a large amounts of fish.
various tests the composition of the fertilizer However this may lead to inaccurate counting
has become relatively clear and the as it has to be done manually.
harmfulness can be predicted thereby.
Table 2 acute toxicology test results
4.3Determining the composition of the Day
produced fertiliser Sample Fish
The total coliform of the produced fertilizer type 1 2 3 4 5 6 7
was analyzed as the raw material sludge was RB 8 8 8 8 8 8 8
Co
rol
nt
detected with coliform. Further fecal coliforms, GF 2 2 2 2 2 2 2

Escherichia coli and Salmonellaspp was RB 8 8 8 8 8 8 8
ad
de
liz
Fe
rti
er
GF 2 2 2 2 2 2 2
d
analyzed. The following standards were used

for each of the parameters.
The survival rate expected is 90% whereas the
Table 1 - Mixed mass ratios of sawdust and resulting survival rate was 100%. This implies
rice husk that the fertilizer will not be of toxic nature in
the short term for aquatic or marine life even if
Parameter Test Method they are very sensitive to toxics.
ISO 4831: 2006 /
Total Coliform SLS 516 Part 3 Section 1 :2013
ISO 7251: 2005 /

Fecal coliforms SLS 516 Part 12 Section 1 :2013
ISO 7251: 2005/

Escherichia col SLS 516 Part 12 Section 1 :2013
ISO 6579: 2002 / Amd 1:2007

Salmonellaspp SLS 516 Part 5:2013
5.1 Results of the acute toxicology test

7 samples of approximately 10g each of
fertilizer was mixed in one of the tanks over 7
days gradually increasing the dissolved Analysis of the composition of the fertiliser
fertilizer in water assimilating the runoff into
streams. The solubility of over 60 g in 4 litres Figure 1 - Acute toxicology test set up
582
5.2 Effects of Heavy metals 5.3 Organic matter from GC-MS
Heavy metals present in compost fertilizers The components present in the fertilizer can be
cause pollution in the soil. Uptake of these interpreted in many ways as many studies can
metals by the plants from the soil leads to be carried out how each chemical behaves in
reduction in crop productivity by inhibiting different environments and how the behavior
physiological metabolism. Metals such as As, of a chemical is altered in the presence of other
Cd, Hg, Pb or Se are not essential for plant chemicals.
growth since they do not perform any The GC-MS test identified over 16 organic
physiological function on the growth of the compounds present in the fertilizer which may
plants. Others, Co, Cu, Fe, Mn, Mo, Ni and Zn have given rise due to the sludge or other
are required for the metabolism and normal additives. The results are presented in figure 2
growth of plants. However, these metals may below. The components detected and toxicity
cause poisoning if the available amounts are and toxic concentration is given in Table 4
higher than their optimal values. below. [4][5]
Heavy metal poisoning by compost is a matter
of concern if applied to food crops since they Table 4 - organic compounds detected and
can easily get transferred to humans and toxicity
animals via edible parts. No Detected Compound
contaminat
Absorption and accumulation of heavy metals
Toxicity
Ground
in plant tissues depends on many factors
Acute
water
ion
including temperature, moisture, organic
matter, pH and nutrient availability. 1 Cyclohexasiloxane
Accumulation of heavy metals in plants 2 2H-Naphthol[1,2-b]pyrane-
depends on the species. Elevated Pb in soils 3-carbonitrile
causes decrease in soil productivity, and a 3 Silane
very low Pb concentration may inhibit some 4 Carbazole
vital plant processes, such as 5 Hexadecanoic acid
photosynthesis, mitosis and water absorption 6 3,6-Dioxa-2,4,5,7-
with toxic symptoms of dark green leaves, tetrasilaoctane
wilting of older leaves, stunted foliage and 7 Hexasiloxane
brown short roots. [2] 8 Octadecanoic acid
9 Mercaptoacetic acid
Heavy metals are phytotoxic to plants and
10 Heptasiloxane
causes chlorosis which decolorizes plant
11 1H-Indole-2-carboxylic
leaves. Furthermore, they have negative effects
acid
such as yield depression, reduced nutrient 12 N-Benzhydrylidene-1-
uptake, disorders in plant metabolism and (2,4,6-
reduced ability to fixate molecular nitrogen trimethylphenyl)ethylamin
in leguminous plants. [2]. The maximum e N-oxide
allowable limits of the heavy metals in any 13 Piperdine
commercial fertiliser and the concentration of 14 Tetrasiloxane
heavy metals present in the produced fertiliser 15 2,4,6- Cycloheptatrien-1-
is given in table 3 below. one
16 4-Methyl-2-
trimethylsilyloxy-
Table 3 - heavy metal composition
acetophenone
Max Amount
Heavy Metal allowable present
Lead (Pb) 30 (mg/kg) (mg/kg)
ND No detected toxicity in past research
Chromium(C 50 11 Toxicity detected
r)
Cadmium 3 ND
Most of the chemicals used do not display an
(Cd)
Arsenic (As) 50 ND acute toxicity even when administered orally.
Mercury (Hg) 1 ND However the lethal dosage varies in each
component. Considering Silane and Piperdine,
though the lethal dosage is relatively low the
ND : Not detected
acute toxicology test proved to be positive
LOQ : Limit of indicating the concentrations were below the
Quantification = 0.04 mg/kg lethal dosage for marine and aquatic life.
Further ground water contamination is not
present in any of the substances.
583
Total coliform and Escherichia coli was not
tested in the sludge. Only the critical
parameter of fecal coliforms and Salmonella
spp were tested. However all four parameters
were tested in the fertilizer that was produced.
The Delta-D process inhibits all pathogenic
microbes and results in the fertilizer being
pathogen free.
When the fertilizer is applied to the earth,

microbial digestion takes place from
microorganisms in the earth further digesting
any large molecules present in the fertilizer
and break it down to smaller fractions to be
able to be absorbed into the plants.
5.6 Results of the FT-IR test:
The FT-IR test was carried out to identify

functional groups in the produced fertilizer.
The Table 6 contains the results of the FT-IR
test. Figure 3 is the graph obtained from the
test.
Table 6 - FT-IR Test Results

Wavenumbe
Type of bond r (cm-1) Intensity
O-H 3650-3200 Strong,
C=O 1750-1650 broad
Strong
C=N 1650-1550 Medium
㻯≡㻺 2260-2100 Medium
C-N 1230-1020 Medium
6. Conclusions
The study revealed that the fertilizer produced
Figure 2 GC-MS results
using the CETPS through DDTech conversion
process was far superior to other fertilizers,
5.4 Pathogenic nature of produced OF available in the local market, that were used in
the experiment.
The pathogenic nature of the feed sludge and
the produced fertilizer were compared by Chemical composition, GC-MS, FTIR and
carrying out microbial tests on the feed sludge microbial test results (carried out by Bureau
and the produced fertilizer according to the Veritas Laboritary, Katubedde, Sri Lanka) )
standards mentioned in the methodology. The indicate that there are no toxic chemicals or
comparison revealed that although the pathogenic microorganisms in the OF
feedstock contained high microbial counts the produced.
final product was free from any form of
microbes. Table 5 contains the results. Acute toxicology test on marine and aquatic
Table 5 - Microbial Test Results life carried out at the University of Moratuwa,
Result of Result of Unit clearly indicate that the OF produced can be
sludge Fertilize applied to agricultural fields, without causing
Parameter
r any harm to marine organism living in
Total Coliform N/A ND MPN/ waterbodies in the vicinity.
Fecal Coliforms 460 ND MPN/
g
Escherichia coli N/A ND g
MPN/
Salmonella spp Absent Absent In g25g
584
Acknowledgement Problem in Sri Lanka, The Proceedings of
the 22nd International Conference on Solid
Waste Technology and Management
The research group is thankful to the great
March 2007, pp. 1312-1319, ISSN 1091-
support and corporation rendered by Mr. 8043 © 2007 JSWTM, Widener University,
Harsha Deraniyagala, The General Manager of Philadelphia, USA.
MAS Fabric Park at Thulhiriya in making the
research a success. 8. Perera, S.A.S. Manufacture of Organic
Fertiliser from Vegetable Market Garbage
(VMG) Using Eppawela Rock Phosphate
References and Its’ Effects on Rice Cultivation,
1. Perera, S.A.S. Aadhil, M.F.H.M. The Proceedings of the 22nd International

Gunarathne, K.K.D.I.K. Dananjina, A.M.U. Conference on Solid Waste Technology
"Conversion of textile industry effluent and Management March 2007, pp. 1301-
treatment plant sludge into a valuable 1311, ISSN 1091-8043 © 2007 JSWTM,
organic fertilizer using Delta-D Widener University, Philadelphia, USA.
technology," Moratuwa, 2017.
9. S.A.S. Perera, P. Ratnaweera, J.N.
2. Singh, J. and Kalamdhad, A. S. "Effects of Meegoda,, Sustainable Agricultural
Heavy Metals on Soil, Plants, Human Practices for Developing Countries,
Health and Aquatic Life," International
Journal of Research in Chemistry and The Proceedings of The First International
Environment, vol. I, no. 2, pp. 15-21, 2011. Conference on Soil and Rock Engineering
August 2007, Colombo, Sri Lanka , pp. 1-7
© The Geotechnical Society of Sri Lanka.
3. US-EPA, "acute-freshwater-and-marine-
wet-manual_2002.pdf,"October 2002.
10. Perera, S.A.S. Delta-D Technology, a
[Online]. Available:
Technically, Economically and
https://www.epa.gov/sites/production/fi
Environmentally Feasible Solution to the
les/2015-08/documents/acute-freshwater- Urban Solid Waste (USW) Problem and
and-marine-wet-manual_2002.pdf. the Fertiliser Problem in Sri Lanka
[Accessed 26 December 2016]. Annual Transactions of IESL-2007, pp.129-
39, © The Institution of Engineers, Sri
4. PAN Pesticide Database, "pesticideinfo," Lanka.
PAN Pesticide Database, 2006. [Online].
Available: http://www.pesticideinfo.org. 11. Perera, S.A.S. Delta-D Initiated
[Accessed 25 5 2017]. Microorganic Digestion of Saw Dust into
Organic Fertiliser – A Technically,
Economically and Environmentally
5. OSHA, "www.osha.gov," [Online].
Available: Feasible Solution to the Saw Dust Problem
https://www.osha.gov/dts/chemicalsamp in Sri Lanka
Engineer (Journal of The Institution of
ling/data/CH_231525.html. [Accessed 31
Engineers, Sri Lanka), Vol XXXX1, No:
05 2017].
04, October 2008, ISSN-1800-1122, pp.50-
6. Perera, S.A.S. Influence of Organic Fertiliser 65, 2007, © The Institution of Engineers,
Fortified with Eppawela Rock Phosphate Sri Lanka.
on the Cultivation of Spinach and Curry 12. Perera, S.A.S. Conversion of Coconut
Chillies. Water Waste Produced in Desiccated
Proceedings of The International Seminar on Coconut Mills into Liquid Organic
―The Engineer in Sustainable, Social and Fertiliser Using Delta-D Technology
Economic Development – Regional by S.A.S. Perera, The Proceedings of the 24 th
Contribution Towards Agenda 21, organised International Conference on Solid Waste
by The Institution of Engineers Sri Lanka Technology and Management March 15-18,
(IESL), World federation of Engineering 2009, pp. 1601-1612, ISSN 1091-8043 © 2009
Organisations (WFEO) and the Federation of JSWTM, Widener University, Philadelphia,
Engineering Institutions of South and Central USA.
Asia (FEISCA), 20 Oct. 2002, pp. 62-74, © The
Institution of Engineers, Sri Lanka 13. Perera, S.A.S. Sustainable Agriculture
Through Delta-D Technology, a Solution to
7. Perera, S.A.S. Patented Process for Rapid Urban Solid Waste and Global Warming
Digestion of All Types of Biomass into The Proceedings of the 24th International
Organic Fertiliser, a Solution to the Urban Conference on Solid Waste Technology and
Solid Waste (USW) and to the Fertiliser Management March 15-18, 2009, pp. 1601-
585
1612, ISSN 1091-8043 © 2009 JSWTM, Widener
University, Philadelphia, USA. 22. I.S. Turovskiy, P.K Mathai, Wastewater
sludge processing, .google.com
14. Perera, S.A.S. Delta-D Technology – A
Patented Technology That Could Be Used To 23. Fullana, A., Conesa, J.A., Font, R.
Prevent Emission Of Green House Gases From Formation and destruction of chlorinated
Urban Solid Waste,Agricultural Waste and pollutants during sewage sludge incineration-
Farm WasteInternational Conference on Environmental science & …, 2004 - ACS
Climate Change, Biodiversity and Food Publications
Security in the South Asian Region, Organised
by Punjab State Council for Science and 24. Dai, J., Xu, M., Chen, J., Yang, X., Z
Technology of Govt. of Punjab and UNESCO – KePCDD/F, PAH and heavy metals in the
New Delhi, pp. 87-89, ISBN No: 978-81-88362- sewage sludge from six wastewater treatment
24-0, © Punjab State Council for Science and plants in Beijing, China
Technology. - Chemosphere, 2007 - Elsevier
Published in Abstract Form / Single Author
Marks Claimed = 1.5
15. Perera, S.A.S. Manufacture of Organic

Fertiliser from Poultry Slaughterhouse Waste
Rendering Plant Sludge Using Delta-D
Technology
The Proceedings of the 24th International
Conference on Solid Waste Technology
and Management March 15-18, 2009, pp. 1601-
1612, ISSN 1091-8043 © 2009 JSWTM, Widener
University, Philadelphia, USA.
Published in Abstract Form / Single Author
Marks Claimed = 1.5
16. Robinson, T. McMullan, G. Marchant, R. P

Nigam, Remediation of dyes in textile effluent:
a critical review on current treatment
technologies with a proposed alternative,
Bioresource technology, 2001 – Elsevier
17. Griffith, J.W., Raymond, D.H. The first

commercial supercritical water oxidation
sludge processing plant
- Waste Management, 2002 – Elsevier
18. Suh, Y.J. Rousseaux, P. An LCA of

alternative wastewater sludge treatment
scenarios,
- Resources, Conservation and Recycling, 2002
– Elsevier
19. KD North, Tracking polybrominated

diphenyl ether releases in a wastewater
treatment plant effluent, Palo Alto, California -
Environmental Science & Technology, 2004 -
ACS Publications
20. Fytili, D., Zabaniotou, A. Utilization of

sewage sludge in EU application of old and
new methods—a review
- Renewable and Sustainable Energy Reviews,
2008 – Elsevier
21. Rulkens, W., Sewage sludge as a biomass

resource for the production of energy:
overview and assessment of the various
options- Energy & Fuels, 2007 - ACS
Publications
586
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[587 594],2017
2017
The Institution
Lanka
Field Experiments on Low-Cost Combined-Media

an Anaerobic Filter System for the Treatment of
Landfill-Leachate
D.M.T.P. Fernando, W.K.C.N. Dayanthi, K.G.D. Chamenuka,
B.S.R. Nanayakkara and K. Kawamoto
Abstract: Anaerobic filter (AF) is commonly applied in the treatment of landfill-leachate, and
one major drawback is the use of expensive packing media. Hence, investigations on the applicability
of low-cost materials as efficient packing materials in anaerobic filters are timely and crucial. Thus,
the present study was focused on assessing the applicability of low-cost materials as packing media in
an anaerobic filter treatment system at field conditions, in which two AFs (AF1 and AF2) were
connected in series. The treatment was processed through the anaerobic filter 1 filled with dewatered
alum sludge (DAS) and washed sea sand (WSS) and then through the anaerobic filter 2 filled with
sawdust (SD) and biochar (BC). Removal efficiencies of organic compounds, nutrients and heavy
metals found in land fill-leachate; and the variation of physical and mechanical properties of the
packing media subjected to the interactions between the packing media and the landfill-leachate were
obtained throughout the experimental period of 133 days. Average overall removal efficiencies of
BOD5 and COD were 43.6 ± 17.7 % and 54.6 ± 17.0%, respectively. Average removal efficiencies of TN
and TP were 61.5 ± 15% and 68.9 ± 16%, respectively. The mean values of the overall removal
efficiencies of the AF system for Cu, Fe, Mn, Pb and Cd were 85%, 79%, 74%, 72% and 67 %,
respectively. The physical and mechanical properties of the packing materials did not undergo a
significant variation as a result of the interaction with the landfill-leachate.
Keywords: Anaerobic filters, Contaminants, Field conditions, Low-cost media, Landfill-leachate
1. Introduction leachate treatment processes can be categorized

Industrialization, population growth and into physical, chemical and biochemical. Due to
urbanization have caused a rapid generation of the reliability, simplicity and high cost-
solid waste. Solid waste can be classified into effectiveness, biochemical treatment is
three main types depending on different commonly used for the removal of the bulk of
sources: municipal waste, industrial waste and leachate containing high biochemical oxygen
biomedical waste [1]. One of the major demand (BOD). When treating biodegradable
pollution problems caused by the municipal leachate, biochemical techniques can yield a
solid waste (MSW) landfill is the leachate, reasonable treatment performed with respect to
which is generated as a result of mixing water the chemical oxygen demand (COD) and NH3–
from precipitation and surface run-off with the N [4].
MSW.
Eng. (Ms.) D.M.T.P Fernando, Student member of IESL,
Generally, landfill-leachate may contain a large B.Sc.(Eng. Ruhuna), Civil Engineer, Nawaloka Piling Pvt(Ltd)
amount of organic matter, which are Email: thanujafernando26@gmail.com
biodegradable, materials refractory to Eng. (Dr) W.K.C.N Dayanthi, MIE(SL), B.Sc.(Eng.
biodegradation, as well as ammonia –nitrogen, Peradeniya), M.Eng.(Moratuwa), M.Eng.(AIT), Dr.Eng.
heavy metals (e.g. copper, iron, zinc, lead, (Kyoto), Senior Lecturer, Department of Civil and
manganese etc.), chlorinated organic and Environmental Engineering, Faculty of Engineering,
University of Ruhuna.
inorganic salts (e.g. Chloride, sulphate, sodium Email: neeth02@gmail.com
etc.) All these contaminants are a great threat to
the surrounding soil, groundwater and even Eng. (Ms.) K.G.D Chamenuka, Student member of IESL,
B.Sc. (Eng. Ruhuna), Civil Engineer, Sanken Construction
surface water [2], [3]. Pvt(Ltd)
Eng. (Mrs.) B.S.R. Nanayakkara, C.Eng.,
Therefore, proper treatment of the landfill- B.Sc.Eng.(Moratuwa), MIE(SL), M.Phil. Student (Ruhuna),
leachate is essential in landfill handling in order Civil Engineer. Ports Authority, Colombo.
to mitigate environmental risks and to protect Prof. K. Kawamoto, Graduate School of Science and
the surface and groundwater. The landfill- Engineering, Saitama University, Saitama, Japan.
1
587 ENGINEER
Landfill-leachate treatment mechanisms can be had been filled with WSS, DAS, SD and BC in
divided into two categories, namely, pump- layers of equal thickness; and plastic pall rings,
and - treat methods and in-situ methods. respectively. The AF filled with plastic pall
Among the advantages of pump - and- treat rings was the control experiment.
method over in-situ method is the less adverse
impacts on the ecological conditions, The aim of the present study was to assess the
groundwater issues, and the less requirement landfill-leachate treatment potential of a system
of the long-term monitoring activities [5]. of two anaerobic filters connected in series,
filled with low-cost, locally and readily
Anaerobic filter (AF) is a notable „pump-and- available materials. The first objective of the
treat‟ method because it is a kind of anaerobic study was to find out the removal efficiencies of
digester containing a packing medium that the organic compounds, nutrients and heavy
anaerobic bacteria can grow in as well as metals. Removal efficiencies of organic
remove contaminants while the leachate flows compounds were obtained in terms of BOD5
through the media. One merit of using an and COD and nutrients in terms of ammonia
anaerobic filter for treating the landfill-leachate nitrogen(NH3-N), total nitrogen (TN), total
is the absence of the secondary clarifiers and phosphorus (TP) and orthophosphate
reduction of all the associated cost and phosphorus (PO43- -P). Similarly, the removal
operational problems [5]. Anaerobic filters have efficiencies of heavy metals were measured in
emerged as a promising technology to treat the terms of Cu, Fe, Mn, Pb and Cd. The removal
landfill-leachate. efficiency of total dissolved solids (TDS) was
also measured. The second objective was to find
However, its application has been limited due out the changes in the physical and mechanical
to the requirement of expensive packing media. properties of the packing media due to them
Plastic pall rings, tubes and flakes are being interacted with the landfill-leachate.
conventional packing materials, which have to
be manufactured in a factory, thus causing As the selected packing materials are low- cost
these materials to be quite expensive. If low- and readily available, the present study is
cost materials can effectively be utilized as significant in the context of developing
packing media in anaerobic filters, it will countries. The study will also find a solution for
enhance the applicability of the AF as a the disposal issue of DAS, which is regarded as
treatment technique for landfill-leachate in a a waste material.
developing country.
Thus, it is a timely need to perform experiments 2. Methodology

on low-cost packing media and their
appropriate quantities and placement 2.1 Experimental Set-Up
techniques in AFs to treat the landfill-leachate. The experimental set-up (Figure 1) includes a
According to a previous research conducted at sequence of components connected to each
field conditions, by the authors of this paper, other in the following order: a lysimeter (1.5 m
materials, namely washed sea sand (WSS), diameter and 5 m height), a receiving tank, an
dewatered alum sludge (DAS), sawdust (SD) overhead storage tank, an anaerobic filter filled
and biochar (BC) are efficient as packing with packing materials of high particle
materials in anaerobic filters to treat organic densities (DAS and WSS), a storage tank, an
compounds, nutrients and heavy metals [6]. In overhead storage tank, an anaerobic filter filled
the above research, the performances of the with packing materials of low particle densities
selected packing materials for the removal of (SD and BC) and a storage tank. The lysimeter
organic compounds were on a par with plastic consisted of five layers of expansive soil,
pall rings, which is a conventional packing metals, sand, MSW and in-situ soil. The
material. Further, the selected materials sectional view of the lysimeter is shown in
surpassed the performances of plastic pall rings Figure 2. This lysimeter acted as a model
in treating the cations, Ca 2+ and Mg 2+ and landfill, of which the leachate was used as the
heavy metals, Cu and Fe, owing to the influent of the AF system.
promotion of treatment mechanisms such as
ion exchange, precipitation and adsorption due The experimental set-up was located in
to different properties possessed by these Hambantota. Hence, the rainfall received by the
materials. The experimental set-up of the above lysimeter was relatively less, and tap water was
research consisted of two AFs, each of which applied onto the lysimeter to increase the
ENGINEER 2
588
leachate production rate. The leachate cycle of leachate onto the respective overhead
generated in the lysimeter was applied on AFs tank of AF1 and AF2. The loading cycle was
in the down-flow direction. The layer thickness approximately 2 weeks. Then, the surface
of each layer packing material was 67 cm. The loading rates were obtained by multiplying the
leachate generated in the lysimeter was stored flow rate with the value of each parameter and
in the storage tank 1, and using a pump it was then dividing by the surface area of the
flown to the overhead tank 1. By gravity, the respective reactor.
flow was passed through AF 1. The effluent of
AF 1 was collected in the receiving tank 1 and Wastewater samples were collected from the
then pumped to the overhead tank 2. The influent of the AF1 and the effluent lines of
effluent in the overhead tank 2 was drained both AF1 and AF2, once in the two-week
onto AF 2 under gravity. The effluent of AF 2 period. Influent characteristics and the surface
was collected temporarily in a storage tank loading rates of the selected wastewater
before being recirculated onto the lysimeter. parameters are mentioned in Table 2 and 3,
respectively. pH of AF1 influent varied
2.2 Packing Configuration of the Media between 6.5 and 9.5. The target wastewater
and the Experimental Run. parameters were organic compounds (BOD5
Figure 3 depicts the packing configurations of and COD), nitrogenous compounds {Total
the media beds of AF1 and AF2. The height and Nitrogen, Ammonia Nitrogen} and
the diameter of the cylindrical AF columns phosphorous compounds {Total Phosphorus
were 2 m and 600 mm, respectively. AF 1 and and Ortho Phosphates Phosphorus}. In
AF 2 consisted of a bed of coarse aggregates addition, heavy metals namely Cu, Fe, Mn, Pb
(200 mm layer thickness) at the bottom in order and Cd, and TDS were also analysed. In the
to collect the drained effluent through the partial run of the experiment (after 98 days), the
media bed, and direct toward the effluent port. influent was added with heavy metals
The packing media of the AF columns were artificially to obtain heavy metal characteristics
arranged considering the particle densities of typical for the landfill-leachate. This was done
the packing materials. In AF 1, packing to find the capability of the treatment system to
materials with relatively higher particle bear high heavy metal loading rates. Table 4
densities, DAS and WSS, were filled with indicates the heavy metal loading rates before
layers. In AF 2, packing materials with and after addition of heavy metals artificially in
relatively lower particle densities, BC and SD, the influent.
were filled in layers. The layer height of each
packing material was 670 mm. After Core samples of the packing media bed were
compacting the packing materials, two AF units collected before applying the influent and at the
were connected in series. end of the experimental run in order to find out
the variation of the physical and mechanical
properties of the packing media with the
progression of the treatment.
Figure 1 - Image and the Isometric View of the

Experimental Set-Up.
After loading the AF system, the influent was

applied, and its flow rate was adjusted to give a
pore volume retention time approximately
equal to 2 d in each AF. Table 1 indicates the
flow rates and the hydraulic loading rates of the Figure 2 - Sectional View of the Lysimeter.
AF units. The flow rate was manually
measured several times during one loading
3
589 ENGINEER
2.3 Analysis of the Physical and Table 3 - Surface Loading Rates on the AF
Mechanical Properties of the Packing System.
Media. Parameter AF 1 AF 2
For the analysis of physical and mechanical
properties of the packing media, specific Surface loading Surface loading
gravity test and hydraulic conductivity tests rate (g/m2. d) rate (g/m2. d)
were conducted. These test results demonstrate
the influence of the treatment mechanisms BOD5 49.7 34.3
inside the packing media on their physical and
COD 199.7 123.6
mechanical properties.
TN 16.8 14.6
TP 0.18 0.08
Table 4 - Heavy Metal Loading Rates Before

and After the Addition of Heavy Metals
Artificially.
Surface loading rate (mg/m2. d)
Before adding After adding
Paramet heavy metals heavy metals
er artificially artificially
Figure 3- Packing Configuration. Cu 4.7 35.9
Pb 11.0 2.5
Table 1 - Flow Rates and the Hydraulic
Cd 0.1 3.1
Loading Rates of the AF Units.
Mn 0.8 9.3
material (m3 /d)
Average flow
loading rate
(m3 /d/ m2)
rate (m3 /d)
Fe 0.3 27.7
individual
Hydraulic
material
Packing
AF unit
Rate of
Flow
TDS 169.0 266.4

AF 1 DAS 0.07 0.06 0.23
WSS 0.06 3.1 Removal of Organic Matters
Removal efficiencies of organic matters with
AF 2 BC 0.06 0.07 0.25 time were obtained in terms of BOD5 and COD
SD 0.08 during the experimental period. Figure 4
indicates the removal efficiencies of BOD5 with
respect to the time. According to the graph,
Table 2 - Influent Characteristics. removal efficiency of AF 1 was in the range of
Parameter AF 1 AF 2 20% -30% in the first half of the experimental
run. However, a continuous increment of the
Average Average influent
removal efficiency was obtained in the second
influent concentration
half of the experimental run. The overall
concentration (mg/L)
removal efficiency of BOD5 in the treatment
(mg/L)
system was 43.6 ± 17.7 %.
BOD5 219.50 137.00
As stated by McCarty (1982) [7], in anaerobic
COD 882.36 493.48 conditions, three physiological groups of
bacteria are involved in the anaerobic
TN 74.10 58.29
conversion of organic materials to methane. In
TP 0.79 0.31 wastewater treatment applications, biomass is
produced continuously as the substrate in the
wastewater is consumed, and thereby
biodegrade [8]. Therefore, the rate of
biodegradation may increase with time. The
low treatment efficiencies of BOD5 at the initial
stage could be caused by the less microbial
ENGINEER 4
590
population in the filtration media. In line with
the above phenomenon, the treatment
efficiency of BOD5 in the present study
increased with time.
COD is another important parameter, which

measures the organic matter content. COD
removal efficiencies of the AF units are
illustrated in Figure 5. According to the graph,
COD removal efficiency was significantly
higher in AF 1, and fluctuations of the removal Figure 5 - COD Removal Efficiency with
efficiencies could also be seen. In the mid Time.
experimental run, the removal efficiencies were
lesser compared to the initial and later stages. 3.2 Removal of Nitrogenous Compounds
The overall COD removal efficiency of the In this analysis of nitrogen compounds, mainly
treatment system was 54.6 ± 17.0%. NH3-N and TN were considered. Figure 6
indicates the removal efficiencies of NH3-N
Sorption refers to the exchange of molecules with time. The mean efficiencies of NH3-N
and ions between the solid phase and the liquid removal by AF 1 and AF 2 were 77.9±17.94%
phase [9]. Adsorption is a very important and 55.1±28.5%, respectively. Nitrification and
method for removing contaminants, the adsorption of ammonium could be the
particularly dissolved organic contaminants possible mechanisms, which can eliminate
from wastewater streams [8]. Hence, adsorption NH3-N [10]. At the initial stage, NH3-N
can be considered a potential removal adsorption could be the process, which
mechanism of organic materials in AF1 and contributed to the removal of most of the
AF2. With time, adsorption capacity decreases, NH3-N. Here, adsorbent could most probably
and hence another mechanism should govern be the packing materials. Large surface areas
the treatment process to keep the consistency in provide high adsorption capacities, and small
treatment. The explanation of Coony (1999) [9] particles have relatively large surface areas. At
on biodegradation process, can be used to the initial stage of the experimental run, the
explain the COD removal in the latter part of dominant mechanism for NH3-N treatment
the experimental run. As the rate of could be adsorption. The adsorption capacity of
biodegradation may increase with time, the bio- the packing materials decreases with the
degradation could be the possible dominant gradual occupation of the vacant sites with
treatment mechanism in both AFs to obtain time.
removal of COD at the latter stage. Average removal efficiencies of TN were
25.63±15.30% and 54.33±11.37% for AF 1 and 2,
respectively as shown in Figure 7. Mainly the
biochemical nitrogen removal is accomplished
by anoxic denitrification process. When
compared to AF 1, AF 2 gained higher removal
efficiency in the latter part of the experimental
run.
Figure 4 - BOD5 Removal Efficiency with

Time.
Figure 6 - NH3-N Removal Efficiency with

Time.
5
591 ENGINEER
Figure 7 - TN Removal Efficiency with Time. Figure 9 - TP Removal Efficiency with Time.
3.3 Removal of Phosphorus Compounds According to Bagchi [9], the surfaces of organic
Phosphorus compounds in terms of TP and matter provide some adsorption sites, and the
PO43- -P were considered in the analysis. A adsorption process has demonstrated to be
considerable treatment occurred in the AF 1 widely effective for removing dissolved organic
unit. According to Figure 8 and 9, it is obvious substances from the wastewater. Therefore,
that AF 1 unit directly contributed to the adsorption could also contribute to the removal
removal. The average overall removal of inorganic dissolved solids in AFs.
efficiency was 75.4 ±14.53%. According to
Razali et al. [11], DAS is an effective absorbent Presence of coagulants such as alum, iron salts,
for phosphorus compounds. Because there are and organic polymers in DAS can cause the
some polymeric forms of aluminium that are removal of heavy metals as a hydroxide
not completely neutralized, the alum sludge precipitate. As microorganisms consume
may carry a small net positive charge. As substrates and carry out oxidation/reduction
phosphorus is normally present in the form of reactions, metal ions can be oxidized/reduced.
negatively charged anions, it is readily In addition, the sulphides available in the
adsorbed by the positively charged aluminium packing media can react with heavy metal ions
hydroxide. It is the principal mechanism for and form insoluble precipitates. The
removal of phosphorus from the wastewater in precipitates such formed can be separated from
sewage treatment plants [12]. Hence these the influent by sedimentation or filtration as it
results verify that DAS is a suitable packing flows through WSS. The removal of heavy
material to treat phosphorus compounds in the metals could occur due to the packing media
landfill-leachate. favouring adsorption such as DAS, SD and BC
[14]. Biosorption activities also may remove
3.4 Removal of TDS and Heavy Metals heavy metals due to the availability of the
with Time microbial biomass in packing materials [15].
The mean values of the overall removal Further BC and SD contain various organic
efficiencies of the AF system for Cu, Fe, Mn, Pb, compounds (lignin, cellulose and
Cd and TDS were 85%, 79%, 74%, 72%, 67 %, hemicellulose) with polyphenolic groups that
and 47%, respectively. Figure 10, 11, 12, 13 and could bind heavy metal ions through different
14 depict the variation of the removal mechanisms. Ion-exchange reactions could also
efficiencies of TDS, Cu, Fe, Pb and Mn, cause the heavy metal removal. Zeolites and
respectively with the progression of the silicate minerals contained in WSS could be
experiment. able to exchange its cations with the heavy
metal ions.
3.5 Variation of Physical and Mechanical

Properties
Table 5 shows the physical and mechanical
properties of the packing materials before and
after loading the influent onto the system. The
physical properties of the packing media did
not undergo much change during the 4-month
long experimental run. The slight changes of
these properties could be attributed to the
Figure 8 - Removal Efficiency with occurrences of treatment mechanisms such as
Time.
ENGINEER 6
592
coagulation, flocculation, sedimentation,
filtration, adsorption and ion exchange. In
addition, a slight wearing off of the materials
could contribute to these variations.
Figure 14 - Mn Removal Efficiency with Time.
Table 5 - Physical and Mechanical Properties

of Packing Materials.
Filter Porosity Hydraulic
Figure 10 - TDS Removal Efficiency with media conductivity(cm/s)
Time. (x10-3)
Initial Final Initial Final
DAS 0.663 0.697 3.8 3.6
WSS 0.401 0.392 6.9 6.5
SD 0.756 0.807 3.8 3.2
BC 0.541 0.446 4.1 3.6
4. Conclusions
The anaerobic filter treatment system described
in this study had the potential to treat the
organic and nutrient compounds as well as
Figure 11 - Cu Removal Efficiency with Time.
heavy metals in the landfill-leachate with
significant efficiencies. There was no significant
variation of the physical and mechanical
properties of the packing media due to the
interactions between packing materials and
leachate. The AF treatment system could
withstand high heavy metal loading rates. It
was also proved to some extent that these low-
cost, locally and readily available filter
materials can effectively be applied in anaerobic
filters to treat landfill-leachate partially,
without rapid deterioration of the physical and
Figure 12 - Fe Removal Efficiency with Time. mechanical properties of the reactive media.
The key feature of this study was the use of

low-cost, locally and readily available materials
instead of the expensive packing media in a
system of anaerobic filters connected in series.
The selected packing media were expected to
provide effective treatment for the organic
compounds and nutrients due to the diverse
properties possessed by these materials.
The study showed that all four materials were
satisfactory in treating landfill-leachate in terms
Figure 13 - Pb Removal Efficiency with Time. of organic compounds and nutrients. As the
anaerobic filter 1 was packed with dewatered
alum sludge and washed sea sand, a significant
removal of phosphorus was obtained. That was
a key finding of this study, which verifies that
the DAS is a suitable adsorbent to treat
7
593 ENGINEER
phosphorus compounds (Total–P and Low-Cost Filter Media”, IESL Annual Transaction
orthophosphate phosphorus) in the leachate at 2017 (In review)
the field conditions prevailing in the dry zone
7. McCarty, P. L. In: (D.E. Hughes, D. A. Stafford,
of Sri Lanka.
B. F. Weatley, W. Beader, G. Lettinga, E. J.
Nuns, W. Verstraete and R. L. Wentworth, eds.)
As the future improvements of this research Anaerobic Digestion. Elsevier Biomedical,
study, better and alternative arrangements of Amsterdam, pp. 3-22. McCoy, J.H. (1962) J.
low- cost and readily available materials in the Appl. Bact. 25:213-224.cited by the Director,
AFs and alteration of the hydraulic retention Publications Division, Food and Agriculture
time can be highlighted in order to enhance the Organization of the United Nations, Viale delle
removal efficiencies. Further experiments with Terme di Caracalla, 00100 Rome, Italy, 1982.
increased packing densities and hydraulic
8. Coony, D. O., Adsorption Design for Wastewater
retention times will give better results.
Treatment, Lewis Publishers, 1999.
Chemically modified packing materials will
also provide higher efficiencies. The effect of 9. Bagchi, A., Design of Landfills and Integrated
the long-run operation of the system on the Solid Waste Management, John Wiley & Sons,
variation of the physical and mechanical Inc., 2004.
properties of the packing materials should be
investigated. 10. Metcalf & Eddy, ”Wastewater Engineering
Treatment and Reuse”, 4th edition, McGraw-
Hill, Inc., New York, chapter 9, pp. 888-889.
Acknowledgement
11. Manoj, V. R. & Vasudevan, N., “Removal of
This work was supported by the research grant Nutrients in Denitrification System using
from the JST/JICA Science and Technology Coconut Coir Fibre for the Biological Treatment
Research Partnership for Sustainable of Aquaculture Wastewater”,J. Environ. Biol.,
Development (SATREPS). Authors would like Vol.33, pp.271-276, 2012.
to acknowledge Ms D.A.M. Nimal Shanthi for
her assistance in laboratory analyses. 12. Razali, M., Zhao, Y.Q., & Bruen, M.,
“Effectiveness of a Drinking-Water Treatment
Sludge in Removing Different Phosphorus
References Species from Aqueous Solution”, Separation and
purification technology, Vol. 55, No. 03, July,
1. Liu, S., Landfill leachate treatment methods and 2007, pp.300-306.
evaluation of Hedeskoga and Måsalycke landfills,
Water and Environmental Engineering, 13. Norvell, W. A., 1982,” Feasibility of
Department of Chemical Engineering, Lund Inactivating Phosphorus with Aluminium Salts
University, Sweden, 2013. in Ball Pond. CT. Conn. Agr. Exp. Sta.
Bull.806.10p.
2. Renou, S., Givaudan, J. G., Poulain, S.,
Dirassouyan, F. & Moulin, P., “Landfill Leachate 14. Wan, N. W. S & Hanafiah, M. A. K. M.,
Treatment: Review and Opportunity”, Journal of “Removal of Heavy Metal ions from
Hazardous Materials, vol. 150, 2008, pp. 468– Wastewater by Chemically Modified Plant
493. Wastes as Adsorbents: A Review”, Bioresource
Technology, Volume 99, Issue 10, July 2008, pp.
3. Robinson A. H., Landfill leachate treatment 3935-3948.
membrane technology, June 6-12, 2005.
15. Friis, N. & Myers-Keith, P., “Biosorption of
4. Kurniawan, T. A., Lo, W. H. & Chan, G. Y., Uranium and Lead by Streptomyces
“Physico –Chemical Treatment for Removal of longwoodensis”, Biotechnol. Bioeng. 28:, 1986, pp
Recalcitrant Contain from Landfill Leachate”, 21–28.
Journal of Hazardous materials,129(1), 2006, pp.80-
100.
5. Madu, J. I., Water and Environmental

Engineering, Department of Chemical
Engineering, Lund University, Sweden, Master
Thesis number: 2008-02
6. Dayanthi, W. K. C. N, Nanayakkara B. S. R.,

Kawamoto K., Premathilaka M.D.B.A. & Dulvin
O.V.N., “Field-Scale Study on Treating Landfill
Leachate Using Anaerobic Filters Packed with
ENGINEER 8
594
AnnualSessions
Annual Sessionsofof IESL,
IESL, pp.pp. [1 - 603],
[595 10], 2017
2017
© The
© TheInstitution
Institutionofof Engineers,
Engineers, Sri Sri Lanka
Lanka
Low-Cost drinking Water Filter

U.A.C. Perera and S.C.S. Karunaratne
Abstract: Sri Lanka could be identified as more fortuitous land owning about 85 % of its
drinking water in the safe zone. But still, in most parts of the country, people are battled against water
borne diseases since they don’t have access to safe drinking water. Hence need of the society is to
have a drinking water filter with the ability to outflow clean potable water at domestic level. Reviews
on existing filter systems in Sri Lanka and contamination levels different physicochemical and
bacteriological parameters were identified in ground water sources. The filter was designed to
fabricate with filter materials such as clay, activated carbon and two different types of ion exchange
resins to remove contaminants, treat total harness and improve physical parameters. The effectiveness
of the filter was analysed for reducing and treating various contaminants and physical parameters in
raw water under provided Sri Lankan Standards. Results showed that designed filter is capable of
treating hardness, removal of Nitrogen based contaminants and improving physical parameters. But
it was found that used filter materials were not capable of treating total alkalinity and removing
bacteria such as Coliform and E-coli. Through the cost analysis, it was rated that fabricated filter
could be manufactured at a unit price of LKR 6030.00 whereas most of the existing filters in Sri Lanka
cost more than LKR 10,000.00 per unit with some having a very high maintenance cost.
Keywords: Low cost, drinking water, filtration
1 Introduction the capabilities of the filter. For that, both

artificial and natural materials were used in
In the present people are being faced the issue
filter and tested for their abilities in treating
of not having safe drinking water with
parameters aforementioned. While ground
adequate quality for the consumption. In
water samples were collected from different
certain situations, people have to pay for their
areas in the western province and they were
water, and it too has limitations. Reasons for
tested before and after filtration through the
unavailability of water have been identified as
designed filter for selected parameters such as
the presence of contaminants such as nitrates,
pH, conductivity, level of several contaminants
nitrites, chlorides, fluorides, hardness due to
and hardness etc. Detailed information in each
calcium and magnesium, heavy metals and
step is given below. Also, two commercially
bacterial contaminants such as coliform and E-
available filters were used to compare results
coli [1]. Presence of this kind of contaminants
with designed filter
affects human body cells if they are consumed
in bulk. Furthermore, it has been identified
2.2 Filter Fabrication
that unavailability of safe drinking water is a 2.2.1 Clay core fabrication.
reason for diseases like Chronic kidney For the fabrication of clay cores, normal brick
diseases, tooth decay due to fluorides and clay was collected from Kaduwela, Ranala and
chlorides, diarrhoea and Cholera [2],[3]. Atigala area. Rice husk was collected from a
Although currently available filters are capable rice mill in Athurugiriya. They were
of treating and removing these contaminants thoroughly washed and dried under sunlight
to some extent, they are priced high and poor prior to fabrication. Initially, clay cores were
people face problems in acquiring one of them. planned to fabricate using clay and rice husk.
Therefore, the objective of the research was to Sample cores were prepared with following
design a Filter at a reasonable cost to remove clay to rice husk proportions given in Table 1.
contaminants such as nitrates, nitrites,
ammonia and treat hardness and capable of
Mr. U.A.C. Perera, B.Sc.Eng. Hons. (SLIIT), Temporary
treating parameters such as turbidity and Assistant Lecturer, Department of Civil Engineering,
colour. During this research scope of the study Faculty of Engineering, Sri Lanka Institute of Information
was based on qualities of ground water in the Technology.
western province.
Eng. (Dr.) (Ms.) S.C.S. Karunaratne, B.Sc.Eng. Hons.
2 Methodology. (Moratuwa), M.Eng. (Saitama), Ph.D. (Saitama), C.Eng.,
MIE(SL), MIEP(SL), Senior Lecturer, Department of Civil
2.1 Approach Engineering, Faculty of Engineering, Sri Lanka Institute of
The research was approached in a way that Information Technology.
filter was fabricated and tested with collected
and prepared water samples for identifying
1
595 ENGINEER
Clay cores were fabricated with dimensions of a surface area of 225 cm2 based on
150mm x 50mm x 50mm yielding surface area specifications in the mineral pot filter.
of 325 cm2 per each core, and four cores were SCR-B cation exchange resin - Trilite is a gel
used instead of one to increase the surface area type strongly acidic cation exchange resin
for filtration and avoid issues due to clogging which is capable of deionization of water,
during operation. These cores were burnt in a separation and recovery of metals, refining of
muffle furnace at 200 oC for 2 hours, then at chemicals, desiccation of organic solvents and
400 oC for 2 hours and another 4 hours at 900 catalysts. The resin was backwashed prior to
oC and these increments were followed to use in order to remove any resin fines and
avoid cracking of cores. small particles. Grade used for the research
was SCR- B and data sheet for the resin is
Table 1 - Mix proportions used for clay cores provided in Table 2. This resin was used at the
Sample Core Core Core Core Core second level of filtration chamber with same
1 2 3 4 5 quantity as activated carbon [4].
Clay % 100% 95% 90% 85% 80%
Rice 0 5% 10% 15% 20% Table 2 - Data sheet for Trilite SCR-B
husk % Type Gel type
Flow 19.5 19.2 20.6 20.0 19.8 Grade Trilite SCR-B
rate Matrix Polystyrene + DVB
(L/h) Functional group SO3-
Ionic form Na
2.2.2 Selection of materials for filter Specific gravity 1.29
fabrication Shipping weight 830
Drinking water parameters which are intended (g/l)
to improve during the filtration process of Moisture retention 43 – 50
water through expected design are as follows. Total capacity (eq/l) 2.0
 Turbidity. Effective size (mm) 0.40
 Colour. Uniformity Max. 1.6
 Chemical contaminants. coefficient
o Nitrates and nitrites Particle size (mm) 0.3 – 1.2
o Phosphates Operating Max 120
 Hardness. temperature
 Electrical conductivity (Celsius)
For the removal of these unpleasant Operating pH range 0 - 14
characteristics and contaminated chemical
compounds following mechanisms and Nitrogen selective ion exchange resin -
filtering, materials were used for the filter. At INDION NSSR is a microporous strongly basic
the first stage, cuboidal shaped clay cores were anion resin which is a ready-made resin used
used. Considering the disadvantages of filters to remove nitrate ions from water. Its proper
mentioned above, Intention was to fabricate mix of physicochemical properties gives ideal
the filter for the research project using a exchange kinetics of nitrate making this
combination of activated carbon, SCR-B ion suitable for removing nitrate ions in the
exchange resin and Nitrate selective ion presence of chloride ions. This resin was used
exchange resin. These were finalized based on as the first filter material in the filtration
their performance in filtration. Normal clay chamber in same quantity as other two filter
was used to cast the clay core used in the materials. Data sheet for the resin is given in
initial stage of filtration. The basis for the Table 3. sequence of filter material usage in the
selection of materials has been discussed filtration chamber is shown in Figure 1 [5].
below.
Table 3 - Data sheet for INDION NSSR
2.2.2.1 Filter materials Type Gel type
Activated carbon - Based on the capability of Appearance Opaque off white to
activated carbon to remove and treat brown beads
contaminants and physical properties in water Matrix Styrene DBC
it was decided to use Granular Activated Functional group Quaternary
Carbon of size 12 x 30 in the cuboidal cage as ammonium
the third level of filtration. 775 cm3 of activated
Ionic form Chloride
carbon was used in 30mm depth chamber with
ENGINEER 2
596
Shipping weight 670 Three of these chambers were planned to rest
kg/m3 on top of each other, and top most chamber
Moisture holding 45 – 55 % would be composed of four clay cores. Middle
capacity chamber would compose of filter materials,
Total capacity Min 0.9 and the bottom chamber has been designed to
(meq/ml) use for storage of filtered water. During
Effective size (mm) 0.40 filtration process water would be seeped
Uniformity Max. 1.7 through clay cores at a rate of 19.5 litres per
coefficient hour and at the second step water would be
Particle size (mm) 0.4 – 0.5 collected at cuboidal shaped storage
Operating Max 100 in Cl form compartment. Then Water would be seeped to
temperature filter materials evenly, and filtered water will
(Celsius) be collected in the bottom chamber with a
Operating pH range 0 - 14 volume of 20 litres at a rate of 18 litres per
hour.
2.2.4 Cost analysis
One of the prominent factor considered during
INDION NSSR For Nitrates, Nitrites
the design of this filter was the cost of
Anion resin and Free Ammonia
production of the filter. Hence detailed Cost
SCR – B Cation For Total hardness
breakdown of the filter has been calculated
resin
tabulated in Table 4. Based on above
Activated carbon For Chlorides, Bacteria
calculation it can be finalized that filter could
and Viruses
be fabricated at a rate of LKR 6030.00 per unit.
Figure 1 - Arrangement of filter materials in
filter chamber
Table 4 - Cost analysis chart
Material Quantity Unit cost Total
2.2.2.2 – Casing and adhesive.
cost
Casing material - Case was fabricated using
Acrylic 1.36 m2 3,000.00 4,080.00
clear acrylic sheets with 3mm thickness and
board for
laser trimming technology was used cut the
chamber
sheets and shapes in the design.
Activated 0.50 kg 340.00 170.00
Adhesive material - Cyanoacrylate
carbon
(Commonly termed as Super glue) was used as
the adhesive to glue acrylic sheets. SCR-B 0.50 litre 700.00 350.00
Cyanoacrylate is known as a non-toxic cation
adhesive in solid form although it shows toxic exchange
properties in its aqueous form [6]. resin
Nitrate 0.5 litre 220.00 110.00
selective
IER
Clay 2.4 kg 50.00 120.00
Adhesive 10 No 20.00 200.00
Water 1 150.00 150.00
dispenser
PVC 1 250.00 250.00
tubes
Plastic 4 25.00 100.00
Sieving
nets
Figure 2 - Side elevation and plan view of the Indirect - - 500.00
designed filter cost and
2.2.3 Fabrication method overheads
The filter was designed and planned to Total cost LKR
fabricate with three main chambers to store 6030.00
raw water, store filter materials and store
filtered water. For each chamber, dimensions 2.3 Sample collection and locations
were selected to make the capacity of cuboid Samples were collected from different areas in
as 20 litres approximately, and dimensions are the Western province according to the
shown in Figure 2 standards given in ISO 566-11 :2009 – Part II to
3
597 ENGINEER
test for different parameters and the Colour (Hazen units) 5 30
responsiveness of filter for those parameters Turbidity (NTU) 2 8
Locations are given in Table 5 [7].
Nitrate (NO3-) (mg/l) 10 45
Table 5 - Sample locations in Western Nitrite (NO2-) (mg/l) - 0.01
Province pH 7.0 – 6.5 –
No Location Latitude Longitude 8.5 9.0
(oN) (oE) Free Ammonia - 0.06
01 Godagama 6.8630639 80.031118 (NH3) (mg/l)
02 Maharagama 6.849661 79.923288
Total Hardness 250 600
03 Athurugiriya 6.891224 80.006153
(as CaCO3) (mg/l)
04 Malabe 6.906096 79.956835
05 Pokunuwita 6.708166 80.042156 Phosphates - 2.0
06 Kaduwela I 6.926053 80.001271 (PO43-) (mg/l)
07 Rajagiriya 6.920425 79.903393
08 Kadawatha 6.994726 79.646420 3 Results and Discussion
09 Kaduwela II 6.924859 79.999439 3.1 Selection of clay core
10 Horana 6.724919 80.093851 Rice husk percentages of 5 - 20 % were
selected for finally produced clay cores due to
2.4 Test methods failures in fabrication for higher percentages of
Collected samples were tested for following rice husk. Cores were tested for treating
parameters based on standard methods given turbidity and colour. Also, Flowrates were
in American Public Health Association tested for each core and had been tabulated in
(APHA, 2005). These test methods are given in Table 1. Based on the test results for flow rate
Table 6. [8]. in Figure 3 and Figure 4, no major variation
was identified between different proportions
Table 6 - Test methods used used in cores for treating turbidity and colour.
Therefore, clay cores were fabricated using
Parameter Test method used
clay as the only material.
Electrical APHA, 2510 B
Conductivity (S/cm)
Colour (Hazen units) APHA, 2120 C
Turbidity (NTU) APHA, 2130 B
Nitrate (NO3-) (mg/l) APHA, 4500 NO3 - E
Nitrite (NO2-) (mg/l) EPA 35.1
pH APHA, 4500 HB
Free Ammonia APHA 4500 NH3 -
(NH3) (mg/l) C
Total Hardness APHA, 2340 B
(as CaCO3) (mg/l)
Phosphates APHA 4500
(PO43-) (mg/l) PE:2012
Figure 3 - Variation of turbidity of water for
2.5 Ground water quality standards. clay cores
Sri Lankan Standards for potable water is
given in SLS 614: 2013 (SLS-Sri Lanka
Standards). Based on the cap values given in
for drinking water filter was tested whether it
could filter raw water to be within given
limits. Highest Desirable level (HDL) and
Maximum Permissible Level (MPL) for
selected parameters are shown in Table 7 [9].
Table 7 - SLS for drinking water

Parameter HDL MPL
Electrical Conductivity 0.75 3.5
(S/cm) Figure 4 - Variation of colour of water for clay
cores
ENGINEER 4
598
3.2 Comparison of results for prepared has shown capabilities of removing nitrites
solutions. with concentrations up to 0.3 mg/while
3.2.1 Nitrate variation Commercial filter 1 has shown its maximum
treatable concentration as 0.1 m/l. But
Commercial filter 2 has not shown any ability
in treating nitrates to permissible limit of 0.01
mg/l. Considering average nitrite removal
percentages given in
Table 8, it can be judged that Designed filter is

highly effective in treating Nitrite
contaminants in water.
3.2.3 Turbidity variation.

Considering the results given in Figure 7, it
can be seen that all the filters are capable of
treating turbidity of water to permissible limit
of 8 NTU. That is all the filters are capable of
Figure 5 - Variation of nitrate concentration
treating water with turbidity varying from 0 –
of prepared solution
150 NTU to the allowable limit. But it can be
Based on the results shown in Figure 5 for the seen that Commercial filter 1 and Commercial
prepared nitrate solutions using NaNO3 with filter 2 has shown slightly better results and
different concentrations given in the graph, reason for this can be identified as fabrication
identification can be made that designed filter errors in clay cores produced in designed
has optimum ability to remove nitrate filter. Percentages of treating turbidity are
contaminants in raw water where filter has given in Table 8.
shown capabilities to treat nitrate .
concentrations up to 500mg/l which is
normally not found in drinking water sources.
While Commercial filter 1 has shown a closer
trend but not as effective as the designed filter.
Commercial filter 2 has not shown the ability
to treat higher level of nitrates which are
greater than the permissible limit. Removal
effectiveness is shown in
Table 8. Hence, the judgement could be made

such that designed filter is highly effective
against Nitrates.
Figure 7 - Variation of Turbidity of prepared
3.2.2 Nitrite variation solution
3.2.4 Colour variation.
Figure 6 - Variation of nitrite concentration in

prepared samples
Figure 8 - Variation of colour for prepared
According to Figure 6, similar pattern can be
solution
identified as in nitrates where designed filter
1.5
Raw 5
599 ENGINEER
1 water
y (NTU)
Core 1
As per Figure 8, a similar pattern can be for this increment can be identified as impact
identified for treatment of colour and as in due to the increment of Total alkalinity after
turbidity removal Commercial filter 1 and filtration. Cause for total alkalinity increment
Commercial filter 2 have shown better could be the use of ion exchange resins and
effectiveness. It could be identified that similar impact due to the exchange of ions to remove
reason as in turbidity has affected that. From nitrates and nitrites. From these results, it is
Figure 8, it can be seen that all three filters clear that designed filter and Commercial filter
tested are capable treating colour with 150 1 is almost similarly capable of adjusting the
Pt/Co to permissible limit of 30 Pt/Co and in pH value of water. That is maximum and
most cases below the desirable limit of 5 minimum adjustment identified through this
Pt/Co. Average treatment percentages are filter was an increment of pH value from 5.22
shown in Table 8. to 5.86 and 5.85 to 6.11.
But during the experiments, only these tests
were performed for prepared solutions due to 3.3.2 Turbidity variation.
constraints in time and resources. Also, these From Figure 10, the ability of each filter to treat
results for Commercial filter 1 and Commercial turbidity in raw water has been shown. It is
filter 2 might have slight variations since old clear that considerable percentage of turbidity
and modelled filters of them were used for has been treated with each type of filter.
tests. Highest removal percentage was identified in
Average capabilities of designed filter and Commercial filter 1 and Commercial filter 2.
selected two commercial filters to The maximum level of treatment can be
remove/treat parameters nitrates, nitrites, identified for the sample 9, where turbidity
turbidity and colour for prepared solutions are level has reduced to 1.65 NTU from an initial
given in Table 8. value of 4.37 NTU. It could be finalized that
. designed filter has adequate capacity to treat
turbidity. Also, it is understandable that for all
Table 8 – Percentages of average capabilities samples repetitive pattern has been shown.
to remove/ treatment for prepared solution
Filter Designe Comm Commerci
d filter ercial al filter 2
filter 1
Nitrate 93.14 % 80.58 % 11.84 %
Nitrite 94.54 % 70.00 % 2.00 %
Turbidity 92.41 % 96.80 % 95.5 %
Colour 94.93 % 96.93 % 96.92 %
3.3 Comparison of test results for raw

water and filtered water.
3.3.1 pH value variation Figure 10 - Variation of turbidity
3.3.3 Electrical conductivity variation
Figure 9 - Variation of pH value

Based on the test results shown in Figure 9 it
Figure 11 - Variation of Electrical
can be identified that pH value of the water
conductivity
samples has approached the provided limits
for suitable drinking water. A possible cause
ENGINEER 6
600
According to Figure 11, it is clear that for all As per Figure 13, the decision could be made
the filter types tested for filtration of raw water that fabricated filter can remove nitrates in the
samples, a considerable change has not been raw water. Possible reason in the filter for this
identified for the electrical conductivity outcome can be identified as use of nitrogen
measurement. Where a maximum change of selective resin in the filtration process. Also,
electrical conductance identified was positive one of the main objectives of the filter design
0.035 for sample location 6 where conductance was to reduce nitrate contaminants in water.
has increased from 0.226 ms/cm to 0.261 Compared to other filters contrast can be seen
ms/cm. Therefore, a conclusion could be made in the ability of nitrate removal from the
that designed filter has a very minimal impact designed filter. In numerical terms for the
on the change in conductance of raw water designed filter, nitrate concentration has been
when filtered. diminished to 4.22 mg/l from its initial value
of 10.5 mg/l for sample 3. While minimum
3.3.4 Colour variation removal was identified for sample 5 where
concentration has been reduced to 0.96 mg/l
from the initial value of 1.3 mg/l.
3.3.6 Nitrite variation

Based on Figure 14, it can be understood that
nitrite presence has been reduced when water
is filtered through a designed filter and
Commercial filter 1. But no considerable
reduction has been detected in nitrite
contaminants when water is filtered through
Commercial filter 2. That is a maximum
reduction of nitrite concentration was 0.007
Figure 12 - Variation of colour intensity mg/l for sample location 2 and reduction were
observed from 0.015 mg/l to 0.008 mg/l for
the designed filter. But more precision was
From Figure 12, decision could be made such observed for Commercial filter 1 where
that intensity of colour measured in Pt/Co has concentration has been reduced to 0.006 mg/l.
been reduced when filtered through designed The main cause for reduction of nitrite when
filter and possible reason for this reduction can filtered through the designed filter can be
be identified as use of clay core in the initial again identified as the presence of nitrogen
step of filtration and the use of activated selective resin as a filter material.
carbon as a filter material in the final step of
filtration. From this figure, it can be
understood that filtration ability of designed
filter is almost similar to other filters used in
the comparison. Maximum colour adjustment
identified was 8 Pt/Co units. This reduction of
colour intensity from 12 Pt/Co to average of
about 4 Pt/Co was identified for sample
location 9.
3.3.5 Nitrate variation
Figure 14 - Variation of nitrite contamination
Figure 13 - Variation of nitrate contamination
7
601 ENGINEER
3.3.7 Phosphates variation
Figure 16 - Variation of total hardness

Figure 15 - Variation of phosphates
3.3.9 Free ammonia variation
According to Figure 15, a reduction in Based on Figure 17, it can be seen that free
phosphate concentration can be seen when ammonia content of the filtered water has been
raw water is filtered through the designed reduced in some samples and some samples
filter. But the conclusive decision could not be have not undergone considerable reduction.
made on which material caused this reduction. But maximum reduction was identified for
Maximum concentration reduction from 0.09 sample 7, and it was 1.6 mg/l to 0.17 mg/l
mg/l to 0.05 mg/l was observed for the while a minimum of zero reduction was read
sample collected from location 1 and minimum for sample location 3. While maximum
concentration reduction observed for sample 8 increment in concentration was observed for
where no reduction was identified. The sample 5 and it was 0.01 mg/l to 0.27 mg/l. A
average percentage of reduction of phosphates potential cause of the reduction could be
is shown below. Based on samples 1,6 and 2 it identified as use of nitrogen selective ion
can be said that reduction capacity of filter for exchange resin in the filter. But totally clear
phosphates is considerably adequate although judgement could not be made on the ability of
the same pattern has not been noticed for other the designed filter to remove free ammonia
sample locations. from raw water due to scenarios discussed
above. But derivation could be made that filter
3.3.8 Total hardness variation can remove them.
From the chart given in Figure 16 decision
could be made such that total hardness in raw
water has been treated through the filtration
process and possible cause for this reduction
could be marked as the ability of SCR-B cation
exchange resin to soften water. Numerically
observation is that maximum reduction in
hardness was from 85 mg/l to 12mg/l for
sample location 7 while the minimum
reduction was observed for sample 5 while
increment was detected for sample location 1
where hardness has been increased to 18mg/l
from 12mg/l. Hence it can be decided that one Figure 17 - Variation of free ammonia
of the primary objective of the design to treat concentration
Hardness has been achieved during the
process.
ENGINEER 8
602
4. Conclusions which use Reverse Osmosis as its principle for
Based on the Results obtained for the test filtration are priced around LKR 20,000 to
procedures carried out conclusion could be 30,000 in the market. Therefore, considering
made in a way that designed and fabricated both variables cost and performance of the
filter can remove nitrates, nitrites and free designed filter which is about LKR 6,000 in
ammonia and treat total hardness, colour and price could be nominated as a low cost and
turbidity prominently based on test results for good in performance filter to be used in the
prepared solutions with above contaminants domestic level.
and physical parameters. Summary of
effectiveness of the designed filter in treating Acknowledgement
these are shown in Table 9.
We would like to acknowledge the financial
Table 9 - Contaminant removal effectiveness
support given by Sri Lanka Institute of
of designed filter
Information Technology (SLIIT) to carry out
Paramet Nitra Nitrite Turbi Colour this study successfully.
er tes s dity
Removal 93.14 94.54 92.41 94.93 References
effective
ness (%) 1. Dissanayake, C. and Weerasooriya, S. The
Hydrogeochemical atlas of Sri Lanka.
Peradeniya: Natural Resources Energy and
while filter has shown capabilities in removing
Science Authority of Sri Lanka, 1985.
certain other parameters such as phosphates to
a certain extent. But it can be concluded that 2. Dissanayake, C. B. ―The fluoride problem in
filter is not capable of treating total alkalinity, the ground water of Sri Lanka —
and the filter has not shown ability in environmental management and health,‖
removing chlorides contaminated in water. Int. J. Environ. Stud., vol. 38, no. 2–3, pp.
Based on the results obtained conclusion can 137–155, Sep. 1991.
be made that bacteriological parameters such
3. Dissanayake, C. B. ―Water quality and
as Coliform bacteria and E-Coli would not be
dental health in the Dry Zone of Sri Lanka,‖
treated at a considerable level although
Geol. Soc. London, Spec. Publ., vol. 113, no. 1,
reduction has been notified for several pp. 131–140, 1996.
samples. Considering the results, Conclusion
can be made on removal percentage of 4. Iontec, ―Trilite SCR-B,‖ 2006. [Online].
contaminants present in raw water. Available:
contaminants reduction percentages are shown http://www.iontec.co.kr/pds_files/130701.
in Table 10. pdf. [Accessed: 30-Nov-2016].
Table 10 - Summary of treatment level
5. Ion resins, ―Ion resins.‖ [Online]. Available:
Contaminant Percentage (%)
http://www.ionresins.com/pds/NSSR
01 Nitrates 55.09 Brochure.pdf. [Accessed: 01-Dec-2016].
02 Nitrites 38.63
03 Total Hardness 63.54 6. Soehendra, N. Grimm, H. Nam, V. C. and
04 pH adjustment 8.44 Berger, B. ―N-Butyl-2-Cyanoacrylate: A
05 Colour correction 43.14 Supplement to Endoscopic Sclerotherapy,‖
06 Phosphates 44.19 Endoscopy, vol. 19, no. 6, pp. 221–224, Nov.
1987.
07 Turbidity 60.67
removal 7. Google, ―Google maps,‖ 2016. [Online].
Available: https://www.google.lk/maps.
Considering the cost of the existing filters, [Accessed: 30-Nov-2016].
commercial filter 1 analysed earlier for its
performance is priced at about LKR 10,000 to 8. Greenberg, A. E. and A. D. (eds) Eaton,
12,000 in the market. Therefore, it is ―Standard Methods for the Examination of
considered as expensive filter although its Water and Wastewater,‖ 1999.
performance is high in treating water.
9. SLS, ―Drinking water standards - First
Commercial filter 2 is priced at LKR 4,000 – Revision (Sri Lanka Standards for potable
6,000 in the market and it was found that for water – SLS 614: 2013),‖ 2013.
most parameters it is not good as a designed
filter in filtration. Also, most of the filters
9
603 ENGINEER
604
pp. [1 -- 10],
[605 615],2017
2017
© The
© The Institution
Engineers,SriSri
Lanka
Lanka
Monitoring Carbon Monoxide (CO) Levels at Five

Major Schools in Kandy City using Real Time Air
Quality Monitoring Sensor Network
R.G.R.N.K. Ariyaratne, D.G.G.P. Karunaratne, M.A. Elangasinghe, A. Manipura,
S.P. Kodithuwakku, A.D. Siribaddana and K.H.N. Abayalath
Abstract: Ambient air quality in Sri Lanka is rapidly deteriorating due to an addition of a large
number of motor vehicles, traffic congestions, combustion of fossil fuel and poor quality of fossil fuels.
Previous researchers have found that Kandy city has a higher degree of ambient air pollution owing
to its geographical location and heavy traffic congestions. However, lack of continuous monitoring
has hindered assessing the impact of air quality on human health in Sri Lanka. In recent years, low-
cost yet reliable air quality monitoring solutions have been developed using open-source hardware
systems. A variety of such electrochemical gas sensors is available for monitoring pollutants such as
NOx, CO, CO2, O3, SO2 and H2S at parts per billion levels (ppb). Six air quality monitoring units using
such electrochemical gas sensors measuring CO, NO, NO2, SO2, O3, temperature and humidity were
developed and setup in Kandy city at selected locations. Emission data from these stations are being
acquired on real time basis using the web based data acquisition system. In this paper we present the
data of CO levels obtained for the months of March and April, 2017 at five different schools in Kandy
city. CO sensors were calibrated to identify the effects of temperature and their linearity. They have
shown 5-7% increase of sensor response with temperature and they were highly linearly correlated
(R2>0.9816) between 0-10 ppm CO concentrations. When comparing the average CO levels at different
schools Mahamaya College has shown the highest 24 hour average of 2.05 ppm and Girls High School
has shown the least value of 0.165 ppm in the above months.
1. Introduction plants, and the poor

quality of fossil fuels [4]. It is observed
Not like many other environmental problems, that Kandy city has a higher degree of ambient
air pollution is directly related to human health
even at low doses, and it is the single largest
Eng. R.G.R.N.K Ariyaratne, B.Sc. Eng. (Peradeniya),
environmental risk to health globally [1].
AMIESL, Research Assistant, Department of Chemical and
According to the World health organization Process Engineering, University of Peradeniya.
(WHO), more than 6 million premature deaths
Eng. (Dr.) D.G.G.P Karunaratne, B.Sc. Eng. (Peradeniya),
were caused due to air pollution exposure in P.hD. (Nova), Senior Lecturer, Department of Chemical and
2012 [2]. Health impacts of air pollution result Process Engineering, University of Peradeniya.
in significant financial and economic
Eng. (Dr.) M.A. Elangasinghe, B.Sc. Eng. (Peradeniya),
consequences, not only in terms of societal costs M.Phil (Peradeniya), P.hD. (Aukland), Senior Lecturer,
of mortality and morbidity but also in terms of Department of Chemical and Process Engineering, University
hospital and public budgets [3]. So, maintaining of Peradeniya.
healthy ambient air quality levels is essential Eng. (Dr.) A. Manipura, B.Sc. Eng. (Peradeniya), M.Eng.
for the well-being of the general community. (Moratuwa), P.hD. (Rhodes), Senior Lecturer, Department of
However, the increasing urbanization, Chemical and Process Engineering, University of Peradeniya.
industrialization and rapid growth of vehicle Dr. S.P. Kodithuwakku, B.Sc. Agri. (Peradeniya), P.hD. (Hong
usage have deteriorated the urban ambient air Kong), Senior Lecturer, Department of Animal Science,
University of Peradeniya.
quality all over the world.
Dr. A.D. Siribaddana, MBBS (Peradeniya), Consultant
Ambient air quality in Sri Lanka is rapidly Respiratory Physician, Teaching Hospital, Kandy
deteriorating due to the addition of large
number Mr. K.H.N. Abayalath, Research Assistant, Department of
Animal Science, University of Peradeniya.
of motor vehicles, traffic congestion,
combustion of fossil fuel in thermal power
605
air pollution owing to its geographic location stationary (point), mobile (line) and area sources.
and heavy traffic congestions [4]. Moreover, in In urban environments, air quality is mainly
Kandy, almost all schools, hospitals, and other deteriorated due to the traffic related air
public working places are situated nearby the pollution. In Colombo, about 60% of the total
main roads increasing the public exposure to air emissions are due to vehicular emissions [7].
pollution for longer time periods. So, taking Since there are no large industries that release
immediate actions to reduce air pollution and gaseous or particulate matter (PM) in Kandy city,
setting up an air quality management plan for it can be assumed that vehicular emissions are the
Kandy is a timely necessity. primary cause of air pollution in Kandy too.
Monitoring air quality and the establishment of Air pollutants can be primary or secondary.
air quality inventory database for the Kandy city Primary pollutants are released into the
is the first step of implementing an air quality atmosphere from ground sources and secondary
management plan. Due to the complexities of the pollutants are formed within the atmosphere
behaviour of air pollutants, emission sources, due to the reactions of primary pollutants with
topographical and meteorological conditions, an natural components of atmosphere. Though
air quality monitoring system should have the there are many such air pollutants, the most
ability to monitor pollutants with broad spatial significant pollutants are particle pollution (often
and temporal variability with having high referred to as particulate matter (PM)), ground-
accuracy of monitored data (at parts per billion level ozone (O3), carbon monoxide (CO), Sulphur
(ppb) levels), low-cost of operation, low power oxides (SOx), nitrogen oxides (NOx), Hydrogen
consumption, capability for mobility and Sulphide (H2S), heavy metals (e.g., lead (Pb)),
selectivity for multiple gases [5]. But, when volatile organic compounds (VOCs) and
compared to traditional air quality instruments polycyclic aromatic hydrocarbons (PAHs) which
they do not meet the above requirements except are found in the ambient air [7].
the higher accuracy.
NO2 is identified as a pollutant which can affect
Nowadays throughout the world, research the mortality rates. Both NO2 and CO have the
studies are being done to introduce low-cost, yet capability to act as respiratory sensitizers. Long
reliable real time air quality monitoring systems. term exposure to NO2 can lead to having adverse
Use of electrochemical gas sensors is found to be effects on lung function. CO can reduce the
a more reliable solution for high-cost air quality body's capacity to transport oxygen to cells [6]. O3
monitoring with having the ability to monitor is formed during the periods of sunny weather
pollutants at ppb levels [6]. In this research study, conditions by the reaction of pollutants such as
we have developed such electrochemical air NOx and VOCs with the sunlight. Ground level
quality monitoring sensor network for Kandy O3 can cause breathing problems, lung function
city. Initially, data are being collected on real time reduction, and lung diseases. SO2 can also affect
basis at six different locations in Kandy city. This the respiratory system and function of the lungs
paper presents preliminary results of this study and causes the irritation of the eyes. Exposure to
with the measures to be taken to improve the PM increases the risk of respiratory infections,
accuracy and reliability of data for future long- cardiovascular diseases and lung cancers [1]. In
term monitoring. most of the developing countries, PM
concentrations are higher at both indoors and
outdoors [8].
2. Literature review Deterioration of the ambient air quality in Sri
Lanka has made significant health impacts to the
2.1 Air pollution and its impacts
people of the country. Respiratory related
Ambient air pollution occurs due to a wide diseases are the most significant issue of the air
variety of sources, which are broadly classified as pollution. In Sri Lanka since 1995, respiratory
606
system related diseases have ranked as the overcome above serious health and economic
second consequences.
highest cause of hospitalization, and it is among
2.2 Traditional air quality monitoring
the first five leading causes of death in most of
methods
the
age groups in Sri Lanka [9]. And also asthma has Air quality monitoring allows us to establish the
become a major respiratory disease in Sri Lanka status of air quality in a particular area, which is
according to the statistical data of hospitalization the preliminary step in setting up air quality
and hospital deaths [10]. Cancer has also become management plans. At present air quality data
one of the main health issues in Sri Lanka, are being collected by environmental or
especially in Kandy [11]. Polycyclic aromatic government authorities using fixed or mobile air
hydrocarbons (PAHs), which can be found at quality monitoring stations. These monitoring
very high concentrations in Kandy is one of the stations use analytical instruments such as optical
major organic pollutants in the atmosphere. This and chemical analysers, gas chromatographs and
can be the prime etiological cause for lung mass spectrometers [5]. Traditional monitoring
cancers too [11]. methods are expensive to operate and maintain.
They are complex in nature in terms design,
The economic costs related to these adverse
operation and maintenance, mostly stationary
health effects are very high. The potential
and consume more power and resources [13]. In
financial loss due to outdoor air pollution in
the UK, setting up of these fixed air quality
Colombo owing to PM2.5 has been estimated to
monitoring stations could cost between £ 5000 to
be about Rs. 886 million, and due to PM10, it was
Table 1 - Comparison of the characteristics of different low-cost sensor technologies available

Analyzer Sensor technology Range Precision Response time (s) Price (USD)
CO
Alphasense B4 Electrochemical 0-1000 ppm 0.004 ppm <25 100
TECO model 48c IR absorption 0.3-3000 ppm 0.1 ppm 60 12K
Aeroqual Series 500 Metal oxide semiconductor 2-100 ppm 0.1 ppm <150 1.5K
Membrapor CF-200 Electrochemical 0-200 ppm 0.1 ppm <40 280
CiTicel 7E/F Electrochemical 0-1000 ppm 0.5 ppm <30 140
NO2
API model 200EM chemiluminescence 0.05 -1 ppm 0.05 ppm 20 8-14K
Aeroqual Series500 Metal oxide semiconductor 0.01-200 ppm 0.001 ppm <180 2K
Membrapor C-1 Electrochemical 0-1 ppm 0.05 ppm <60 320
CiTicel 7NDH Electrochemical 0-20 ppm 0.1 ppm <40 225
NO
Citicel 7NT Electrochemical 0-100 ppm 0.5 ppm <15 200
Unloaded WO3 2-300 ppm 2 ppm <160 N/A
O3
API Model 400 UV absorption 0.6-200 ppb 0.001 ppm <20 8-14K
2B technologies, 202 UV absorption 1.5 -200 ppb 0.0001 ppm 10 5K
Aeroqual Series500 Metal oxide semiconductor 8-50 ppm 0.001 ppm <60 2K
OMC - 1108 Electrochemical 0.01-100 ppm 0.01 ppm <70 1.2K
SO2
Envirocel A3Oz Electrochemical 0-10 ppm 0.02 ppm <40 500
Unloaded SnO2 5-100 ppm 5 ppm <35 N/A
about Rs. 30 billion per annum in 2003 [12]. Thus, £ 60000 (Reference). Though these have proven
air pollution mitigation plans are required to higher accuracy and precision of the
measurements, above drawbacks associated with
607
these, have hindered the capability to use them in four electrodes namely a working electrode (WE),
networks with higher densities [6]. Currently, in a reference electrode, a counter electrode and an
Sri Lanka only one such air quality monitoring auxiliary electrode (AE). These electrodes are
station is located in Colombo Fort [7]. So, with the separated by hydrophilic wetting filters, which
traditional air quality monitoring methods it is allow the electrolyte to transport via capillary
difficult to understand the behaviour of air action and keep the ionic contact in between
pollutants in the atmosphere. Therefore, there is electrodes [6].
an urgent need to introduce air quality
Measuring gas that comes in to contact with the
monitoring methods with flexible and affordable
sensor diffuses through a gas permeable
techniques having the capability to use them for
membrane to reach the WE surface. The WE
both scientific and legislative purposes with the
either oxidizes or reduces generating an
aim to improve the understanding of health
electronic charge which is balanced by the
impacts of air pollution.
reaction at the counter electrode, forming a redox
2.3 Low-cost air quality monitoring networks pair of reactions. These reactions are catalysed by
the electrode materials specifically developed for
Due to the different types of emission sources
the gas of interest. Current output by the sensor
and atmospheric flow conditions, air quality
is directly proportional to the concentration of the
varies over a relatively small-scale environment
measuring gas. During the operation, the
and with temporal scale resulting the high level
WE is kept at a fixed potential while the counter
of unpredictability. To predict the behaviour of
electrode potential is allowed to change. Working
air flow masses sophisticated air quality
electrode potential is kept at a defined value
modelling tools and real time high-resolution air
during the operation using the third reference
pollution concentration maps are required for
electrode. Three-electrode sensor design means
covering a whole urban area [5]. Hence, to fulfil
that the sensor sensitivity is stable
the above requirements, low-cost air quality
and operates linearly within the range of use.
monitoring methods with remote data acquisition
To measure the effect of temperature and
capabilities are being developed around the
background current arising from solid electrode
world nowadays [5]. Gas sensor devices tied with
processes or from electrochemistry involving the
the advances in computing and communication
electrolyte other than the gas of interest the AE is
have provided the ability to monitor air
used. This AE current could then be subtracted
pollutants with high spatial and temporal
from the WE total current to give a corrected WE
resolutions having relatively very low costs. Most
current corresponding solely to the
widely used sensor types are based on two major
electrochemical reaction of the sampled gas.
principles: (1) those depend on interactions with
However, the sensing electrode potential does
sensing material and detecting gas (e.g.,- metal
not remain constant due to the continuous
oxide semiconducting sensors, electrochemical
electrochemical reaction taking place on the
sensors) and (2) those that measure the
surface of the electrode causing a deterioration
absorption of light (e.g.,- non-dispersive infrared
of the performance of the sensor over an
sensors, chemiluminescence) [13]. Comparison of
extended period of time [6]. With the use of
these different low-cost air quality monitoring
correction algorithms, these errors can be
methods is given in Table 02.
minimized. Sensors for this research were
2.4 Electrochemical sensors purchased from Alphasense (Pvt) Ltd. of United
Kingdom, which is the largest manufacturer of
Use of electrochemical sensors in ambient air electrochemical air quality monitoring sensors.
quality monitoring is widely used around the
world due to its low cost and capability to
measure various air pollutants at parts per
billion (ppb) levels with the use of low noise
circuitry [5]. An electrochemical sensor consists of
608
3. Methods 3.2 Calibration of sensors
3.1 Sensor unit design and data acquisition Issues concerned with the application of
method electrochemical sensors include: calibration of the
sensor sensitivity to the target gas, the drift of
Six sensor units were made with each including the sensor sensitivity, cross sensitivities within
CO, NO2, NO, SO2 and O3 pollutant sensors and complex gas mixtures, the effect of temperature,
DHT 11 temperature and humidity sensors. and humidity on sensor response and
Sensor unit enclosure was designed to operate at competition between the various gases at the
extreme weather conditions, and they were active site of the detector which reduces the
placed about 1.75m above the ground. Power sensitivity [14]. These factors need to be
supply for the sensor units was obtained using evaluated in detail for accurate detection of gas
AC power supply from nearby. concentrations and to characterize the emissions.
All the sensors purchased were calibrated

initially by the sensor manufacturer. For each
sensor, they have given their calibration data
done at 23 0C. But the percentage change of
sensor sensitivities with the temperature was
given as a typical value for all sensors. And the
effect of humidity on the sensors was not given.
To identify two parameters our in-house
developed calibration procedure was conducted.
Those calibration parameters were 1). Effect of
temperature on the sensor sensitivity and 2).
Linearity of the sensor responses within the
range.
Figure 1 - Sensor unit placed at Faculty of
Engineering, University of Peradeniya Sensors were tested using different gas
concentrations inside a gas tight chamber.
Data acquisition was made using an open source Standard calibration gases used were obtained
hardware unit called Arduino Mega 2560. The from GASCO, LTD, USA. The gas tight chamber
acquired data were then sent to the web based was fabricated using rubber enclosure with the
server created using MySQL and PHP ability to change the volume and to create
(kandyair.eu5.org) via a GPRS module. Until the vacuum condition. Six gas sensors and DHT 11
sensor characteristics were clearly identified data sensors were inserted into the chamber using a
were downloaded, and further processing was wooden frame. Initially, the chamber was
done manually to obtain the concentration vacuumed using a pump and filled with zero
values. grade air. Then different pollutant gas volumes
were injected using syringes to the chamber filled
with zero air to create step increases in the gas
concentrations, and the respective sensor
responses were obtained subsequently.
Temperature effects were also tested changing
the temperature of the chamber filled with zero
grade air. And humidity effects were tested using
a dehumidifier by changing the humidity of the
exposing environment.
Figure 2- Schematic of the sensor unit and data
acquisition method Since the CO levels at the ambient conditions
vary between 0.2-50 ppm and CO sensor cross
609
interferences are less compared to the other of the people.
sensors CO concentration values were finalised
36 monitoring locations were identified to obtain
using the calibration data and data given by the
a good spatial distribution of monitoring
sensor manufacturer. According to the
locations. However, to cover all these locations
manufacturer data sheets, CO sensors are affected
using only six sensor units monitoring will be
by having cross sensitivity percentages of < 1 to
taken for about two months at each location up to
NO2, <0.1 to SO2 and < -3 to NO. Since the NO,
a period of one year. As our initial monitoring
NO2, O3, and SO2 concentrations are at ppb levels
locations, we have selected five schools and
(<100 ppb) their cross sensitivities are significant.
Faculty of Engineering. These initial locations
Thus, those data are yet to be finalised comparing
were selected considering the index values and
with the high accurate air quality monitoring
safety, power supply and accessibility for the
station. The results present here are of those CO
sensor units.
sensors only.
3.3 Monitoring locations and their 4. Results and discussion

characteristics 4.1 Calibration results of CO sensors
Kandy town is about 465 m above the sea level  Linearity of the sensors responses
and has an area of about 26 km2. The city area is
surrounded mainly by Hanthana, Bahirawa, According to the Figure 4, all the six sensors were
Kurusa, and Dharmaraja mountain ranges and its highly correlated linearly (R2>0.9816) with gas
condition is like a bottom of a basin. The concentrations varying from 0-10 ppm. Data were
residential population of the city area is about obtained at 27C0 and 60-65% humidity range.
120,000 and about 100,000 people arrive the city Therefore, the linear relationship between sensor
for daily activities. Daily vehicle arrivals are response and gas concentrations was verified,
about 100,000 approximately causing heavy and the manufacturer's data and linear equations
traffic congestions, which eventually leads to were used to convert sensor responses into
heavy air pollution in the city area [15]. corresponding concentration values.
Figure 4 - Linear correlation plots of CO sensor
To identify the importance of different places in responses with calibration gas mixing ratios
Kandy city for monitoring air quality city area is
 Effect of temperature
gridded according to figure 3 and index was
created by considering the emission sources that It can clearly be seen (Figure 5) that sensitivities
are the road network and traffic conditions, of the sensors have shown a linear (R2>0.953)
receptors that are the locations where people are increment with the temperature. Sensitivity
mostly engaged in their work and the age limits varied between 5 - 7% with the temperature.
Figure 5- Variation of CO sensor responses with

temperature
Figure 3- Gridded map of the Kandy city along with

the developed guide map to select optimal
monitoring locations on the left side. Five of the
initial monitoring sites are shown.
610
According to the manufacturer's data sheets, the Figure 6 shows sample data obtained at 3
average change of sensitivity for a CO sensor is minutes intervals during a week at Kingswood
6% with the temperature. The results obtained College, Kandy. Location of the sensor unit was
from the calibration has shown similar values. about 15m away and about 8m above from the
main road. There were no obstacles between the
Further significant changes of sensor responses
sensor unit and road. So vehicle density at the
due to humidity changes could not be identified.
road directly affected the pollutant levels near the
4.2 Variation of CO concentrations at sensor unit. Especially due to the variations of
different locations the wind fields around the sensor units
significant variations of air quality data could be
Following equation was used to convert the observed. So when averaging the air quality data,
sensor responses into concentration values. having data with high temporal resolution is
....(1) essential.
Of the equation 1, C is the concentration (ppb),

WEu is the uncorrected WE output voltage (mV)
which is the measured value, WE0 is the WE
output voltage (mV) in zero air. n is a factor
calculated for temperature corrections using
temperature values and manufacturer’s data, AEu
and AE0 are respective AE output voltages (mV),
and S is the sensitivity (mV/ppb). n is a function
of temperature and calculated using the
temperature measurements. S values were
corrected with the calibrated data and
temperature measurements.
According to Sri Lanka national air quality

standards 1 hour, 8 hour and anytime values are
26 ppm, 9 ppm, and 50 ppm respectively. Sensor
units were programmed to obtain data at every 3
minutes intervals so that high temporal variation
of the data could be obtained.
Real time variations of CO and temperature at Kingswood College, Kandy

3/17 Friday 3/18 Saturday 3/19 Sunday 3/20 Monday 3/21 Tuesday 3/22 Wednesday 3/23 Thursday Temperature
4500 36
33
4000
30
3500
27
Concentration (ppb)
3000 24
Temperature
21
2500
18
2000
15
1500 12
9
1000
6
500
3
0 0
18:53
0:01
0:28
0:56
1:23
1:50
2:17
2:52
3:19
3:46
4:14
11:33
4:41
5:08
5:35
6:03
6:30
6:57
7:24
7:52
8:19
8:47
9:13
9:44
10:12
10:39
11:06
12:01
12:28
12:59
13:26
13:53
14:20
14:48
15:15
15:43
16:10
16:37
17:04
17:32
17:59
18:26
19:21
19:48
20:15
20:43
21:10
21:37
22:04
22:32
22:59
23:26
23:53
Time
Figure 6 – Sample real time data of CO and temperature values at Kingswood College,
Kandy obtained for a week period
611
VARIATION OF HOURLY CO AVERAGES IN MARCH AND APRIL
Total (March) Weekday (March) Weekend (March) Total (April) Weekday (April) Weekend (April)
1200
1000
CONENTRATION (PPB)
800 Kingswood College
600
400
200
0
1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00 0:00
1000
900
800
700
CONENTRATION (PPB)
Trinity College
600
500
400
300
200
100
0
1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00 0:00
800
700
600
CONENTRATION (PPB)
Girls High School

500
400
300
200
100
0
1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00 0:00
1800
1600
Pushpadhana College
1400
CONENTRATION (PPB)
1200
1000
800
600
400
200
0
1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00 0:00
3000
2800
Mahamaya College
2600
CONENTRATION (PPB)
2400
2200
2000
1800
1600
1400
1200
1000
1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00 0:00
TIME
Figure 7 – Hourly averages of CO levels in the months of March and April, 2017 at five
different schools in Kandy city
612
Table 2 - Sensor locations, their characteristics and averages of CO levels in March and April, 2017
Location Surrounding of the sensor Traffic Distance, March (ppb) April (ppb)
unit condition Elevation 24 hour average 24 hour average
to road Day time average Day time average
(m) Peak value (time) Peak value (time)
Kingswood Nearby main road, outside Average 16, 8 470 430
College (KC) city area, no disturbing traffic and 460 450
trees or buildings, bus stop vehicle 945 (6-7) 780 (8-9)
speed
Girls High Nearby main road, bus Heavy 40, 5 220 165
School (GHS) stop, near city area have traffic, Very 250 200
disturbing trees and slow vehicle 660 (7-8) 515 (8-9)
buildings speed
Pushpadhana Nearby main road, road Heavy 40, 20 790 880
College (PC) junction, inside city area, traffic, Very 820 1000
have disturbing buildings, slow vehicle 1160 (11-12) 1460 (10-11)
no trees speed
Trinity Nearby main road, inside Heavy 10, 6 355 310
College (TC) city area, road crossing, traffic, slow 370 330
have disturbing trees, no vehicle 770 (6-7) 640 (7-8)
buildings speed
Mahamaya Nearby main road, bus Heavy 5, 8 2050 1970
College (MC) stop, no disturbing trees or traffic, slow 2030 1930
buildings, outside city area vehicle 2600 (12-13) 2460 (12-13)
speed
In figure 6 it could be clearly observed that in the peaks, evening peaks and diurnal variation of the
morning and late evening significant increases in CO levels could be clearly visible at KC, GHS and
the CO levels, mainly due to the high traffic TC. Further peak and average CO levels have
congestions on the road. Further the diurnal decreased in April compared to March. And the
variation of pollutants can be clearly seen. During peak hours has also moved towards day time
the morning due to the low temperature in the during the month of April. This is clearly due to
environment, the mixing height of the the school vacation period in April. At KC in both
atmosphere is low, so that with the influx of March and April during the weekend CO levels
vehicles the pollutant levels increase quickly. have decreased and in March the peak values
When the environment temperature increases, have moved towards day time clearly indicating
the mixing height increases and air pollutants the effect of school time heavy traffic conditions.
disperse over a large volume of atmosphere Since TC is in the middle of the city weekend CO
decreasing the ground level pollution levels. levels have increased in the month of March
During the evening again the pollutant levels compared with the weekday values. One other
increase due to the influx of vehicles. But due to reason can be the private tuition classes at
the decreasing temperature, CO levels increase weekend in nearby. CO levels in GHS during
slowly and pollutants tends to be more stable in weekends have decreased compared with
the air than the morning due to the decreasing weekday values.
atmospheric mixing height.
Sensor unit of the PC was situated about 20m
Figure 7 shows the variation of hourly averaged height and about 40m distance from the road.
CO levels at five different schools in March and Further there were disturbing buildings in front
April, 2017. The data were obtained at every 3 of the sensor unit. So due to above reasons
minutes intervals daily and hourly averages were pollutants take time to travel and peak CO levels
calculated for the whole month. Table 2 shows were obtained during mid-day period. Further
the location characteristics and summary of the when comparing the CO level in weekdays and
average CO levels in different locations. Morning weekends, during the morning time CO levels
613
have decreased in weekends and during mid-day Ambient air in urban environments has complex
period CO levels have increased clearly mixtures of gases where the gas abundances can
indicating the effect of traffic conditions. Since fluctuate rapidly due to the inhomogeneous
the location is at a junction entering to the city dilution by the variations in the wind field or by
relatively heavy traffic could be observed. turbulent eddies and due to the changes in traffic
flows. Considering the overall data, this could be
Near the MC highest CO levels could be
clearly observed that the surrounding of the
obsereved. Since the surrounding is covered with
sensor units and distance from the road
trees and buildings wind variation is less around
significantly affect the CO levels monitored. So,
the sensor unit. So throughout the whole day
by changing the monitoring locations
high levels of CO could be observed. In March
continuously can generate high-resolution air
morning peak CO levels could be observed
pollution maps according to macro
during 6.00 am and 8.00 am. Then the CO levels
environmental conditions.
have reduced significantly between 8.00 am and
10.00 am. After CO levels have again increased Further, the sensor responses drift over the time
during the mid -day period. During weekends in due to the decaying of the electrolyte. To
the month of March there is a significant understand the issue of this long term data need
reduction of CO levels due to lack of school and to be obtained (>6 months) and an algorithm will
office vehicles. In April CO levels have compared be developed to correct the sensor responses
with CO levels in March. comparing data with high precision instrument.
When comparing the overall data, national air Since this is the first time we have used such
quality standards have not been violated any sensor network to monitor air quality in Sri
location. Mahamaya College had the highest Lanka, there are a lot of improvements need to be
levels of CO levels among all the locations. done before enabling such systems in the
Variation pattern of the CO levels clearly indicate country. Development of standard calibration
the traffic conditions of the road. Atmospheric method for these sensor units is the major future
mixing height and wind velocities directly affects work that we plan to do with long term data.
the spread of air pollutants in the environment. These real time air quality monitoring systems
are a promising solution for the higher costs
5. Conclusions and future work associated with air quality monitoring. And also
The results presented here are the preliminary they provide real time data which is highly useful
results of our study. We have developed sensor when understanding the behaviour of pollutants.
units to monitor pollutants CO, NO, NO2, SO2 As our future work we are working on
and O3. Accuracy of the sensor data is highly developing a comprehensive air quality model
important in these kind of studies. Compared to for the Kandy city. For that we aim to monitor
CO other four pollutants present in the weather and traffic data simultaneously with air
environment at very low concentrations (below quality data to identify the effects of them on air
100ppb), so that those sensors need to be pollution in Kandy city. Weather instruments are
accurately calibrated comparing with high being setup now. But monitoring real time traffic
precision air quality monitoring instrument. data is difficult since there are no established
Currently we are in the process of verifying the traffic monitoring systems in Sri Lanka.
sensor data. Since CO concentrations are at
higher levels and their cross interferences are As our observations clearly indicate that the
negligible the results present here are accurate. traffic congestions in the Kandy city area is the
Further to increase the data accuracy main source of air pollution, implementing
interchanging of monitoring instruments and methods to reduce the number of incoming
using multiple monitoring instruments at one vehicles to the city can be the best possible
location could be done. solution. Having good public transportation
systems can reduce the number of incoming
614
vehicles significantly and also public community BMC Public Health. 10 (2010) 300.
can be motivated for using public transportation doi:10.1186/1471-2458-10-300.
at least once a week. Further better maintenance 8. M.A. Elangasinghe, R. Shanthini,
of vehicles, specially the old busses to reduce air Determination of atmospheric PM
pollution. By assessing and comparing the concentration in Kandy in relation to traffic
impacts of public, private and electric intensity, 36 (2008) 245–249.
transportation methods or air pollution levels in 9. S. Nandasena, A.R. Wickremasinghe, N.
the Kandy can clearly aware the public about this Sathiakumar, Air pollution and public health
issue. in developing countries  : Is Sri Lanka
different?, J. Coll. COMMUNITY PHYSICIANS
Acknowledgement SRI LANKA.17(2012)21–42.
10. M.J. Bunch, V.M. Suresh, T.V. Kumaran, C.S.

Authors would like to acknowledge kindly the
Hons, P. Merit, AN OVERVIEW OF AIR
funding provided for the project by Vehicle POLLUTION AND RESPIRATORY
Emissions Testing Trust Fund, Department of ILLNESSES IN SRI LANKA, in: Proc. Third Int.
Motor Traffic, Central Province Education Conf. Environ. Heal. Chennai, India, 2003: pp.
Department and all the schools involved for their 489–501.
cooperation for this project. 11. A.P. Wickramasinghe, D.G.G.P. Karunaratne,
R. Sivakanesan, PM(10)-bound polycyclic
References aromatic hydrocarbons: biological indicators,
lung cancer risk of realistic receptors and
1. WHO, WHO Expert Consultation  : Available ―source-exposure-effect relationship‖ under
evidence for the future update of the WHO different source scenarios., Chemosphere. 87
Global Air Quality Guidelines ( AQGs ), (2015). (2012) 1381–7.
doi:10.1016/j.chemosphere.2012.02.044.
2. Evolution of WHO air quality guidelines: past,
present and future. Copenhagen: WHO 12. S. Yalegama, G. Bathan, Clean Air in Sri Lanka:
Regional Office for Europe; 2017. Summary of progress on improving air quality,
Country Network Sri Lanka, Clean Air Sri
3. World Health Organisation, Economic cost of
Lanka: (2008) 7.
the health impact of air pollution in Europe,
2015. 13. E.G. Snyder, T.H. Watkins, P. a Solomon, E.D.
Thoma, R.W. Williams, G.S.W. Hagler, D.
4. O.A. Ileperuma, V.D.K. Abeyratne,
Shelow, D. a Hindin, V.J. Kilaru, P.W. Preuss,
MONITORING AIR POLLUTION LEVELS IN
The changing paradigm of air pollution
KANDY USING PASSIVE AND ACTIVE GAS
monitoring., Environ. Sci. Technol. 47 (2013)
SAMPLING TECHNIQUES, Ceylon J. Sci.
11369–77. doi:10.1021/es4022602.
Phys. Sci. 9 (2002) 54–61.
14. T.J. Roberts, J.R. Saffell, C. Oppenheimer, T.
5. P. Kumar, L. Morawska, C. Martani, G. Biskos,
Lurton, Electrochemical sensors applied to
M. Neophytou, S. Di, M. Bell, L. Norford, R.
pollution monitoring  : Measurement error
Britter, The rise of low-cost sensing for
and gas ratio bias — A volcano plume case
managing air pollution in cities, Environ. Int.
study, J. Volcanol. Geotherm. Res. 281 (2014)
75 (2015) 199–205.
85–96. doi:10.1016/j.jvolgeores.2014.02.023.
doi:10.1016/j.envint.2014.11.019.
15. H.D.S. Premasiri, I.H.K. Samarasinghe, K.M.N.
6. M.I. Mead, O. a. M. Popoola, G.B. Stewart, P.
Lakmali, Population exposure risk assessment
Landshoff, M. Calleja, M. Hayes, J.J. Baldovi,
to air pollution in Kandy city area,
M.W. McLeod, T.F. Hodgson, J. Dicks, a.
Environmental Division, National Building
Lewis, J. Cohen, R. Baron, J.R. Saffell, R.L.
Research Organisation (2010).
Jones, The use of electrochemical sensors for
monitoring urban air quality in low-cost, high-
density networks, Atmos. Environ. 70 (2013)
186–203. doi:10.1016/j.atmosenv.2012.11.060.
7. Y.L.S. Nandasena, A.R. Wickremasinghe, N.

Sathiakumar, Air pollution and health in Sri
Lanka: a review of epidemiologic studies.,
615
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[617 622],2017
2017
The Institution
Lanka
Dense Graded Asphalt for Sri Lanka:Considerations

related to Climate and Different Traffic Conditions
W.W. Bandara and R.P.M.M. Premathilaka
Abstract: High standard crushed rock aggregates, and 60/70 penetration grade bitumen which
have a high stiffness at 25°C (testing temperature) are used to produce dense graded asphalts (DGAs)
in Sri Lanka. Hence it is considered in pavement structure design that asphalts have a high stiffness
(Modulus) approximately 2500MPa. Pavement temperature is measured in terms of weighted mean
annual pavement temperature (WMAPT).WMAPTs of different geographic locations in Sri Lanka are
calculated to be between 35°C and 41°C.It is revealed that 60/70 bitumen has a low stiffness and
DGAs made with 60/70 bitumen also have a low stiffness (Modulus) at WMAPTs of Sri Lanka and
low heavy vehicle speed. Further, this is verified by in-situ asphalt moduli which are back-calculated
from surface deflection measurements. Nevertheless, 60/70 bitumen can be used up to heavy traffic
load which is 20 Million Equivalent Standard Axles (ESAs).Then modulus of DGAs can be taken from
1000MPa to 2200MPa for WMAPTs of Sri Lanka and 30km/h to 70km/h heavy vehicle speed.
Moreover, an equivalent viscosity grade bitumen class shall be used to mitigate reduction of modulus
due to the poor coating of bitumen with aggregate, compaction and consistency in the mix.
Keywords: Asphalt Pavement, Bitumen Grade, Stiffness, Modulus, Defects, Sri Lanka Climate,
Traffic Conditions
1. Introduction Pavement Technology” (Austroads) has been

principally used for the study.
In basic terms, asphalt is a mixture of coarse
and fine aggregates, filler and bitumen. Dense 2. Methodology
graded asphalts (DGAs) are widely used in the
construction of new road pavements and Asphalt stiffness modulus values along a good
strengthening of existing road pavements in Sri condition road section for known asphalt
Lanka. DGAs are manufactured from 60/70 thickness are estimated from in-situ surface
penetration grade bitumens and20mm nominal deflections measured by a falling weight deflect
size aggregates[4]. DGA is placed on the meter(FWD). Here rutting roughness data are
aggregate base course or an existing used to evaluate the condition of the pavement
bituminous pavement as a wearing course in and coring results in the road pavement are
conjunction with or without an asphalt binder used to determine asphaltthickness.
course.
A viscosity grade classification has been
Alligator (or map) cracking and formation of adopted by countries like Australia, South
waves and corrugation (caused by stop-and-go America and India [1]. For Sri Lanka climate
vehicles) are observed on asphalt surface of and different traffic conditions, a viscosity
moderate-to-heavily trafficked road pavements. grade bitumen class is selected from
This kind of defects occurs on loaded areas Austroads[3]. Selected bitumenclass is
even after several years from construction. The compared with 60/70 bitumen class use in Sri
defects occur due to insufficient resistance to Lanka. Here stiffness and operating
fatigue and permanent deformation effected by temperature limits of both bitumen are
the asphaltmodulus. Modulus of DGAs made estimated from their basic properties given in
with selected nominal size aggregate depends specifications [3] and [4].It is documented the
on the stiffness of bitumen at pavement significance of the use of a viscosity grade
temperature and heavy vehicle speed, a coating bitumen class for road pavements in Sri Lanka.
of bitumen with aggregate, compaction and
Eng. W.W. Bandara, M.Eng. (Moratuwa), B.Sc.Eng.
consistency in the mix. Hence the study is Hons. (Peradeniya), Director, Engineering Design and
conducted to review a selection of the bitumen Project Management Consultants (Pvt) Ltd.
class for Sri Lankan climate and different traffic
conditions, and mixing and compaction Eng. R.P.M.M. Premathilaka, B.Sc. Eng.Hons.
(Peradeniya), Engineer, Engineering Design and Project
temperatures of DGAs. “Austroads Guide to Management Consultants (Pvt) Ltd.
1
617 ENGINEER
Finally, design modulus of DGAs is estimated The selection of binder for hot mix asphalt
for pavement structure designs. applications is generally based on traffic levels
and expected maximum temperatures. A
3. Literature Review summary of binder type for common dense
graded asphalt wearing course and base
The performance of asphalts, in both the short course(binder course) applications is provided
and long term, is influenced by the bitumen in Table-2 and Table-3 respectively for the
properties, aggregate and grading categories of traffic in equivalent standard
characteristic, and the composition. Here the Axles (ESAs) and maximum pavement
bitumen play a critical role [1]. In Sri Lanka, temperature[3]. Mampearachchi in 2012
DGAs are manufacturedfrom20mm nominal revealed that 7-day average maximum
maximum size aggregates of high standard pavement temperature in different
crushed rocks and 60/70 penetration grade geographical areas of Sri Lanka is between 52°C
bitumen. Properties of 60/70 bitumen are given and 58°C [5].
in Table_1. The bitumen and aggregate are
mixed at temperatures within absolute limits of Table 2 – Australian Binder Types for Dense
145 °C and 170 °C. The final rolling are Graded Asphalt Wearing Course [3]
completed before the temperature of mix falls Maximum Pavement
below 90°C. Air void incompacted aggregate Traffic Category Temperature (°C)
are from 3% to 7%. However, it is generallyless >58 52-58 < 52
than 5% in the mix design. [4] Light C170 C170
C320
(<0.5x106 ESAs) C320 C320
Table 1 - Properties of 60/70 Grade Bitumen Medium C170
6 6 C320 C320
(0.5x10 -5x10 ESAs) C320
Penetration at 25°C 100g, 5s 60/70
Heavy C320 C320
Softening point (°C) 48/56 C320
(5x106-20x106 ESAs) PMB PMB
Loss on heating for 5 hrs. at 163°C Very Heavy
PMB PMB PMB
(1) Loss in weight percent 1.0 max (>20x106 ESAs)
(2) Loss in penetration,
(% of original value) 75 min Table 3 – Australian Binder Types for Dense
Solubility in trichloroethylene (%) 99 min Graded Asphalt Base Course [3]
Maximum Pavement
Flash point (°C) 232 min
Traffic Category Temperature (°C)
In Australia, bitumen is classified by its
consistency expressed as viscosity at 60°C and >58 52-58 < 52
comprise three major grades: Class 170, Class Light C320 C170 C170
320 and Class 600. Class 170 bitumen is the (<0.5x106 ESAs) C320 C320
softest grade and provides the greatest Medium C320 C320 C320
flexibility and durability for use in asphalt in (0.5x106-5x106 ESAs)
cool climates, and for light traffic. Class 320 is a C320 C320 C320
Heavy
slightly harder grade of bitumen that is the 6 6 C600 C600
(5x10 -20x10 ESAs)
most versatile and commonly used binder for PMB PMB
hot mix asphalt in a wide range of applications. C320 C320 C320
Very Heavy
Compared to Class 170 bitumen, it provides C600 C600 C600
(>20x106 ESAs)
greater resistance to rutting and deformation at PMB PMB PMB
high service temperatures or heavy traffic.
Class 600 is a harder grade of bitumen than The bitumen test data chart developed (BTDC)
Class 320. The use of Class 600 is generally by Heukelom in 1969, can be used for
confined to heavy-duty asphalt base layers comparing the temperature, penetration and
where greater asphalt mix stiffness is required. viscosity characteristics of different type of
Class 600 is not generally recommended for use bitumen. The chart consists of one horizontal
as a wearing course layer as increased stiffness scale for the temperature and two vertical
can result in reduced resistance to cracking and scales for the penetration and viscosity. The
fatigue in thin layers. Where greater stiffness temperature scale is linear, the penetration scale
for rutting resistance is required in wearing is logarithmic, and viscosity scale has been
course layers, it is generally more effective to devised so that penetration grade bitumen with
usepolymer modified bitumen (PMBs). [3] penetration indices give straight line
relationship. Then the test data for a large
ENGINEER 2
618
group of bitumen can be represented by a Modulus of DGAs made with known bitumen,
straight line on BTDC. [1] and nominal size of aggregate can be calculated
from the percentage of air void, temperature
The monographis used to estimate the stiffness and heavy vehicle traffic speed (V).Modulus of
modulus of a bitumen from the time of loading AustralianDGAs, made with 20mm nominal
(heavy vehicle speed), the pavement maximum size aggregate, determined on
temperature, the temperature (°C) at which the laboratory manufactured samples using the
penetration of the bitumen is 800 (0.1mm) indirect tensile test procedure and standard test
(T800Pen)and temperature susceptibility conditions and 5% air voids ( ) are shown
measured in terms of the penetration index (PI). in Table 04. Standard test conditions are 40ms
[1] rise time and 25 °C test temperature. This
modulus is corrected for the in-service air voids
( , temperature (WMAPT) and loading
rate by Equation 5, Equation 6 and Equation 7
respectively. [2]
where A is the change in log (penetration) per
°C change in temperature.
Thus, A and T800Penfor conventional bitumen

can be figured out from the BTDC. Also, the
BTDC is a useful tool for ensuring that the
correct operating temperatures are selected to [ ]
achieve the appropriate viscosity of any grade
of bitumen. When the aggregate and the
bitumen are mixed together, the viscosity
should be approximately 0.2Pa.s for satisfactory
coating and not be below 0.5Pa.s.Also, it is Table 4 - Laboratory Moduli of Australia
widely recognised that the optimum bitumen DGAs
viscosity for compaction is between 2Pa.s and Modulus, MPa
20Pa.s. [1] Bitumen Range Typical
Class 170 2000-4500 4000
Pavement temperature is measured in terms of Class 320 3000-7500 5500
weighted mean annual pavement temperature Class 600 4000-9500 7000
(WMAPT).The monograph is used to estimate
the stiffness modulus of bitumen at the Roughness (longitudinal unevenness) describes
WMAPT and the rate of loading. Heavy vehicle the surface condition of a road. Roughness can
traffic speed (V) is used to calculate the rate of be measured in terms of International
loading (1/V).The WMAPT is estimated from Roughness Index (IRI). The IRI is a property of
the monthly average air temperatures by the true profile of the pavement. An overall
averaging the maximum and minimum air description of the pavement condition is
temperatures (Tair) in an area. Firstly, measured in terms of present Serviceability
temperature weighting factor (WF) is calculated Index (PSI). Serviceability of pavement is its
for each month by substitutingTair in Equation ability to serve traffic in its present condition.
2, and secondly, the weighted mean annual air According to AASHTO Guide for Design of
temperature (WMAAT) are estimated by Pavement Structures (AASHTO Guide), when
substituting WFs in Equation 3 and averaging. PSI reduces to 2.0, the road pavement shall be
Finally, the WMAPT is estimated by anoverlay. Also, PSI equal to 4.2 is considered
substituting the WMAAT in Equation 4. [2] as initial serviceability index of newly
constructed pavement [6]. Since the PSI is
( )
mostly affected by roughness, measured IRI
values can be converted to PSI by Equation 8
developed by Paterson, 1986.Rutting
(Transverse Unevenness) is load associated
deflection will appear as longitudinal
depressions in wheelpaths. Permissible rut
depth for asphalt pavement surface is 10mm
[8].
3
619 ENGINEER
Historical average monthly temperature data of
all districts in Sri Lanka is obtained from the
The Falling Weight Deflectometer (FWD) is an URL https://www.msn.com/en-
excellent device for evaluating moduli of gb/weather/today/ [5].Validation is done for
pavement layers and subgrade. BAKFAA the figures obtained from the given URL
program hosted by Federal Aviation against the actual data obtained from
Administration (FAA) are used to calculate Metrological Department of Sri Lanka.
moduli of pavement layers from FWD data [9]. Furthermore the WMAPT values calculated
from the URL is compared with the actual
4. Analysis figures given for Melbourne City. In both the
cases results closer to the actual figures are
The section from 58+000 to 62+400 (Variyapola given by the data obtained from the URL.
to Padeniya) of Katugastota - Kurunegala - WMAPTs of all districts are calculated and
Puttalam (A10) road is selected to calculate in- presented in Table_5. WMAPTs in Sri Lanka is
situ asphalt stiffness moduli as road pavement between 35°C, and 41°C. WMAPT of several
is visually in good condition and insignificant areas is 40°C.Since WMAPTs in Sri Lanka are in
cracks are observed on the road surface. FWD a narrow region, single bitumen class is
data at 50m intervals and roughness and sufficient for Sri Lanka. Also, there are many
rutting data at 100m intervals are obtained from practical difficulties to use two grade of
RDA.10th percentile values of the PSI bitumens in deferent layers of pavement or
calculated from IRI values are3.13and 3.18 for different regions in Sri Lanka. Since maximum
LHS and RHS respectively, and 90th percentile pavement temperature is between 52°C and
value of rut depth is5.07mm. Then calculated 58°C, according to Austroads refer Table_2 and
IRI and rutting values are well matched to a Table_3, a suitable viscosity grade bitumen
good condition road (i.e. 4.0 IRI and 10mm class for Sri Lanka climate and traffic up to 20
maximum rutting for a newly constructed million ESAs(heavy traffic) isC320. Its
road). Hence, it is confirmed that the road properties are shown in Table_6.
pavement is in good condition. According to Table 5 - Weighted Mean Annual Pavement
asphalt coring result that is carried out at Temperatures (WMAPTs) of Sri Lanka
250mm, thicknesses of asphalt at coring Group WMAPT District
location are more than 200mm. Then modulus
1 35°C Nuwara Eliya, Badulla,
values of upper 200mm asphalt layer at test
Rathnapura
locations are back calculated by BAKFAA
2 36°C Kandy, Matale,
program using FWD data.10th percentile value
Monaragala
of Modulus is 880MPa.Modulus values of
3 39°C Kegalle, Kurunegala,
upper 200mm asphalt layer along the selected
Anuradhapura,
section of A10 road are presented in Figure
Polonnaruwa,
1.Measured temperatures of the pavement
Trincomalee, Vavuniya
surface along this road section are varied from
4 40°C Colombo, Gampaha,
31°C and 37°C.
Kalutara, Galle, Matara,
Hambantota, Jaffna,
5
Kilinochchi, Mullaittivu,
Mannar
4
Modulus, MPa/1000
5 41°C Puttalama, Ampara,

Batticaloa
3
Table 6 - Properties of C320 Grade Bitumen [3]
2
Viscosity at 60°C (Pa.s) 260–380
1 Penetration at 25°C 100g, 5s 40 min
Viscosity at 135°C 0.40-0.65
0 Viscosity at 60°C after Rolling 300 max
58
60 59
61 62 Thin Film Oven Test (% of
Location, km
original)
Figure 1 - Modulus values of Upper 200mm of
asphalt in selected locations of A10 Road Matter insoluble in toluene (%) 1.0 max
Flashpoint (°C) 250 min
Loss on heating (%) 0.5 max
ENGINEER 4
620
In order to compare C320 bitumen class with Lower Mixing
60/70 bitumen class, four bitumen from each Limit,°C 122 163 156 127
class are selected considering their boundary Upper
limits. (Bitumen A, B, C and D for Bitumen Compaction
Class C320 and Bitumen E, F, G and H for Limit,°C 103 135 130 106
Bitumen Class 60/70 respectively) Boundary Lower
limits of C320 bitumen class are taken as upper Compaction
and lower limit of viscosity at 60°C (Vis@60) Limit,°C 78 99 96 80
and 135°C(Vis@135). As well as Boundary
limits of 60/70 bitumen class are taken as upper A stiffness range of both class of bitumen at
and lower limit of penetration at 25°C 40°C is almost equal. Hence it can be concluded
(Pen@25)and softening point (T800Pen). The that modulus of asphalt made with both classes
bitumen stiffness at 40°C is estimated from of bitumen is equal. Then modulus of Sri
Monograph given in Austroads, and upper and Lankan DGA, made with 60/70 bitumens and
lower limits of mixing temperature compaction 20mm nominal maximum size aggregates, are
temperature are estimated from BTDC. To considered as the modulus of DGA given
estimate bitumen stiffness, viscosity at mixing forC320 bitumen in Table_4. However given
and compaction limits are taken as 0.2Pa.s to modulus DGA should be corrected for the in-
0.5Pa.s and 2Pa.s and 20Pa.s respectively, and service air voids, WMAPT and heavy vehicle
heavy vehicle traffic speed is assumed traffic speed by Equation 5, Equation 6 and
practically as 30kmphto.All specified and Equation 7. Modulus of DGAs in different
estimated properties are shown in Table_7 and WMAPTs in Sri Lanka and different heavy
Table_8. vehicle traffic speed are calculated and
presented in Figure_2. Also, modulus of Sri
Table 7 - Properties of Selected C320 Bitumen Lankan DGAs in different zones shown in
Bitumen A B C D Table_5 and different heavy vehicle traffic
Vis@60,Pa.s 260 380 260 380 speeds are presented in Table_9 for the purpose
of pavement structure designs.
Vis@135,Pa.s 0.40 0.65 0.65 0.4
A/1000 44 41 39 46 There are no large variations in upper and
PI -0.7 -0.2 0.1 -0.9 lower temperatures of mixing and compaction
T800Pen,°C 49 51 48 52 of DGA made with C320 viscosity grade
T800pen-40°C 9 11 8 12 bitumen. Hence temperature limits can be
Stiffness@40, specified for C320 bitumen. However variations
MPa 0.4 0.5 0.3 0.7 of mixing and compaction of DGAs made with
Upper Mixing 60/70 penetration grade bitumen are huge, and
Limit,°C 148 159 160 148 some limits are out of specification limits
Lower Mixing (Specified mixing limits are 145°C and 170°C,
Limit,°C 131 140 140 131 and specified lower compaction limit is 90°C).
Upper Then there is a risk to produce poor DGA mixes
Compaction due to the poor coating of bitumen with
Limit, °C 109 116 116 110 aggregates, compaction and consistency in the
Lower mix and it leads the insufficient resistance to
Compaction fatigue and permanent deformation affected by
Limit, °C 82 93 85 90 the modulus of DGAs. Defining new specified
limits are not possible because of huge
variations. In order to mitigate the mixing and
Table 8 - Properties of Selected 60/70 Bitumen
compaction issues, viscosity grade bitumen
Bitumen E F G H class smiler to 60/70 bitumen class should be
Pen@25,dmm 60 70 60 70 used in Sri Lanka for quality service of road
Softening Pt.,°C 48 56 56 48 pavements.
A/1000 49 34 36 46
PI -1.3 1.1 0.7 -0.9
T800pen-40°C 8 16 16 8
Stiffness@40,MPa 0.5 0.6 0.7 0.4
Upper Mixing
Limit,°C 138 185 178 144
5
621 ENGINEER
2500 Lanka together with a poor coating of bitumen,
30km/h compaction and consistency in the mix, causes
50km/h insufficient resistance to fatigue and permanent
70km/h deformation in trafficked road pavements and
Modulus, MPa
2000 90km/h leads to alligator cracking and formation of

waves and corrugation caused by stop-and-go
vehicles. In this condition, defects can occur on
1500 loaded areas even after a few years from
construction. Hence, it is very essential to
control viscosities of bitumen at mixing and
1000 compaction temperatures. Since viscosity
35 36 37 38 39 40 41 grade, bitumen calcification controls viscosities
at 60°C, and 135°C near mixing, viscosity grade
WMAPT, °C
bitumen calcification shall be introduced to Sri
Figure 2 - Modulus of Sri Lankan Dense Lanka for the performance of roads.
Graded Asphalt
Table 9 - Modulus of Dense Graded Asphalt

Acknowledgement
of Deferent Zones in Sri Lanka The Authors wish to acknowledge Prof W.K.
WMAPT Heavy Vehicle Traffic Speed, Mampearachchi and other engineers for
Group km/h sharing knowledge. The Authors express their
30 50 70 90 gratitude to Asian Development Bank (ADB)
1 1600 2000 2200 2400 Division of Road Development Authority
2 1500 1800 2000 2200 (RDA) and Engineering Design and Project
3 1200 1400 1600 1800 Management Consultants (Pvt) Ltd for
4 1100 1300 1500 1600 providing relevant data for the study.
5 1000 1200 1400 1500 References
5. Conclusions 1. Robert N. Hunter, Andy Self, & John Read, The

Shell Bitumen Handbook,6th ed., ICE Publishing,
One Great George Street, Westminster, London,
The thickness of the asphalt for decided ESAs
2015,pp.87-145, 503-542, 750-752.
depends on the modulus of the asphalt. Since
Sri Lankan DGAs varies from 1000MPa to 2. Austroads, Guide to Pavement TechnologyPart 2:
2400MPa and precise modulus shall be selected Pavement Structural Design, Sydney, 2012,pp.64-
according to pavement temperature (WMAPT) 83.
and heavy vehicle traffic speed from Table_9. 3. Austroads, Guide to Pavement TechnologyPart 4F:
Then only road pavements in Sri Lanka provide Bituminous Binders, Sydney, 2008,pp.10, 23-25.
their service without a premature failure. 4. ICTAD, Standard Specification for Construction and
Further laboratory and field studies are Maintenance of Road and Bridges, 2th ed., Colombo,
required to improve modulus of Sri Lankan 2009,pp.143-151, 459-463.
DGAs given in Table_9. 5. Mampearachchi, W. K., "Review of asphalt binder
grading systems for hot mix asphaltpavements in
Although 60/70 bitumen can be used to Sri Lanka”, Journal of the National Science
produce GDAs up to 20 million ESAtraffic Foundation of Sri Lanka 40 (4), 2012, pp. 311-320.
loads, excessive asphalt thickness has to be
used for high traffic loads with low heavy 6. https://www.msn.com/en-
vehicle speed or high WMAPT. High-grade gb/weather/today/,Visited, January 2017.
bitumen cannot be recommended for use as a
wearing course layer as increased stiffnesscan 7. AASHTO, Pavement Guide for Design of Pavement
result in reduced resistance to cracking and Structure, Washington, 1993, ii-10 p.
fatigue in thin layers. Then a polymer modified 8. Transport & RoadResearch Laboratory, Overseas
bitumens or composite pavement structure can Road Note 18: A guide to the pavement evaluation and
be used to reduce asphalt thickness. For this maintenance of bitumen-surfaced roads intropical and
sub-tropical countries, Berkshire, 1999,34 p
reason, typical pavement structures used in Sri
Lanka for high traffic have to be reviewed. 9...http://www.airporttech.tc.faa.gov/Downoad/Ai
rport-Pavement-Papers-Publications-
Low asphalt modulus at low heavy vehicle Detail/dt/Detail/ItemID/34/BAKFAA-version-
20-update-04012013, Visited, February 2017.
speed and high pavement temperature in Sri
ENGINEER 6
622
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[623 630],2017
2017
The Institution
Lanka
Suitability Recycled Pet Fiber and Carbonized Wood

Ash in Hot Mix Asphalt Concrete
N. Jegatheesan, T.M. Rengarasu and W.M.K.R.T. Wasala Bandara
Abstract: This study was aimed at investigating the suitability of Polyethylene Terephthalate
(PET) fibres having a diameter of 0.5 mm and a length of 4 to 6 mm and Carbonized Wood Dust
(CWD) having a nominal size of 1.18 mm, as alternative materials of Hot Mix Asphalt (HMA) by
following the Marshall Method. Accordingly, conventional materials of HMA which are binder and
aggregates were partially substituted with PET and CWD in three different ways. Frist, PET was
added as a binder additive in a wet process. Next, stone dust (Filler) was replaced by CWD. Finally
combined modification was done using PET and CWD which were added as binder additive and filler
respectively in same HMA mix. Due to the less specific gravity of CWD, bulk density and void of
Modified HMA (MHMA) with CWD were reduced. However, the Marshall properties of MHMA
with CWD were impaired, due to the lack of bond between aggregates and CWD. Maximum 2.73% of
CWD can be added to satisfy the minimum stability requirement given in CIDA specification.
Modification with PET was increased the Marshall Strength, bulk properties and bonding ability as
well. Therefore, to improve the Marshall Stability of MHMA with 2.73% of CWD, PET was added as
an additive that resulted in 30% - 103% increment in Marshall Stability compared to MHMA with
CWD only. However, a reduction in stability was observed in comparing with the control mix.
Furthermore, the bulk density of combined MHMA was reduced with the addition PET modification.
The absorption rate of combined MHMA was increased at 5% to 10% addition of PET. Beyond 10%
addition of the PET, the absorption rate was considerably reduced. Eventually, it was discovered that
minimum 5% PET and maximum 2.73% of CWD can be added in the combined MHMA to satisfy the
medium and the low traffic limitation of CIDA specification.
Keywords: PET fibre, Carbonized wood dust, Modified Hot Mix Asphalt, Wet process
1. Introduction attention on using alternative materials like

waste or recycled materials to reduce the usage
The rapid development of the country amplifies of convention aggregate and the overall cost of
the number of expressway, highway and road HMA.
construction to enhance the connectivity
between lagging and developed area. Around 2. Literature review
80% of the roads are paved with Hot Mix
Asphalt Concrete (HMA) by using more than In developing countries like Sri Lanka, more
90% of aggregates by the total weight of the mix efforts have to be taken to tackle the waste
[6, 7, 12, 29]. However, materials used in HMA disposal problem, due to lack of space. About
can be able to withstand the traffic load that is 85% of collecting waste in Sri Lanka is subjected
usually spread out through the structural layer to open or ocean dumping [11, 18, 30]. At
of the pavement [14] in such a way that the present, nearly 4000 saw mills including pit-
stress at the sub-grade soil is low enough not to sawing units and excluding 680 mills owned by
cause major damage to the pavements [1, 7, 8]. Eng. N.Jegatheesan, B.Sc.Eng.(Hons) (Ruhuna),
Postgraduate Student (MPhil), Department of Civil
The surface wearing course makes direct Engineering, Faculty of Engineering, University of Ruhuna.
contact with the traffic load in the flexible Eng. (Dr.) Terrance M. Rengarasu, Ph.D.(Hokkaido),
pavement [6, 8]. Therefore, it should consist of M.Eng. (Hokkaido), B.Sc.Eng.(Hons) (Peradeniya),
proper materials to withstand traffic load AMIE(SL) Senior Lecturer, Department of Civil and
Environmental Engineering, University of Ruhuna.
without undergoing to any failure during their
design life [6, 15, 20]. Because of the high cost of Eng. (Dr.) W.M.K.R.T. Wasala Bandara,
aggregates, conventional method with usual Ph.D.(Hokkaido), M.Eng. (Hokkaido), B.Sc.Eng.(Hons)
materials is not an economical way to follow in (Peradeniya), AMIE(SL), AMSSE(SL), MBEA(SL), Senior
Lecturer, Department of Civil and Environmental
the case of medium and the low traffic class Engineering, University of Ruhuna.
pavement. Therefore, Engineers are paying
1
623 ENGINEER
furniture manufacturers, are running in Sri using as filler and PET using as binder additive
Lanka. Annually, around 112,000 MT sawdust separately and combine modification.
is produced from the saw mills [26]. In general,
12% of the input wood is wasted as sawdust 3. Materials
during milling operations [23, 31]. Only a
considerable amount of sawdust is used as fuel, 3.1 Conventional materials
ultimately remaining part is dumped or burned In this study, crushed rocks were taken at the
in the direct surroundings of the mills [26]. In required nominal size (Coarse – 20 mm, Fine –
Sri Lanka, it was estimated that, around 10.5% 14 mm and Filler – Quarry dust). The materials
of total waste are non-biodegradable plastic were tested based on American Standards.
and polythene in 2015 [11]. Approximately 4.8 Based on particle size distribution of aggregate,
to 12.7 million MT plastic waste is floating in the mix proportion of three types of aggregates
the world’s oceans which causes negative (20%: 20%: 60% of course, fine and filler
impacts on the environment [13]. respectively) and binder percentage were
selected. According to the CIDA specification
As an environmentally friendly construction mix design was categorized as Type 3 wearing
method, a considerable number of researches course. The 80/100 penetration grade binder
have been conducted on HMA with different with a melting point at 43˚C was used in this
kind of waste materials such as waste glass, study.
rubber, steel slag, marble and ceramic dust,
cement dust, rice husk ash, sawdust , sawdust 3.2 Waste materials
ash, recycled concrete aggregates, tyres, Wood dust which was belonged to one type,
demolished building waste, plastic, polythene Albesia (Albizia Julibrissin), was burned in a
etc [2, 3, 5, 12, 25]. Wood dust ash having 5– sealed oxygen-less incinerator (Figure 1).
30% of carbon, can be used as filler or modifier, Carbonized Wood Dust (CWD) having a
because it can be ground finer and become nominal size of 1.18 mm was used as filler in
more activated due to the metastable state of the order of 1.77%, 2.73% and 3.75% of total
surface structure [21, 27]. Wood dust ash as weight of aggregate (respectively 1133.1g,
asphalt binder modifier, improves the rutting 1099.7g and 1066.2g). Because of less specific
resistant, viscosity and complex modulus. gravity (0.61) of the CWD, the filler
Further, sawdust as filler produces light weight replacement was done at volume basis to keep
concrete; however, stability was reduced by the standard sample size.
lack of bonding ability and high absorption of
the binder [17, 22].
Polyethylene Terephthalate (PET) recycled

material form plastic also can be used as
aggregate and binder modifier [2, 9, 20]. PET
replaced with aggregate, improves performance
of pavement in rutting and fatigue cracking
compared to conventional one by the reduction
of permanent strain [16, 20]. The viscosity of
modified binders was simultaneously increased
with both mixing time and adding PET
Figure 1 - Carbanized wood dust (1.18 mm)
percentages [4, 9]. PET coated aggregate
MHMA resulted in higher Marshall Stability
(MS) and less penetration [16, 28]. The 20%
(weight of coarse aggregate) of PET as
aggregate reduce the bulk density. However,
result obtained from Marshall Test was almost
same for control and modified mixture. The
flow value of the modified mixture was
somewhat lower than conventional one [2, 10].
PET as fine aggregate also reduced the
permanent deformation of HMA mixture [19,
24]. This study aimed at investigating the
Figure 2 - Polyethylene Terephthalate (PET)
properties of single modified HMA with CWD PET (having a length of 4-6 mm and a diameter
of 0.5 mm, specific gravity 1.2 and melting
ENGINEER 2
624
point - 252˚C) was collected from PET Control mix was prepaid to compare the result
manufacturing company (Figure 2). It was of MHMA. As a first modification, PET was
mixed with bitumen as an additive at 5%, 10% properly mixed with bitumen at high
and 15% of total binder weight. temperature (200 ˚C to 250 ˚C) to make the
homogeneous modified binder. After that, PET
4. Methodology modified binder was commingled with
convention aggregate mix. In the second
In this study, Marshall Mix design was modification, pure bitumen was mixed with
followed to investigate the bulk and Marshall aggregate partially replaced with CWD. Finally,
properties of modified HMA (MHMA) with modification was done with PET modified
PET and CWD. The major scope of this mix bitumen and CWD replaced aggregate.
design is to find out the optimum requirement
of materials. Figure 3 shows the summary of 50 blows (For medium traffic load) per side
the experimental study. According to the CIDA were applied at 145˚C to compact a sample.
specification, binder content was selected as 4 Bulk properties (Specific gravity, Void in Mix
to 6% at 0.5% interval with respect to the (VIM), Voids in Mineral Aggregate (VMA) and
aggregate gradation (from Sieve analysis test). Void filling Bitumen (VFM)) was determined
before test the stability of the sample. Marshall
Sieve Analysis Properties (MP) which are Marshall Stability
Test for (MS) and Marshall Flow (MF) value, of the
aggregates preheated specimens at 60˚C for 30 minutes,
Select the proportion of were determined by using the Marshall Test
each aggregates type apparatus. Finally, optimum requirement of
materials were determined by analysing the
No obtained result. Absorption of wood dust is
Compare higher than the aggregate. Therefore,
with CIDA absorption test was conducted to determine the
specification absorption rate of MHMA.
Yes
5. Result and discussion
Select the waste 5.1 HMA with PET as binder additive
materials in
Conventional
The properties of the binder and the overall

suitable phase performance of MHMA were improved with
mix
addition of PET. When the PET content was

Select the Select adding increased, the specific gravity of MHMA was
replacement percentage of gradually reduced. However, compared to the
rate of CWD PET control sample, the specific gravity was less.
Variation of the specific gravity of control mix
Casting the sample for each proportion and the modified HMA with binder content are
shown in Figure 4. There was no considerably
Determine the bulk properties different was observed, because of the specific
(Check with CIDA specification) gravity of the PET (1.2) which was nearly equal
to the bitumen (0.94). In the case of air void
percentage, the PET modified HMA resulted in
Test the Marshall properties
greater value compared to control sample. Even
(Check with CIDA) though the air void percentage was greater, the
specification) value of air void at optimum bitumen content,
Determine the was between the allowable ranges (3< VIM <5,
requirement of CIDA specification). The air void of the
materials modified HMA concrete was decreased with
the addition of PET (Figure 5). The proper
variation was observed on voids in mineral
Optimum aggregate of PET fibre MHMA with respect to
material the addition of PET. However, modified HMA
Figure 3 - Methodology flowchat with PET resulted in greater percentage which
was within allowable range, compared to the
3
625 ENGINEER
control sample. Void filling ability also Marshall Stability of the PET modified HMA
somewhat less compared to the control sample. was increased with the addition of the PET.
Even though, optimum bitumen percentage of
the modified binder was less compared to
2.4
control; it was increased with the addition of
PET. PET improved the stiffness and the
Specific Gravity
2.3 bonding strength of the binder which improves

the load bearing ability of the pavement (Figure
6). At 5% of PET modification, the stability was
2.2 increased by maximum 14%. Likewise, 10% of
PET addition showed 25% increment in
stability at 4.65% of modified binder content.
2.1
3.5 4 4.5 5 5.5 6 6.5
18
Bitumen Percentage (%)
5% PET 10% PET 15% PET Control
1.77% CWD 2.73% CWD 3.75% CWD
8 < MF < 18
16
Figure 4 – Specific gravity of single MHMA
Marshall Flow
14
6.0
12
5.5
Void In Mix (VIM -%)
5.0 10
3.5 4 4.5 5 5.5 6 6.5
4.5 Bitumen Percentage (%)
4.0 1.77% CWD 2.73% CWD 3.75% CWD
Figure 7 – MF of single MHMA
3.5
At 4.7% of modified bitumen, the 15% of PET
3.0 modified sample resulted in 67% improvement
3.5 4.5 5.5 6.5 in stability compared to HMA. Similarly,
Bitumen Percentage (%) Marshall Flow (deflection) was increased due to
the increment of air void. However, the
occurred deflection was within the accepted
1.77% SDA
CWD 2.73% CWD
SDA 3.75% SDA
CWD
range (8 < Marshall Flow < 18, CIDA
Figure 5 –Void of single MHMA specification) (Figure 7). Flow value was
increased with binder content like control
sample. However, the increasing percentage
25
was decreased with percentage of modified
bitumen. According to Sengoz & Topal [25]
20 research, that variation also was acceptable. Up
MS > 5.33kN
to 15% of PET modification, MHMA showed

Marshall Stability (kN)
15 better performance compared to conventional

HMA.
10
5.2 HMA with CWD as a filler – replacement
The replacement of CWD as filler produced the
5 light weight concrete due to the low bulk
density of CWD (Figure 4). When the amount
0
of CWD was increased in the sample, they
3.5 4 4.5 5 5.5 6 6.5 became as low-quality mix due to lack of
Bitumen Percentage (%) binding ability of the binder. Therefore, with
5% PET 10% PET 15% PET the increment of binder content, the samples
Control 1.77% CWD 2.73% CWD
3.75% CWD displayed more consistency in the visual
inspection due to the increased requirement of
Figure 6 – MS of single MHMA
ENGINEER 4
626
binder to bind up the aggregates together. compared to the modification with CWD only.
However, the CWD particles were affixed to Due to the addition of CWD and the PET,
the aggregate surface during the mixing and combine MHMA showed less bulk density
casting stages of the samples. Therefore, a compared to all other modification. Maximum
better surface finish was achieved. Due to the 15% of bulk density was reduced compared to
low bulk density (0.61) of CWD, it was replaced the control mix at 15% addition of PET. The
in volume basis (Weight of CWD having variation of bulk density of combined MHMA
volume equal to volume of filler aggregate) to was considerably small at 5% to 15% of PET
keep the standard sample size. The addition of (Figure 8). When the bitumen percentage was
1g CWD replaced the 4.3g of conventional filler increased the air void of the modified sample,
(Stone dust). Maximum 24% of the weight of was decreased like control. Air void was also
filler was substituted with 40g CWD. Due to the reduced with the addition of PET. At the low
absorptive nature of CWD, the bitumen added binder content, the air void was a little high
to the mix was partially absorbed by the CWD compared to the control sample. Beyond 4.5%
particles. After being saturated, the rest of the of modified binder percentage, the air void was
bitumen added, was acted as the binder to tie less (Figure 9). VMA was within the allowable
up the aggregate skeleton. range of specification (VMA > 13); however,
compared to the control that was significantly
Addition of CWD reduced the specific gravity less.
by 5-10% at 1.77% to 3.75% of CWD
replacement. Due to the fineness of the 2.35
particles, void in the modified HMA was
reduced (Figure 5). Moreover, the tiny particles 2.25
of CWD filled the small cracks of aggregates.
Specific Gravity
Compared to the control mix, the VIM also was

2.15
less. However, it was within the allowable
range. VMA of the modified HMA was less
compared to the control mix and that satisfied 2.05
the specification (VMA > 13).
1.95
The stability of MHMA with CWD was
reduced by the addition of the CWD. The 1.85
replacement of 1.77% (20g) showed better result 3.5 4.5 5.5 6.5
compared to the other replacement. Even Bitumen Percentage (%)
though, 2.73% of CWD replacement had less 5% PET + 2.73% CWD 10% PET + 2.73% CWD
stability which satisfied the minimum 15% PET + 2.73% CWD Control
requirement (MSmin = 5.33kN, CIDA
specification) at 4.8 to 5.2% of binder content. Figure 8 – Specific gravity of combine MHMA
MHMA at 3.75% of CWD replacement was 5.0
completely failed under the loading (Figure 6).
Flow value was also higher due to the softness
Void In Mix (VIM - %)
4.5
of the CWD particles (Figure 7). Except for
3.75% of replacement, other flow values was
4.0
within the allowable range. The flow value was
increased by the increment of replacement rate
of CWD. As the result of this modification, 3.5
optimum replacement rate of CWD should be
2.73% (around 30g). 3.0
5.3 HMA with CWD as filler and PET as 2.5

binder additive
In the previous case, due to the lack of bonding 2.0
ability, CWD particles were removed from the 3.5 4 4.5 5 5.5 6 6.5
sample during the casting process. In the Bitumen Percentage (%)
combined modification, due to the modified 5% PET + 2.73% CWD 10% PET + 2.73% CWD
binder with PET, bonding ability was improved 15% PET + 2.73% CWD Control
well. In the low binder percentage also, the Figure 9 – VIM of combine MHMA
samples showed better finishing surface
5
627 ENGINEER
traffic load condition at 5.65 to 6 % of modified
bitumen. The flow value of the modified HMA
14 with PET and CWD was higher than the control
sample (Figure 11). However, the flow value
was within the allowable range. Compared to
Marshall Stability (kN)
12
modified sample with CWD only, deflection
was less due to the addition of PET. Because of
10 PET modification, the stiffness of the combined
modified mix was improved. It was found that
to improve the properties of the 2.73% of CWD
8
MHMA, minimum 5% of PET is required.
6 5.4 Absorption
Minimum Specification = 5.33kN The absorption of the single modification of
HMA with CWD was increased with the CWD
4 content. With time, the absorption of the
3.5 4.5 5.5 6.5
Bitumen Percentage (%) modified sample was increased until reach to
5% PET With 30g SDA
CWD 10% PET With 30gCWD
SDA the saturation point. The period for the
15% PET With 30g CWD
SDA Control saturation (15days) did not vary with the CWD
content. At the saturation stage, the absorption
Figure 10 – MS of combine MHMA of MHMA having 2.73% of CWD was increased
by 26.7% compared to the MHMA with 1.77%
18
of CWD. When the PET and CWD added in the
same mix, the absorption was increased with
16 the addition of the 5% to 10% of PET.
5.5
Marshall Flow
14 5.0
4.5
Absorption (%)
12
4.0
3.5
10
3.0
2.5
8
3.5 4 4.5 5 5.5 6 6.5 2.0
Bitumen Percentage (%) 1.5
5% PET + 2.73% CWD 10% PET + 2.73% CWD 1.0
15% PET + 2.73% CWD Control 0 5 10 15 20
Days
20g 30g
Figure 11 – MF of combine MHMA 5% PET + CWD 10% PET + CWD
The addition of the modified binder with PET

improved the Marshall properties of the Figure 12 – Absorption rate
modified HMA with CWD and PET. Even The 5% of PET modification resulted in 50%
though the stability was lesser than the control, increment in the absorption, compared to the
while adding the PET with CWD, stability was 1.77% of CWD modified HMA mix. The
increased compared to the single MHMA addition of PET increased the percentage of the
concrete with CWD as a filler (Figure 10). Even void, because the molecular size of the binder
though the 2.73% of CWD modified sample particles may become as bigger after the
partially satisfied the stability requirement by addition of the PET. During the mixing process,
single modification, the 5% of PET modified bitumen was absorbed by the CWD. However,
binder improved the stability by 30%. The CWD absorbed more conventional bitumen
addition of 15% of PET modified binder than modified bitumen with PET. Up to 10% of
improved the stability of the modified mix with the PET addition, the absorption was increased.
CWD by 103%. However, compared to HMA Then beyond the 10% of PET, the absorption of
stability was reduced by 13.3%. In the 5% of the combine modified HMA was decreased.
PET MHMA with CWD, sample was not Maximum 73% of increment was resulted at the
satisfied the CIDA specification for the medium 10% of PET modification. At 15% of PET
ENGINEER 6
628
modification, absorption rate was reduced by The CWD can undergo to the carbon caption
18.2% compared to 10% of PET modified HMA which may reduce the amount of the CWD
mix. When the PET content increased, the CWD with the time. Therefore, the design life of this
particles were covered with impermeable modified sample should be investigated to
modified binder film. Due to this process, the fulfil the design requirements.
absorption of the combined modified HMA
was reduced. Therefore, the absorption can be Acknowledgement
controllable with the addition of the PET in the
combined MHMA with CWD and PET. Special gratitude of the Authors goes to the
Province Road Development Department
6. Conclusions (Southern Province) who provided various in
kind support to conduct this study.
The PET modified HMA showed better
performance against Marshall and bulk References
properties. PET as an additive, improved the
bonding ability. The 15% of PET modification 1. Mallick, R. B. and El-Korchi, T., 2013. Pavement
resulted in maximum 67.4% increment in engineering: principles and practice. CRC Press.
stability. Marshall Flow was increased with
PET. However it was satisfied the specification. 2. Ahmadinia, E., Zargar, M., Karim, M.R.,
Abdelaziz, M. and Shafigh, P., 2011. Using waste
plastic bottles as additive for stone mastic
In the CWD modified HMA, even though the
asphalt. Materials & Design, 32(10), pp.4844-4849.
bulk density was improved by 10%, Marshall
Stability was reduced by 60% at 2.73% of CWD 3. Akbulut, H. and Gürer, C., 2007. Use of
when compared to control. However, stability aggregates produced from marble quarry waste
requirement was satisfied. Maximum 2.77% of in asphalt pavements. Building and environment,
CWD can be replaced with conventional filler. 42(5), pp.1921-1930
The deflection of MHMA with CWD was
increased with the addition of CWD. 4. Al-Hadidy, A. I. and Yi-qiu, T., 2009. Effect of
polyethylene on life of flexible
pavements. Construction and Building
The modification with both PET and CWD in Materials, 23(3), pp.1456-1464.
the same mix, improved the Marshall Stability
of the sample, compared to MHMA with CWD 5. Arabani, M., Tahami, S. A. and Taghipoor, M.,
only. When Marshall Stability was compared 2017. Laboratory investigation of hot mix
with 2.73% of CWD only, stability was asphalt containing waste materials. Road
increased by 30% to 103% due to the addition of Materials and Pavement Design, 18(3), pp.713-729.
5 to 15% of PET respectively. However, stability
6. Arasan, S., Yenera, E., Hattatoglu, F.,
of combined MHMA was reduced by 13.3% at Hinislioglua, S. and Akbuluta, S., 2011.
15% PET content compared to control. The bulk Correlation between shape of aggregate and
properties also showed better improvement mechanical properties of asphalt concrete:
because of the CWD having tiny particles. Light Digital image processing approach. Road
weight MHMA was produced with maximum Materials and Pavement Design, 12(2), pp.239-262.
15% of weight reduction.
7. Chen, J. S., Shiah, M. S. and Chen, H. J., 2001.
Further, the absorption of combined MHMA Quantification of coarse aggregate shape and its
effect on engineering properties of hot-mix
was controlled by the addition of PET modified
asphalt mixtures. Journal of testing and evaluation,
binder. Beyond 10% of PET modification, the 29(6), pp.513-519.
absorption reduced well compared to single
MHMA with CWD. 8. Chen, J.S. and Liao, M.C., 2002. Evaluation of
internal resistance in hot-mix asphalt (HMA)
7. Future direction concrete. Construction and Building
Materials, 16(6), pp.313-319.
In this study maximum 5 to 10% of PET was
9. Earnest, M.D., 2015. Performance Characteristics
good enough to modify the MHMA with CWD
of Polyethylene Terephthalate (PET) Modified
to the medium traffic load. However, Up to Asphalt.
15% the stability increased with the addition of
the PET. Therefore, the optimum PET content 10. Hassani, A., Ganjidoust, H. and Maghanaki,
should be greater than 15%. A.A., 2005. Use of plastic waste (poly-ethylene
terephthalate) in asphalt concrete mixture as
7
629 ENGINEER
aggregate replacement. Waste Management & ashes. Cement and Concrete Research, 38(6),
Research, 23(4), pp.322-327. pp.861-869.
11. Hikkaduwa, H. N., Gunawardana, K. W., 22. Osinubi, K., Edeh, J. and Onoja, W., 2012.
Halwatura, R. U. and Youn, H. H., 2015. Sawdust ash stabilization of reclaimed asphalt
Sustainable Approaches to the Municipal Solid pavement. In Testing and Specification of Recycled
Waste Management in Sri Lanka. 6th International Materials for Sustainable Geotechnical Construction.
Conference on Structural Engineering and ASTM International.
Construction Management 2015, Kandy, Sri Lanka,
11th - 13th on. December 23. Perera, K. K. C. K., Rathnasiri, P. G., Senarath, S.
A. S., Sugathapala, A. G. T., Bhattacharya, S. C.
12. Huang, Y., Bird, R. N. and Heidrich, O., 2007. A and Salam, P. A., 2005. Assessment of
review of the use of recycled solid waste sustainable energy potential of non-plantation
materials in asphalt pavements. Resources, biomass resources in Sri Lanka. Biomass and
Conservation and Recycling, 52(1), pp.58-73. Bioenergy, 29(3), pp.199-213
13. Jambeck, J. R., Geyer, R., Wilcox, C., Siegler, T. 24. Rahman, W. M. N. W. A. and Wahab, A. F. A.,
R., Perryman, M., Andrady, A., Narayan, R. and 2013. Green pavement using recycled
Law, K. L., 2015. Plastic waste inputs from land polyethylene terephthalate (PET) as partial fine
into the ocean. Science, 347(6223), pp.768-771. aggregate replacement in modified asphalt.
Procedia Engineering, 53, pp.124-128.
14. Kandhal, P., Lynn, C. and Parker, F., 1998.
Characterization tests for mineral fillers related 25. Sengoz, B. and Topal, A., 2005. Use of asphalt
to performance of asphalt paving mixtures. roofing shingle waste in HMA. Construction and
Transportation Research Record: Journal of the Building Materials, 19(5), pp.337-346.
Transportation Research Board, (1638), pp.101-110.
26. Shyamalee, D., Amarasinghe, A. D. U. S. and
15. Kandhal, P. S. and Parker, F., 1998. Aggregate Senanayaka, N. S., 2015. Evaluation of different
tests related to asphalt concrete performance in binding materials in forming biomass briquettes
pavements (No. 405). Transportation Research with saw dust. International Journal of Scientific
Board. and Research Publications, 5(3), pp.2250-3153.
16. Liang, Z. Z., Woodhams, R. T., Wang, Z. N. and 27. Sinsiri, T., Kroehong, W., Jaturapitakkul, C. and
Harbinson, B. F., 1993. Utilization of recycled Chindaprasirt, P., 2012. Assessing the effect of
polyethylene in the preparation of stabilized, biomass ashes with different finenesses on the
high performance modified asphalt binders. compressive strength of blended cement paste.
In Use of Waste Materials in Hot-Mix Asphalt. Materials & Design, 42, pp.424-433.
ASTM International.
28. Sojobi, A. O., Nwobodo, S. E. and Aladegboye,
17. Marteano, D., 2002. The performance evaluation of O.J., 2016. Recycling of polyethylene
hot rolled asphalt mixed with sawdust ash as a filler, terephthalate (PET) plastic bottle wastes in
Ph. D. Doctoral dissertation, Magister Teknik bituminous asphaltic concrete. Cogent
Sipil. engineering, 3(1), p.1133480.
18. Menikpura, S.N.M., Gheewala, S.H. and Bonnet, 29. Tillson, G. W., 1900. Street Pavements and
S., 2012. Sustainability assessment of municipal Paving Materials: A Manual of City Pavements:
solid waste management in Sri Lanka: problems the Methods and Materials of Their
and prospects. Journal of Material Cycles and Construction.
Waste Management, 14(3), pp.181-192.
30. Viswanathan, C., 2006. Domestic solid waste
19. Modarres, A. and Hamedi, H., 2014. Effect of management in South Asia. Environmental
waste plastic bottles on the stiffness and fatigue Engineering and Management Program Asian
properties of modified asphalt mixes. Materials Institute of Technology Thailand. http://www.
& Design, 61, pp.8-15. iges. or.
jp/en/ltp/pdf/activity08/07_visvanathan. pdf.
20. Moghaddam, T. B., Soltani, M. and Karim, M. R.,
2014. Experimental characterization of rutting 31. Introduction to Uva Province (2009), Waste
performance of polyethylene terephthalate Quantification and Characterization – Sri Lanka,
modified asphalt mixtures under static and visited 10th May 2016
dynamic loads. Construction and Building http://www.unep.or.jp/ietc/spc/activities/GP
Materials, 65, pp.487-494. WM/data/T1/AB_3_WasteQC_SriLanka.pdf
21. Nair, D. G., Fraaij, A., Klaassen, A. A. and

Kentgens, A.P., 2008. A structural investigation
relating to the pozzolanic activity of rice husk
ENGINEER 8
630
Annual SessionsofofEngineers,
© The Institution IESL, pp. Sri
[1 -Lanka
10], 2017
Evaluating Road user Cost for Highway Work Zones –

Case Study for Urban Upgrading Projects in Sri Lanka
M.H.M. Hamsath and H.R. Pasindu
Abstract: A work zone is defined as a segment of highway in which maintenance and

construction operations affect the number of lanes available for traffic or affect the operational
characteristics of traffic flowing through the segment. Work Zone Road User Cost (WZRUC) is
the additional cost borne by the road users as a result of the work zone activities. It is a
combination of Value of Time Cost (VoTC), Vehicle Operating Cost (VOC), and Vehicle Emission
Cost (VEC), Accident Cost (AC) and some non-monetary values such as noise and impact on
social inconvenience. This paper is to identify the impact of Work Zone Road User Cost from
upgradation projects in Sri Lanka compared with the real economic costs.
This study evaluated the five road upgrading projects in Colombo district and based on the study
finding the economic cost during the construction in upgradation. WZRUC is in the range of 13-
45 % of the total project cost. The major cost component of which is delay cost which is in the
range of 8-27% of the total project cost. By evaluating the WZRUC, the real economic cost of
upgradation projects can be assessed. This can be used in evaluating its economic impact of the
project, in assessing work zone management strategies that may reduce the additional economic
cost during construction stage. Thereby, the contractors can be given more specific guidelines to
implement a proper work zone management plan.
Keywords: Road user cost, Value of time cost, Vehicle operating cost, Emission cost
1. Introduction of the projects and disincentives to the

contractors for the delay in projects as well[1].
There has been a rapid increase in trip Delay in completions is one of the major
generation in urban areas in Sri Lanka and problem in highway construction projects in Sri
urban upgradation projects are very common in Lanka and all around the world as well[2]. As a
urban areas. As a result of increased result of delay in completion a huge amount is
upgradation projects in urban area, the traffic wasted by the government (by additional traffic
congestion has been increased and this has control measures & maintenance) and road
caused severe concern regarding the liveability users (Road User Cost - RUC). Also Work
in urban areas especially in Colombo area and zones formed by construction works lead to
its suburbs. As a result of increased road works delay in travelling to road users. Main
in urban areas, these activities cause traffic inconvenience causes by the work zones to the
disruptions, safety aspects to road users and road users in urban area is the delay in
construction workers and negative impact on travelling. Proper management of available
local business and community. lanes to road users is a major role of the
contractors and they fail to maintain it properly
The impact of the WZRUC are neglected in Sri in Sri Lankan practice because the contractors
Lankan practice, in the estimation of benefits do not consider the inconvenience to the public
from an upgradation project. As it is neglected and additional cost borne by the road users.
in the benefit cost analysis, the contractors do This study will estimate the additional cost
not consider the impacts at the construction borne by the road users as a result of high way
stage. Since they do not consider about the work zones in urban upgrading projects in Sri
impacts, the work zone management is very Lanka.
poor in Sri Lankan urban upgradation projects.
But in developed countries, the WZRUC is M.H.M. Hamsath B.Sc. Eng. Hons (Moratuwa)
estimated through various tools and it is used
Eng. (Dr.) H.R. Pasindu, B.sc. Eng. Hons (Moratuwa), Ph.D.
to estimate the incentives to the contractors to (NUS)
decrease the project duration, early completion Senior lecturer, Department of Civil Engineering, University of
Moratuwa.
631 1
2. Objectives was by the road user charges studies, which
calculated roughness – cost relationships for
Since there are no prior studies available on the eight representative vehicle type using HDM-lll
estimation of WZRUC in Sri Lanka, the models. Emission cost was estimated according
objective of this research is to estimate the to the fuel consumption rate (from HDM-lll
WZRUC as a percentage of total construction model) and the amount of pollutants per liter of
cost of the project in urban upgradation projects fuel consumed.[4] The cost of pollutants was
in Sri Lanka. A case study is done in 05 selected estimatedfor four pollutants (CO, NOx, SOx&
urban upgrading roads with work zone and the PM) – emitted by thermal power generating
WZRUC is estimated as a percentage of total projects. Current exchange rate was used to
construction cost of the project. convert the value for Sri Lankan values[5].
Accident cost was also estimated according to
3. Literature Review the damage type from the accident. Since there
are no studies available on estimating accidents
WZRUC is a combination of Travel delay cost,
in work zones, this cost is neglected in this
Vehicle operating cost, emission test and
research.
Accident cost as described in previous
literatures.[2]&[3]
3.2 Application of WZRUC
3.1 Estimation of WZRUC

Feasibility studies –At the feasibility analysis
There are no specific studies available on the WZRUC analysis should be considered
evaluation of WZRUC in Sri Lankan practice. whether the project is going to be a real benefit
This research produces a method to evaluate to the society. Likewise, at the Benefit cost
the WZRUC for Sri Lankan urban upgrading analysis WZRUC should be included to
projects. When a road is upgrading, at the estimate the real benefits from the upgradation.
economic analysis of benefit cost analysis the
major benefits are considered as Travel time Selecting the contract type – Delay in
reduction, reduction in Vehicle operating cost, completion has been a major issue in the
reduction in Vehicular emission and reduction delivery of highway construction projects in Sri
in accident costs. For the evaluation of WZRUC Lanka. Delay in highway projects do not only
this approach is used to estimate the additional result in cost over turn but also cause adverse
cost borne by the road user as a result of work economic impacts and disruptions on public
zones. travelling. The Highway agency in United
The benefits are estimated according to the States provides measure to shortening project
report of Assessing Public Investments in the delivery to manage the overall impacts of
Transport Sector [2] the estimations are construction delays and associated road user
available in Rupees value in 2000. Estimation of costs by evaluating WZRUC and shorten the
Travel time saving – It is estimated by the construction time by selecting various contract
potential time saving after the project types.
implementation, Estimation of Value of time –  Incentives/Disincentives for early/late
The report described the value of time for completion -
passenger vehicles and freight consignee In this approach, the Contractor is required
differently. Vehicle operating cost was to complete the project by the Engineer’s
estimated by HDM –lll model for passenger car. estimation of the contract time specified in
The following inputs were added as cost of the bid document. Upon completion the
inputs – Fuel cost, Lubricant cost, Tires, contractor is rewarded with bonus payments
Maintenance labor, Maintenance parts and for early completion and penalized with
crew costs. Relationship between speed, disincentives for late completions. The
roughness and costs has not been developed for incentive payments should be higher than
Sri Lanka. The most detailed work done to date the additional costs incurred by the
contractor for accelerating the work.
632 2
a single contract. This method involves two
components –
Cost (A) – The dollar amount of contract
items (Equipment, material& Man
power) for all works to be conducted
under traditional low – Bid contract.
Time (B) – The dollar for the component
of a contract, estimated by multiplying
the number of calendar days to complete
the work by the daily road user cost.
Figure 1- Relationship between project cost Bid Value = (A) + (B * Daily road user
and duration cost)
The contract is then awarded to the lowest
 Lane rental – The contractor pays a rental fee bid value for contract award. This formula
for the time period for a lane when it is is not
closed for traffic as a result of work zones. used in determining the payment to the
This is intended to minimize the disruption contractor. The contractor receives
of the work zone traffic and to motivate incentives for
minimum use of lanes for construction early completion and is required to pay
activities. The owner agency determines the disincentives (and liquidated damages) for
numbers and the duration of the lane delaying beyond the completion date
closures. The lane rental fee is estimated agreed in the contract. Generally,A+B
using the WZRUC of the closure period. bidding is suitable for time-critical projects
Therefore, contractors are motivated to close such as high-traffic volume roadways,
the lane in off peak hours because of low business, tourist, and environmentally
WZRUC in off peak hours. In some cases, sensitive areas. Typical projects include
the contractor may be allowed to propose new/reconstruction, rehabilitation projects,
the required amount of closure time and simple bridge replacement projects, detour
number of closures in their bid submissions. projects, intersection upgrades, and bridge
Generally, lane rental is suitable when rehabilitation projects.
detours are long, unavailable, impractical, or
when peak hour traffic is impacted 4. Methodology
adversely. It is well suited for multiple lane
roads with high traffic volumes where there 4.1 Scope
is flexibility for intermittent or temporary  05 numbers of urban upgrading roads were
lane closures to keep at least one lane open selected for the case study
to traffic through the work zone. Typical
projects include mill and overlay, temporary o B295 – (2+620 km – 5+120 km)
widening, patching, diamond grinding, o B389– (5+700 km – 7+840 km)
dowel retrofitting, reclamation and o B267– (0+000 km – 6+150 km)
recycling, guardrails, striping, signing, o AB10 – (8+650 km – 9+100 km)
bridge painting, crack sealing, signal o Old Kottawa road – (0+000 km – 1+900
systems, and traffic management projects. km)
Lane rental is not appropriate for the
projects where long-term permanent lane  02 Segments were identified along each road
closures are required, such as bridge re-deck – Normal segment (Before upgradation),
or concrete rehabilitation projects. Work zone.
 A+B Bidding – Allow the agency to request  All other relevant details of the project were
bids for the cost of work items and the time collected (Type of upgradation, duration,
to complete the work and procedure them in project cost & length of the road to be
upgraded)
633 3
 Speed survey was done by video matching Bus 32 4% 37 12
method in each road, each segment in a Truck 17 2% 35 10
peak and off peak hours.
MGV 65 7% 40 15
 From the speed survey – Vehicle count,
Average Speeds of different vehicles, and
From the speed survey – Speeds were close to
reduction in speeds from a normal segment
the speeds in the B295 road in other roads as
to work zone was calculated in peak and well.
off peak hours.
 Road user cost values are updated to 2016 4.3 Estimation of RUC in 2016 values
values by the guideline to estimate the road
4.3.1 Travel Delay cost
user costs from the Department of National
Planning (DNP). Travel delay cost is calculated by multiplying
 WZRUC cost was evaluated and compared the additional delay to passenger vehicle and
with the total project cost. freight consignees caused by the work zone (hr)
by the value of time (Rs / hr).
4.2 Data collection
4.3.2 Delay time calculation
Table 1 - Result of speed survey in B295 road
in peak hour Delay time = Time taken to travel in effective
work zone – Time taken to travel in normal
Peak hour 5.00 pm - 6.00 pm road
Percent Average Speed Some assumptions were made to calculate the
Vehicle Nos age (kmph) delay time cost in this research case study. They
Normal Work zone are,
 Effective work zone length = (150 + 100 + 150)
Car 416 29% 25 10 = 400 m (Average queue length observed in
3 peak time was 150 m & length of work zone
Wheel 373 26% 30 13 is 100 m)
 Queue were formed in peak hours only & no
Mo.
queue in off peak times
Bike 458 32% 40 25  Peak day time (07.00 -10.00), off peak day
Bus 57 4% 22 8 time (05.00 - 07.00 & 10.00 – 16.00), Peak
evening time (16.00 – 19.00) & off peak
Truck 43 3% 22 7
evening time (19.00 – 23.00)
MGV 85 6% 25 10  Evening 23.00 pm – morning 5.00 am no
value was assumed
Table 2 - Result of speed survey in B295 road  Its assumed that Saturdays = 75% of week
in off peak hour day value and Sunday = 25% of week day
value ( 6 week days per week)
Off Peak hour 10.00 am - 11.00 am
 Only one work zone exist along the road at a
Average Speed day and construction works equally interrupt
Vehicle Nos Percentage (kmph) the users along the entire construction period.
Work
4.3.3 Monetary Value of time
Normal zone
The monetary value of travel time is defined in
Car 180 20% 40 15
Federal Highway Administration [1] as it is
3 Wheel 283 31% 35 20 based on the concept that time spent traveling
Mo. Bike 326 36% 50 30 otherwise would have been spent productively,
whether for remunerative work or recreation.
634 4
Saving in travel time is major concern in Table 3 - Monthly income rate comparison
upgrading a road. The system used in Sri Lanka
User Assumed Income Mean monthly HIR
to estimate the saving in travel time while
group Group Income (Rs) Rs/hr
upgrading a road is based on a survey/
estimation done in 2000 by the DNP[2]. This 1999 2015 1999 201 2015
research updates the monetary values 5
according to the DNP guideline to 2016 values. Car 10th 8th - 21,465 685 411
The monetary value of time is calculated in two
Decile 10th 73
categories, Passenger Vehicle and Freight
Decile
consignees.
Motor 7th&9th 4th – 7th 4770 316 190
4.3.3.1 Passenger Vehicles bike Decile Decile 92
Three NA 5th – 8th NA 385 231
Travel time saving for the passengers is the wheel Decile 68
most important aspect when considering the
Bus 1st to 8th 1st – 7th 3387 243 146
benefits from the urban upgrading road
projects. When we consider the work zone Decile Decile 35
stage of an upgrading work, the additional *NA – Not Available

travel time spend in the road is more
 Weighted Hourly Income (WHI) – HIR
significantly affects the road users than the
depends on the average income of the
other factors. From the analysis done by the
respective group. However, mobility indices
DNP of the Ministry of finance and planning,
for travel time in Sri Lanka, which means
the value of travel time depends on the
higher income earners within a particular
passenger income and the purpose of the trip.
user group make relatively more trips than
The value of time of the same person
lower income earners in the group. Therefore,
undertaking trips for different purposes can
Hourly income rate is multiplied by 1.5 to
vary accordingly. VoTCis calculated on the
adjust for mobility
basis of
WHI = Mean monthly Income of user group
 Hourly income rate of the passengers
(Rs/Month) x1.5x 12 (Months/year) /
 Trip purpose of the passengers
2000(hrs/year)
 Quantum of travel time saved
 Trip purpose
 Hourly Income Rate(HIR) of the passengers
As the literature reveals trips undertaken in
Hourly income rate is determined by obtaining
working time and hence referred to as work
the average annual income for the user and
time trips, should be valued at the hourly
dividing by the number of working hours per
income rate plus an allowance for employer’s
year.
overhead.
HIR = Work time VoTC = Hourly income rate x

1.2
Non Work time VoTC = Hourly income rate x
0.2
Table 4 - Assumption of work time trips
percentage
User In In 2016
group 1999 Mor. Day Eve. Night
– Peak Off peak time
peak
Car/Je 15 % 85 % 40 % 75 % 5 – 10 %
635 5
ep Travel time is an important function in the
Mo. 10 % 80 % 55% 70 % 5 – 10 % movement of freights as well. Increased travel
Bike
time induces higher transport cost.
3 NA 80 % 45 % 65% 5 – 10 %
 Value of time (VoT) = Number of trucks x
wheel (Corresponding) Percentage x
Bus 4% 60 % 15% 55% 5 – 10 % (Corresponding) Capacity
x(Corresponding)Cost per ton per hour.
Value of User group = WHI of user group x (1.2
x Percent of Work Trips + 0.2 x Non - Percent of  VoT for Perishable Commodities
Work Trips)/100 Perishable commodities can have an additional
 Quantum of travel time saved VoT arising from the prevention of
In some circumstances even a smaller value of deterioration in quality of the product being
travel time can have very high value (E.g.: - transported. From the user estimates the
Salary reduction for late arrivals) this reflect following values assumed.
value of reliability rather than value of time per
second. But in this study it is not considered Table 7 - Cost of refrigeration
and assumed that the value of time is randomly Size of truck Cost of Refrigeration (₨
distributed. per hr )
Table 5 - Monetary value of time in 2016 Small (0.5 -2 ton ) 150
Medium (2- 10 ) 250
User WHI Value of time ( Rs/hr) – 2016
group Morning Day Eve Nig
It is assumed that 50 % of Delivery vans and 25
Peak Off ning ht % of Small trucks are transporting Perishable
peak pea time commodities.
k
Car 617 687 392 621 180  Average VoT of a freight vehicle (MGV)
Motor
= (0.35 x 0.5 x 1200 + 150) + (0.65 x 1.5 x
285 302 227 272 83 1000 + 250) = 1640
Bike
Rs/hr
Three  Average VoT of a freight vehicle (LGV)
356 377 245 321 104
wheel = (0.75x10x335) + (0.25 x 25 x 215) =
Bus 219 186 81 174 64 3850 Rs/hr
4.3.3.2 Freight Consignees 4.3.4 Vehicle Operating Cost

Table 6 - Capacity of freight vehicles & Value of
freight Vehicle operating costs are fuel cost, lubricants
cost, tire depreciation, labour and parts for
Type of vehicle Percentage Capacity Cost/
repair and maintenance, operator (crew) labour
(tons) ton/hr if applicable, depreciation and interests and
Medium Delivery ₨ overheads such as insurance and license fees
35 % 0.5
Good Vans 1200 and administration. From a HDM – 4 model
Vehicle
developed in 2016,it is concluded that, VOC in
Small ₨
65 % 1-2 2016 is 3.165 times greater than VOC value in
(MGV) truck 1000
2000[6].
Large Medium
75 % 5-15 ₨335
Good Truck
Vehicle Large
25 % 15-40 ₨215
(LGV) Truck
636 6
100
 Estimation of emissions per liter of fuel
90 consumed
80 The estimation of emissions per liter of fuel
70 consumed is based on DNP guideline and the
VOC (Rs/km)
60 monetary value of emissions were estimated

from the Caltrans report of California
50
Department of transportation.[8].The monetary
40 value of emissions is defined as “Emission costs
30 are social costs that are borne directly by the
road users, but are estimated based on the
20
10 20 30 40 50 60 70 80 90 100 impact borne by the society in general”. There
Speed (kmph) is no consensus on how to assign a dollar value
VOC at IRI = 6 for car
to quantify the impact of each pollutant type.
VOC at IRI = 3 for car Unit cost of emissions typically used in practice
VOC at IRI = 6 for Bus is derived based on the economic analysis of
Figure 2 - VOC comparison of Passenger cars health impacts caused by air pollutants and
and Bus greenhouse gases.
4.3.5 Vehicular Emission Cost  Estimation of cost of emission per liter of fuel
Transport activities lead to air pollution directly consumed was calculated by multiplying
from emission and indirectly from refining grams of emission per pollutant, by
petrol and diesel. The following components respective costs of damage.
are from vehicle emission to the air – Ozone,
Carbon Monoxide (CO), Particulate Matter  Estimation of cost of Emission per kilometer
(PM), Lead, Carbon Dioxide, Methane, by vehicle type at 40 kmph were calculated
Hydrocarbons and Sulphur oxides. by dividing the cost of emission per liter fuel
(Rs/l) by fuel consumption rate (km/l)
From a research of Vehicle Emission Inventory [(Rs/l) / (km/l) = (Rs/km)]. Fuel
for Sri Lanka[7] conducted by Liyange, the consumption rate was calculated from the
following graph was drawn for the Carbon HDM – 4 model
Monoxide (CO) emission from passenger car at
700.0
different speeds. Other emission gases also
600.0
follow very similar pattern. Since there are no
500.0
researches available on vehicle emission
400.0
studies, this study assumed that the emission
300.0
ratio at speed 10-15 kmph is 3 times greater
200.0
than speed 40 kmph. For the estimation of
100.0
Emission cost, it is assumed that vehicles are
0.0
moving at 40 kmph speed in normal road
Car Three Motor LGV/bus MGV
segment and 10-15 kmph speed in Work zone. wheel bike
25
Figure 4 - Emission cost at kmph speed (Rs/
CO emission g/km
20 1000km/veh)
15
5. Results
10
5  Speed survey & Speed reduction percentage
0
0 50 100 150 From the speed survey – Speed reduction from
Speed - kmph normal segment to work zone was analyzed
and there was a significant reduction in speeds
Figure 3 - CO Emission rate with speed for in work zones were identified.
passenger car
637 7
40  Estimation of work zone travel delay cost -
The speeds in work zone and normal road
35
segment are calculated from the speed survey
30
and the work zone travel delays for different
25 vehicle are estimated. The travel delay cost is
Kmph
20 calculated by multiplying the delays by

15 monetary value of time for different vehicle
10 types.
5 Table 8 - Estimation of work zone travel

0 delay cost
Car
3Wheel
Mo.Bike
Bus
Truck
MGV
Road Travel delay Cost
Normal Work zone B295 Rs. 58,159,960.00
B389 Rs. 53,389,771.00

Figure 5 - Speeds in Normal road and work B267 Rs. 25,877,947.00
zone in B295 road
Old Kottawa road Rs. 20,158,396.00
AB 10 Rs. 41,933,546.00
68%
64%
60% 60%
57%
 Estimation of additional VOC - The speeds in
work zone and normal road segment are
38% calculated from the speed survey and the
road roughness (IRI measurement) is
measured by Roadroid Android application
and it is concluded that, the road roughness
CAR THREE MOTOR BUS TRUCK MGV in work zones are IRI = 6 and road roughness
WHEEL BIKE in normal segment IRI = 3. The vehicle
Figure 6 - Speed reduction from normal road to operating cost for different speed and
work zone in peak hour in B295 different vehicles are adjusted to 2016
monetary values.
71% Table 9 - Estimation of additional VOC

68%
63% 63%
Road VOC
43% B295 Rs. 18,967,084.00

40%
B389 Rs. 21,463,383.00
B267 Rs. 14,971,848.00

CAR THREE MOTOR BUS TRUCK MGV AB 10 Rs. 12,277,143.00
Figure 7- WSpeed
HEEL reduction
BIKE from normal road to
work zone in off peak hour in B295
 Estimation of additional vehicular
emission cost - The speeds in work zone and
normal road segment are calculated from the
speed survey and additional emission cost is
calculated as described above in methodology.
638 8
100%
Table 10 - Estimation of additional vehicular 90%
emission cost 80%
70%
60%
Road Emission Cost 50%
40%
B295 Rs. 13,391,817.00 30%
20%
B389 Rs. 17,520,958.00 10%
0%
B267 Rs. 9,858,991.00 B295 B389 B267 Old AB 10
Kottawa
road
AB 10 Rs. 14,690,224.00 WZRUC Construction Cost
Figure 9 - Comparison of WZ RUC with

construction cost
 Estimation of WZRUC - Is the addition of
Travel delay cost, Vehicle operating cost & 6. Conclusion
Emission cost caused by the work zones.
Following table elaborates the summary of
The WZRUC is varies from 13 – 45 % compared
Work zone Road user cost in different work
to the total project cost and the major
zones.
component of the cost is travel delay cost which
WZRUC is around 60% of the WZRUC and 08- 27 % of
the project cost. The WZRUC is affected by the
Emission traffic volume of the road, Vehicle composition,
20% Project duration and Type of upgrading works.
Dealy When we consider the first four roads (B295,
VOC Cost B389, and B267 & Old Kottawa road) the
21% 59%
percentage of WZRUC compared to total
project cost is almost same and which is in the
range of 12.7 – 15.1 % of the total project cost. It
is huge amount borne by the road users as a
result of the work zone activities in urban
Figure 8 - Estimation of WZRUC upgradation projects. Since these roads are
rehabilitating roads (road widening) it can be
 Comparison of WZRUC with Total concluded that the WZRUC is within the range
Construction cost is the major indication of of 12.7 – 15.1 % of the project cost. Since the
the impacts of the WZRUC Evaluation and total project cost is comparably low for pipe
the scope of this research as well. When laying projects (Water and waste water
comparing the real economic benefits from an management) the WZRUC is very highin
upgradation project, this kind of negative percentage of the total project cost. It is
impacts should be included to the benefit cost recommended to develop a proper work zone
analysis of upgradation projects and with the management plan for all the upgradation
identification of the impacts by the work projects in urban areas in Sri Lanka. When
zones, the work zones can be managed considering two, similar traffic volumes with
efficiently. Following table shows the different upgradation works performing roads,
comparison of WZRUC with construction WZRUC for a unit duration is same but when
costs in prices for different work zones. comparing with the project cost, the percentage
of WZRUC is getting higher for low project cost
works. The condition of temporary lanes
allowed for traffic movement is maintained at
very poor level and traffic management
schemes are not implemented often.
639 9
7. Recommendation 5. Munasinghe and Meier, “Incorporation
Environmental Concerns into Power sector
Decision Making - A case study of Sri Lanka,”
 Improving monitoring mechanism to ensure
Environmental Policy and Research Division,
work zone management practices are 1992.
implemented.
Since The Road Development Authority is the 6. Ranawaka, D. “Estimating the Vehicle Operating
client for all the road projects, they have to Cost used for Economic Feasibility Analysis of
Highway Construction Projects,” in Moratuwa
implement a proper monitoring mechanism
Engineering Research Conference 2017, Moratuwa,
to confirm the work zones are properly 2017.
managed in a way to reduce the WZRUC by
increasing the mobility in work zones. The 7. Preethika, L. U. “Vehicle Emission Inventory for
mobility cannot be increased at all the times Sri Lanka,” 2005.
but it should be maximized.
 Providing incentives to contractors to work 8. Department of Transportation -California,,
“Caltrans estimates,” Californiya, 2011.
during off peak hours and expediting project
completion.
9. Highway capacity Manual, Washington DC,
Motivating the contractors to complete the 2000.
project with optimum duration by providing
incentives if they complete the project early 10. Mallela J, and S. Sadasivam, “Work zone road
and deduction of disincentives for the delays. user cost Concept and Application,” US
Department of Transport, 2011.
 Selecting a proper contract method when
requesting for bids. The impact of WZRUC is
11. D. o. N. Planning, “Assesing Public Investment in
included in some contract types described in the Transport sector,” 2000.
Application of WZRUC. With the
identification of the impacts of the WZRUC,
contractors will be encouraged to reduce the
WZRUC as much as possible.
 Reviewing and updating the existing
guidelines on work zone management.
Acknowledgement
Authors would like to express their gratitude to

all the individuals and organizations, which
help to accomplish this research.
References
1. FHA, F. H. A. “Work Zone Moblity and Safety

program - Work Zone Road User Cost,” U.S.
Department of Transportation, Washington, 2012.
2. Department of National Planning, “Assesing

Public Investent in the Transport Sector,”
Ministry of Finanace and Planning, 2000.
3. Federal Highway Administration, “Work Zone

Moblity and Safety program - Work Zone Road
User Cost,” U.S. Department of Transportation,
Washington, 2012.
4. NBRO - Environmental Division, “Air pollution

from motor vehicles,” 1998.
640 10
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[641 647],2017
2017
The Institution
Lanka
Development of a Methodology for Road Maintenance

Planning of Low Volume Roads Based on Roughness
Data
M.A.D. De Silva and H.R. Pasindu
Abstract: Sri Lanka has one of the highest road densities in the region, with most regions in the
country having connectivity to a road network. Successive governments have invested billions of
rupees to upgrade or expand the national road network, especially at provincial or local government
level to improve the accessibility in rural areas. However, there is no systemic process in place to
maintain these roads. Often maintenance is done on ad-hoc basis and there is high political influence
and other societal pressure. The main reason is the lack of a maintenance management system,
budgetary constraints and lack of resources to evaluate the condition of the road and make objective
decisions with regard to implementing maintenance work. The study proposes a new methodology to
assist local authorities make decisions on maintenance works considering the existing condition and
cost of repair works. The existing condition is measured via smartphone based pavement roughness
measurement method. These roughness values are correlated to representative cost values using HSR
based on distress and roughness data. This can be used to select roads for maintenance works based
on the current road condition as well considering the budgetary constraints. Using a prioritizing
criteria and optimization methods such as integer programming, proposed in the study, candidate
roads for maintenance can be selected. The methodology will provide an objective process to make
maintenance decision making which allow the authorities to optimize their budget utilization and
improve the network road condition.
Keywords: Low volume roads, pavement roughness, pavement maintenance, decision support
system, maintenance budget
1. Introduction agencies are unable to define priorities,

quantify disinvestment, or support
Low volume roads (LVR) provide the primary maintenance needs [4].
links to the national highway network. They
provide connectivity from homes and farms to There is no adequate maintenance planning for
markets, and they provide public access to low volume roads in Sri Lanka. Most of the
essential health, education, civic facilities etc. time maintenance is carried out often on ad-hoc
These are critical to economies locally and basis, due to external influence, lack of proper
nationally in all countries of the world. Low plans, and irregularity of funds. One reason for
volume roads are defined as roads having an this is the lack of consistent objective method to
average daily traffic (ADT) less than 400 evaluate the pavement condition and a simple
vehicles per day but typically roads with ADT tool/guideline to help local agencies to make
up to 2000 is considered in this category [4]. decisions with regard to maintenance work.
Many low volume roads consist of single lane In consideration of the above, road roughness,
with gravel or even more native surfacing. In can be considered as a good indicator of
Sri Lanka there are about 70% of low volume pavement condition, provided that roughness
rural roads among all the roads while 30% is measurements can be gathered at a low expense
the other highways. Very few data concerning and with lesser manpower. Using the
LVR performance, use, cost and so forth is roughness data, road agencies are able to
available within road agencies [7]. Fewer the decide on the required mainteance operation
road users, lesser the funding for LVR for the road and thus, perpare a prelimanry
maintenance, restoration. Insufficient funding budget for the maintenance work with
and inconsistent fund allocation is the most
relevant issue associated with LVRs. Another Mr. M.A.D. De Silva, Undergraduate, Department of Civil
Engineering, University of Moratuwa.
major issue with financing and planning LVRs
is the shortage of sufficient data. With Eng. (Dr) H.R. Pasindu, Senior Lecturer, Transportation
Engineering Division, Department of Civil Engineering,
incomplete data, investment strategies and University of Moratuwa, pasindu@uom.lk
funding justifications are fairly weak, and
1
641
prioritizing of the road considering its accessible and simple method is applicable for
condition. measurements.
2. Literature Review As a solution to this, there are new methods

developed in order to obtain road roughness
2.1 Pavement Roughness data. One such popular example is using
smartphones for road roughness estimation [5].
Pavement roughness is generally defined as an Using smartphones to collect data is a
expression of irregularities in the pavement promising alternative because of its low cost
surface such as potholes, depressions, cracking, and ese of use. Moreover, smartphones are very
rutting, ravelling etc. Pavement roughness popular devices all over the country, thus the
affects in many ways such as ride quality, technology easily accessible to any road agency.
vehicle delay cost, safety, asset management, Furthermore, a non-technical employee of an
fuel consumption and maintenance cost etc. organization can operate a simple mobile
Globally accepted indicator to measure application and collect the required roughness
pavement roughness is International data, thus reducing the burden on the technical
Roughness Index (IRI). It is also possible that staff which is often in shortage in the provincial
future pavement conditions might be more level road agencies.
accurately predicted from the components of
the index [3]. 2.2 Low volume road maintenance
management
International Roughness Index was developed
by World Bank in 1982. The IRI was first There are many issues when authorities try to
recommended as a standard for roughness implement these maintenance works. Some of
measurements at the International Road the issues would be [2],
Roughness Experiment conducted in 1982 [12].  Insufficient road maintenance funding
International Roughness Index (IRI) is a scale  Limited construction material
for roughness based on the simulated response  Limited capacity and number of
of a generic motor vehicle to the roughness in a qualified staff
single wheel path of the road surface. Its true  Inadequate maintenance
value is determined by obtaining a suitably  Lack of decision supporting system.
accurate measurement of the profile of the road,
processing it through an algorithm that Typical types of distresses that were observed
simulates the way a reference vehicle would on the sample of roads selected for the study
respond to the roughness inputs, and are fatigue cracking, block cracking, edge
accumulating the suspension travel. cracking, longitudinal and transverse cracks,
potholes, rutting, ravelling etc. The evaluation
Roughness can be measured in several ways of current condition of the pavement and
depending on the accuracy of the measurement. selection of the correct maintenance type is vital
Profilometers, Response type road roughness to ensure the functionality of the roads. By
meters (RTRRMs) are examples that require introducing a proper pavement management
high capital cost and high maintenance cost, system, better decisions can be made in order to
but these methods produce very accurate optimize the pavement condition with
measures. Profiling devices are used to provide maintenance work.
accurate, scaled, and complete reproductions of
the pavement profile within a certain range. Pavement management system performs an
They are available in several forms, and can be analysis of the collected rating data and reports
used for calibration of RTRRMs [8]. But these on the current and projected condition of the
methods have few limitations when road system in order to evaluate the
implementing to low volume roads. These effectiveness of the planning and funding
devices are mounted on to a standard vehicle priority and provide guidance in decision
and the vehicle has to travel in the respective making process [9].
road to obtain the measures. Most of the time, it
is not practical to travel in such vehicles in low Using a pavement management system, there
volume roads as they do not provide sufficient will be effective and efficient directing of the
space for the vehicle to travel. For low volume various activities involved in providing and
roads in Sri Lanka, a cost effective, more sustaining pavements in a condition acceptable
to the traveling public at the least life cycle cost.
2
642
[1] According to research, it has shown that it is
far less expensive to keep a road in good
condition than it is to repair it once it has
deteriorated. Pavement management system
performs the tasks of inventorying pavement
conditions, identifying good, fair and poor
pavements, assigning importance ratings for Figure 2 - Capturing the
road segments, based on traffic volumes, road road condition using video
functional class, and community demand,
scheduling maintenance of good roads to keep
them in good condition and scheduling repairs
of poor and fair pavements as remaining Figure 1 - Roughness
available funding allows. Measuring from
“Roadroid Classic”
There are key measures used to define the
condition of a pavement [13].
 Roughness
 Surface Distress
 Structural Capacity Figure 3 - Accelerometer data transferred to
 Surface Friction spread sheet
Roughness is directly related to both user While traveling, the road surface is captured by
satisfaction and user cost. It is an integral a video camera for cross checking a validation.
component of the pavement quality assessment Roadroid Classic application is available in
in a PMS. In addition, as discussed earlier with Google Play Store for free. This app registers
the smart phone base application, roughness the vibrations from the road and correlates to
measurements can be gathered within a budget the International Roughness Index. It displays
constraint and less manpower in data eIRI, cIRI and location data (speed of the
collection. vehicle, longitude, latitude and Altitude). The
Application calculates eIRI for every 20-meter
3. Evaluation of roughness of low road section. Roadroid application can only
calculate eIRI when driving speed of
volume roads and assessing experiment vehicle is 20 kmph or faster. After
road condition calibrating the smartphone, accelerometer data
from the app is obtained and transferred to
Roughness and distress data collection of low spread sheet. Then those spreadsheets are
volume roads were carried out covering few thoroughly checked and IRI values are matched
districts in Sri Lanka such as Kandy, Galle, to the location of the road for the respective IRI
Matara, Kurunegala, Kaluthara. Roads for data using the images from the video camera. This is
collection were selected to include, low volume done for assessing of the current condition of
roads with varying lengths, pavement the road. Table 1 shows how sample roads are
conditions and pavement types. 20 number of evaluated using Roadroid Classic app, which
roads were selected for the purpose of the shows that it gives satisfactory roughness
study. values.
For the data collection the study follows the

method proposed by Gamage et.al., 2016 [6].
The android mobile application, “Roadroid
Classic” is used in getting roughness
measurements. Smartphone equipped with this
app is fixed on the windshield of the vehicle
keeping in a steady position. Thus, we assume
that the acceleration coordinates of the vehicle
and smartphone are the same. A Toyota Hilux
4WD cab is used in data collection as most of
the road agencies in Sri Lanka use such vehicles
in their field work.
3
643
Table 1 - Sample of estimated IRI from mobile This will give representative corrective
application and condition of the Kuliyapitiya maintenance cost for each road section
- Bowatta road evaluated, along with its roughness value.
IRI from Furthermore, sections are categorized into IRI
Road
Road section mobile
Condition ranges such as IRI <2, 2-5, 5-7, 7-10, 10+ to
application
indicate the condition of the roads that are
selected such as very good condition,
satisfactory, poor condition, very poor
1.5 Excellent condition respectively.
From the data available, the average

maintenance cost variation with IRI is plotted
in Figure 4. The analysis also be extended to
estimate average cost for each IRI range
13.0 Very Poor (representing different road condition
categories), which is given in Table 3.
The information can be used to make

preliminary estimates on the maintenance
4. Determination of typical budget requirement for each road section,
maintenance cost for low based on its roughness value.
volume roads
Along with the roughness of the road, the

distresses that are present in the road section
are recorded as well. Based on the existing
distresses, the corrective maintenance treatment
required can be decided by the maintenance
engineer or decision criteria as given in the
Road Development Authority National Road
Master Plan [10].
The representative cost of the identified Figure 4 - IRI vs. Maintenance Cost (Rs. /sq.m)
maintenance treatment for each road section is Plot
estimated using the HSR (Highway Schedule of
Rates) [11]. Sample rates in HSR are as follows. Table 3 - Cost per sq.m in Rs. for IRI ranges
Table 2 - Sample rates in HSR Avg.

IRI Range
Cost
Proposed HSR-2016 Cost (Rs.) <2 6.43
maintenance Item No 2 to 5 1,158.41
operation 5 to 7 2,172.57
Pothole MS1-002 209 per sq.m 7 to 10 3,097.85
patching
>10 3,766.13
MS1-007 949 per sq.m
MS1-008 905 per sq.m 5. Selection of roads for

Patching MS1-011 1202 per sq.m maintenance based on
ruts roughness data
MS1-010 1,433 + 24
+MS1- =1457per sq.m 5.1. Prioritizing roads based on pavement
010A condition
Sand S1-036 135 per sq.m
Sealing In order to select the priority of a road to be
Edge Repair MS1-015 1485 per sq.m selected for maintenance work, there should be
a proper prioritization criteria. Roughness,
safety, importance of road and traffic etc. can be
the criteria which are considered in
4
644
prioritization. Among them roughness data can Other aspects such as safety, road hierarchy
be used in the preliminary stage as data (roads used to access civic centres) can be
collection and analysis is rather easy and cost incorporated into the objective function with
effective. This is especially relevant for local priority factors.
authorities who have resource limitations to Constraints can be included to represent the
implement large scale pavement evaluation other requirements the authority seeks to
schemes. implement. For the model proposed in this
section the following constraints are given.
Considering the IRI values of the road sections,
they are ranked from worst roads to best roads. IRIi(1)* Xi >= 4
The roads can be selected based on the three *Ci <= B
ranks until the budget is utilized.
where IRIi(1) is IRI before maintenance, Ci is
Table 4 - Maintenance Priority the cost of maintenance of road section i based
on the maintenance treatment selected and B
IRI Range Condition Priority Rank total budget.
>10 Very Poor 1
7-10 Poor 2 6. Illustrative Example
5-7 Fair 3
2-5 Good 4 To illustrate the proposed methodology the
<2 Excellent 5 following case study is presented using the
roads included in the study. There are 20 roads
However, simple ranking of road sections selected for the sample, and they have varying
based on road condition alone does not pavement conditions.
accurately represent the overall network
condition, the minimum standards that should Total cost of maintenance for each road is
be maintained etc. Furthermore, this method determined using the regression model
may not be able to account for other parameters developed earlier. IRI after maintenance is
that should be incorporated in prioritizing formulated by considering types of operations,
roads. and the input of experienced highway
engineers.
5.2. Selection of candidate roads using
optimization methods Table 4, shows the Worst first ranking of the
road network is formulated considering the
Optimization based approach alleviates some current average IRI of the road sections. The
of the problems identified in earlier section. For selection of the roads for maintenance depends
example, the proposed Integer Programming on the budget available. For example, selecting
model can be used to select candidate roads for all roads for maintenance would require
maintenance. Lingo software was used as the approximately Rs. 950 mn. For the given
Integer Programming tool in this research. It is budget of Rs. 500 mn road sections 6, 19, 16, 4, 2
a comprehensive tool designed to make and 3 is selected. Alternatively, the engineer
building and solving Linear, Nonlinear, could propose alternative maintenance
Quadratic etc and Integer optimizations models treatments (which would change the cost) and
[14]. change the number of roads selected
considering the budgetary constraints
Objective function of the model is to minimize
the network roughness value, i.e. improve the
network road condition.
Objective function: min *

where i is the road section no, n is number of all
road sections in the network, IRIi(2) is IRI after
maintenance (to be defined at the time of
selecting the maintenance treatment for road i),
L(i) is the length of road section i and X(i) is 1 if
the road is selected for repair and 0 if the road
is not selected for repair.
5
645
Table 5 - Worst first ranking of road sections
for a budget of Rs. 500 Mn.
Figure 6 - Network IRI for various budget
7. Conclusion
The study has proposed a novel methodology

to prioritize roads for maintenance planning for
a low volume/provincial road network.
The proposed optimization model is then
Pavement condition evaluation is based on
applied to select the roads. The selected and not
road roughness data that can be obtained easily
selected roads from the Integer Programming
using mobile application. Estimating budget
model are given below.
requirement is based on the roughness value of
the roads. A simplified prioritization method is
Table 6 - Selected and not selected roads from given to select roads for maintenance planning
Integer Programming considering both the condition and the budget
availability. This can also be used to determine
budget requirement and other factors can be
Selected Roads Not Selected Roads incorporated to the optimization model to
3*,4*,5,7,8,9,11,12,15,16* 1,2,6,10,13,14,171,18,19 represent variables such as importance of the
*- roads selected under both methods road, safety considerations etc.
This proposed approach provides a logical and
The results show that the both methods objective methodology which can be
produce different results and second method implemented easily to select roads sections for
includes a higher number of roads in the maintenance planning. This study presents a
selected category. decision support system for a maintenance
engineer or decision makers to implement
The proposed methodology can also be maintenance work in a local/ provincial road
adopted to estimate the required budget to, network without any external influence but
a) maintain all roads above a min IRI value – with proper planning.
using the first approach, i.e prioritizing based
on road condition; Acknowledgement
b) considering the network road condition
(using a parameter such as network IRI) Assistance received from Mr. Lalith Sumanasiri
(Road Development Authority) and Mr. Ajith
As illustrated in Figure 6, as the budget Panagoda (Sabaragamuwa Provincial Road
allocation is increased the network IRI value Development Authority) is acknowledged.
can be reduced, thus indicating an overall
improvement to the road condition in the References
network.
1. AASHTO. (1985). Guidelines on Pavement
Management. American Association of State
Highway and Transportation Officials,
Washington D.C.
2. ADB. (2013). 2013 Annual Evaluation Review.
Asian Development Bank.
3. Camahan, J. V., ASCE, M., Davis, W. J., Shahin,
M. Y., Keane, P. L., & Wu, M. I. (n.d.).
6
646
4. Coghlan, G. T. (2000). Opportunities for Low-
Volume Roads. U.S. Department of Agriculture
Forest Service. Transportation Research Board.
5. Douangphachanh, V., & Oneyama, H. (2013). A
Study on the Use of Smartphones for Road
Roughness Condition Estimation. Proceedings of
the Eastern Asia Society for Transportation Studies,
9.
6. Gamage, D., Pasindu, H., & Bandara, S. (2016).
Pavement Roughness Evaluation Method for
Low Volume Roads. Proc. of the Eighth Intl. Conf.
on Maintenance and Rehabilitation of Pavements.
Research Publishing, Singapore.
7. Kinigama, G. (2010). Development of
methodology to estimate esal values for low
volume roads in provincial sector.
8. Pavia Systems, Inc. (n.d.). Roughness. Retrieved
from Pavement Interactive:
http://www.pavementinteractive.org/roughne
ss/
9. Public Works Maintenance, Eugene. (2015).
Pavement Management Report. Public Works
Maintenance, Eugene.
10. Road Development Authority. (2007). National
Road Master Plan 2007-2017.
11. Road Development Authority. (2015). Road
Development Authority – HSR Analysis
(Metric).
12. Sayers, M. W., Gillespie, T. D., & Queiroz, C. A.
(1986). The International Road Roughness
Experiment. Establishing Correlation and a
Calibration Standard for Measurements.
13. US Department of Transportation Federal
Highway Administration. (2002). Use of PMS
Data For Performance Monitoring. US Department
of Transportation Federal Highway
Administration.
14. Lindo Systems. (2017).
http://www.lindo.com/index.php/products/li
ngo-and-optimization-modeling , (Accessed
1.4.2017)
7
647
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[649 657],2017
2017
The Institution
Lanka
Travel Time Estimation Based on Dynamic Traffic

Data and Machine Learning Principles
Sakitha P. Kumarage, V.M. Jayawardana, Dimantha De Silva and J.M.S.J. Bandara
Abstract: Travel time estimation is an important function in Transport Planning Activities.

Applications such as Advanced Transport Information Systems (ATIS), Dynamic Route Guidance
(DRG) and Network Performance Monitoring require travel time data. These applications, when
deployed for an urban area, require models that estimate travel times on urban arterial corridors with
temporal and spatial variations. Most of the travel time estimation models in the literature are
inductive in nature, which require vehicle travel time observations for calibration. Obtaining vehicle
travel time data is both a time consuming and costly process for a single link and it is impractical to
collect this information for all the links across the road network. Hence, there is a need for travel time
estimation models that do not require reference vehicle travel time data for calibration.
This paper presents a model for addressing the commodity of effective travel time estimation based
on dynamic traffic data on top of data mining and machine learning foundations. Rather based on
aforementioned reference vehicle travel time data, the proposed model incorporates dynamic and
concurrent traffic data collection on multiple links based on google maps distance matrix application
programming interface. Streaming data being fed into machine learning model, a rigorous estimation
of travel time and formation of bottlenecks is calculated. The proposed solution performed well within
an experimental workbench and the model is constantly learning with experience leading to better
calculations with time.
Keywords: Advanced transport information systems, Data mining, Urban traffic networks, Machine
learning
1 Introduction floating car data occur at irregular intervals and

are typically much less frequent for any
Urban traffic congestion is one of the most particular road segment [2]. Hence integrated
severe problems of everyday life in stochastic estimation methods based on
Metropolitan areas. In an effort to deal with this machine learning principles are proposed to
problem, intelligent transportation systems adopt in estimation of travel time.
(ITS) technologies have concentrated in recent Most of the travel time estimation models in the
years on dealing with urban congestion [1]. One literature are inductive in nature, which require
of the most critical aspects of ITS success is the vehicle travel time observations for
provision of accurate real-time information and calibration[3]. Obtaining vehicle travel time
short-term predictions of traffic parameters data is both a time consuming and costly
such as traffic volumes, travel speeds and process for a single link and it is impractical to
occupancies [1]. collect this information for all the links across
the road network[3]. Hence, there is a need for
Predicting car travel times along road segments travel time estimation models that do not
is an increasingly important component of require reference vehicle travel time data for
today‟s car navigation systems [2]. Johns et.al calibration.
explored the problem of predicting travel times Mr. Sakitha P. Kumarage, B.Sc.Eng(Hons)(Moratuwa),
from floating car data (FCD) using geospatial B,Sc(Hons)London, Research Assistant, University of
inference method [2]. Floating car data are Moratuwa .
traces of GPS positions from actual cars driving Mr. V.M. Jayawardana, Undergraduate, Faculty of
on a road network. It provides speed or travel Engineering, University of Moratuwa.
time estimates across the entire road network Dr. Dimantha De Silva, B.Sc.Eng.(Hons) (Moratuwa),
M.Sc (Moratuwa), Ph.D. (Calgary), P. Eng.(Alberta),
this is an advantageous feature in geospatial Senior Lecturer, Department of Civil Engineering,
inference method when compared to stationary University of Moratuwa.
sensors, which typically only provide speed or Eng. (Prof.) J.M.S.J. Bandara, BSc. Eng (Moratuwa),
flow estimates on highways or major roads [2]. Ph.D. (Calgary), FCILT, Professor, Department of Civil
A major disadvantage of floating car data is Engineering, University of Moratuwa.
that, unlike data from stationary sensors, the
1
649 ENGINEER
The main objective of the paper is to identify contribution can be utilized in large-scale urban
the use of google travel time data in planning and transit planning[6].
and management of traffic in urban road Venessa and Enrique identified that individuals
networks. On this regard the possibility of generate vast amounts of geo-located content
using Google travel time data in identifying through the use of mobile social media
bottlenecks on an urban road network. Data applications[7]. The results have shown that
mining for travel time data and weather data geolocated tweets can constitute a good
was supported by the Google distance matrix complement for urban planners to model and
API and Dark-Sky weather API which analyzed understand traditional land uses) in an
using Support Vector Machine for prediction of affordable and near real-time manner[7].
bottlenecks on an urban road network. Another informative way of data mining is by
using Foursquare place data[8]. Spyratos et al
Use of travel time data to analyze the efficiency suggest Foursquare place data for estimating
of utilizing traffic management plans was nonresidential building blocks and land use
discussed with reference to the pilot project on mapping purposes[8].
implementing bus priority lanes for public
buses of Colombo metropolitan area. On the aspects of travel time estimation using
crowed sourced data Dewulf et al. (2015) took
2 Literature Review Floating Car Data (FCD) from the Be-Mobile
system to calculate car travel times[9]. In which
2.1 Data Mining the research estimated travel time by using
geolocated positions of 400,000 vehicles
On the aspects of collecting data related to equipped with track and trace devices. The
traffic models the literature provides many obtained results were further aggregated to
solutions. Such solutions are elaborated here. produce a generic travel time for peak and off-
Real travel time data is typically measured and peak periods[9].
recorded in one of two ways: using stationary
observers or moving observers[2]. Stationary Further on this regards Araghi et al discusses
observers include loop detectors and video the Accuracy of Travel Time Estimation using
surveillance. Moving observers, involving Bluetooth Technology in which results show
floating cars or probe cars, are becoming that Bluetooth Technology can be used as a
popular travel time collection methods since relatively new cost-effective source of travel
they can cover almost any road segment as time estimation[10].
needed[2].
As stated by Jones et al, one of the
The information on transport networks has disadvantages in use of crowd sourced data is
improved with the emergence of crowed frequent unavailability of data[2]. When
sourced data generated by social media, compared with travel time estimation, frequent
smartphones, Satellite navigation and other floating car data is not available both on the
technologies[4]. According to Geurs et al link being predicted and succeeding links[2].
websites like Here, Bing Maps, Google Maps, Thus a geospatial inference method to deal
Google Transit, Open-Street-Map and the with such scenarios, which utilizes the data
public availability of Transit Feed Specification from geographically nearby links to predict the
(GTFS) data from transit authorities, enables travel time of the designated link need to be
vast area of research on evaluating accessibility incorporated[2]. On this regards, Giannotti et
and travel time. [5] al presents the theory of trajectories which
elaborates on a sequence of geo-referenced
Georgios et al shows that use of crowd sourced objects S of size „m‟ and a list of Temporal
data is an efficient way of data mining with the Annotations A of size (m - 1), whose values
wide spread availability of smart mobile represent the temporal distance between two
phones and the multi sensing features such as consecutive geo-referenced objects belonging to
geolocation light, movement, audio and visual S. Developing on this principle[11] Masiero et
sensors[6]. al propose a mechanism based on machine
learning principles for estimating the travel
Data gathered from crowd sourced applications time between any two points that: pre-
could be used in identifying mobility patterns processes raw trajectory data to create
or popularity of a given trajectory. Such a aggregated data; and learns the behavior of the
vehicles using this aggregated data[12].
ENGINEER 2
650
2.2 Data analysis and inferences than Artificial Neural Networks because SVR is
more amenable to generalization than Artificial
It was identified that there are many sources in Neural Networks[17].
which data could be mined and stored for
analysing purposes. The basic theory was 3 Methodology
presented by Cho et al. To predict the travel
time he identifies a function which relates the 3.1 Data gathering
predicted travel time for an interval on a link to
the available parameters such as speed, flow From the previous literature, it was identified
data[13]. The optimisation criteria would be to that travel time given by crowd sourced data
reduce the percentage error of predicted travel could be utilized in estimating the travel time
time with the observed [13]. between origin and destination points with a
significant accuracy and reliability.
In prediction of the travel time data many
approaches could be observed in such Zhang In obtaining data for travel time variation
and Jones propose to use historical means of the between two points google ETA given by
training data set at time t as the predictor for Google Maps could be incorporated into the
travel time[14]. Cho et al propose the current data mining procedure[13]. Travel time
time predictor method in which the travel estimations given by this method is quite
predictor is obtained by assuming the speeds of reliable and highly efficient source of data
all segments at time t remain the same as the collection for developing countries such as Sri
speeds at t − ∆ throughout the entire travel[13]. Lanka which face a broader gap in funding for
Further Cho et al suggest time varying linear dedicated traffic sensors which can cover the
regression method in which they identify the whole urban arterial network[14].
travel time predicator as a linear combination
of a time dependant constant and the the Although google does not reveal the exact
current travel time predictor of the previous algorithm which they use for travel time
segment[13]. Improving the time varying prediction, the basis on which the travel time is
regression methods Kwon and Varaiya calculated is crowd sourced data[15]. When a
suggests time varying linear regression per user enables Google services on a smart phone
segment by considering the current time it sends anonymous bits of data back to Google
predictor of the entire route as the key describing how fast the user is moving[15]. The
feature[15]. travel time prediction algorithms on google
servers combine users speed with the speed of
other phones on the road, across thousands of
Deriving from these basic concepts Ide et al. phones moving around a city at any given time
proposed the probabilistic prediction of the in which a significantly reliable estimate on
travel time along an unknown path such that traffic conditions could be gained[15]. This
the similarity between paths is defined as a process is continuously carried out to make
kernel function[16]. In this method authors predictions accurate based on machine learning
emphasised on application of the Gaussian principles[15].
process regression for predicting the travel
time[16]. According to ex-googler Richard Russel, to
come up with the best prediction for travel
Incorporating machine learning principles Wu times the google consider data available in a
et al suggest to use support vector regression particular area[16]. The data range from official
(SVR) for travel-time prediction[17]. Since speed limits and recommended speeds, likely
support vector machines have greater speeds derived from road types, historical
generalization ability and guarantee global average speed data over certain time periods
minima for given training data, it is believed (sometimes just averages, sometimes at
that SVR will perform well for time series particular times of day), actual travel times
analysis[17]. According to authors comparative from previous users, and real-time traffic
to other baseline predictors which were information[16]. Hence the gained travel times
mentioned above , their results show that the are scientifically justifiable to be accurate and
SVR predictor can significantly reduce both reliable [16].
relative mean errors and root-mean-squared
errors of predicted travel times[17]. Moreover In this paper, data was gathered from Google
authors emphasis that SVR had better results Distance Matrix API where, it is possible to
3
651 ENGINEER
collect travel time and distance values of a temperature, average speed of the road. It is
given route between start and end points. The possible to extend the model into parameters
collected data over a period of 3 months were such as road pavement conditions, vehicle
analyzed. distribution, road geometry, accessibility of the
road etc. Such extension could be included in
The main objective of the paper is to identify the future work with the availability of data.
the usability of google travel time data in
planning and management of traffic in urban Based on past research it is suggested to
road networks. On this regard the possibility of identify of bottlenecks is by training a support
using Google travel time data in identifying vector machine which runs a machine learning
bottlenecks on an urban road network and algorithm in identifying extreme outliers of
using data to analyze the efficiency of utilizing travel time data along a road segment varying
traffic management plans are discussed. with space and time. In inducing the SVM
algorithm the factors which can cause
3.2 Identification of bottlenecks bottlenecks were considered. Support Vector
Machines (SVM) is a supervised learning tool
A bottleneck on a road is a point space that has well known for its discriminative power. SVM
a reduction in capacity as compared to the starts separating the data with a hyperplane
upstream and downstream capacity. Usually and then extends this procedure to non-linear
these are at facilities such as a bridge or an decision boundaries.
intersection at a given point along the road
considered. When the road segments on each 3.3 Effectiveness of traffic management
side of the bottleneck have sufficient capacity plans
then no congestion occurs on them. The
bottleneck, whose capacity is finite, is subject to It was identified with the literature review;
congestion. Queuing will develop when the travel time data can be used to evaluate the
arrival rate of flow at the bottleneck exceeds its traffic management of a particular link or a
capacity [17]. The travel time along such a network[2]. As an application of the concept the
segment will increase when the bottle neck travel time variation of Rajagiriya area with the
reaches its capacity[17]. Therefore, observing implementation of a Bus Priority Lane (BPL)
the increase in travel time along road segments was studied to test the variations traffic
will indicate the existence of any speeds/times that can be found with the
bottlenecks[17]. The study focuses on method which could be used to evaluate the
identifying the formation of bottlenecks along effectiveness of the implemented plan.
road segments subjected to the spatial and
temporal variations. On that regard the such BPL was suggested in the CoMTrans Master
identification will enable efficient traffic plan [14] and the Megapolis Transport Master
management plans considering the spatial and Plan[19]. Hence BPL was implemented as a
temporal variations suitably. pilot project at Parliament Road from 12th to
23rd March 2017 by the Ministry of Megapolis
According to Chien et al travel time in an urban and Western Development with the assistance
traffic environment is highly stochastic and of Korean International Cooperation
time dependent due to random fluctuations in Agency[20]. At the pilot project a single lane of
travel demands, interruptions caused by traffic the study area was separated for public
control devices, incidents, and weather buses[20]. The effect on overall traffic
conditions[18]. As discussed by Leong et al movement was analyzed from the collected
weather conditions tend to have a significant data by Google Distance Matrix API. A
impact on the driving conditions, temperature situation analysis on variation of travel times
precipitation and visibility tends to increase the before and after the implementation of BPL was
road traffic and reduce roadway capacity[18]. studied.
Therefore, when formation of bottlenecks, these
interruptions will have a significant influence. 4 Data Verification
Hence its is required to consider weather
conditions on propagation of bottlenecks[18]. The data collected from Google Map APIs was
validated using Floating Car data collection
Thus, travel time variation data was analyzed method. Floating car data collection is a
by taking following parameters into commonly used traveltime collection method.
consideration; Day of the week, time, rainfall, The test is conducted by dispatching a vehicle
ENGINEER 4
652
to travel with the traffic stream and collect the Collected data was structured on a data set
travel time in-between specified locations[21]. with following attributes: Road-segement, date,
The speed of the floating car was controlled to time, distance, travel-time, avg-speed,
follow the usual traffic to set the driving precipitation, tempMax, tempMin. The data set
guidelines of an average moving vehicle [21]. was prepared to identify patterns of variations
Data collected from the manual method was in traveltime based on machine learning
compared with the API travel times obtained principles.
parallel to floating car data collation. The
results observed showed a minor variation
from the traveltime collected from floating car
data and Google Distance Matrix API. The
correlation in between the FCD and API travel
time was observed very high with a R2 value of
0.9802, which indicates that the traveltime
estimates given by Google Distance Matrix API
do not indicate a significant difference from the
observed floating car data. (see Figure 1)
FCD�vs�API�Traveltime
Figure 2 – Map of segments Used
120
100 y�=�1.07x 6 Analysis

API�Traveltime�(min)
80
R²�=�0.9802 6.1 Identification of bottlenecks
60
SVM is used in the process of identifying the
40
formation of traffic bottlenecks. Collected data
20 set was fed into the SVM algorithm to predict
0 the formation of bottlenecks. The algorithm was
0 20 40 60 80 100
FCD�Travel�Time�(min) based on attributed identified earlier.
The basic algorithm was fine tuned to the
Figure 1 - FCD vs API Traveltime situation in which it enables to train separate
model for each road segment in the urban road
5 Data Collection network considered. Based on the trained
Data collection for bottleneck identification and models, it was checked the severity of traffic
evaluation of traffic management plans was condition using the test data.(see Figure 3)
conducted using a data set collected on travel
time and weather at Rajagiriya Junction.
Travel time data was collected using Google
Distance Matrix API. Data was collected in 15
minute intervals from 0530h to 2359h. A PhP
script was run on a server which collected data
for the scheduled time. At every data collection
interval travel time data between two links
were recoded for 3 pre-identified road
segments. Figure 2 shows the segments used
for the analysis. Weather data which included
precipitation, max. temperature and min.
temperature of the Rajagiriya area was
collected from Dark Sky API which was
recorded in hourly intervals for the observed
duration. Road profile data was not included in
the estimation model as it is a generic
parameter of the considered road segment in
which the predictions were made. As the
objective was to estimate traveltime of a
particular link, road profile parameters will be Figure 3 – Flow diagram of analysis
kept a constant value for short term estimates.
5
653 ENGINEER
The prediction gave following results which are In Figure 4 the spatio-temporal variation of
quite satisfactory. speeds(km/h) is illustrated for the considered
segments. According to the Figure 4 it could be
Table 1 – Summery of estimation results
identified that the segment S1-S2 (kotte road to
HSBC junction) forms a bottleneck in the road
Segment Precision Recall F1
segment considered where the segment speed
M1-S1 91.84% 100 % 95.75
is considerably less than the upstream and
S1 –S2 98.90% 99.45% 99.17
downstream segments, The speeds have
S2 – S3 94.88% 100 % 97.37 localised minimum between at 0715h and 0830
most likely to the movement of school traffic
precision also called positive predictive value is along the segment. The segment M1-S1 speed
the fraction of relevant instances among the reduction from 0800h to 0815h shows that the
retrieved instances, while recall also known as capacity of the road is been reached during the
sensitivity is the fraction of relevant instances said time period. The results show that in the
that have been retrieved over the total number morning peak hours, M1-S1 segment and S1-S2
of relevant instances. F1 is a measure calculated segment should be given high priority in traffic
by using both precision and recall. Thus, the management to improve the traffic flow in the
estimates have given highly accurate results in segment considered.
all three instances.
6.2 Effectiveness of traffic management
Identification of bottlenecks along the three plans
road segments
Sheet 1 were carried out using the SVM
algorithm and spatial and temporal variation
Segment The variation of travel time along the M1-M6
bottleneck formation was identified.
Minut e of
road segment from Polduwa bridge to

Hour of
Polduw a Br dg Kot t e Rd - HSBC HSBC Junc-NAITA

Time
Time
-Kot t e Rd [M1-S1] Junc [S1-S2] [S2-S3]

Senanayake Junction (see Figure 5) was
0 36.46 37.88 51.36
recorded from the data collection method
5 15 36.12 36.97 48.99
AM 30 35.20 36.04 51.51 described above.
45 34.32 35.19 46.22
0 32.58 32.91 40.34
6 15 28.60 28.52 35.11
AM 30 18.05 17.03 23.51
45 14.18 13.81 19.45
0 14.14 13.19 18.28
7 15 14.17 12.29 16.86
AM 30 14.52 12.71 16.74
45 12.52 11.62 15.61
0 11.99 11.35 15.63
8 15 12.28 11.08 16.18
AM 30 14.39 12.18 16.71
45 16.24 12.82 15.67
0 18.44 12.91 16.97
9 15 20.18 14.20 18.58
AM 30 21.74 14.96 21.32
45 23.88 15.92 21.75
0 24.31 16.26 22.92
10 15 24.07 15.36 21.88 Figure 5 – Map of road segment M1-M6
AM 30 23.83 14.86 21.27
45 24.50 14.52 22.82
0 23.66 14.68 21.82
11 15 22.41 13.10 21.69
AM 30 20.24 12.16 21.60
45 18.48 11.01 20.80
0 16.89 11.10 20.88
12 15 16.41 11.53 20.01
PM 30 15.14 10.94 19.04
45 13.70 10.76 20.89
Average of Speed (km/ h) br oken dow n by Segment vs. Time Hour and Time
Minut e. Color show s average of Speed (km/h). The marks are labeled by
average of Speed (km/h). The view is fi lt ered on Segment and Time Hour.
The Segment fi lt er keeps Polduw a Brdg -Kot t e Rd [M1-S1], Kot t e Rd - HSBC
Junc [S1-S2] and HSBC Junc-NAITA [S2-S3]. The Time Hour fi lt er excludes 12
members.
Avg. Speed (km/ h)
8.91 40.00
Figure 4 – Bottlenecks along road segments Figure 6 – Variation of traveltime along M1-
M6
ENGINEER 6
654
Figure 6 above shows an illustration on the 7 Discussion
variation of travel time along time for several
days with and without BPL. A BPL was The use of crowdsourced data in estimation of
implemented for a 1km distance between travel time was emphasized on the paper and
Rajagiriya Junction and Ayuvedha Junction the applications of travel time estimation by
from 12th March (Sunday) to 17th March crowed sourced data were evaluated. From the
(Friday) for the whole day. The analysis shows literature review it could be identified that
that there is a significant difference in travel information required for transport analysis
time with the weekday and weekend travel could be obtained from data sources such as
patterns where the weekends are represented APIs provided by Google, Twitter and other
by the 11th March and 12th March data social media. Data mining based on these
platforms will enable to identify mobility
The data shows that there has been daily patterns of vehicles and people, travel time and
variation in travel times with the BPL accessibility. Although the data provided by
implementation and could be compared with mobile phone companies for research work is a
the 9th March data as base data for weekday primary source of data reliability and
and 11th March as base data for weekend where consistency of such data is questionable. Hence
the BPL was not operational. use of public data APIs provided by Goggle
and other social media platforms will enable
consistent collection of data. One of the main
advantage of using travel time data provided
by Google APIs, that it is cost effective and
available consistently. Further availability of
travel time data on any given time for a very
large network of urban road networks will
enable further advancement of travel pattern
analysis. Incorporation of machine learning
algorithms could be identified as a constructive
approach. It is confirmed by the past research
that use of machine learning is the novel
approach as it will reduce the errors more than
Figure 7 - Average Speed variation before and that of probabilistic models when the analysis
after implementation of BPL spreads to a large urban network.
Figure 6 and Figure 7 illustrate the variation in Based on the data mining principals of frequent
travel time and average moving speed of the collection of current traffic conditions and
passengers before and after the implementation travel time data for a longer period of time has
of BPL. The illustration shows that there is a enabled to identify the traffic patterns of a
marginal decrease in average travel time, thus a given network which is dependent on spatial
marginal increase in speed during 6:30 am to and temporal parameters. Data collected on the
8:30am with the implementation of the Bus Rajagiriya junction has showed how the
Priority Lane (BPL), but a considerable increase bottlenecks are formed on selected road
in average travel time, thus a considerable segments with time. The use of support vector
decrease in speed during other times of the day machine (SVM) has enabled to extend the
with BPL implementation. Therefore, it could analysis to a larger urban network in which
be noted that when the existing speeds is cluster data of each road segment with a
already low the BPL has performed marginally hyperplane and then extends this procedure to
positively during the peak time period. non-linear decision boundaries and predict the
formation of bottleneck on a road segment for
However, it should be noted caution should be given conditions. As suggested in the
made when making judgement based on just a methodology the algorithm could be extended
1km of section of BPL since the impact to the to incorporate other parameters which can
overall network will only be evident with cause bottlenecks but in our analysis, we have
sufficient length of such implementation. limited only to parameters such as day, time,
distance, travel-time, average-speed,
precipitation, maximum-temperature,
minimum temperature. Incorporation of these
attributes have shown overall improvement in
7
655 ENGINEER
identification and prediction of bottle necks. on an urban road network subjected to spatial
But the influence of maximum and minimum and temporal variations.
temperature has shown less significance as they
are parameters which have lesser variation. In The use of data to analyze the impact of
future work, it is expected to incorporate utilizing traffic management plans was
parameters such as road pavement and discussed by evaluating the pilot project on bus
geometry, vehicle distribution, road priority lane.
accessibility etc
References
Regarding the study on travel time variation 1. Stathopoulos, A., and Karlaftis, M. G., “A
with the implementation of bus priority lane, Multivariate State Space Approach for Urban
the analyzed data reveals the pilot project has Traffic Flow Modeling and Prediction,”
reduced the overall link speeds in the Transp. Res. Part C Emerg. Technol., vol. 11,
considered link after implementation. no. 2, pp. 121–135, 2003.
Therefore, it is questionable whether bus
2. Jones, M. Geng, Y. Nikovski, D., and Hirata,
priority lane is a suitable solution for the
T. “Predicting Link Travel Times from
considered route. Hence, it cannot be justified Floating Car Data,” IEEE Conf. Intell. Transp.
the opinion that there will be a considerable Syst. Proceedings, ITSC, no. Itsc, pp. 1756–
saving of time or money by implementing the 1763, 2013.
bus priority lane.
3. Herring, R. J. “Real-Time Traffic Modeling
As the limitations of the study it is identified and Estimation with Streaming Probe Data
that the study area used for data collection was using Machine Learning,” Vasa, 2010.
not a large urban network. Further the collected
4. van Wee, B., “Accessible Accessibility
data set cannot portray the variations which are
Research Challenges,” J. Transp. Geogr., vol.
dependent on months as the data was collected 51, pp. 9–16, 2016.
only for a one month duration. The experiment
was carried out in the dry season therefore 5. Geurs, K. T. L. La Paix, and Van Weperen,
influence of high rainfalls and changing S., “A Multi-Modal Network Approach to
weather was not possible to identify. The Model Public Transport Accessibility
extension of the methodology to rural areas has Impacts of Bicycle-Train Integration
limitations as the data was collected using Policies,” Eur. Transp. Res. Rev., vol. 8, no. 4,
pp. 1–15, 2016.
Google distance matrix API which collect data
of moving smart phones. Hence availability of
6. Chatzimilioudis, G., Konstantinidis, A.,
many people in a road network is a significant Laoudias, C., and Zeinalipour-yazti, D.,
factor which can cause reliability issues. “Crowdsourcing with Smartphones,” pp. 1–
7.
8 Conclusion
7. Frias-Martinez, V., and Frias-Martinez, E.,
In this paper, we reveal the use of google travel “Spectral Clustering for Sensing Urban Land
Use using Twitter Activity,” Eng. Appl. Artif.
time data for effective estimation of travel time
Intell., vol. 35, pp. 237–245, 2014.
and identify bottlenecks based on data mining
and machine learning foundations. 8. Spyratos, S., Stathakis, D., Lutz, M., and
Tsinaraki, C., “Using Foursquare Place Data
Use of google Distance Matrix API can surpass for Estimating Building Block Use,” Environ.
the cumbersome work of developing models to Plan. B Plan. Des., 2016.
estimate travel time data. Further improving
the prediction based on these estimates can was 9. B. Dewulf et al., “Examining Commuting
constructive approach in prediction of Patterns using Floating Car Data and
Circular Statistics: Exploring the Use of New
bottlenecks of an Urban network.
Methods and Visualizations to Study Travel
Times,” J. Transp. Geogr., vol. 48, no. August,
It could be concluded that the main objective of pp. 41–51, 2015.
the paper is reached as it was able to identify
the usability of google travel time data in 10. Araghi Namaki, B., Pedersen Skoven, K.,
planning and management of traffic in urban Christensen Torholm, L., Krishnan, R., and
road networks. The approach was confirmed by Lahrmann, H., “Accuracy of Travel Time
successful use of data in identifying bottlenecks Estimation Using Bluetooth Technology:
Case Study Limfjord Tunnel Aalborg,” pp.
ENGINEER 8
656
166–191, 2012.
11. F. Giannotti et al., “Unveiling the Complexity

of Human Mobility by Querying and Mining
Massive Trajectory Data,” VLDB J., vol. 20,
no. 5, pp. 695–719, 2011.
12. Masiero, L. P., Casanova, M. A., and de

Carvalho, M. T. M., “Travel Time Prediction
using Machine Learning,” Proc. 4th ACM
SIGSPATIAL Int. Work. Comput. Transp. Sci. -
CTS ’11, no. November, pp. 34–38, 2011.
13. Massimino, B., “Accessing Online Data:

Web-Crawling and Information-Scraping
Techniques to Automate the Assembly of
Research Data,” J. Bus. Logist., vol. 37, no. 1,
pp. 34–42, 2016.
14. Japan International Coporation Agency;

Oriental Consultants Co LTD, “Ubnan
Transport System Development Project For
Colombo Metropolitan Region.”
15. Google, “The Bright Side of Sitting in Traffic:

Crowdsourcing Road Congestion Data,”
Googleblog, 2009.
16. E.-G. (2006-2012) Richard Russell, “How

Does Google Maps Calculate Your ETA,”
FORBES, 2013. [Online]. Available:
https://www.forbes.com/sites/quora/2013
/07/31/how-does-google-maps-calculate-
your-eta/#241f6c01466e.
17. Yang, H., and Hai-Jun, H., “Analysis of the

Time-Varying Pricing of a Bottleneck with
Elastic Demand using Optimal Control
Theory,” Transp. Res. Part B Methodol., vol.
31, no. 6, pp. 425–440, 1997.
18. L. W. Leong et al., “Improving Traffic

Prediction by Including Rainfall Data,” no.
January 2016, 2014.
19. Ministry of Megapolis and Western

Development, “Western Region Megapolis
Transport Master Plan,” 2016.
20. Department of Government Information Sri

Lanka, “DECISIONS TAKEN BY THE
CABINET OF MINISTERS AT ITS MEETING
HELD ON 25-04-,” State Print. Coporation,
2017.
21. Turner, S. M., Eisele, W. L., Benz, R. J., and

Douglas, J., Travel time data collection
handbook, vol. 3, no. 5. 1998.
9
657 ENGINEER
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[659 665],2017
2017
The Institution
Lanka
Infrastructure Feasibility Analysis of Inland Water

Transportation from Peliyagoda to Hanwella along
Kelani River
K.P.H.S. Kalahe and R.M.N.T. Sirisoma
Abstract: In the ancient times, people used to settle near waterways for their agricultural
activities as well as to fulfil their mobility requirements. With the urbanization and development of
technology, people began to shift to land transportation neglecting the waterways. When the land
transportation system prevailing in the country is declining gradually, and it arises a need for a
transport mode which does not use the existing road network. Therefore, transport planners have
again eyed to identify the potential of inland waterways to cater transportation needs and recreational
purposes. Inland water transport is a preferred mode throughout the world because it is a low cost,
fuel efficient, environment-friendly, less capital needed, safe and a pleasant transportation mode. This
paper focuses on the infrastructure feasibility of the proposed Mattakkuliya- Hanwella Inland
waterway line. Through a detailed literature review, specifications of the vessel that should be
operated, environmental impacts and safety measures related to water transportation and facilities at
the stations were identified. A site survey was carried out to find the minimum horizontal and vertical
clearance at the overhead structures along the river. Using the bathymetry data gathered for the
Kelani River, navigable path and areas to be dredged were recognized. Water Intake and salinity
barrier at Ambatale are major concerns in this waterway line and therefore, solutions have been
identified to overcome these impacts. Data regarding on bathymetry of the river, has found that the
entire water stream is navigable. Optimizing the number of boat stations with respect to the access,
demand and the availability of transfer modes it was identified that 13 boat stations are required.
Demand for the system was identified using a Logit Model. Highest and lowest demands per stations
are at Malwana with 9700 passengers per day at Embulgama with 2000 passengers per day
respectively. The new system will help to minimize the traffic congestion on the High-Level Road
(A4) providing passengers with a safer journey.
Keywords: Navigable Path, Least Available Depth, Inland water transport
1. Introduction transportation through rivers, canals and lakes.

This paper focuses on, public transportation
Sri Lanka owns a rich history in inland water and recreational aspects of inland waterways.
transportation where inland waterways such as For a transport mode to be usable, the factors
rivers, canals and lakes were recognized as vital such as time, cost and intermodality are critical.
arteries for transportation. During ancient times Inland water transportation can serve as a
in Sri Lanka, inland waters were used for convenient mode of transport in congested
international trade. There have been ports cities. Inland water transport has many inborn
situated at estuaries of rivers such as direct and indirect advantages. As for the direct
Salwathura at Deduru Oya, Wattala at the advantages, its cost is low, capacity is high,
Kelani Ganga, Kalathilaka at the Kalu Ganga, safe, energy efficient, the capital need is less
Nilwalatittha at the Nilwala Ganga and Kirinda and pollution is limited. The indirect
at Kirindi Oya. [4]. But with globalization, advantages are it attracts new industries,
traditional transport using inland waterways creates new business opportunities, boosts the
were substituted by road, rail, air and marine economy in the area and generates tourism and
systems. recreation activities [1].
Waterborne transport is one of the oldest Eng. (Dr.) R.M.N.T. Sirisoma, Ph.D., B.Sc. Eng (Hons),
modes of transport compared to other modes of C.Eng, CMILT(UK), MITE(US), MIE(SL)
transportation prevailing in the world. Inland Senior Lecturer, Department of Management and Finance,
General Sir John Kotelawala Defence University.
water transport can be divided into three
sectors like public transport, freight transport Mr. K.P.H.S Kalahe, undergraduate, Department of Civil
Engineering, General Sir John Kotelawala Defence
and recreational. And it comprises University.
1
659 ENGINEER
Because of these reasons, inland water Malwana Road 47,718
transport is being used all around the world, Awissawella Road 25,518
especially in the countries such as Bangladesh, Colombo-Hanwella 15,742
Germany, USA, UK, Russia, China, Thailand Low-Level Road
etc. In the Western Province of Sri Lanka, water Hanwella- Urapola 16,476
ways are exploited in small scale for Road
transporting commodities like bricks, tiles, sand (Source - Western Region Megapolis
and timber in the construction industry [11]. Transport Master Plan)
The Kelani is the second longest river in Sri

Due to reasons, such as traffic congestion in the
Lanka. The Kelani River and its tributaries
area, shortage of parking spaces and pedestrian
provide the portable and industrial water
walkways, high rate of road traffic accidents
requirements for the people in Greater
and lack of comfort, safety and efficiency in
Colombo area. Starting from the Sri Pada
public transport service the prevailing transport
mountain range in the Central hill country, it
flows mainly in the western direction and falls system has failed to cater to the mobility
into the sea at Modara, Colombo. The river requirements of the people. Therefore, wastage
section passing via Western Province has the of valuable man hours, wastage of fuel and
most potential to use in public transport mode other resources and environmental pollution is
due to the land use pattern in the area. The unbearable.
study area consists of fifty-six Grama Niladhari
Divisions, and the main cities are Peliyagoda, For declining of the level of service of
Kaduwela, Kelaniya, Malwana and Hanwella. commuters, constructing new roads is not a
Along the study area the river runs for a total solution. Construction of new road networks,
distance of 31 km, and on the bank, on either railway networks is always costly. The
side of the river, there are a number of investment needed for inland waterways is 5%-
important economic, religious and recreational 10% of the investment for an equivalent road or
places such as Kelani Maha Vihara, Water rail line. Its maintenance cost is 20% of road
World Lanka and Water Board Pump Division, maintenance cost [1]. Therefore, Colombo
Ambatale. Metropolitan Region needs a transport mode
which does not depend on the present road
2. Need for an Inland Waterway structure. Hence developing inland water
Line transportation as a transit mode is a better
solution to reduce the traffic congestion and to
Colombo, being the capital of Sri Lanka, has an provide a pleasant, time-saving, economical
average annual growth ratio of 8%, the number and safer journey for the commuter.
of vehicles in the Western Province has
increased by a factor of 2.5 in 12 years and Therefore, with the aim of reducing the heavy
average speed of vehicles in major transport traffic burdens on the main road corridors and
corridors during the peak time have fallen to capitalize the existing water bodies as a
below 10 km/h. Colombo attracts more than minor public transport medium Western
one million daily commuters by 160,000 Region Megapolis Transport Masterplan has
vehicles from the suburbs of Colombo [9]. The identified 3 potential inland waterway lines.
following table shows the average daily And the inland waterway line from
passenger flows along the corridor. Mattakkuliya to Hanwella is one of them [8].
Table 1 - Average Daily Traffic: Mattakkuliya- 3. Literature Review

Hanwella Line [8]
Corridor Name Average Daily 3.1 Station Facilities
Traffic-
2013(Passengers/day) To attract passengers to the new system, station
Negombo Road 20,455 facilities are important. Jugović, Mezak and
Negombo- Colombo 92,662 Lončar(2007) state that terminal capacities
Main Road should be sized and profiled for quick flow of
New Kelani Bridge 119,903 passengers. To provide passenger comfort,
Biyagama Road 25,945 factors such as service and maintenance shops,
New Kandy Road 55,344 sanitary facilities, parking lots, refreshment and
2
660 ENGINEER
service areas, level of service of employees at some negative impacts. Pollution is the major
the station and offer in the station is vital. [6]. negative impact, and some of the others are
bank erosion, loss of habitat, barriers to
During the daytime, the system can be used for migration and species dispersal. Further, the
public transportation minimizing the traffic pollution can be due to 6 major sources, and
congestion in the road network. During the off- they are operational oil pollution, solid waste
peak time where the demand is low, they can disposal, accidental spills, air pollution and
be used to promote recreational activities. A construction and maintenance of stations and
study done by Transportation Engineering the waterway channel. Dredging poses a direct
Division of the University of Moratuwa in 2003 threat to the areas where it is carried out as it
has suggested that with the gradual introduces sediment into the adjacent water
development of the system, restaurants, column where sediment is redeposited at the
children parks, leisure activities will develop bottom. This has short-term effects on pelagic
around the river bank. This will booze the fish and the benthic community. Dredging also
economy of the people living nearby. Hence a increases turbidity because of the suspended
station has the ability to promote local and sediments. It decreases light penetration and
regional development. Local employment photosynthetic activity [5].
opportunities are vast when a wide range of
value-added services are provided. 4. Horizontal and Vertical
Furthermore, Information Technology related Clearance
infrastructure, aggressive marketing,
public/private cooperation are some important Navigational channel with desired width,
assets for the success of a station/jetty. [2] depth and clearance height is one of the basic
Inland water transport related infrastructure for
3.2 Safety and Environmental Pollution the development of waterways. In the study
area, the Kelani River passes through 6
In inland water transportation, the probability highway bridges, a footbridge and a railway
of accidents is less, and the costs of the bridge. These bridges were identified using the
accidents are low in economic and human “Google Earth Pro” software, and a site survey
terms. Awal (2015) states that accidents cannot was carried out to measure the clearance levels.
be caused by a single factor, it is a result of The results showed that the river has sufficient
various complex interactions. They may be horizontal and vertical clearance for navigation.
mechanical failures, human errors and natural This survey was done during the low tide, and
cases. According to most of the studies that the results are as follows.
have been carried out, especially in Bangladesh,
the main factors related to accidents are human Table 2 - Minimum horizontal and vertical
factors, operating environment factors, clearance
enforcement and educational factors and vessel Bridge Min: Horizontal
design factors. Some of the causes related to Vertical Clearance(m
human factors are overloading and Clearance(m )
overcrowding, stability problems due to the )
rush of passengers and collision due to New Kelani 4.90 25.84
handling errors. By making overloading Bridge
passengers unlawful, instructing service Railway 5.85 18.74
providers about the safety of the passengers, bridge-
establishing vessel traffic monitoring systems Peliyagoda
and enhancing manpower and infrastructural Kelansiri 6.5 35.58
facilities needed for the system, the number of Bridge at
accidents can be minimized. [7] Kelanimulla
Footbridge at >5m 30
Water is a scarce resource in the country. Use of
Mabima West
water for transport is sometimes difficult to
Bridge at 8.4 33.5
justify. Kalyani, Vidyasagar and Srinivas (2015)
Outer Circular
express that Inland water transport is a more
Highway
environment-friendly mode of transport
Kaduwela 8.6 20.5
compared to other modes of transport. But
bridge
depending on the individual characteristics of
the river, its physical and ecological state it has
3
661 ENGINEER
Nawagamuwa 9.45 11.1 navigation was not found. The navigable line
bridge at was then buffered for 2 paths of 10 m and 20 m
Nawagamuwa in size, where 10 m path is the safest path for
- Mapitigama navigation.
Road
Hanwella 13.14 10.2
bridge at
Hanwella-
Urapola Road
5. Determining the Navigable

Channel
The least available depth (LAD) is a critical Figure 2 - Section of the navigational path
parameter for navigability of a waterway. From
the literature studied, it was found that for From the river depth analysis, it was found that
transportation, the least available depth must areas such as downstream to salinity barrier at
be 1.5 m. To decide the navigable path, the Ambatale, below the Outer Circular Highway
bathymetry data is important. In order to bridge, near to Navagamuwa Bridge and some
obtain accurate information, these data were areas in Hanwella have to be dredged for safe
collected from two institutes namely, Survey navigation especially during low flow periods.
Unit of Sri Lanka Navy and Irrigation
department of Sri Lanka. 6. Salinity Barrier at Water Intake
Ambatale
From Sri Lanka Navy, centreline river depth
variation values (Figure 1), river cross section at Impact of the salinity barrier and the two water
different locations and depth values at different intakes namely, Water intake Ambatale and
locations in the river of the entire study area Kelani Right Bank (KRB) Intake are critical for
were collected. 1:10,000 digital data in the study the navigability of this inland water line. That is
area were obtained from the Survey because on average 540,000 m3 of water is
Department of Sri Lanka. Using the software abstracted at the Ambatale Intake and another
“ArcGIS 10.3.1,” the bathymetry data obtained 180,000 m3 of water is abstracted at Kelani
from Sri Lanka Navy were merged with the 1: Right Bank intake per day accounting for more
10,000 digital data obtained from the Survey than 70% of the drinking water supply in the
Department. Using the river depth details, the Colombo Metropolitan Region. [3]
contours were obtained differentiating the bed
levels of the river into 4 depth ranges. These The water level of the Kelani River is below 0
depth ranges are depths above 4 m, depths MSL up to about Hanwella during the dry
from 4-3 m, depths in the range 3-2 m and river season. When there is high tide, salt water of
depths below 2 m. the sea propagates along the river and reach
Ambatale water intake inducing a limitation in
the treated water production experienced
almost every year. After examining several
solutions to overcome this problem, authorities
have decided to erect a salinity barrier across
the Kelani River in the vicinity of Ambatale
Water Intake. At present, a sandbag barrier is
used to control the salinity intrusion and
control the water flow. But this has not been
effective as maintenance is difficult and also as
salinity intrusion cannot be stopped
Figure 1 - Centreline Depth variation in
Kelani River completely. Thus, they have decided to
construct a permanent barrier across the river
The navigable line was then drawn connecting with the main objectives of ensuring the quality
points with high bed levels, and it was of water by preventing salinity intrusion during
discontinued at several locations as given in low flow periods, maintaining minimum flow
Figure 2, where least available depth for requirement during low flow situations and
4
662 ENGINEER
ensuring a more flexible operation and 7. Locations of the Stations/Jetties
maintenance of the barrier where maintenance
of a sandbag barrier is extremely difficult. Inland water stations are always and should be
inter-modal. People travelling by water must
The barrier is situated 20 m to 30 m arrive and depart by another transportation
downstream of the Water Intake, Ambatale and mode. Typically, roads and rail services which
its construction have been proposed to be provide access to the inland stations/jetties can
executed in 2 phases, and the phase I, which is be called intermodal connectors to inland water
the construction of the foundation has already stations. Improvements to the access of the
been implemented. The foundation has been ports can increase the efficiency of the ports,
constructed by filling boulders and sand in benefit trade and contribute to employment
between 2 sheet piles. The top of the foundation growth and regional productivity.
is at -0.5 m from MSL level. At one corner of the
barrier, a raft passage has been designed for the To attract more passengers to the system, the
movement of vessels. It is a trapezoidal location of jetties and the distances between
passage, and the top of its base is at -2 m from them are vital. When deciding the locations and
MSL level. The dimensions and the location of distances of the stations, the demand generated
the barrier are as follows as given in Figure 3. in the area, land use details, access to the
identified stations/jetties, connectivity with
other transport modes, availability of main
transport corridors and amount of dredging
which need to be done at the proposed location
were considered. In the analysis, the top
priority was given to the demand generated in
the area. These demand data were obtained
from the demand estimation model use by
Figure 3 - Section of the raft passage Megapolis Transport Planning team. The
demand data were forecasted for 2035 with the
Phase II of the project is the installation of an assumption that other modes such as rapid
inflatable rubber balloon across the river. It is transit systems, electrified railways etc. will
going to be installed in 2 sections; a larger come up simultaneously.
balloon across the raft passage and a smaller
balloon across the rest of the structure. The When locating the jetties, the area around the
inflatable balloon will only be inflated during river was analysed using the 1: 10,000 digital
the dry season up to +1 m MSL level usually data from the software “ArcGIS 10.3”. Then the
from January to March, and it will be deflated possible locations of the stations were marked
during normal conditions. considering the demand in the area and the
land use. The land use information was
Hence during normal conditions movement of obtained from the Survey Department of Sri
vessels through the salinity barrier is possible, Lanka, and they were classified to identify the
but movement during the dry season is suitable lands for the stations/jetties. In order
completely restricted. In this time of the year, to minimize land acquisition costs, an attempt
the river will be navigable only from was made to use bare lands at each possible
Peliyagoda to Gonawala West and from location. Further to reduce the construction
Hanwella to Mabima West. Therefore, the only costs, dredging need to be done was also
possible solution that can be implemented is considered at this stage.
having a shuttle bus service connecting 2
stations on either side of the barrier along the To identify the connectivity with other public
Biyagama Road and Sedawatta- Ambatale Road transportation modes, information regarding
via Avissawella Road. Although dredging a bus routes in and crossing the study area were
new channel is an option it is too costly, and it collected from Road Passenger Transport
can increase the salinity intrusion in fresh water Authority, and their routes were drawn on the
in the upstream intakes. Furthermore, another map using “Google Earth Pro” software. The
objective of having a barrier is to store water main bus routes in the area are Kaduwela-
and to control the flow. If another channel is Biyagama (237), Bambalapitiya- Wattala (104),
dredged, controlling of water flow will not be Kohuwela- Kelaniya (135), Fort-
possible. Kadawatha(138/1), Kiribathgoda- Angulana
(154), Dekatana- Pettah (226), Kaduwela-
5
663 ENGINEER
Gampaha(228), Colpity- Kohilawatta (175), Table 3- Locations of stations/jetties
townhall- Salmal Uyana (164) and Pettah-
Kaduwela (143). The marked locations in the
“Arcgis” document, were imported to the
“Google Earth Pro” software in order to
recheck the suitability of the locations marked.
As given in figure 4, marked locations were
adjusted considering the connectivity to other
transport modes, access facilities available at
each marked location and availability of main
transport corridors.
In the stations 1,3,4,6,7,8,9,10,11 and 13, two

jetties/stations were marked on either bank in
order to provide the facility for people on both
Figure 4 - Access to the proposed
sides of the river. At stations 2 and 5, for one
stations/jetties
side of the bank, access facilities are
unavailable. Therefore, to supply the demand
After the analysis, the locations of the proposed
on the side where access facilities are not
stations and corresponding expected daily bi-
directional demand obtained from Megapolis
Transport Planning Team are as follows in 8. Features of the Boat/Vessel
Table 3.
The type of vessel that can be used for
available, the solutions are the construction of a waterways depend on the water body,
bridge connecting river banks or construction passenger capacity requirements, length of trip,
of a station while developing access facilities to depth and width of the waterway and docking
it. Out of these solutions, the most economical requirements. There are different kinds of
solution must be taken considering the vessels operating in the inland waterways
characteristics at the location. In the station 12, throughout the world. Some of them are
land use is very high on only one side of the passenger trawlers, passenger steamers,
river, which is Embulgama area and in the hydrofoil boats, electric boats where it can be
other side land use and the population density solar powered, battery powered or hybrid,
is very low. Thus having 2 stations on either passenger launches, hovercraft and ferries.
bank is an additional cost.
Considering the features of the vessel to be
After locating the stations, it was seen that the used in the Kelani River, the most important
station to station distance between first 11 aspect to consider will be the pollution the
stations ranges from 1.5 km to 2 km and in the vessel creates. Since two major water intakes
last few stations, the distance between station to namely Ambatale water intake and Kelani
station is more than 3 km. That is because Right Bank intake which supplies 70% of the
population density and demand around last 2-3 total amount of water supply in the area are
stations are very low compared to other located in the study area, special attention must
stations. Therefore, having stations in between be given to minimizing the adverse effects on
them will not be effective in the present them. Since fuel powered boats emit various
conditions. types of chemicals to the water body, a solar
6
664 ENGINEER
powered or battery powered boat will be facility. And also, a detailed survey of the river
feasible to the Kelani River as they do not emit must be carried out to find locations of the rock
any harmful substances to the water body. outcrops that can hinder the movement of
Since the depth variation in the Kelani River is boats.
significant, a low draught boat like a hydrofoil
will be more suitable. The demand for References
Mattakkuliya-Hanwella line is not high when
compared to other 2 inland waterway lines. 1. De Silva, G. L. D. I., 2003. The Study of the
And another significant feature of this line is Feasibility of Water Transportation in Colombo
the distance between 2 stations. Compared to Metropolitan Region, Colombo: University of
other waterway lines station to station distance Moratuwa.
is greater in Mattakkuliya-Hanwella line.
2. Bandara, J. M. S. J., Sirisoma, R. M. N. T.,
Therefore, a boat with a capacity around 30-40 Pushpakumara, T. D. C., Jayarathne, P., and
passengers and speed up to 30 km/h will be Cooray, L. H. C. (2004). Canal Commuter Transit
ideal for this line. And also, the boats should be Study. Colombo: Transportation Engineering
of good quality, unsinkable and air- Division, Department of Civil Engineering,
conditioned. University of Moratuwa.
9. Conclusion & Recommendations 3. Abeykoon, A. M. H. K. and Nawarathna, N. M. A.

L. B., 2011. Enhancing Water Quality by Salinity
Although the potential of water bodies was Barriers and its Impact on Environment. XIVth
neglected in the recent past, traffic planners IWRA World Water Congress, p. 9
have realized its potential as it is a good
solution for the existing traffic in the road 4. Jayawardane, C., 2009. Ancient Modes of Inland
structure and also as it is a cost-effective, Water Transportation in Sri Lanka; Stability
environment-friendly, flexible transportation Analysis of Craft Excavated from Attanugala
mode. Oya. The Institution of Engineers, Sri Lanka,
43(3), pp. 14-21.
In this paper, the infrastructural aspects of the
inland transportation line from Peliyagoda to 5. Kalyani, T., Vidyasagar, D. and Srinivas, V., 2015.
Hanwella along Kelani River was focused. Accident analysis of river boats capsize in Indian
Bathymetric data, land use data, data regarding Inland Waters and safety aspects related to
other transportation modes were collected from passenger transportation. International journal of
various institutes and using “ArcGIS” and innovative research and development, 4(7).
“Google Earth Pro” software’s, navigational
6. Jugović, A., Mezak, V. and Lončar, S., 2007.
path was drawn. A site survey was carried out Organization of Maritime Passenger Ports, s.l.:
to examine the horizontal and vertical clearance PREGLEDNI RAD.
of the channel. Impact of salinity barrier at
Ambatale was studied, and it was found that 7. Awal, Z., Islam, M. & Hoque, M., 2015. Analysis of
the navigational path would be blocked during Passenger Vessel Accidents in the Inland Waterways
the dry season. Therefore, a shuttle bus service of Bangladesh. s.l.:ResearchGate.
was proposed to mitigate that issue.
8. Western Region Megapolis Planning Project
Transport Master Plan, Ministry of Megapolis
It can be concluded that Kelani River is a
and Western Development,
navigable river although there are some areas https://www.megapolis.gov.lk/, Visited, 20th
to be dredged. After developing the system, September 2016.
traffic congestion in the road network will be
minimized, passengers will be able to have an 9. Perera, D., 2014. Colombo Metropolitan
efficient, safe and a pleasant journey, air and Transport Master Plan $ Areas for International
sound pollution will be minimized, and the Cooperatioon. Colombo: Ministry of Transport.
area will be economically upgraded.
10. http://www.google.com/earth/download/ge/,
Visited, 10th September 2016.
Further studies must be done to reduce the
siltation and bank erosion which makes the 11. Senanayake, I., Pushpakumara, T. and
navigable sections shallow and hinder the Puswewala, U., 2013. A GIS Based Analysis
movement of the boats. A detailed on Use of Inland Waterways for Public
environmental impact assessment must be Transportation in the Western Province of Sri
carried out. IT related applications must be Lanka. The Institution of Engineers Sri Lanka,
introduced to improve the quality of the 46(2), pp. 1-11.
7
665 ENGINEER
AnnualSessions
IESL,pp.
pp. [1 -- 10],
[667 674],2017
2017
The Institution
Lanka
Hydroplaning Risk Evaluation in High Speed Low

Traffic Volume Highways
M.D.M. Sheranie and S. Senevirathne
Abstract: During high-intensity rainfall events, a significantly thick water film builds up on the
surface of a highway. Hydroplaning occurs when water pressure builds up in front of a moving tire
resulting in an uplift force sufficient to separate the tire from the pavement. The loss of steering and
drag force produced during hydroplaning. During the rain, due to the decreased visibility, the
traveling speed usually exceeds 80 kmph can drop by ten percent (10%) of its initial value. According
to the current studies, the effect of hydroplaning can be significant even at a speed less than 70 kmph.
Since the travel speed is well above this value in Sri Lankan expressways there is a higher probability
of occurrence of hydroplaning hazard in expressways in Sri Lanka. A preliminary evaluation was
performed to compare and test the wet weather crash frequencies and rates across different pavement
materials. In this study, Florida crash database used for the analysis to identify the critical crash
locations in Sri Lanka. The effects of the generated water film are considered as primary sources of
investigation in crown and transition sections. It is found that the conditions are quite different for
superelevation transition sections compared to normal crown sections. The combined effect of lateral
and longitudinal slopes in superelevation transition sections causes a significant increment in Water
Film Thickness (WFT) and the magnitude of the maximum Water Film Thickness (WFT) increases at
superelevation transitions compared to normal crown conditions. Finally, the current study evaluates
the hydroplaning effect by each factor and identified the safe speed limit for high crash locations in
expressway network of Sri Lanka based on a case study of Southern Expressway. Criterion for
pavement surface design must include a hydroplaning risk evaluation in the design stage of
expressways and posted speed limits can be redefined for already constructed expressways.
Keywords: Heavy rain, water film thickness, travel speed, crown and transition sections, safe
speed limit
1. Introduction
Expressway network in Sri Lanka (Figure 1) was
Driving risk under inclement weather conditions designed to have a high speed. Even though there
has long been considering higher than under are some expressway projects are taking place,
normal weather conditions. Among several Southern expressway and Katunayake expressway
variables pertaining to weather conditions, rainfall are the operational and implemented
is recognized as a critical factor that affects expressways out of the entire expressway
traffic crash occurrences [1]. Heavy rainfall network in Sri Lanka.
often results in reduced visibility, increased
stopping distance and hydroplaning. When the travel speed exceeds 80 kmph, the
speed can be dropped by ten percent (10%)
During highly intense rainfall events, a during the rain due to the decreased visibility
significantly thick water film builds upon [2]. According to the studies in developed
highway surfaces. When a vehicle is countries, the effect of hydroplaning can be
hydroplaning, the driver will lose control of the significant even at speed less than 70 kmph [3].
vehicle. Hydroplaning is caused when water on
a roadway is deep enough to lose contact Eng. (Ms.) M.D. M. Sheranie, BEng (Hons, UK),
between tires of the vehicle and the road. Many AMIE(SL), ICIOB(UK), AMCPM(SL).
people assume that water has to be deep enough
Eng. (Mrs.) S. Seneviratne, C. Eng., MIE(SL), B.Sc. Eng.
to cause hydroplaning, but even a thin film of (Moratuwa).
water can cause problems, especially if
someone is driving at a higher speed. It is
obvious that the risk of hydroplaning is higher
on high-speed roadway facilities.
667
Since the travel speed on Southern Expressway hydroplaning accidents. Since the Southern
is well above this value (i.e. minimum travel Expressway is the first expressway in Sri Lanka,
speed is 80 kmph), it is very important to people have less awareness of the effect of
evaluate risk factors that contribute to hydroplaning.
hydroplaning. Even though there are several
kinds of situations which produce the An accident occurred in Southern Expressway at the
hydroplaning effect in expressway, mostly 8th-mile post in Kahatuduwa on the very first day
where water is assembling as a pond such as after the expressway was declared open. This was
the bottom of sag vertical curves, horizontal the first accident occurred in Expressway and it was
transition curve sections, and pavements with a viscous hydroplaning crash [6].
wheel tracks or ruts are the most common Practical methods have to be followed in
situations [4]. experimentation and analysis of the effect of
hydroplaning in wet weather condition in
The water film conditions are different in expressway network of Sri Lanka since Sri Lankan
superelevation transition section compared to normal expressway network has been designed using
crown sections. The magnitude of the maximum ASSHTO and TRL Road Design.
Water Film Thickness (WFT) increases at
superelevation transition section when compared The objective of this research is to identify the
with normal crown sections. Therefore, the potentially high crash locations in a road section
hydroplanning can present a server safety issue at and propose a safe speed to minimize the
transition sections than at normal crown sections in consequences of hydroplaning.
highways.
But unfortunately, this has not been considered in 2. Literature Review

most highway geometric design guidelines (ie: -
AASTO, TRL Road note and Australian guideline). 2.1 Impact of Weather and Climate
As mentioned by Torbic [5], even though the design Fwa and Ong [7] have stated that “impact of
speed has been estimated based on the mechanistic
the weather and climate is a major part of the
properties of the vehicle at curved segments, there is
hydroplaning accident”.
a high probability for accidents to be occurred in
curved sections due to high speed.
The wet climate of a major part of Sri Lanka is
strongly governed by the monsoon seasons, the
associated air masses and planetary wind
regimes over South Asia. Due to alternating
monsoon circulation systems through the year,
two principal monsoon rainfall seasons and
two transitional periods, called Inter-monsoon
seasons can be identified in Sri Lanka.
Rainfall intensity at any time scale is one

important aspect of monsoon climatology.
Mean statistics reveal comparably high rainfall
intensity with a maximum of 25 mm in the
interior part of the south-western region, where
the mean number of rainy days is higher than
100 [8].
2.2 Water Film Thickness with Rainfall

Intensity
The thickness of the water film that contributes
to hydroplaning is the Mean Texture Depth
(MTD) plus the water film above the tops of the
surface asperities. The flow path (L) for a
particle of water falling on a pavement surface
Figure 1 - Expressway Network in Sri Lanka is simply defined as the line determined by the
(RDA) slope (S) of the pavement surface. Thus, the
maximum flow path for a pavement section is
There were a number of accidents occurred in the longest flow path for the section. The
the expressway during the rainy time including maximum distance that a rain droplet can flow
668
between the point of contact with the water In this transition sections, cross slope reaches to
film and its point of exit from the pavement. zero value. With a longitudinal slope, the flow
For a given quantity of rainfall, with known path of water can be extended and it leads to
pavement parameters, the following relation higher water film thickness near to transition
can utilize to evaluate water film thickness t, sections.
from PAVDRN program [2].
0 .6 2.4 Water Film Thickness in Transitions

 nLI 
t 0 .5 
 MTD Sections
 36.1S  …(1) When pavement rotates with a superelevation
Where, section, the flow path of water get extended
I = Rainfall intensity and as a result of that, it can lead to higher
n = Manning coefficient water film thickness.
Maximum runoff depth (Water Film Thickness,
WFT) was estimated using numerical
2.3 Effect of Cross Slope on Water Film simulation modelling for different roadway
Thickness configurations [10]. Figure four shows the
Typically, the cross slopes on either side are results for a roadway with 4 travel lanes for an
symmetric about the centre line and they are Asphalt roadway surface with a Manning (n) =
denoted as normal crown. But at a curve 0.015 and rainfall intensity of 100 mm/hr. The
(super-elevation) the cross slopes about the roadway alignment is based on a design speed
centre line changes as shown in Figures 2 and 3. of 96.6 kmph, which corresponds to a
The cross slopes of the lanes between the centre maximum relative gradient of 0.45 %. The „zero‟
line and inside edge remains constant and is distance corresponds to the Zero Cross Slope
equal to the normal crown. However, the cross (ZCS) station.
slopes between the centre line of the road and
the outside edge change and its‟ magnitude
becomes equal but higher that the cross slope
on the other side in super-elevation sections.
Figure 2 - Lateral alignment of superelevation

transition with cross slope = 4% [9]
Figure 2 shows that the slope of the normal
crown is 2% and the slope of the full super-
elevation is 4%. In this, fully superelevated Figure 4 - Variation of the WFT with distance
section can be seen at a distance of L2 and from ZCS (100 mm/h rainfall, longitudinal
cross-sectional details can be seen as Figure 3. slope varies from 0.5% to 6% [10]
Figure
Figure 3 - Typical
3 – Typical CrossCross Section
Section of Southern
of Southern Experssway
Expressway
669
It can be seen that the water film near to ZCS Where,
can be increased up to 4.3 mm at outer lanes. HPS = Hydroplaning speed (mi/h)
Further, it can be seen that when longitudinal WFT = Water film thickness (in)
slope varies from 0.5% to 6% effect of that does
not contribute to much change of maximum  Water film thickness greater than or
water film. equal to 2.4 mm,
With regard to vehicle safety at superelevation HPS = 3.09 A …(3)

transitions, it is not clear whether the
magnitude of WFT or the changes in WFT in Where,
the longitudinal and lateral directions are the A is the greatest of the values
most critical variables. According to Figure 4, it calculated using equations (4) and (5);
is clear that maximum water film that can be
generated by the super elevated transition
section is always less than 5 mm due to a
rainfall intensity of 100 mm/h.
…(4)
2.5 Friction Number Vs Water Film
Thickness
With a known water film thickness, the effect of
dynamic friction has been evaluated as shown …(5)
in Figure 5 [10]. The dynamic friction lowered Where,
by 35% once the pavement surface gets wet. MTD – Mean Texture Depth (in)
Further, it can be seen that the dynamic friction
coefficient can be further lowered up to 0.15
with the increment of water film thickness. 3. Methodology
As described in the section 1, Sri Lankan

expressway network has been designed using
the road manuals in the USA and the UK. In
this research, significant efforts have been
devoted to data acquisition, processing, and
compilation. A preliminary evaluation was
performed to compare and test the wet weather
crash frequencies and rates across different
pavement materials to compare the pavement
parameters with Sri Lankan expressways.
Project Methodology flowchart shown in Figure

6 has been used for the current study to
Figure 5 - Variation of Friction Coefficient
evaluate the hydroplaning effect by each factor
with Speed and Water Film Thickness [10]
and also to identify the safe speed limit for high
crash locations in expressway network of Sri
2.6 Hydroplaning Speed Vs Water Film
Lanka based on a case study of Southern
Thickness
Expressway.
As stated by Glennon [4], the vehicle speed at
incipient hydroplaning depends on the storm
Florida crash database used for the analysis of
water runoff depth, tire pressure, tire tread
crash locations in Sri Lanka since it is difficult
depth, average pavement texture depth and
to find a highway crash database in Sri Lanka.
other factors. From above-mentioned factors,
Florida is a tropical state as Sri Lanka, but few
storm water runoff depth (water film depth) is
assumptions have made such as both Southern
the primary variable.
expressway and Florida state road segment
have similar driver behavior, demographic,
Following equations are used in PAVDRN [2];
environmental and vehicle factors.
 Water film thickness less than 2.4 mm,
In this analysis, data for four years (2007 to
2010) obtained from Florida Department of
HPS = 26.04 WFT-0.259 …(2)
Transportation (FDOT) was considered. Two
670
major data source used in this evaluation as deliver the most reliable and real-time weather
follows; data for Florida State. Considering crash
 Crash Analysis Reporting System distribution in the study area and weather
(CARS) database of Florida station proximity to each crash location, 72
 Pavement Condition Survey (PCS) airport weather stations were selected to obtain
database microscopic weather data measured on hourly
based intervals in this study from year 2007 to
Data compilation process used to identify wet 2010. For each crash observation, three weather
weather crashes and rainfall intensity at the stations out of the 72 weather stations were
crash location. Filtering criteria has been used used by their proximity to the crash location to
to identify the wet weather crashes from the estimate the hourly weather data.
crash analysis reports.
However, when it was difficult to find out three
weather stations nearby for these one-
dimensional points must be applied to two-
dimensional areas. This was accomplished by
creating Thiessen Polygons around the weather
station location points [11].
4.2 Southern Expressway

As described in section 2.1, since the Southern
Expressway is located in South-western region
rainfall is an utmost important factor to
consider. Rainfall intensities are calibrated on
the basis of accumulated storm water during 15
min, 30 min or during 1-hour rainfall. Irregular
variation can be observed many times as shown
in Figure 8.
According to the Figure 8, it can be seen that

rainfall intensity can be such different during
that considered time period. So it is important
Figure 6 – Methodology Flow Chart to consider the immediate rainfall effect with
respect to total rainfall. The effect of generating
4. Weather Data Collection water film thickness can be varied on the
pavement geometry and the pavement
4.1 Florida State materials.
All the identified crashes have been assigned
with geological locations. According to that, all 5. Results & Analysis
the crashes were located in high-speed
pavement facilities like interstate roadway 5.1 Pavement Properties Comparison
network. Sample data set used to compare the pavement
Figure 7 - Fluctuation in Rainfall Intensity (mm/h) Vs Time at Interstate Highway (I-275) Segment
properties of Southern Expressway (SE) and
National Climatic Data Centre (NCDC) in the Florida Department of Transportation (FDOT)
USA affiliated „Weather Source‟ was found to highways (Table 1).
671
A statistical analysis was done for both data curves. In this analysis, two crash rates have
sets to find the mean and standard deviation. been considered,
The results are shown in Table 2. 1. Straight sections and
2. Curved section.
Table 1 – Extracts of the SE and FDOT Formulation of the null hypothesis and
comparing data set alternative hypothesis are shown in below;
H0: μ1 = μ0
 During wet weather conditions, the
nature of hydroplaning related crashes
that have occurred on straight roadway
segments is the same as that in curved
segments.
SE roadway sections
H1: μ1 ≠μ0
 A significant difference in
hydroplaning related crashes on curved
sections with corresponding to the
straight roadway segments.
Reject the null hypothesis if there is no

FDOT roadway sections difference between means at designated
significance levels. The four-year crash data
Table 2 – One Sample Statistics from 2007 to 2010 were considered. Then the
data set was screened to identify the
hydroplaning related accidents as shown in
Table 4.
Table 4 - Hydroplaning Related Crash

Screening and Analysis
A sample t – Testing to compare pavement

parameters of SE and FDOT highways. The
result is shown in Table 3.
Table 3 – One Sample t – Test
Since the variances and means of the two

populations are known, hypothesis testing has
been considered to avoid any Type I error, i.e.
considering the very low significant level of
0.1%. It was assumed that data populations are
normally distributed.
From Table 3, it is seen that variance of lower
and upper values of FDOT data set is in From Table 5 it is seen that there is a significant
between 95% confidence interval of SE data set.
hydroplaning risk in curved sections compared
Therefore, from above t-testing that pavement to straight portions. It could be concluded from
parameters of FDOT are similar to the SE
the above hypothesis testing that the
pavement properties. hydroplaning risk at curvature segment is three
times higher than on normal crown straight
5.2 Critical Section Identification
roadway segments.
Hypothesis testing to compare wet-weather
crashes Vs crashes during dry condition on
672
It should be noted that hypothesis tests Table 6 – Typical Roughness Parameters Used
performed in this preliminary evaluation in PAVDRN [2]
considered pavement width as the sole factor to
explain the difference in wet weather crash
frequencies and rates. It can be viewed as a
qualitative analysis. More sophisticated
multivariate analyses, such as Poisson and
Negative Binominal regression, will be
employed to evaluate and quantify the
potential risk of hydroplaning and the effects of
other contributing factors, such as pavement
condition, speed, visibility, curvature, etc.
Table 5 - Hypothesis Test Verification of

Hydroplaning Related Crashes
Figure 9 - Probable Water-film Thickness Vs

Rainfall Intensity Based on Southern
Expressway Pavement Characteristics
5.3 Case Study Based on Southern

Expressway
In this analysis, I-275 roadway segment of the
Figure 10 – Safe Vehicle Speed Limit for
Southern expressway had been considered at a
Southern Expressway Road Segment
chainage of 0+700 distance, which highway
behaves with a 0.1% cross slope and 0.667%
longitudinal slope. Based on the probable
rainfall pattern occurring on the southwestern 6. Conclusions
part of Sri Lanka, water film thicknesses have
been predicted as shown in Figure 9. Hydroplaning occurs when water pressure
builds up in front of a moving tire resulting in
The PAVDRN equation (1) stated in section 2.2 an uplift force sufficient to separate the tire
has been used to predict the water film from the pavement. During rain, due to the
thickness for given parameters. Typical decreased visibility, the traveling speed which
roughness parameters for different pavements usually exceeds 80 kmph can drop by ten
used in PAVDRN are shown in Table 6. percent (10%) of its initial value. But according
to the current study, the effects of hydroplaning
It is obvious that dynamic friction reduces with can be significant even at speeds around 70
the increment of water film thickness. With kmph. This is applicable for Southern
that, the safer vehicle speed calculated using Expressway since the design speed is well
equations (2), (3), (4) and (5) stated in section above 70 kmph.
2.6. It can be seen that, when rainfall intensity
increases more than 20 mm/h, the safer speed The effects of the generated water film
reduces than the design speed. Based on the thickness on superelevation sections are quite
lane geometry and pavement parameters it can different with compared to normal crown
be further reduced i.e. 74 kmph. Likewise sections. When compared regard to vehicle
following Figure 10 was developed to identify safety at superelevation transitions, it is not
the safer speed limit at a specific transition clear whether the reduction of friction across
location (chainage 0+700 of I-275 roadway the pavement or changes in WFT in the
segment) of the Southern expressway. longitudinal and lateral directions are the most
critical variables.
673
It is realized that hydroplaning can present a 6. http://www.dailymirror.lk/article/accident-
severe safety issue at transition sections than on-southern-expressway-15065.html. Visited 21st
the normal crown sections. Due to the numbers September 2017.
of different causative factors are involved in
7. Fwa, T. F., & Ghim P. O., "Wet-Pavement
hydroplaning and their relative contributions,
Hydroplaning Risk and Skid Resistance:
there is a great deal of uncertainty in the Analysis." Journal of Transportation
initiation of hydroplaning. Engineering, ASCE, Volume134, No. 05, 2008,
pp. 182-190.
The safe speed limit for the critical sections
such as curve sections and super elevated 8. Ranatunge, E., et al., "Changes in the Southwest
transition sections has been identified using Monsoon Mean Daily Rainfall Intensity in Sri
factors. Redefine the posted speed limit can be Lanka: Relationship to the El Niño–Southern
done for already constructed expressways. Oscillation." Palaeogeography, Palaeoclimatology,
Palaeoecology, Volume 197, No. 01, 2003, pp.
To increase the safety level of the expressways, 1-14.
it is an advantage to identify critical locations in
the expressway network by analysing many 9. http://onlinemanuals.txdot.gov/txdotmanuals/
sections. Based on the analysis, high priority hyd/pavement_drainage.htm,
Visited, 25th August 2016.
must be given for drainage design at horizontal
curve sections with superelevation in order to 10. Charbeneau, R. J., Jaehak J., & Michael E.
minimize the water level builds on the surface. B., “Highway Drainage at Superelevation
It is always better to do the skid test for Transitions”, Centre for Transportation
identified critical locations and skid proof if Report, The University of Texas at Austin,
necessary and also pavement grooving will Report No. FHWA/TX-08/0-4875-1, 2008.
help to reduce the thick water film builds up on
the surface of a highway [12]. 11. Kyte, M., et al., "Effect of Weather on Free-Flow
Speed." Transportation research record: Journal
Consideration of the effect of hydroplaning of the transportation research board, Report
when designing the highway geometry and 1776, 2001, pp. 60-68.
pavements will lead to a safe drive during the
wet weather. 12. Enustun, N. “Study to Identify Hydroplaning
Crashes” TSD 330-77. Final Report submitted to
Michigan Department of State Highways and
References Transportation, 1976.
1. Gallaway, B. M., et al., “Tentative Pavement and

Geometric Design Criteria for Minimizing
Hydroplaning”, No. FHWA-RD-75-11 Final
Report, 1975.
2. Anderson, D. A., et al., "Improved Surface

Drainage of Pavements”, The Pennsylvania
Transportation Institute, The Pennsylvania State
University, No. Project 1-29 Final Rpt., 1998.
3. Cheng, G., et al., "Speed Analysis of the Radial

Tire on Hydroplaning Pavement”, 3rd
International Conference in Material, Mechanical
and Manufacturing Engineering, IC3ME, 2015, pp.
1485-1489.
4. Glennon, J. C., Roadway hydroplaning – A

framework to determine critical wheel rut depths,
2015.
5. Torbic, D. J., et al., "A Guide for Reducing

Collisions on Horizontal Curves." NCHRP,
Report 500, Vol. 7, 2004.
674

IESL Transaction Partb2017 Technical Paper PDF

Uploaded by

Copyright:

Available Formats

IESL Transaction Partb2017 Technical Paper PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

IESL Transaction Partb2017 Technical Paper PDF

Uploaded by

Copyright:

Available Formats

“To promote the acquisition and interchange of technical knowledge,

THE INSTITUTION OF ENGINEERS

CEO / Executive Secretary

The Institution of Engineers, Sri Lanka

Tel : 2 698426, 2685490, 2 699210

Karunaratne & Sons (Pvt) Ltd.

1. Rainfall Thresholds for Initiation of Landslides in Ratnapura District

2. Experimental Study on the Effect of Presence of Deep Embedded Retaining Walls on

3. Shrinkage Behavior of a Liner Material with Expansive Properties for use in an

5. Design of Soil–Cement–Bentonite (SCB) backfill and construction of SCB slurry walls

7. Analysis of Energy Dissipation in Post-Tensioned Hybrid Precast Concrete Frames

8. Study on Filler Slab Construction

9. Energy Dissipater for Post-Tensioned Steel Frame System

13. Sub-Contracting and Accident Risk in Construction Industry : A Case Study in

15. Multi-Functional Floor Cleaning Robot for Domestic Environment

18. Assessing the Performance of a Destructive Measurement System: A Case of an

21. Mechanized Fertilizer Applicator Operated by Two-wheel Tractor for Coconut

22. Water Loss Reduction by Understanding the Reticulation A Case Study

24. A Study on Recontamination Risks in Water Distribution Systems, using A Network

26. Analysis of Long-term Trends of Climatological Parameters in Kigali, Rwanda

29. Application of 3D Physical Modelling for Verification and Optimization of Hydraulic

30. Numerical Model Investigations on Dispersion of Industrial Effluents in East Coast of

31. Numerical Modelling Study for RahfalhuHuraa Resort Development in Maldives

33. Verification of Wave Transformation to Establish the Nearshore Wave Climate

34. Fabrication and Mechanical Property Analysis of Microcrystalline Cellulose Based

35. Synthesis and Characterization of Nanoparticle-Incorporated Antimicrobial Coating

37. Antecedents of Work Engagement : Case Study of Outsourced Employees in

38. The Structure of Land Prices in Colombo and its Implications

42. Condition Assessment of Power Transformers using Dissolved Gas Analysis

45. Museum Navigation System with Beacons

51. Energy Management System of a Solar Driven DC System

53. A Novel Handover Mechanism for Visible Light Communication Network

54. Technical, Economic, Operational and Dispatch Analysis of Optimum Integration of

56. Smart Electronic Analyzer for Determination of Chicken Meat Spoilage

57. Power Electronic Interface of a Solar Driven DC System

59. Energy Efficiency Benchmarking of Pumps in Water and Wastewater

60. Feasibility of Hybrid Reactors for Removal of Heavy Metals in Wastewater

61. Feasibility of UASB Reactor for Removal of Heavy Metal in Wastewater

62. GIS-Based Site Suitability Analysis for Wastewater Treatment Plants

65. Upgrading of the Existing Sedimentation Tank at Hemmathagama Water Supply

69. Life Cycle Assessment Based Tool to Visualize Environmental Impact of

70. Safety Analysis of Anaerobic Digestion Plant Operation

75. Low-Cost drinking Water Filter

80. Development of a Methodology for Road Maintenance Planning of Low Volume

82. Infrastructure Feasibility Analysis of Inland Water Transportation from Peliyagoda to

Rainfall Thresholds for Initiation of Landslides in

Keywords: Disaster Management, Landslide, Rainfall Threshold, Ratnapura District

1. Introduction created frequent landslide events in over 10

4. Results and Discussion

The curvilinear Regression Model obtained

Figure 3 - Summary of Thresholds

17. Cannell, R. Q., Belford, R. K., Blackwell, P. S.,

18. Kim, S., Hong, W., and Kim, Y., “Prediction of

Experimental Study on the Effect of Presence of Deep

Keywords: Shallow foundations, Deep embedded retaining walls, Settlement

1. Introduction are time consuming, expensive and complex.

According to [5]“earth retaining structures are

It can be seen that the load–settlement

 Experimental investigation on the effect of