Book (Assessment of Soil Liquefaction Safety..)

Lecture Notes in Civil Engineering
T. G. Sitharam
Ravi Jakka
L. Govindaraju Editors
Local Site
Effects and
Ground
Failures
Select Proceedings of 7th ICRAGEE 2020
Volume 117
Series Editors
Marco di Prisco, Politecnico di Milano, Milano, Italy
Sheng-Hong Chen, School of Water Resources and Hydropower Engineering,
Wuhan University, Wuhan, China
Ioannis Vayas, Institute of Steel Structures, National Technical University of
Athens, Athens, Greece
Sanjay Kumar Shukla, School of Engineering, Edith Cowan University, Joondalup,
WA, Australia
Anuj Sharma, Iowa State University, Ames, IA, USA
Nagesh Kumar, Department of Civil Engineering, Indian Institute of Science
Bangalore, Bengaluru, Karnataka, India
Chien Ming Wang, School of Civil Engineering, The University of Queensland,
Brisbane, QLD, Australia
Lecture Notes in Civil Engineering (LNCE) publishes the latest developments in
Civil Engineering - quickly, informally and in top quality. Though original
research reported in proceedings and post-proceedings represents the core of
LNCE, edited volumes of exceptionally high quality and interest may also be
considered for publication. Volumes published in LNCE embrace all aspects and
subfields of, as well as new challenges in, Civil Engineering. Topics in the series
include:
• Construction and Structural Mechanics
• Building Materials
• Concrete, Steel and Timber Structures
• Geotechnical Engineering
• Earthquake Engineering
• Coastal Engineering
• Ocean and Offshore Engineering; Ships and Floating Structures
• Hydraulics, Hydrology and Water Resources Engineering
• Environmental Engineering and Sustainability
• Structural Health and Monitoring
• Surveying and Geographical Information Systems
• Indoor Environments
• Transportation and Traffic
• Risk Analysis
• Safety and Security
To submit a proposal or request further information, please contact the appropriate

Springer Editor:
– Pierpaolo Riva at pierpaolo.riva@springer.com (Europe and Americas);
– Swati Meherishi at swati.meherishi@springer.com (Asia - except China,
and Australia, New Zealand);
– Wayne Hu at wayne.hu@springer.com (China).
All books in the series now indexed by Scopus and EI Compendex database!
More information about this series at http://www.springer.com/series/15087

T. G. Sitharam · Ravi Jakka · L. Govindaraju
Editors
Local Site Effects

and Ground Failures
Select Proceedings of 7th ICRAGEE 2020
Editors
T. G. Sitharam Ravi Jakka
Indian Institute of Technology Guwahati Department of Earthquake Engineering
Guwahati, Assam, India Indian Institute of Technology Roorkee
Roorkee, Uttarakhand, India
L. Govindaraju
Department of Civil Engineering
Bangalore University
Bengaluru, Karnataka, India
ISSN 2366-2557 ISSN 2366-2565 (electronic)

ISBN 978-981-15-9983-5 ISBN 978-981-15-9984-2 (eBook)
https://doi.org/10.1007/978-981-15-9984-2
© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature
Singapore Pte Ltd. 2021
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Preface
The local soil condition can amplify the seismic waves passing through the soil and
this effect is termed as local site effects. This book volume contains the latest research
papers on ground response analyses and local site effects, seismic slope stability
and landslides, and GIS and remote sensing applications for geo-hazards, selected
from the proceedings of the 7th International Conference on Recent Advances in
Geotechnical Earthquake Engineering and Soil Dynamics, 2021.
We thank all the staff of Springer for their full support and cooperation at all the
stages of the publication of this book. We do hope that this book will be beneficial to
students, researchers, and professionals working in the field of earthquake hazards.
The comments and suggestions from the readers and users of this book are most
welcome.
Guwahati, India T. G. Sitharam

Roorkee, India Ravi Jakka
Bengaluru, India L. Govindaraju
v
Acknowledgements
We (editors) want to thank all the authors, who have contributed to the book. We
could bring this book out due to all the authors’ timely contribution and cooperation.
We thank and acknowledge the service of the following reviewers for their valuable
time and efforts.
Ajay Chourasia, CSIR-CBRI
Amarnath Hegde, Indian Institute of Technology Patna
Amit Verma, IIT (BHU)
Anil Cherian, Strainstall
Anitha Kumari S. D., Ramaiah University of Applied Sciences
Arvind Kumar Jha, Indian Institute of Technology (IIT) Patna
Asha Nair, CMR Institute of Technology, Bengaluru
Babloo Chaudhary, NITK Surathkal
Bal Rastogi, Indian Society of Earthquake Science
Chittaranjan Birabar Nayak, Vidya Pratishthan’s Kamalnayan Bajaj Institute of
Engineering & Technology
Dauji Saha, Bhabha Atomic Research Centre; Homi Bhabha National Institute
Deepankar Choudhury, Indian Institute of Technology Bombay
Dhanaji Chavan, IISc
Gopal Santana Phani Madabhushi, University of Cambridge
Jagdish Sahoo, IIT Kanpur
Kalyan Kumar G., NIT Warangal
Karthik Reddy Konala S. K., IIT Hyderabad
Ketan Bajaj, Risk Management Solutions
Manas Kumar Bhoi, PDPU
Md. Mizanur Rahman, University of South Australia
Padmanabhan G., Indira Gandhi Center for Atomic Research
Pradeep Kumar Singh Chauhan, CSIR-Central Building Research Institute, Roorkee
Premalatha Krishnamurthy, Anna University
Prishati Raychowdhury, IIT Kanpur
Purnanand Savoikar, Goa Engineering College
Rajib Saha, NIT Agartala
vii
viii Acknowledgements
Rajib Sarkar, IIT(ISM) Dhanbad

Ramkrishnan R., Amrita Vishwa Vidyapeetham
Rangaswamy K., NIT Calicut
Ravi K., IIT Guwahati
Renjitha Varghese, National Institute of Technology, Calicut, Kerala, India
Sanjay Verma, Indian Geotechnical Society, Jabalpur Chapter
Sarat Kumar Das, Indian Institute of Technology (ISM) Dhanbad
Shreyasvi C., National Institute of Technology Karnataka
Snehal Kaushik, Girijananda Chowdhury Institute of Management and Technology,
Guwahati
Sreevalsa Kolathayar, National Institute of Technology Karnataka, Surathkal
Supriya Mohanty, Indian Institute of Technology (BHU), Varanasi
Surya Muthukumar, Amrita School of Engineering, Amrita Vishwa Vidhyapeetham
Vinay Srivastava, Retd. IIT ISM Dhanbad
Vipin K. S., Swiss Re
Visuvasam Joseph Antony, Vellore Institute of Technology
Contents
Ground Response Analysis with Deep Bedrock Depth

in Indo-Gangetic Plains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Priyanka Sharma, M. L. Sharma, and V. A. Sawant
Soil Amplification Study for Kalyani Region, Kolkata . . . . . . . . . . . . . . . . . 13
Himanshu Shukla, Pradeep Muley, and Sajjan Kumar
Simulated Annealing Algorithm for Subsurface Shear Wave
Velocity Investigation Using Ground Vibration Data . . . . . . . . . . . . . . . . . . 23
Anushka Joshi, Sateesh K. Peddoju, and Mohit Pandey
Nonlinear Soil Amplification Models for a Moderately Active
Seismic Zone in India . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
C. Shreyasvi and K. Venkataramana
Prediction of Future Surface PGA in the States of Indo-Gangetic
Basin Considering Site Specific Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
P. Anbazhagan, Mohammad Rafiq Joo, Meer Mehran Rashid, and Ketan Bajaj
Synthetic Ground Motion Simulation for Varanasi City . . . . . . . . . . . . . . . 67
Manjari Singh, S. K. Duggal, V. P. Singh, and Keshav Kumar Sharma
Dynamic Study of Existing Structure Influenced by Adjacent Deep
Excavation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
L. Geetha, M. N. Hegde, and M. Mohammed Ayaz
1D and 2D Dynamic Site Response of Landfill Site Through
Numerical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
Parul Rawat and Supriya Mohanty
A Study on Characteristics of Soil Profile of Guwahati City Against
Different Ground Motions: 1D NonLinear Ground Response
Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Amar F. Siddique, Anusuya Acharjee, and Binu Sharma
ix
x Contents
One-Dimensional Ground Response Analysis to Arrive at Surface

Peak Ground Acceleration—A Case Study of Golaghat District
in Assam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
A. F. Siddique, D. Dutta, and A. Deka
Elastic Seismic Design Response Spectra for Deep and Shallow
Basin of the Indian Subcontinent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Ketan Bajaj and P. Anbazhagan
Case Studies on Preshaking and Reliquefaction Potential
for Different Earthquakes in Japan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
Gowtham Padmanabhan and B. K. Maheshwari
Seismic Response of Basal Geogrid Reinforced Embankments
Supported on a Group of Vertical and Batter Piles . . . . . . . . . . . . . . . . . . . . 145
Radhika M. Patel, B. R. Jayalekshmi, R. Shivashankar, and N. R. Surya
Seismic Response of Low Height Embankment over Soft
Foundation Clay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
Debabrata Ghosh, Narayan Roy, and R. B. Sahu
Influence of Toe Cutting on Seismic Response of a Typical Hill
Slope in North-East India . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
Rubi Chakrabory and Arindam Dey
August, 2019 Landslide Events in Kinnaur, H.P.—An Assessment
of Earthquake and Landslide Consequences Using Satellite Data . . . . . . . 175
Madan A. Mohan, Vidya Sagar Khanduri, and Amit Srivastava
Geotechnical Investigation of Landslide of Atharamura
and Baramura Hill, Tripura . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Kousik Adak and Sujit Kumar Pal
Stability Analysis of Kattipara and Meppady Regions of Hill Soil
Slope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
P. Aswathi and K. Rangaswamy
Numerical Studies on the Behavior of Slope Reinforced with Soil
Nails . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
Akhil Pandey, Sagar Jaiswal, and Vinay Bhushan Chauhan
Seismic Stability of Non-homogenous Cohesive Soil by Using
Calculus of Variation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
Sourav Sarkar and Manash Chakraborty
Site Characterization Using Satellite Data and Estimation
of Seismic Hazard at Ground Surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
Naveen James, Sreevalsa Kolathayar, and T. G. Sitharam
Contents xi
GIS-Based Landslide Hazard Zonation and Risk Studies Using

MCDM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
Ankit Tyagi, Reet Kamal Tiwari, and Naveen James
A Comparative Study on Landslide Susceptibility Mapping Using
AHP and Frequency Ratio Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
Malemnganbi Lourembam Chanu and O. Bakimchandra
Support Vector Machine for Evaluation of Liquefaction Potential
Using SPT Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
Dev Kumar Pradhan, Suvendu Kumar Sasmal, Vamsi Alla,
and Rabi Narayan Behera
Assessment of Seismic Liquefaction of Soils Using Swarm-Assisted
Optimization Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
T. Vamsi Nagaraju, Ch. Durga Prasad, Babloo Chaudhary, and B. M. Sunil
Liquefaction Potential Evaluation: Necessity of Developing Energy
Correction Factor for SPT N-value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
S. M. Ali Jawaid
Assessment of Soil Liquefaction Safety Factors Under Different
Earthquake Magnitudes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
Shiva Shankar Choudhary, Avijit Burman, and Sanjay Kumar
Reliability-Based Assessment of Liquefaction Potential Using SPT
Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
G. S. Bhatia, K. Bhargava, and A. Mondal
Building Performance and Geotechnical Failures in 7.5M Palu
Earthquake and Tsunami 28 September 2018 . . . . . . . . . . . . . . . . . . . . . . . . 335
Sumedha Koul and Harish Mulchandani
Numerical Modeling of Three-Tiered Reinforced Soil Wall
with Different Offset Distances Subjected to Dynamic Excitation . . . . . . . 345
Sudipta Sikha Saikia and Arup Bhattacharjee
Study on Earth Pressure Distribution and Displacements
of Narrow Reinforced Earth (RE) Wall Under Static and Cyclic
Loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
Shivani R. Patel, P. J. Mehta, and M. V. Shah
Seismic Mitigation Liquefaction––An Extensive Study on New
Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
Jiji Krishnan and Shruti Shukla
About the Editors
Prof. T. G. Sitharam is currently the Director of Indian Institute of Technology

Guwahati, India. He is also a KSIIDC Chair Professor in the area of Energy and
Mechanical Sciences and Senior Professor at the Department of Civil Engineering,
Indian Institute of Science, Bengaluru (IISc). He was the founder Chairman of the
Center for Infrastructure, Sustainable Transport and Urban Planning (CiSTUP) at
IISc, and is presently the Chairman of the AICTE South Western Zonal Committee,
Regional office at Bengaluru and Vice President of the Indian Society for
Earthquake Technology (ISET). Prof Sitharam is the founder President of the
International Association for Coastal Reservoir Research (IACRR). He has been a
Visiting Professor at Yamaguchi University, Japan; University of Waterloo,
Canada; University of Dolhousie, Halifax, Canada; and ISM Dhanbad, Jharkhand,
and was a Research Scientist at the Center for Earth Sciences and Engineering,
University of Texas at Austin, Texas, USA until 1994.
Prof. Ravi Jakka is working as Associate Professor in the Department of

Earthquake Engineering, Indian Institute of Technology, Roorkee. He is also
currently serving as Secretary, Indian Society of Earthquake Technology(ISET). He
has graduated in Civil Engineering from Andhra University Engineering College in
the year 2001. He has obtained masters and doctorate degrees from IIT Delhi in the
years 2003 and 2007 respectively. His areas of interest are Dynamic Site
Characterization, Soil Liquefaction, Seismic Slope Stability of Dams, Landslides,
Foundations & Seismic Hazard Assessment. He has published over 100 articles in
reputed international journals and conferences. He has supervised over 35 Masters
Dissertations and 6 Ph.D. thesis, while he is currently guiding 10 PhD Thesis. He
has received prestigious DAAD and National Doctoral fellowships. He has
obtained University Gold Medal from Andhra University. He also received ‘Young
Geotechnical Engineer Best Paper Award’ from Indian Geotechnical Society. He
was instrumental in the development of Earthquake Early Warning System for
northern India, a prestigious national project. He is also the Organizing Secretary to
7th International Conference on Recent Advances in Geotechnical Earthquake
Engineering.
xiii
xiv About the Editors
Prof. L. Govindaraju is currently Professor in the Department of Civil

Engineering, University Visvesvaraya College of Engineering (UVCE), Bangalore
University, Bangalore. He obtained his Bachelor’s in Civil Engineering from
Mysore University, India (1986), Master’s in Geotechnical Engineering from
National Institute of Technology Karnataka (NITK), India (1994) and Ph.D. in
Civil Engineering from Indian Institute of Science (IISc), India (2005). He was
awarded fellowship from National Programme on Earthquake Engineering
Education (NPEEE) instituted by the Ministry of Human Resources &
Development, Government of India to pursue post doctoral research in the area of
earthquake engineering at the Department of Engineering Science, University of
Oxford (UK) from March 2007 to August 2007. He is the member of many
professional bodies including IGS, ISTE and ISET. He has published more than
150 technical papers and two books.
Ground Response Analysis with Deep
Bedrock Depth in Indo-Gangetic Plains
Priyanka Sharma, M. L. Sharma, and V. A. Sawant
Abstract Local site effects play an important role in causing damage to structures
during earthquakes. Thus, one of the aim of seismologists and geotechnical engineers
is to characterize the soil for the region prior to seismic hazard assessment. In the
present work, an endeavor is made to study the depth of bedrock in Indo-Gangetic
Plains from Seismotectonic Atlas of India [6]. A huge variation of bedrock depth
ranging from 0 to 4000 m indicates the presence of thick soil cover in the study region.
Roorkee city, situated in the foothills of Himalayas has a bedrock depth of around
3000 m and due to the presence of this huge soil cover, the occurrence of any great
seismic event will pose a threat to both life and property as properties of propagating
waves change as they travel toward the surface. The site characterization is carried out
by MASW and Microtremor methods and shear wave velocity profiles are estimated
with the thickness of soil. The region has lower values of shear wave velocity and
falls in Site Class D. Ground Response Analysis (GRA) of the site is performed using
STRATA [9] to Uttarkashi Earthquake (1991) and Chamoli Earthquake (1999) and
is compared with the suggested response spectrum in (IS, in (Part 1): 2002 Indian
Standard, Criteria for earthquake resistance design of structures, Fifth Revision, Part-
I, Bureau of Indian Standard, New Delhi, 1893):2002 (Part 1). The comparison of
the response spectrum from two earthquakes inhibits higher variation as provided in
the IS Code.
Keywords MASW · Microtremor · Joint fit inversion · Response spectra
P. Sharma (B) · M. L. Sharma

Department of Earthquake Engineering, IIT Roorkee, Roorkee 247667, India
e-mail: pssharma.iitr@gmail.com
V. A. Sawant
Department of Civil Engineering, IIT Roorkee, Roorkee 247667, India
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 1
T. G. Sitharam et al. (eds.), Local Site Effects and Ground Failures, Lecture Notes
in Civil Engineering 117, https://doi.org/10.1007/978-981-15-9984-2_1
2 P. Sharma et al.
1 Introduction
About 59% of India’s land mass is at a risk of moderate to rigorous seismic hazard
covering many important cities, as given in seismic zoning of the country. The surface
ground motion in comparison to bedrock motion is highly influenced by the local site
conditions. The presence of loose material on the bedrock amplifies the incoming
waves as they travel toward the surface causing damage to both life and property. The
recently experienced large earthquakes caused heavy damage at epicentral distances
ranging from 250 to 500 km or so due to local site conditions. For example, in 1985
Michoacan earthquake, major damage was observed at 200 miles away from the
epicenter in Mexico City. Similarly, in Bhuj earthquake, 2001, heavy damage was
seen in multistory building at an epicentral distance of 350 km in Ahmedabad City.
Himalayan region is one of the most seismically active regions due to the accu-
mulation of heavy strains. Many researchers [5, 2] have predicted the presence of a
seismic gap in this region. Moreover, the probability of occurrence of a great seismic
event is 0.59 in the next 100 years [8]. The Indo-Gangetic Basin (IGB) lies parallel to
the Himalayas and is highly populated. The basin is formed by the loose soil deposits
and any future earthquake in the Himalayas will lead to massive destruction and will
pose a threat to both life and property. Many studies in India to estimate the effect
of local site conditions have been done by many authors [1, 4]. The site response
study for Delhi region was done by [10]. Kirar et al. developed a correlation between
VS and N for all types of soils, clay soils and sandy soil for Roorkee region. The
insitu tests and laboratory tests were used to obtain the dynamic properties of soil in
Italy by Cavallaro et al. (2017). Ray et al. (2019) used the GR relationship and the
Gumbel’s annual extreme value method to see the effect on b-value of Sylhet region.
Thuamthansanga et al. (2019) reported the anomalous behavior of Radon isotope
pairs at Mat fault, Mizoram (India) in different depths of soil. Hence, it becomes
necessary to consider these local site conditions in order to predict the seismic hazard
before the occurrence of an earthquake and proper mitigations measures can be taken
for the infrastructures.
In the present paper, firstly, the variation of bedrock depth in Indo-Gangetic Plains
is estimated using GSI [6]. Secondly, shear wave velocity is estimated up to depth
500 m using the joint fit inversion technique from dispersion curves obtained from
multichannel analysis of surface wave (MASW) and horizontal to vertical spectral
ratio curves from microtremor measurements using [15] method for Roorkee city
having deep bedrock depth of around 3000 m. The analysis is done at different sites
of the city and the city falls in Site Class D as per NEHRP Site Classification based on
the shear wave velocity at a 30 m depth of soil. After obtaining the velocity profiles,
ground response analysis (GRA) is performed using STRATA [9] by equivalent
linear method. The response spectra from the two earthquakes, namely, Uttarkashi
Earthquake (1991) and Chamoli Earthquake (1999) is compared with that given in
[10]:2002 (Part 1).
Ground Response Analysis with Deep Bedrock Depth … 3
2 Estimation of Bedrock Depth in Indo-Gangetic Plains
The depth of bedrock is estimated from Seismotectonic Atlas of India [6] in which
contours represent the bedrock depth of a place. In AutoCAD, the study area (longi-
tude 77.4–78.4° E and latitude 28.5–30.0°N) covering an area of 18,150sq km (length
and breadth are 165 km and 110 km, respectively) was marked. Further, this rect-
angular area was divided in such a way that six profiles, namely, AG, BF, CE, AC,
HD, GE were obtained. Figure 1 represents the variation of bedrock depth of the six
profiles obtained. It can be observed, that a huge variation of bedrock depth ranging
from 0 to 4000 m indicates the presence of thick soil cover in the study region. The
length of the AG profile is 165 km. The location of this profile is from 30°00 20 N,
77°81 E to 28°50 45 N, 77°40 E. A wide variation in depth of bedrock from 100
to 3000 m is seen in this profile. From 100 m depth at Shahdara, the depth varies
to around 3000 m at Saharanpur. In Khatauli, depth varies from 1000 to 2000 m.
The length of BF profile is 165 km. The location of this profile is from 30°09 11"N,
78°20 12 E to 28°62 87 N, 77°74 92 E. A wide variation in depth of bedrock is
observed ranging from 0 to 3000 m. There is a sharp increase in the bedrock depth
Fig. 1 Variation of bedrock depth for six profiles AG, BF, CE, AC, HD and GE
4 P. Sharma et al.
from 0 to 1000 m and then a slow rate is seen from 1000 to 2000 m and again a steep
increase to 3000 m bedrock depth near Roorkee.
The length of CE profile is 165 km. The location of this profile is from 29°64 69 N,
78°44 89 E to 28°45 78 N, 78°12 07 E. A wide variation is observed ranging from
1000 to 4000 m in this profile. A sharp increase in bedrock depth is observed from
1000 to 2000 m, but a slow trend is observed from 2000 to 3000 m and again a
sharply bedrock depth increases to 4000 m. The length of AC profile is 110 km. The
location of this profile is from 30°20 72 N, 77°90 40 E to 29°99 54 N, 78°42 41 E.
It can be seen from the Figure, that bedrock depth of Dehradun city lies between
2000 and 4000 m and at Hardwar, bedrock depth is around 4000 m. The length of
HD profile is 110 km. The location of this profile is from 29°44 87"N, 77°64 39 E
to 29°29 84 N, 78°12 07 E. An increase in bedrock depth is observed from 1000
to 3000 m toward Bijnor. The length of HD profile is 110 km. The location of this
profile is from 28°44 41"N, 78°12 07 E to 29°73 12 N, 77°40 24 E. In this profile,
the contour of 0 m intersects the profiles at two places where there is an exposure of
bedrock. The variation of bedrock in this profile is from 0 to 1000 m. The depth of
bedrock from 0 m near Delhi varies to 100 m in about 10 km of distance and then,
increases gradually to 1000 m near Bulandshahar. The absence of soil cover in this
region will not cause any amplification of seismic waves.
3 Study Region: Roorkee (Uttarakhand)
In the present study, the region selected for site characterization is Roorkee attributing
to its high seismicity and having local site characteristics of soil. Roorkee is situ-
ated at latitude 29° 51 and longitude 77° 54 is a city in the state of Uttarakhand.
Himalayan Frontal Thrust lies very close to Roorkee and to the north of it lies the Main
Boundary Thrust and Main Central Thrust. Moreover, it is located around 30 km from
Shiwaliks range which can cause the city to experience strong ground motion, if, a
moderate earthquake occurs on Main Boundary Thrust or on Main Central Thrust.
Karakoram Fault is partly exposed in the study area and extends for almost 1000 km
from Central Pamir to Kumaon Himalaya [6]. The Main Central Thrust (MCT) is
among the important tectonic features of the Himalayas which terminates against the
Kishtwar fault in Jammu and Kashmir. MCT acts as a geological boundary between
the Lesser and Higher Himalaya. The geology of the area is represented by alluvial
soil and forms a part of Gangetic alluvial plains. Important tectonic features which
are Almora, Berinag, Ramgarh, Munsiary Thrust, etc., also affect the seismicity of
this region. As per the seismic zoning map of India as incorporated in Indian Stan-
dard Criteria for Earthquake Resistant Design of Structures: General Provisions and
Buildings, Roorkee lies in seismic Zone IV (IS: 1893-Part I; 2002). Indian Mete-
orological Department has reported that numerous earthquakes of medium to large
size have occurred in the study region as per the data obtained from historical and
instrumental records.
Fig. 2 Roorkee city division

into three main areas (i)
Roorkee East (ii) Solani
River and (iii) Roorkee West
Shear wave velocity is an important parameter for understanding the dynamic

behavior of soil. Moreover, it can be used for the determination of shear modulus (G)
of soil as well as for the site characterization applications of geotechnical earthquake
engineering. In this paper, 15 sites of Roorkee city, having a deep bedrock depth
of 3000 m is explored using geophysical methods. These sites were located at a
distance of about 2–3 km from each other. Figure 2 shows the division of Roorkee
city namely, as, Roorkee East, Solani River and Roorkee West. In Roorkee East, six
sites are taken (Sites 1–6), in Solani River, three sites are selected (Sites 7–9) and in
Roorkee West, six sites are chosen (Sites 10–15). In the field set up, two tests were
performed. Firstly, active MASW consisting of Soil Spy Rosina as a data acquisition
system. Secondly, HVSR of Microtremor was performed in which Tromino was used
as a data acquisition system.
4 Methodology
4.1 Multichannel Analysis of Surface Waves (MASW)
The setup used for investigating the subsurface materials is 9-channel MASW setup.
For many applications, such as, microzonation and dynamic response analysis of
sites, Multichannel Analysis of Surface waves (MASW) is increasingly being used in
earthquake geotechnical engineering. It is generally used for the estimation of shear
wave velocity and dynamic properties. Moreover, subsurface material boundaries
identification and spatial variations of shear wave velocity can also be measured.
The basic steps are as follows:
• In this technique, the recorded Rayleigh waves are transformed into the frequency
domain from the time domain using the Fast Fourier Transform (FFT) techniques.
6 P. Sharma et al.
• The corresponding phase difference (f ) between each receiver pair of channels
is computed for every component of frequency measured.
• The travel time (t) between receivers is computed using the following relation

t( f ) = ( f ) 2π f (1)
• The distance between each pair of receivers d = d 2 –d 1 is known which can be

used in estimation of the Rayleigh wave phase velocity [VR(f )] as follows:

VR ( f ) = d t( f ) (2)
• After the estimation of Rayleigh velocity, the wavelengths (λR ) are determined as

λ R ( f ) = VR ( f ) f (3)
• Finally, a dispersion curve can be plotted between V R -λ or VR -depth.
4.2 Micro Tremor Method
Nakamura [15] proposed the use of microtremors for microzonation and in the
study of side effects, and, is therefore also known as Nakamura technique. In the
Horizontal-to-Vertical Spectral ratio (HVSR) technique, a sharp peak of the H/V
spectrum at a particular site corresponds to the fundamental frequency of the sedi-
ments. In the present scenario, it is widely accepted that this frequency of the HVSR
peak reflects the fundamental frequency of the underlying sediments and the ampli-
tude is mainly dependent on its impedance contrast with the bedrock. The major
advantages of HVSR method included its low-cost and simple measurement tech-
nique and providing the direct estimates of the resonance frequency of sediments.
The shear-wave velocity structure cannot be obtained directly from the Microtremor
HVSR method, but can be derived by synthetic modeling of spectral ratio curve.
It can be inferred from the various researches carried out in the field of H/V spec-
trum that by applying the inversion technique of H/V spectrum, shear wave velocity
structure can be obtained by estimating the thickness of every soil layer. H/V can be
efficiently used to estimate the shear wave velocities upto bedrock if there is some
methodology adopted to constraint the shear wave velocity of shallow layers [3].
Parolai et al. [17] used the surface wave technique for constraining the estimation of
shear wave velocity in shallow layers. Hence, in the present study, the same method-
ology has been adopted, where for constraining the shear wave velocity profile in
shallow layers MASW technique is used and for deeper depths HVSR technique is
used for the estimation of shear wave velocity.
4.3 Experimental Setup
In the field setup, in order to characterize the sites of the present study, two tests were
performed. Firstly, active MASW consisting of Soil Spy Rosina as a data acquisition
system. Secondly, HVSR of Microtremor was performed in which Tromino was used
as a data acquisition system. The basic components and the methodology are briefly
explained as follows:
• Soil Spy Rosina [12] is the hardware and software platform of multichannel digital
system for carrying out active and passive seismic surveys.
• The setup used in the present study consisted of nine sensors with 2 m inter-
geophone spacing. For setting out a trigger, first geophone from the source was
used.
• Tromino user’s manual [13] has three channels connected to three orthogonal elec-
trodynamic velocity meters with selectable gain for seismic tremor acquisition.
For HVSR analysis, this instrument has been used to record ambient vibrations
for 20 min at each location.
The setup used for investigating the subsurface materials for upper layers was
9channel MASW setup. It is generally used for the estimation of shear wave velocity
and dynamic properties for shallow depths (upto 30 m). In order to characterize deeper
soil, the use of microtremors with passive source was explored. In this method, a
sharp peak of the H/V spectrum at a particular site corresponds to the fundamental
frequency of the sediments is obtained. The major advantages of HVSR method
included its low-cost and simple measurement technique and providing the direct
estimates of the resonance frequency of sediments. By applying the inversion tech-
nique of H/V spectrum, shear wave velocity structure can be obtained by estimating
the thickness of every soil layer. For more deeper depths, the available bore hole
data has been used. The information was obtained from the construction division
(IIT Roorkee) for bore holes that were drilled for installation of tube wells in IITR
campus. In Roorkee East, the bore hole data is available upto depth of 150 m. It
represents that, poorly graded sands, sands with gravels with fines or without fines
are overlaid on a layer of gravelly clays, inorganic clays, silty clays of low plasticity
of 10 to 15 m thickness. In Roorkee West and Solani River, the available standard
penetration tests (SPT) data shows inorganic silts consisting of very fine sand rock
flour, clayey silts with zero to low plasticity at the surface laid on silty sands. The SPT
N values have been used for estimating the shear wave velocity profiles by using the
available empirical relations. The empirical relationships considered in the present
study for different types of soils are by Shibata [18], Ohba and Toriumi [16] and Lee
[11].
8 P. Sharma et al.
5 Results and Discussions
Figure 3a, b and c shows the graphical representation of shear wave velocity profile of
all the sites where vertical axis represents the estimated depth (m) and the horizontal
axis represents the shear wave velocity VS (m/s). Figure 3a shows the estimated shear
wave velocity profiles for Roorkee East. As expected and seen from the Figure, the
shear wave velocity increases as depth increases. The shear wave velocity (Vs30 ) of
all the sites in this region is about 250–300 m/s Fig. 3b shows the estimated shear
wave velocity profiles for sites near Solani River. The shear wave velocity (Vs30 ) of
all the sites in this region is about 200–250 m/s. Figure 3c shows the estimated shear
wave velocity profiles for Roorkee West. The shear wave velocity (Vs30 ) of all the
sites in this region is about 240–300 m.
6 Ground Response Analysis
In order to estimate the response spectrum at a particular site, for any input time
history at the base, it is essential to conduct the ground response analysis. The prop-
erties of the soil at a given site and the acceleration time history are required for
the analysis. The shear wave velocity profiles which are obtained from the joint fit
inversion technique of the Grilla Software have been used to carry out the ground
response analysis. STRATA Software is used for the ground response analysis of
each velocity profile. In this software, one can generate multiple profiles with vari-
ations in a single profile by incorporating the uncertainties. The input parameters
are the thickness of the layer, unit weight, shear wave velocity of the layer and the
bedrock depth.
All the site studies lie in the seismic zone IV having zone factor of 0.24 (Bureau
of Indian Standards [7]: 2002 Part-1). For the generation of response spectra, two
earthquakes (Uttarkashi Earthquake 1991 and Chamoli Earthquake 1999) have been
chosen and were applied at bedrock of the Roorkee East division. A brief description
about the earthquakes is given in Table 1. The acceleration time histories of these
earthquakes have been applied at the bedrock as an input motion in the STRATA
software for the ground response analysis. For the estimation of dynamic soil prop-
erties such as G/Gmax and the damping ratio values, the inbuilt curves in the software
have been used. The output obtained is thus compared with the response spectrum
given in [7]:2002 (Part-1).
Figure 4 shows the comparison of the response spectrum obtained as output from
the applied input ground motion at the bedrock of the study region with that provided
in the [7]:2002 (Part 1). It can be clearly seen that the response spectrum obtained after
applying the strong ground motion inhibits higher variation as seen in that provided
in [7]:2003 (Part 1). Large amplification is observed for the Chamoli earthquake at
period around 0.4 s in reference to that provided in the [7]:2002 (Part 1). In the case
(a) Shear Wave Velocity (Vs) in m/s

0 200 400 600 800 1000 1200
0
Site 1
100 Site 2
Site 3
200
Depth (m)
Site 4
300 Site 5
Site 6
400
500
(b) Shear Wave Velocity (Vs) in m/s

0 200 400 600 800 1000 1200
0
100 Site 7
Site 8
Depth (m)
200
Site 9
300
400
500
(c) Shear Wave Velocity (Vs) in m/s

0 200 400 600 800 1000
0
Site 10
Site 11
100
Site 12
Site 13
200 Site 14
Depth (m)
Site 15
300
400
500
Fig. 3 a, b and c Graphical representation of shear wave velocity profiles where vertical axis
represents the estimated depth (m) and the horizontal axis represents the shear wave velocity VS
(m/s) of all the six sites
10 P. Sharma et al.
Table 1 Description about Input Motions

Name Date & time Latitude (N) Longitude (E) Magnitude PGA (g)
Uttarkashi Earthquake 20.10.1991 30.9 78.1 6.8 0.309
02:53
Chamoli Earthquake 29.03.1999 30.5 79.4 6.5 0.359
00:35
Type II(Medium Soil) U arkashi Earthquake Chamoli Earthquake

4
Spectral Accelera on Coefficient
3.5
3
2.5
(Sa/g)
2
1.5
1
0.5
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Period (s)
Fig. 4 Response Spectra comparison for given earthquakes by ground response analysis in STRATA
Software with response spectrum given in [7]:2002 (Part 1)
of Uttarkashi earthquake, the amplification is observed around 0.3 s in reference to

that provided in the [7]:2002 (Part 1).
7 Conclusion
Seismic site characterization of Roorkee (Uttarakhand) was carried out by using

the geophysical methods and the available bore log data for deeper soil characteris-
tics. The results indicate the increase in shear wave velocity with depth, which was
expected. The range of VS30 observed lies between 224 and 357 m/s. The deepest
layer observed was at a depth of 801 m from the top with half space shear wave
velocity of 1000 m/s. From the analysis of the data, it was found that all the six sites
fall under class D as given by NEHRP site class classification. However, according to
Mittal et al. [13] all of the studied sites fall under class C which corresponds to sites
having the shear wave velocity less than 375 m/s. The comparison of the response
spectrum obtained from STRATA for Uttarkashi Earthquake (1991) and Chamoli
Earthquake (1999) applied at the bedrock inhibits higher variation as seen in that
provided in [7]:2003 (Part 1).
Large amplification is observed for the Chamoli earthquake at a period around

0.4 s in reference to that provided in the [7]:2002 (Part 1). In the case of Uttarkashi
earthquake, the amplification is observed around 0.3 s in reference to that provided
in the [7]:2002 (Part 1). Thus, it requires the engineer’s or scientist’s attention for
considering the IS Code modification or site specific studies.
The above conclusions are based on the limited test data and to arrive at quantita-
tive outcome, more number of tests needs to be conducted in the city. Nonetheless,
the research presented has very much practical significance and will provide data for
carrying out microzonation or hazard studies in future.
References
1. Anbazhagan P, Sitharam TG (2008) Spatial variability of the weathered and engineering

bedrock using multichannel analysis of surface wave survey. Pure Appl Geophys 166(3):409–
428
2. Bilham R, Blume F, Bendick R, Gaur VK (1998) Geodetic constraints on the translation and
deformation of India: implications for future great Himalayan earthquakes. Curr Sci 74(3):213–
229
3. Castellaro S, Mulargia F (2009) VS30 estimates using constrained H/V measurements. Bull
Seismol Soc Am 99:761–773
4. Desai SS, Choudhury D (2015) Site-specific seismic ground response study for nuclear power
plants and ports in Mumbai. Nat Hazards Rev 16(4)
5. Gahalaut VK, Chander R (1997) On interseismic elevation changes and strain accumulation
for great thrust earthquakes in Nepal Himalaya. Geophys Res Lett 24:1011–1014
6. GSI (2000) Seismological atlas of India and its environs. Geological survey of India
7. IS 1893 (Part 1) (2002) Indian standard, criteria for earthquake resistance design of structures,
Fifth Revision, Part-I. Bureau of Indian Standard, New Delhi
8. Khattri KN (1999) An evaluation of earthquakes hazard and risk in northern India. Himalayan
Geol 20:1–46
9. Kottke AR, Wang X, Rathje EM (2013) Technical manual of STRATA. Department of Civil,
Architectural and Environmental Engineering, University of Texas, Geotechnical Engineering
Center
10. Kumar A, Baro, O, Harinarayan NH (2016) Obtaining the surface PGA from site response
analyses based on globally recorded ground motions and matching with the codal values. Nat
Hazards 81(1):543–572
11. Lee SH (1990) Regression models of shear wave velocities. J Chin Inst Eng 13:519–532
12. Micromed S. p. A. (2008a) Soil spy rosina user’s manual Ver. 2.0, Micromed S.p.A., Treviso,
Italy
13. Micromed S. p. A. (2012) Tromino user’s manual, Micromed S.p.A., Treviso, Italy
14. Mittal H, Kumar A, Ramhmachhuani R (2012) Indian national strong motion instrumentation
network and site characterization of its stations. Int J Geosci 3(06):1151
15. Nakamura Y (1989) A method for dynamic characteristics estimation of subsurface using
microtremors on the ground surface. Quart Rep Railway Tech Res Inst 30(1):25–33
16. Ohba S, Toriumi I (1970) Dynamic response characteristics of Osaka plain. In: Proceedings of
the annual meeting AIJ (in Japanese)
12 P. Sharma et al.
17. Parolai S, Picozzi M, Richwalski SM, Milkereit C (2005) Joint inversion of phase velocity
dispersion and H/V ratio curves from seismic noise recordings using a genetic algorithm,
considering higher modes. Geophys Res Lett 32(1). 1 Jan 2005
18. Shibata T (1970) Analysis of liquefaction of saturated sand during cyclic loading. Disaster
Prev Res Inst Bull 13:563–570
Soil Amplification Study for Kalyani
Region, Kolkata
Himanshu Shukla, Pradeep Muley, and Sajjan Kumar
Abstract The main objectives of the present paper are: (i) to identify the areas in
Kalyani region, Kolkata, i.e. AIIMS Kolkata campus in which soil formations are
prone to amplifying ground motions and (ii) to analytically evaluate amplification
ratio using computer program for earthquake site response analysis. Peak ground
acceleration at different depth is calculated for all the selected six sites with the help
of soil profile. For this purpose, one-dimensional linear ground response analysis is
done by the help of DEEPSOIL V-7.0 computer program. It can be observed that
from the results peak ground acceleration of different site is varying from 0.159 to
0.190 g and the soil amplification value is ranging from 0.996 to 1.190 for peak
ground acceleration of 0.16 g.
Keywords Amplification factors · Peak ground accelerations · Ground motion

amplification
1 Introduction
Site response analysis is considered an important and primary phase of any seismic
soil structure study and is also required for the microzonation studies of a region.
Ground motion characteristics are greatly influenced by geological formation of
soil. For the design of any structures ground response analysis and seismic analysis
are compulsory to know the various factor of safety. Effects of earthquake can be
estimated by predicting the site response and liquefaction potential as the local ground
are affected by seismic waves when earthquake occurs [1].
H. Shukla · P. Muley (B) · S. Kumar

Department of Civil Engineering, Madan Mohan Malaviya University of Technology Gorakhpur,
Gorakhpur 273010, U.P., India
e-mail: pmce@mmmut.ac.in
14 H. Shukla et al.
Soil formation of any site highly affects the peak ground acceleration (PGA)
value so their amplification factor and de-amplification factor must be determined.
Soil amplification of ground motion is generally calculated by the specifying site
dependent response spectra or by specifying amplification co-efficient for various
soil formations.
According to various researchers [2–4] soil classification should give more proper
information of soil formation through the use of several site response parameters,
e.g., soil grain size, standard penetration test blow-count (N-SPT), shear resistance
modulus, material damping, seismic impedance contrast between the bedrock and
the overlying soil deposits.
The most used various equivalent linear computer programs are DEEPSOIL devel-
oped by [5], EERA [6], SHAKE [7], SHAKE-91 [8], SHAKE-2000 [9], ProShake
[10] and CyberQuake [11]. All these programs use for computational programming.
Previously number of studies have been completed by several researchers on soil
amplification for the various region of India like for Roorkee region it was carried
out by [12]. Similar study on soil amplification for Bangalore region and Ahmedabad
region was conducted by [13, 14] respectively.
To estimate the realistic earthquake ground motion and seismic hazard of region,
evaluation of site amplification functions is mandatory. Technique used for estima-
tion of site response requires a reference site. For ideal condition, a complete ground
response analysis will decide the rapture phenomenon at the earthquake source,
beneath a particular site, seismic wave traveling through the earth toward the top of
bedrock and the influence soil formation above the bedrock on the ground surface
motion measured using seismic wave [1]. But in actual case the phenomenon of
fault rupture is very complicated, and the uncertainty is found in energy transmis-
sion between source and site. This may be the reason that this method is not used in
engineering field. Various empirical relations are developed based on realistic earth-
quake which was recorded. These developed relations are used to predict the motion
of bedrock at the site for seismic hazard analysis. So, in-ground response analysis
problem motion of the bedrock is determined instead of response of soil formation.
Seismic waves can penetrate 10 km through rock and are able to go through soil up
to depth of 100 m.
The main objective of this study is used to evaluate the soil amplification factor
for Kalyani region, Kolkata, India. This factor is used to identify the soil forma-
tions which are prone to amplification. This aim is achieved by performing analyt-
ical evaluation of amplification ratio using DEEPSOIL V-7.0 computer program for
earthquake site response analysis.
To determine the response of motion of the bedrock closely below the soil site
specific ground response investigation is carried out. This also shows the effect on
amplification of seismic waves of local soil condition [1]. These results may help in
further study and microzonation of the study area.
Soil Amplification Study for Kalyani Region, Kolkata 15
Table 1 Location of sites in AIIMS Campus in Kalyani region

S. No Name of Bore log Latitude Longitude Water table No of SPT
sites depth (m) sample
(m)
1 BH-01 18.00 N22º5812.62 E88º3138.44 2.8 11
2 BH-02 18.00 N22º5812.62 E88º3138.44 3.1 11
3 BH-03 18.00 N22º5812.62 E88º3138.44 1.6 11
4 BH-04 18.00 N22º5812.62 E88º3138.44 3.3 13
5 BH-05 18.00 N22º5812.62 E88º3138.44 1.4 11
6 BH-06 18.00 N22º5812.62 E88º3138.44 1.8 12
2 Study Area
In this paper soil amplification ratio is evaluated for Kalyani region, Kolkata India.
AIIMS Kalyani, Kolkata Campus was selected for soil amplification study purpose.
The present study will predominantly be concerned with identifying the areas in
Kalyani region, Kolkata in which soil formations are susceptible to amplifying
ground motions. AIIMS Kalyani, Kolkata is a medical institute which is located
at Kalyani city of Nadia district located around 50 km from the capital city of West
Bengal, India. Kalyani region lies in seismic zone-III [15] which is susceptible to
earthquake. Six locations in AIIMS Kalyani campus were selected for this study.
Locations of sites, i.e., latitudes and longitudes, borehole depth, location of water
table and number of samples collected through SPT are mentioned in Table 1.
3 Geological Investigation
In this paper the amplification factors of AIIMS Kalyani campus of West Bengal
are evaluated. In-situ tests are performed to collect the geotechnical data and soil
samples of the study area which are required for the evaluation of soil amplifica-
tion. Geotechnical investigation was performed by Centre for Advanced Engineering
Kolkata [16]. For geotechnical investigation purpose Standard Penetration Test (SPT)
was performed as per guidelines of Indian standard [17]. Liquefaction potential for
the same locations has been evaluated by [18] and details of soil profile and SPT N
value with the depth are mentioned in [18].
Based on geotechnical investigation, soil profile may be divided into three layers.
First layer up to 4.5 m, second layer started around 5.0–13 m and further third layer
is exploring up to depth of 20 m. First layer consists of clayey silt which is soft
to medium brownish grey and presence of sand and kankar in small amount was
observed. Values of SPT N values lie between 4 and 19 in this layer. Similarly soil
of second layer carries loose to medium grey silty fine sand and due to presence of
sand, ranges of SPT N values varying from 11 to 41. Soil deposits of the third layer
16 H. Shukla et al.
Table 2 Summary of key point of all the six sites in AIIMS Kolkata campus
S. No Name of site Depth N Vs FC Ud Specific Gravity (G)
(m) Value (m/Sec) (%) (kN/m3 )
1 BH-01 18.00 5–44 145–211 10–88 1.45 2.64–2.67
2 BH-02 18.00 7–47 157–204 76–88 1.42 2.63–2.65
3 BH-03 18.00 4–29 138–190 2–86 1.45 2.64–2.68
4 BH-04 18.00 4–55 136–218 79–89 1.45 2.64–2.65
5 BH-05 18.00 10–39 170–205 3–90 1.44 2.63–2.66
6 BH-06 18.00 4–45 138–203 7–87 1.42 2.64–2.66
consists of dense to very dense grey fines and a trace of silt was found in medium
sand. Exploration of last layer was done up to depth of 20 m. SPT N value is high
for this layer and is varying from 30 to 55.
Geotechnical investigation of all the six sites is surmised in Table 2. N value is
varying from 4 to 55 and the N values are higher for BH-04. As the depth increases
value of shear wave velocity also increases due to increase in N value [18]. The
values of shear wave velocity (Vs) are higher for BH-04 i.e.136–218 m/s, it can be
attributed due to the fact of high SPT N values. For the calculation of the corrected
shear wave velocity equations given by [19] used in this analysis (Fig. 1).
Properties of soil, i.e. unit weight, specific gravity, presence of fines content for
Kalyani region, Kolkata are mentioned in Table 2. Variation in fines content for
BH-04 is 79–89%.
The Shear wave velocity profile for AIIMS Campus is shown in Fig. 2. It can be
observed from Fig. 2 that the maximum and minimum values of Vs are obtained at
higher (15.0 m) and at shallow depth at BH-04 respectively. Also, it can be observed
from Fig. 2 that the value of Vs increases with the depth for all the six sites. This
variation may be due to the increasing values of SPT N with the depth [18].
4 Ground Response Analysis
Ground response analysis is used to forecast site natural period, to access ground
motion amplification, in the development and design of response spectra. It can
also be used in evaluation of liquefaction potential and determination of earthquake
induced forces which can lead to instability of earth and earth slope [1]. For one-
dimensional ground response analysis three methods are used, i.e. linear method,
equivalent linear method and nonlinear method.
Linear approach to one dimensional ground response analysis of layered soil
has been coded into widely used computer program named as DEEPSOIL [5]. This
approach is being used in the present study for the development and analysis of the
peak ground acceleration for six sites in Kalyani region, Kolkata. Linear frequency
BH-01 BH-02 BH-03

BH-04 BH-05 BH-06
Shear Wave Velocity (m/sec)
100 125 150 175 200 225 250
0
2
4
6
8
Depth (m)
10
12
14
16
18
20
Fig. 1 Shear Wave Velocity with the depth for all the six sites in AIIMS Kalyani, Kolkata
(a) (b)
Fig. 2 Layered soil profile a Soil profile model b Layered soil profile of BH-01
18 H. Shukla et al.
Fig. 3 Acceleration time 0.2

history (sinusoidal wave for
10 s and 0.16 g) 0.1
Acceleration (g)
0
-0.1
-0.2
0 1 2 3 4 5 6 7 8 9 10
Time (Sec)
domain analysis for damped elastic rock is being used in this study, stiffness of layer
in shear wave velocity (Vs) or shear modulus (G) is also used.
DEEPSOIL computer program includes three major components—soil profile,
input motion and result [5]. Soil profile model is shown in Fig. 2a. Layered soil
profile of BH-01 is shown in Fig. 2b which describes the different layers of the soil
according to their depth, density and shear wave velocity. It can be observed from
Fig. 2b that the thickness of each layer is uniform (1.5 m) except third layer which
is of 3.0 m for BH-01. Shear wave velocity plays an important role in the analysis.
Assessment of potential ground motion of each site is essential after the charac-
terization of site. Suitable rock motion is selected to symbolize design rock motion
for the study area. Kalyani region, Kolkata lies in seismic zone-III. For this seismic
zone, PGA = 0.16 g [15] is considered in the present study. In the present analysis
input motion is applied in the bedrock for 0.16 g acceleration for 10 s. The accelera-
tion time history used in this analysis is shown in Fig. 3. Geotechnical investigation
is done up to the depth of 20 m and further bedrock is considered in this analysis.
5 Results and Analysis
Study of ground response analysis is important which may help in controlling the
damage situation during an earthquake or cyclic loading. The destruction of earth-
quake at any location does not only depend on its magnitude and distance from the
epicenter but also on the sub-soil properties of the location [1].
In this study one-dimensional ground response analysis is done by DEEPSOIL
computer program. This program can perform one dimensional nonlinear time
domain analysis with and without pore water pressure generation [5]. This soft-
ware was developed by [5]. Ground response analysis with DEEPSOIL includes
three steps—profile tab, motion tab and analysis tab [5].
In the present study soil amplification for AIIMS Kalyani Campus is evaluated
for PGA = 0.16 g. For this purpose, soil amplification factor and peak ground accel-
eration with the depth for all the six sites are calculated using DEEPSOIL computer
program. It can be observed from Table 3a and b that the peak ground acceleration
value and the soil amplification factor are maximum for all the sites at the shallow
depth and are decreasing with depth.
The maximum value of soil amplification factor is 1.190 observed at top surface in
case of BH-03 (Table 3a). Ranges of Peak ground acceleration for all the six sites vary
from 0.190 to 0.159 g. From Table 3a it is observed that range of soil amplification
factor is 1.086–1.002, 1.145–1006 and 1.190–1.012 for BH-01, BH-02 and BH-
03 respectively. Similarly, from Table 3b it can be concluded that the amplification
factor varies in the range of 1.053 to 0.996, 1.053–1.002 and 1.080–1.000 for BH-04,
BH-05 and BH-06 respectively.
6 Summary and Conclusion
Study of ground response analysis is necessary which may help in controlling the
damage situation during an earthquake. Amplification factor shows the measure of
increase in certain frequency of ground motion of soft soil deposit that causes incre-
ment of earthquake damage. The one-dimensional linear ground response analysis of
different location of the AIIMS Kalyani campus, Kolkata is evaluated using DEEP-
SOIL V-7.0 computer program. Peak ground acceleration with depth is evaluated
for all the six sites for Harmonic Motion of PGA = 0.16 g for 10 s duration. It can
be observed that from the results, peak ground acceleration for all the six sites is
varying from 0.159 to 0.190 g and the soil amplification value is ranging from 0.996
to 1.190 respectively. This analysis may be further refined analysis using different
earthquake motions and also for Non-Linear site response analysis.
Table 3 a PGA and Soil

Site Depth (m) Acceleration (g) Soil
amplification Factor for three
Identification amplification
sites (BH-01 to BH-03).
factor
b PGA and Soil amplification
Factor for other three sites BH-01 0.0 0.174 1.086
(BH-04 to BH-06) 1.5 0.173 1.083
3.0 0.171 1.067
6.0 0.168 1.05
7.5 0.167 1.043
9.0 0.166 1.037
10.5 0.165 1.031
12.0 0.164 1.025
13.5 0.163 1.020
15.0 0.162 1.014
16.5 0.161 1.008
18.0 0.160 1.002
(continued)
20 H. Shukla et al.
Table 3 (continued)
factor
BH-02 0.0 0.183 1.145
3.5 0.179 1.116
5.0 0.176 1.100
6.5 0.174 1.086
8.0 0.172 1.073
9.5 0.170 1.063
11.0 0.168 1.052
12.5 0.167 1.041
14.0 0.165 1.032
15.5 0.164 1.023
17.0 0.162 1.014
18.0 0.161 1.006
BH-03 0.0 0.190 1.19
3.0 0.185 1.155
4.5 0.182 1.138
6.0 0.179 1.122
7.5 0.177 1.106
9.0 0.175 1.093
10.5 0.173 1.080
12.0 0.171 1.066
13.5 0.168 1.052
15.0 0.166 1.039
16.5 0.164 1.025
18.0 0.162 1.012
BH-04 0.0 0.169 1.053
1.5 0.168 1.051
3.0 0.167 1.044
4.5 0.166 1.040
6.0 0.165 1.034
7.5 0.164 1.027
9.0 0.163 1.020
10.5 0.162 1.015
12.0 0.162 1.010
13.5 0.161 1.007
(continued)
Table 3 (continued)
factor
15.0 0.161 1.004
16.5 0.160 1.000
18.0 0.159 0.996
BH-05 0.0 0.169 1.053
3.0 0.168 1.048
4.5 0.167 1.045
6.0 0.167 1.041
7.5 0.166 1.036
9.0 0.165 1.031
10.5 0.164 1.026
12.0 0.163 1.022
13.5 0.163 1.017
15.0 0.162 1.012
16.5 0.161 1.008
18.0 0.160 1.002
BH-06 0.0 0.173 1.080
3.0 0.168 1.052
4.5 0.167 1.046
6.0 0.167 1.042
7.5 0.166 1.038
9.0 0.165 1.033
10.5 0.165 1.028
12.0 0.164 1.023
13.5 0.163 1.017
15.0 0.162 1.012
16.5 0.161 1.006
18.0 0.160 1.000
References
1. Kramer S (1996) Geotechnical earthquake engineering. Pearson Education Pvt. Ltd., Singapore
2. Medvedev J (1962) Engineering seismology. Academia Nauk Press, Moscow
3. Zhao J, Irikura K, Zhang J, Fukushhima Y, Somerville P, Asono A, Takashi T, Ogawa H (2006)
An empirical site classification method for strong motion stations in Japan using H/V response
spectral ratio. Bull Seism Soc Am 96, 914–925
4. Pitilakis K, Gazepis C, Anastasiadis A (2006) Design response spectra and soil classification
for seismic code provisions. In: Proceedings, ETC-12 workshop
5. Hashash YMA, Musgrove MI, Harmon JA, Ilhan O, Groholski DR, Phillips CA, Park D (2017)
DEEPSOIL 7.0, nonlinear and equivalent linear seismic site response of one-dimensional soil
columns user manual
22 H. Shukla et al.
6. Bardet JP, Ichii K, Lin CH (2000) EERA: a computer program for equivalent-linear earthquake
site response analyses of layered soil deposits. University of Southern California, Los Angeles
7. Schnabel PB, Lysmer J, Seed HB (1972) SHAKE: a computer program for earthquake response
analysis of horizontally layered sites. Report n° EERC72-12, University of California at
Berkeley
8. Idriss IM, Sun JI (1992) User’s Manual for SHAKE91, A computer program for conducting
equivalent linear seismic response analysis of horizontally layered soil deposits, Center for
geotechnical modeling, Department of civil and environmental engineering, University of
California, Davis
9. Ordonez GA (2002) SHAKE2000: a computer program for 1-D analysis of geotechnical earth-
quake engineering problems, User’s manual, Geomotions, LLC, Lacey, Wash, 264. https://
www.geomotions.com
10. EduPro civil system Inc. (2003) Proshake, version 1.12. Sammamish, Washington. https://pro
shake.com
11. Modaressi H, Forerster E (2000) CyberQuake. User’s Manual, BRGM, France
12. Kirar B, Maheshwari B (2015) Soil amplification studies for Roorkee region. In: Interna-
tional geotechnical symposium on disaster mitigation in special geoenvironmental conditions
Chennai, India, vol 6, 21–23
13. Anbazhagan P, Sitharam T, Divya C (2006) Soil amplification and liquefaction studies for
Bangalore city, Indian geotechnical conference 2006. Channai, India, pp 14–16
14. Trivedi S (2011) Soil amplification study for Ahmadabad region. In: Institute of technology,
International conference on current trends in technology, vol 2011, 382–481
15. IS: 1893:2016 (Part-I), Criteria for earthquake resistance design for structure: general
provisions and building bureau of Indian standards (BIS), New Delhi
16. Centre for Advanced Engineering (2015) Report on geotechnical investigation for setting up
of AIIMS at Kalyani West Bengal. 1–77. Centre for Advanced Engineering, Kolkata
17. IS 2131 (1981) Method for standard penetration test for soils. Bureau of Indian Standards (BIS)
New Delhi
18. Kumar S, Srivastava T, Muley P (2019) Assessment of SPT-based liquefaction potential of
kalyani region, Kolkata. In: Proceedings of the Indian geotechnical conference, SVNIT Surat
19. Andrus RD (2000) Stokoe, KH-II: liquefaction resistance of soils from shear-wave velocity. J
Geotech Geoenviron Engg 126(11):1015–1025
Simulated Annealing Algorithm
for Subsurface Shear Wave Velocity
Investigation Using Ground Vibration
Data
Anushka Joshi, Sateesh K. Peddoju, and Mohit Pandey
Abstract Subsurface shear wave velocity plays an important role in designing

earthquake-resistant structures. Average shear wave velocity up to 30 m depth, known
as Vs30 , is used as a common design parameter. Ambient noise data that is generated
by ground vibrations due to the passing of vehicles or other passive sources carry
an important information about the subsurface shear wave velocity structure of the
region. The seismic instrument records these disturbances along with three different
directions i.e. two horizontal and one vertical. The ratio of the spectral amplitude of
the horizontal to the vertical component is an important factor that is dependent on
subsurface velocity structure of the investigating area. Ground vibration data from
ambient noise recording system has been recorded at Bayasi site in the Garhwal
Himalaya. A computer code DISHV in FORTRAN has been developed in this work
for obtaining HV curve having a limited data from a large sampled HV curve having
similar spectral property. The obtained output from DISH has been effectively used
as an input to the HV-Inv software developed by Garcia et al. (Comput Geosci 97:67–
78, 2016). The HV curve obtained from recorded data is modelled with a simulated
annealing algorithm. The obtained velocity model from Simulated Annealing algo-
rithm is further refined using forward modelling approach to match obtained and
theoretical HV curve.
Keywords HVSR · Simulated annealing · Shear wave velocity
A. Joshi
Department of Computer Science and Engineering, Govind Ballabh Pant Institute of Engineering
and Technology, Pauri, Uttarakhand, India
e-mail: anushkazinc@gmail.com
S. K. Peddoju
Department of Computer Science and Engineering, Indian Institute of Technology Roorkee,
Roorkee, Uttarakhand, India
e-mail: drpskfec@iitr.ac.in
M. Pandey (B)
Department of Earth Sciences, Indian Institute of Technology Roorkee, Roorkee, Uttarakhand,
India
e-mail: pmohit42@gmail.com
24 A. Joshi et al.
1 Introduction
Shear wave velocity is a valuable indicator of the dynamic behaviour of the soil
due to its relationship with the shear modulus (i.e. μ = ρ · Vs 2 ), which depends
on strain level. Therefore, it is the important parameter in construction purposes and
engineering studies. Shear waves are the type of elastic waves, where particle motion
is perpendicular to the direction of wave propagation. These waves are also called
S-wave or Secondary wave because of its appearance on the seismogram immediate
after the first arrival P-wave. Shear waves travel slowly in the rock as compared to the
P-wave. The P-wave velocity is 1.37 times the S-wave velocity. Shear wave velocity
is used to classify the soil types based on Vs30 (S-wave velocity over top 30 m of
depth) by characterizing the effect of sediment stiffness on the ground motion. There
are many soil classifications based on Vs30 like NEHRP, EUROCODE8, Uniform
Building Code (UBC), etc. which is used for earthquake engineering design with
the expectation that sites in the same class will respond similarly to a given earth-
quake. The shear wave velocity can be estimated by many geophysical methods.
Out of which, Horizontal to Vertical Spectral Ratio (HVSR) technique is one of the
geophysical method to estimate the shear wave velocity and for the study of ground
motion amplification. This technique is non-destructive and cheapest. However, there
is no evidence of the particular composition of the noise field at HV curve. The spec-
tral ratio curve has been assumed to be dominated by the body waves around the
peak of the H/V [6, 10, 15]. The HVSR ratio curve from ambient noise on other
hand, has been related to the Rayleigh wave ellipticity by Lermo and Chávez-García
[11], Malischewsky and Scherbaum [13]. When Rayleigh and Love waves come
from various directions an ellipticity analysis becomes very complicated. However,
successful inversion schemes based on surface wave ellipticity have been used [2] .
The description of HV curve by taking into account the combined effect of complete
seismic ambient-vibration wave field [21]. This is called diffuse field assumption.
The purpose of this paper is to get the best shear wave velocity model by modelling
Horizontal to Vertical Spectral Ratio curve by using Simulated Annealing Method
based on software developed by García-Jerez et al. [9]. The input to this software
require HV spectral ratio data in very limited samples. In the present work, a program
has been developed to model the spectral ratio curve from ambient noise data and
obtain a reliable limited sampled spectral ratio data.
2 Methodology
It has been verified by many workers that earthquake ground motion is strongly
effected by local site conditions [1, 19]. Site effects is characterized by the means of
spectral ratios of recorded motions with respect to reference rock site (e.g. Cadet et al.
[7]). On the other hand, for more quiescent areas seismic noise severe as a major tool
Simulated Annealing Algorithm for Subsurface Shear Wave … 25
to provide useful information about spectral ratio contents of ground motion. Naka-
mura [14] was among initial workers who have related spectral ratio of horizontal
to vertical component of recorded ground motion with the site effects. Lermo and
Chávez-García [11] has obtained reliable estimates of the soil resonance frequen-
cies from the spectral ratio of horizontal with vertical motions of microtremors. The
HV curve has been studied to explain its strengths and limitations by different workers
viz. [4, 6, 12, 17]. The HV ratio curve reveals the site dominant frequency f 0 but the
amplitude of this HV curve is not well understood [17]. The recent work of Lunedei
and Albarello [12] is a significant step towards the understanding of the amplitude
as well as frequency part of HV spectrum.
In the present work, noise is assumed to be diffuse wavefield containing all types
of elastic waves. Underlying this assumption is the connection between diffuse fields
and the Green’s function [8, 20, 23–25], which implies the proportionality between
the average energy densities of a diffuse field and the imaginary part of Green’s
Function at the source [16, 20]. This can be summarised as:
u i (x A , ω)u ∗j (x B , ω) ∝ Im[G i j (x A , x B ; ω)] (1)
where G i j (x A , x B ; ω) is the Green’s function defined as the displacement in the ith

direction at point x B due to the application of a unit harmonic point force in the
jth direction applied at point x A , u i (x A , ω) is the displacement field component in
i direction at point x A , ω = 2π f is the angular frequency. For coinciding source
and receiver, the autocorrelation of both the components gives the directional energy
density E i (x, ω) at point x as:
E i (x, ω) = ρω2 u i (x, ω)2 ∝ Im[G ii (x, x; ω)] (2)
In this expression ρ is the mass density. Therefore, the HV can be expressed as

the square root of the ratio of the corresponding imaginary parts of Green’s tensor
components [2, 22]:

Im[G 11 (x, x; ω)] + Im[G 22 (x, x; ω)]
H V (x, ω) = (3)
Im[G 33 (x, x; ω)]
In above expression the subscripts 1 and 2 represent the horizontal directions

while the subscript 3 corresponds to the vertical one, respectively.
HV-Inv is a computer code for forward calculation and inversion of H/V spec-
tral ratios of ambient noise (HVSRN) based on the diffuse field assumption (DFA)
developed by García-Jerez et al. [9]. The software supports joint inversion of
HVSRN and dispersion curves by using several local and global algorithms which
include simulated annealing algorithm.
The definition of an optimal solution for any optimisation problems can be an
extremely challenging task or some time it is practically difficult. This is because
26 A. Joshi et al.
when a problem gets very large, we need to search a vast number of the possible solu-
tion to find the optimal solution. This type of problem can be solved using Simulated
Annealing Technique (SAT). This is a probabilistic technique for approximating the
global optimum of a given task. It is efficacious and commonly used form of optimiza-
tion problems, which is used for solving the unconstrained and bound-constrained
optimization problems. Simulated annealing is not affected by the initial guess, as
it is a global optimization technique and gives the global optima instead of local
optima, which is an optimal solution [5]. This algorithm was originally motivated by
the process of annealing of metal. “Annealing” discusses the process of thermody-
namics, which involves the cooling and heating of the material to modify its physical
properties due to change in its internal structure [18]. The algorithm is a hill-climbing
except instead of picking the best move, it picks a random move. At each iteration of
the simulated annealing technique, a new point is randomly generated. If the selected
iteration improves the solution, then it is accepted. Otherwise, the algorithm makes
the iteration with some probability less than one. The difficulty with this approach
is that while it rapidly finds a local minimum, it cannot get from there to the global
minimum. There are so many uses of Simulated Annealing (SA technique in different
domains. It has a broad range of application that is still being explored. In the present
work the spectral content of HVSR curve is mainly dependent on the P and S wave
velocity profile of region and density of the subsurface layers. Initial range of these
parameters is selected from the heuristic guess based on geophysical and geolog-
ical observations. The iterative computation of HVSR curve is based on Simulated
Annealing algorithm. In the present work, HV-Inv has been used to study the site
specific case study of the Bayasi region in the Garhwal Himalaya.
The ground vibration created by any disturbance at the surface of the earth can be
recorded by three-component recorders. These recorders record ground vibrations
in two horizontally and one vertically oriented sensor. Nakamura [14] has shown
that the ratio of the amplitude spectrum of the horizontal component to the vertical
component is directly related to the shear wave velocity profile of the subsurface
of the earth. Normally in this technique, the ground disturbance created by passing
vehicle is recorded for a specific duration of time in three-component sensors i.e. two
horizontal components along north–south and east–west direction (ans (t), aew (t))
and one vertical component (av (t)) as shown in Fig. 1. Fourier transform of these
components are as follows:
ans (t) −→ Ans (f) (4)
aew (t) −→ Aew (f) (5)
av (t) −→ Av (f) (6)
Amplitude spectrum of each component is calculated as:

0.16
0.08
ans(t) 0
-0.08
-0.16
0 20000 40000 60000 80000
No of samples (Sampling interval = .0078 sec)
(a)
0.16
0.08
aew(t)
-0.08
-0.16
0 20000 40000 60000 80000
(b)
0.16
0.08
av(t)
-0.08
-0.16
0 20000 40000 60000 80000
(c)
Fig. 1 Time series record of the a North South ‘ans (t)’, b East West ‘aew (t)’ and c vertical ‘av (t)’
component of ground vibrations created by passing vehicle
1/2
Ampns (f) = [Re(Ans (f))2 + Im(Ans (f))2 ] (7)
1/2
Ampew (f) = [Re(Aew (f))2 + Im(Aew (f))2 ] (8)
1/2
Ampv (f) = [Re(Av (f))2 + Im(Av (f))2 ] (9)
As we have two horizontal and one vertical component, so there are two amplitude
spectrums of NS and EW component. The amplitude spectrum of ans (t), aew (t) and
av (t) is shown in Fig. 2.
28 A. Joshi et al.
Fig. 2 Amplitude spectrum of time series shown in Fig. 1 along North South, East west and Vertical
component of ground vibration
We can calculate the average of horizontal components either by geometrical or

root mean square amplitude of horizontal components by the following formula:
2 2 1/2
Root mean square = AmpNS (f) + AmpEW (f) /2 (10)
or
Geometrical mean = AmpNS (f) × AmpEW (f) 1/2

(11)
In the present work, Geometrical mean has been used for obtaining the H/V curve
at any site. H/V curve can be obtained by the following expression and is shown in
Fig. 3.
1
HV(f) = AmpNS (f) × AmpEW (f) 2 /AmpV (f) (12)
Fig. 3 Ratio of amplitude spectrum of horizontal by vertical component of record shown in Fig. 1
Fig. 4 Flow graph of Input (x(t)) Output (y(t))

forward modelling approach
System (α,β,σ,ρ)
Fig. 5 Flow graph of Output (y(t)) Input (x(t))

inverse modelling approach
System (α,β,σ, ρ)
According to system theory, the output ‘y(t)’ is obtained when any input ‘x(t)’ is
passed through a system as shown in Fig. 4.
In any mathematical approach, when input is known and system response is known
and we are calculating the output, then such system is termed as the forward approach.
Similarly, when the output of the system is known and we are interested to calculate
input from the system, then this approach is termed as inverse approach. In both
approaches, we know the system response. Inverse approach is shown in Fig. 5.
The obtained HVSR curve at any site is theoretically represented by the formu-
lation given which is mainly dependent on parameters of the earth model. In the
formulation, system response mainly depends on P wave velocity (α), Shear wave
velocity (β), the density of the medium (ρ) and Poisson ratio (σ ). The purpose of
present work is to select final earth parameters using Simulated Annealing algorithm
developed by García-Jerez et al. [9]. Different authors have attempted to explain
the H/V phenomenology in terms of SH waves [10, 14], of Rayleigh waves [3, 11,
13], and by adding the effects of Love. Recent studies have considered the role
of all waves, known as a total field [21, 22]. It was found that results differ not
only as a function of the type of waves but also as a function of the temporal and
spatial distribution of the sources and their strength. The inversion of the H/V curve
requires not only the knowledge of the specific sources acting at the site but also of
several other soil parameters (e.g., Poisson’s ratio, damping of each layer, 2D effects,
etc.) which cannot be easily determined. Under these condition various algorithms
helps in selecting most preferred model out of various possible range of modelling
parameters.
Modelling of HV curve can be made by Simulated Annealing algorithm in HV-Inv
software. This software requires HV curve as an input function which is modelled
by using appropriate parameter space that can be estimated using several algorithms
supported in the software. In the present work the acquired data consist of a time series
of 12 min length that has sampling frequency 128 Hz. Therefore, the data of ground
vibration consist of 92,160 samples along three different mutually perpendicular
directions. This gives a similar number of samples in HV curve in frequency domain.
However, software HV-Inv needs very less number of data point in the HV curve as
an input to the algorithm. Therefore, the challenging task before us is to re discretize
92,160 samples into very less samples retaining the spectral property of HV curve.
We need to discretize it so that all peaks should remain preserved in an HV curve of
30 A. Joshi et al.
fewer samples. For this purpose, a software DISHV is prepared in FORTRAN that
prepares an HV curve of small length i.e. of nearly 30 samples in frequency range
0.1–100. Input to this program is a time series in NS, EW and Vertical component
of ground vibrations. The program calculates FFT of these input ground vibration in
various small windows which can be decided by the user. Presently 38,000 samples
at sampling frequency 128 Hz has been used as input of ground vibration along with
NS, EW and Vertical direction. Figure 6 shows the time series along with NS, EW
and vertical direction. This entire ground vibration is divided into 10 windows and
HV curve in each window is calculated. Mean and Standard deviation of all HV
curve is calculated which is used as input to the algorithm. Figure 7a shows the HV
curve when window length consists of 3800 samples. The software DISHV further
check the number of points in each logarithmic window and discretize uniformly
10 samples in each window. The discretize HV curve having 39 points is shown in
0.16
0.08
ans(t)
-0.08
-0.16
0 20000 40000 60000 80000
(a)
0.16
0.08
aew(t)
-0.08
-0.16
0 20000 40000 60000 80000
(b)
0.16
0.08
av(t)
-0.08
-0.16
0 20000 40000 60000 80000
(c)
Fig. 6 a NS, b EW and c vertical component of ground vibration

Fig. 7 a HV curve obtained from time series containing 3800 samples, b recreating HV curve
from obtained curve using 39 samples and c comparison of HV curve having 3800 and 39 samples
respectively
Fig. 7b. Comparison of HV curve that is having 3800 samples and 39 points clearly
shows that peak are preserved and both curves has characteristic similarity shown
in Fig. 7c. This reduces approximately 100 times the sample present in actual data,
thereby retaining the main spectral contents of the HV curve. The output from DISHV
program is further used as an input to HV-Inv program for modelling of HV curve
using Simulated Annealing algorithm.
3 Case Study for the Bayasi Site, Uttarakhand
The ambient noise data has been recorded at the Bayasi site, Uttarakhand of the
Garhwal Himalaya in two horizontal and one vertical direction shown in Fig. 8.
The amplitude spectrum of time series of recorded noise data is shown in Fig. 9a
along North–South, East–west and vertical component of ground vibration. The H/V
curve obtained by the ratio of averaged horizontal amplitude spectrum to that of the
vertical spectrum is shown in Fig. 9b. The obtained H/V curve is then used as input
to the DISHV program to discretize the HV curve for less number of samples. The
discretize HV curve having 36 samples is shown in Fig. 9c.
The parameter ranges of P-wave velocity, S-wave velocity and density are chosen
for the inversion of HV curve using simulated annealing algorithm is shown in
Fig. 10. The average velocity profile and best fit model of VP and VS obtained after
32 A. Joshi et al.
Amplitude 0.4
-0.4
0 20000 40000 60000 80000 100000

No of samples (s.i. = .0078125 sec)
(a)
0.4
Amplitude
-0.4
0 20000 40000 60000 80000 100000

(b)
0.4
Amplitude
-0.4
0 20000 40000 60000 80000 100000

(c)
Fig. 8 Time series record of the a North South ‘ans (t)’, b East west ‘aew (t)’ and c vertical ‘av (t)’
component at Bayasi site
Fig. 9 a Amplitude spectrum of time series, b ratio of amplitude spectrum of horizontal by vertical
component of record, c HV curve from 36 samples
Fig. 10 Set of tested models (blue and yellow lines), average and best-fit models showing with
black and red colour line
the inversion process is shown in Fig. 10 with black colour line and red colour line
respectively. Figure 11 is showing the best-modelled HV curve in red colour after
inversion and the yellow and blue lines showing the set of forward HV curves for
the different test model. The output obtained from Hv-Inv software by using input
spectral ratio at 36 points using Simulated Annealing algorithm is shown in Fig. 12.
Fig. 11 Best modelled HV curve for the best model after inversion shown in red colour line
34 A. Joshi et al.
Fig. 12 HVInv program window of the inversion of HV curve using HV-Inv
4 Results and Discussion
The best-fit S-wave velocity model obtained from the simulated annealing algo-
rithm is further refined by iterative modelling of HV curve using forward modelling
approach. The minute change in the thickness and velocity according to the user gives
the best model after the refinement. The modelled HV curve after the refinement is
shown in Fig. 13. The one dimensional Shear wave velocity structure obtained after
the refinement of model obtained from Simulated Annealing algorithm is shown in
Fig. 14.
The final model obtained after refinement shows that the first layer is of thickness
0.45 m with S-wave velocity 100.0 m/s and the second layer of thickness 9 m with
S-wave velocity about 320.1 m/s and the half-space velocity is about 645.2 m/s.
The final model obtained after refinement in the model through iterative forward
modelling of the model selected from Simulated Aneling algorithm is shown in Table
1. The work done in this paper shows that the HV-Inv program for the inversion of
HV curve effectively model the HV curve by a global Simulated Annealing technique
and allow the user to further refine the model obtained from the inversion process.
The geological input of the area obtained from the field investigation of the region
also support this observation.
Fig. 13 Final HV curve obtained after the refinement
5 Conclusion
The ratio of the amplitude of horizontal to the vertical component of ground vibration
created by an ambient noise is dependent on the subsurface velocity structure of the
investigating area. Ground vibration data from ambient noise recording system has
been recorded at the Bayasi site in the Garhwal Himalaya. A computer code DISHV
in FORTRAN has been developed in this work has been used for obtaining HV
curve with a limited number of data points that can be effectively used as an input
to the HV-Inv software developed by García-Jerez et al. [9]. The HV curve obtained
from DISHV is further used to obtained theoretical HV curve corresponding to a
model of earth by using simulated annealing algorithm. The obtained velocity model
from simulated annealing is utilized for the forward modelling approach that further
refines the obtained shear wave velocity model by comparing the simulated and the
observed HV curves.
36 A. Joshi et al.
Fig. 14 Final S-wave

velocity model after the
further refinement in the
model
Table 1 Final model parameter after the refinement

Thickness (m) P-wave velocity (m/s) S-wave velocity (m/s) Density (Kg/m3 ) Poisson ratio
0.5 350.7 100.0 2000.0 0.45
9.0 792.8 320.1 2000.0 0.40
0.0 1415.9 645.3 2000.0 0.36
References
1. Aki K (1988) Separation of source and site effects in acceleration power spectral of California
earthquakes. In: Proceedings of 11th world conference on earthquake engineering, vol 8, pp
163–167
2. Arai H, Tokimatsu K (2004) S-wave velocity profiling by inversion of microtremor H/V
spectrum. Bull Seismol Soc Am 94(1):53–63
3. Arai H, Tokimatsu K (2005) S-wave velocity profiling by joint inversion of microtremor disper-
sion curve and horizontal-to-vertical (H/V) spectrum. Bull Seismol Soc Am 95(5):1766–1778
4. Bard PY (2008) The H/V technique: capabilities and limitations based on the results of the
SESAME project. Bull Earthq Eng 6(1):1–2
5. Billings SD (1994) Simulated annealing for earthquake location. Geophys J Int 118(3):680–692
6. Bonnefoy-Claudet S, Köhler A, Cornou C, Wathelet M, Bard PY (2008) Effects of Love waves
on microtremor H/V ratio. Bull Seismol Soc Am 98(1):288–300
7. Cadet H, Bard PY, Rodriguez-Marek A (2010) Defining a standard rock site: propositions based
on the KiK-net database. Bull Seismol Soc Am 100(1):172–195
8. Campillo M, Paul A (2003) Long-range correlations in the diffuse seismic coda. Science
299(5606):547–549
9. García-Jerez A, Piña-Flores J, Sánchez-Sesma FJ, Luzón F, Perton M (2016) A computer
code for forward calculation and inversion of the H/V spectral ratio under the diffuse field
assumption. Comput Geosci 97:67–78
10. Herak M (2008) ModelHVSR—a Matlab® tool to model horizontal-to-vertical spectral ratio
of ambient noise. Comput Geosci 34(11):1514–1526
11. Lermo J, Chávez-García FJ (1994) Are microtremors useful in site response evaluation? Bull
Seismol Soc Am 84(5):1350–1364
12. Lunedei E, Albarello D (2010) Theoretical HVSR curves from full wavefield modelling of
ambient vibrations in a weakly dissipative layered Earth. Geophys J Int 181(2):1093–1108
13. Malischewsky PG, Scherbaum F (2004) Love’s formula and H/V-ratio (ellipticity) of Rayleigh
waves. Wave Motion 40(1):57–67
14. Nakamura Y (1989) A method for dynamic characteristics estimation of subsurface using
microtremor on the ground surface. Railway Tech Res Instit Quart Rep 30(1)
15. Nakamura Y (2000) Clear identification of fundamental idea of Nakamura’s technique and
its applications. In: Proceedings of the 12th world conference on earthquake engineering, vol.
2656. New Zealand, Auckland
16. Perton M, Sánchez-Sesma FJ, Rodríguez-Castellanos A, Campillo M, Weaver RL (2009) Two
perspectives on equipartition in diffuse elastic fields in three dimensions. J Acoust Soc Am
126(3):1125–1130
17. Pilz M, Parolai S, Leyton F, Campos J, Zschau J (2009) A comparison of site response tech-
niques using earthquake data and ambient seismic noise analysis in the large urban areas of
Santiago de Chile. Geophys J Int 178(2):713–728
18. Press W, Teukolsky S, Vetterling W, Flannery B (1992) Numerical recipes in C, 2nd edn
19. Sanchez-Sesma FJ (1987) Site effects on strong ground motion. Soil Dyn Earthquake Eng
6(2):124–132
20. Sánchez-Sesma FJ, Pérez-Ruiz JA, Luzon F, Campillo M, Rodríguez-Castellanos A (2008)
Diffuse fields in dynamic elasticity. Wave Motion 45(5):641–654
21. Sánchez-Sesma FJ, Rodríguez M, Iturrarán-Viveros U, Luzón F, Campillo M, Margerin
L, García-Jerez A, Suarez M, Santoyo MA, Rodriguez-Castellanos A (2011) A theory
for microtremor H/V spectral ratio: application for a layered medium. Geophys J Int
186(1):221–225
22. Sánchez-Sesma FJ, Rodrıguez M, Iturrarán-Viveros U, Rodrıguez-Castellanos A, Suarez M,
Santoyo MA, Garcıa-Jerez A, Luzón F (2010) Site effects assessment using seismic noise. In:
Proceedings of the 9th international workshop on seismic microzoning and risk reduction
38 A. Joshi et al.
23. Sato H (2010) Retrieval of Green’s function having coda waves from the cross-correlation
function in a scattering medium illuminated by a randomly homogeneous distribution of noise
sources on the basis of the first-order Born approximation. Geophys J Int 180(2):759–764
24. Weaver RL, Lobkis OI (2004) Diffuse fields in open systems and the emergence of the Green’s
function (L). J Acoust Soc Am 116(5):2731–2734
25. Yokoi T, Margaryan S (2008) Consistency of the spatial autocorrelation method with seismic
interferometry and its consequence. Geophys Prospect 56(3):435–451
Nonlinear Soil Amplification Models
for a Moderately Active Seismic Zone
in India
C. Shreyasvi and K. Venkataramana
Abstract The dynamic stiffness and damping of the soil material, depth of the soil
profile, impedance between the soil and the underlying bedrock and soil nonlin-
earity are the factors influencing the local site response. The important parameter in
assessing the site response is the “amplification factor,” which is usually correlated
to Shear Wave velocity in the top 30 m (Vs(30) ). Though using VS(30) as an index for
amplification is simple and robust, it is not recommended for site-specific applica-
tions. In the present study, two distinct soil types i.e. “Sand” and “Clay” with the same
value of VS(30) demonstrated variable amplification characteristics. Hence, distinct
site amplification models were derived for the two soil types considering the intensity
of the input bedrock motion as the primary independent variable. The borehole data
from nearly 50 locations in North Kerala, an intraplate region in the Southern part
of India was collected. The ground response was simulated in 1-dimension consid-
ering equivalent linear behavior of soils on the SHAKE 2000 platform. The ground
motions used in the simulation were scaled to the target spectrum obtained from the
regional seismic hazard assessment. The average spectral amplification observed is
5 for “Clay” and 3.5 for “Sand” in the study region. The soil profiles categorized
as “sand” exhibits nonlinear behavior. “Clay” deposits reveal sustained amplifica-
tion at longer periods and hence, can significantly influence ground response during
longer duration ground shaking. The empirical amplification equations developed
from the study can be used to modify the generic ground motion prediction models
to region-specific applications.
Keywords Spectral matching · Amplification function · North Kerala · Ground

response simulation
C. Shreyasvi (B)
Department of Earth Sciences, IIT Gandhinagar, Palaj 382355, Gujarat, India
e-mail: shreyasvic@gmail.com
K. Venkataramana
Department of Civil Engineering, NITK Surathkal, Mangalore 575025, India
40 C. Shreyasvi and K. Venkataramana
1 Introduction
The seismic source, wave propagation, and site effects are the fundamental factors
that need to be accounted for when estimating ground motion at a point of interest
on the surface. The seismic wave originating from the source traces multiple paths
undergoing amplification and in some cases attenuation depending on the dynamic
properties of the constituent medium. The top few meters of the soil deposits have
the potential to alter the transmitted seismic wave both in terms of magnitude as
well as the direction of propagation. The impact of local site effects has been amply
demonstrated during past earthquakes such as Kutch (1819), Michoacan (1985), Bhuj
(2001), Sendai (2015) and many more. The local site effect is a broad term collec-
tively describing various phenomena such as local ground response, basin effects, and
topographic effects. The present study addresses the local ground response aspect
for a region that is treated as a moderately active seismic zone. The local ground
response assesses the impact of shallow sediments on the ground motion by consid-
ering nonlinear behavior, impedance, and possible resonance effects of the soils.
The objective of ground/site response studies is to determine the amplification of the
seismic waves and the shape of the resulting response spectra in order to improve
the seismic resistance of the structures.
Earlier attempts by Raghukanth and Iyengar [1] presented site amplification as a
function of shear wave velocity in the top 30 m (VS(30) ) for India. A similar study by
Choi and Stewart [2] presented nonlinear site amplification as a function of VS(30) for
the USA. In the present study, it was observed that multiple factors are responsible
for site amplification such as density, soil type, cementation, depth to bedrock, and
natural period of the soil column. Most of the site response investigations in the study
region undermine the influence of soil nonlinearity in ground motion estimation and
has not received much importance. The present study is an attempt to capture the
nonlinear soil amplification has a function of soil type in contrast to the conventional
VS(30) approach. The nonlinear behavior of soils has been estimated using an equiva-
lent linear analysis as it is more efficient and requires lesser computation time when
compared to nonlinear analysis.
The distinct amplification pattern of two soil types (clay and sand) have been
presented under various time frames. Finally, a nonlinear amplification model has
been proposed for each soil type as a function of input bedrock motion. The ampli-
fication model has been developed to accommodate various earthquake scenarios
and to provide an overall rational estimate of site amplification. These amplifica-
tion models can be directly incorporated into the existing ground motion prediction
models to provide surface-level estimates of ground motion parameters such as PGA,
PGV and Sa (at 5% damping). Further, the proposed models can be updated with
supplemental field data in the study region.
Nonlinear Soil Amplification Models for a Moderately Active Seismic Zone in India 41
2 Study Region
The study region belongs to the Northern part of Kerala stretching from Kasargod to
Calicut. The data corresponding to soil strata was obtained in the form of borelogs.
The topographical map of the study area, as well as the location of boreholes, has been
shown in Fig. 1. As evident from Fig. 1, the topography of the study area has higher
elevation indicating hilly terrain with gradually sloping toward the beach. The study
area is outlined by coastal line on one side and hills on the other side. These borelogs
consist of penetration resistance value, i.e. SPT “N” value, unit weight (γ, kN/m3 ),
location of groundwater table, index and engineering properties, and soil description.
The statistics of the collected field data have been compiled and presented in the form
of a histogram as shown in Fig. 2. Most of the soil samples belong to categories III
(N < 15) and I (N > 15) highlighting the existence of competent as well as weaker
soil material.
Fig. 1 Topographical map in the vicinity of borehole locations with the borehole depths dropped
as a shadow [3]
Fig. 2 Histogram of the soil

bore logs collected for the
study
The collected soil data has been further classified into two soil types, i.e. sand and
clay, depending on the grain size distribution and index properties. The representative
stratigraphy of the two soil types is shown in Fig. 3. The soil formations typically
vary from red fine loamy to clayey soils apart from red laterite with large variation
in overburden thickness. Laterite rich in Iron and aluminum is found near to coastal
plains. These laterites can be observed at higher elevations in the form of plateaus
as well. Due to the absence of field measurements of VS , an indirect method of
Fig. 3 Stratigraphy of the Depth Clayey

Depth Sandy
two representative soil types
(m) Deposit (m) Deposit
used in the study [3]
1.5 0.9
3.4 3
5.1
7.0 7.6
10.3
10.6
11.8
14.7
15.0
16.7 18.6
18.6
21.6
Legend
Filled up earth 24.4
Sandy clay
Silty Sand 30
Lateritic Clay
32.6
Clay mixed with
Quartz
37
Silty clay
38.1
Lithomerge
Clay with high 41.1
plasticity
Laterite 44.8
Weathered Rock
46.6
Hard Rock
estimation of a VS profile for different soil columns was adopted. The SPT “N” value
has been correlated to VS by many researchers as shown in Table 1. These correlations
have been developed for similar topographical conditions as our study region. Further,
the use of multiple correlations aids in reducing the epistemic uncertainty in the
estimation of VS .
A statistical comparison of estimated VS profiles from various correlations for a
regional soil profile revealed that few correlations such as HR, KMM, TK estimated
extreme values and had to be discarded while finalizing the VS profile for a given
soil column. Finally, a total of 6 correlations were used for estimating the soil profile
and a sample of this estimation has been shown in Fig. 4.
3 Soil Modeling and Ground Motion Records
The dynamic characteristics of the soil material need to be estimated and modeled
to estimate the local site response. The modulus reduction (M-R) curves, i.e. shear
modulus reduction (G/Gmax) versus cyclic shear strain (γc) and damping (D) curves
(i.e. damping versus γc) are essential in modeling the dynamic behavior of soils.
Based on experimental investigation, many classical M-R and damping curves have
been proposed in the past. In the present study, the standard empirical relations of
G/Gmax (Fig. 5) and D as a function of cyclic shear strain (γc) for sand and clay
[19–22] have been used to model the dynamic soil response.
The second phase deals with the selection of ground motion records compatible
with the target spectrum. The third phase involves the modification of the chosen
records with respect to the target spectrum. The practice of selecting ground motion
varies widely and no definitive guidelines or strict procedures exist. However, the
codal provision provides a general guideline to consider a minimum of five recorded
or simulated rock outcrop horizontal ground motion records from events with magni-
tude and distance range consistent with those controlling Maximum Considered
Earthquake (MCE) ground motion [14].
The uniform hazard spectrum (UHS) for the study region has been adapted from
Shreyasvi et al. [15]. The objective of the study is to quantify the seismic site ampli-
fication and the UHS resulting from a single hazard level may not be sufficient for
developing the amplification equation. Hence, multiple hazard levels were consid-
ered. The design ground motions specified in the Indian seismic code correspond to
10% POE in 50 years, i.e. Design Basis Earthquake (DBE). With DBE as the stan-
dard design condition, two more hazard levels, i.e. 10 * DBE and 0.25 * DBE were
defined based on the recommendations from Stewart and Afshari [16]. The DBE
(abbreviated as HM) represents the median hazard level whereas 10 * DBE (HL)
and 0.25 * DBE (HH) represent lower and higher hazard levels of the considered
range. HL corresponds to 65% POE and HH corresponds to 2.5% POE in 50 years.
The accelerograms selected for the study have been presented in Table 2 corre-
sponding to DBE. The selected accelerograms are spectrally matched to the UHS
of the study region in order to generate spectrum compatible ground motion records
Table 1 Vs–N Correlations investigated in the study [3]

Authors Correlation Soil type Region
Maheshwari et al. Vs = 89.31 * N0.358 Clay Chennai
(2010)a —UM1 [4] Vs = 100.53 * N0.265 Sand
Vs = 95.64 * N0.301 All soil
Hanumanthrao and Vs = 79 * N0.434 Sand Delhi
Ramana Vs = 82.6 * N0.43 All soil
(2008)—HR [5] Vs = 86 * N0.42 Silty sand/sand silt
Unal Dikmen Vs = 58 * N0.39 All soil Turkey
(2009)—UD [6] Vs = 73 * N0.33 Sand
Vs = 60 * N0.36 Silt
Vs = 44 * N0.48 Clay
Chatterjee and Vs = 78.21 * N0.38 All Soil Kolkata
Choudhury Vs = 77.11 * N0.39 Clay
(2013)a —CC1 [7] Vs = 54.82 * N0.53 Silty Sand
Vs = 58.02 * N0.46 Silt
Kirar, Maheshwari Vs = 100.31 * N0.348 Sand Roorkee
et al. Vs = 94.4 * N0.379 Clay
(2016)—KMM [8] Vs = 99.5 * N0.345 All Soil
Maheshwari et al. Vs = 90.75 * N0.304 All Soil Chennai
(2010)a —UM2 [4] Vs = 96.29 * N0.266 Sand
Vs = 83.27 * N0.365 Clay
Hasaneebi and Vs = 90 * N0.309 All Soil Turkey
Ulusay Vs = 90.8 * N0.319 Sand
(2007)a —NR [9] Vs = 97.9 * N0.269 Clay
Anbazhagan et al. Vs = 68.96 * N0.51 All Soil Lucknow
(2012)—AZ [10] Vs = 60.17 * N0.56 Sand
Vs = 106.63 * N0.39 Clay
Sil and Haloi Vs = 75.478 * N0.3799 All Soil Any region
(2017)a —SH [11] Vs = 79.217 * N0.3699 Sand
Vs = 99.708 * N0.3358 Clay
Mhaske and Vs = 72 * N0.4 All soil Mumbai
Choudhury
(2011)a —MC [12]
Thokchom et al. Vs = 2.641 * N + 189.6 Sand Dholera, Western India
(2017)a —TK [13] Vs = 3.925 * N + 143.1 Silt
Vs = 3.395 * N + 156.8 Clay
Vs = 3.311 * N + 160.5 All soils
Chatterjee and Vs = 78.63 * N0.37 All soil Kolkata
Choudhury Vs = 78.03 * N0.38 Clay
(2013)a —CC2 [7] Vs = 58.62 * N0.45 Silt
Vs = 56.44 * N0.51 Silty sand
a Correlationsfinalized for estimating the VS profile. Nomenclature ending with numbers 1 and 2
represents correlation derived based on uncorrected and corrected SPT ‘N’ values respectively
Shear Velocity (m/s)

100 200 300 Mean
400 + SD
Mean - SD
Mean
5
10
15
Depth (m)
20
25
30
35
40
45
50
Fig. 4 Shear velocity profile of a typical clay deposit
1.2
SAND, Upper Bound (Seed & Idriss 1970)
Clay upper range (Seed and Sun, 1989)
Modulus Reduction, G/Gmax
1.0
Rockfill (Gazetas, 1992)
ROCK (Schnabel 1973)
0.8
0.6
0.4
0.2
0.0
1E-4 0.001 0.01 0.1 1 10
Shear Strain (%)
Fig. 5 Shear modulus reduction curves as a function of cyclic strain for different soil types
Table 2 Details of ground motions used in the study [3]

Earthquake Acronym Date Mw Rrup (km) PGA Scaled Hazard
name (g) PGA level
(g)
RiviereDuLoup RSN1688 06-03-2005 4.65 19.05 0.045 0.100 Median
RiviereDuLoup RSN1771 06-03-2005 4.65 41.75 0.070 0.124 hazard
level
Greenbrier RSN6934 28-02-2011 4.68 6.27 0.030 0.135
Greenbrier RSN7052 28-02-2011 4.68 54.07 0.001 0.126
ValDesBois RSN4027 23-06-2010 5.1 52.94 0.018 0.102
Sicily IT.NOT 13-12-1990 5.6 48.30 0.090 0.111
Central Italy 3A.MZ14 26-10-2016 5.9 36.60 0.515 0.108
Central Italy 3A.MZ19_30 30-10-2016 6.5 22.60 0.363 0.114
for the site response analysis. Individually matched accelerograms, original UHS
termed as Target Spectrum and the mean of the matched accelerograms are plotted
in Fig. 6.
IT.NOT
3A.MZ29
0.3 3A.MZ21
3A.MZ19_30
3A.MZ19
Spectral Acceleration (g)
3A.MZ14
RSN7052
0.2 RSN6937
RSN4027
RSN1771
RSN1688
Mean
0.1
Mean SD
Target
0.0
0.01 0.1 1
Period (s)
Fig. 6 Plot of 5% damped rock acceleration spectrum of ground motions scaled to HM [3]
4 Site Response Analysis
The dynamic simulation of the ground motion propagated through the layered soil
medium was realized through the SHAKE2000 program. The compiled borehole
information along with the selected and scaled input motions served as input to
equivalent linear analysis. The dynamic characteristics of the local soil are explained
through different output parameters such as amplification and surface response spec-
trum. Due to space constraints, the results of all the simulated soil columns are not
presented. It was observed that the two soil profiles of nearly the same value of
VS demonstrated a distinct amplification pattern and hence, an alternate predictor
variable for site amplification was mandatory.
Figures 7 and 8 represent the median amplification function along with the stan-
dard deviation and 95% confidence interval as a function of rock spectral acceleration
assessed at different spectral periods. The amplification function varying with the
period (AF(f)) for different input motion acceleration values (PGA and Sa (f)) was
compiled for all the numerically modeled soil deposits belonging to clay and sand site
category. Each soil profile generated 33 data points from the input ground motions
for each period window. The data points from all the simulated soil columns have
been presented in Figs. 7 and 8. Nonlinear regression was performed on these data
points as shown.
The observed amplification data points are sparse and widely distributed for Sand.
However, they were sufficient to draw a nonlinear amplification function for different
2.5 T = 0.01 s T = 0.2 s 0.5 T = 1.5 s

2
1.5
Amplification, AF(f)
1 0.1
1
0.05
Observed Amplification
0.5 +/- 95% confidence interval
Median AF(f) 0.5
Median +SD
Median - SD
0.01
0.01 0.05 0.1 0.01 0.05 0.1 0.5 0.005 0.01 0.05 0.1
Rock Spectral Acceleration (g) Rock Spectral Acceleration (g) Rock Spectral Acceleration (g)
Fig. 7 Amplification factors regressed against rock spectral acceleration for ‘sand’ sites [3]
T = 0.01 s T = 0.2s T= 1.5 s

1
1
0.1
Observed Amplification
+/- 95% Confidence interval
Median
0.1 Median + SD 0.01
Median - SD
0.1
0.01 0.1 0.01 0.1 0.01 0.1
Rock Spectral Acceleration (g) Rock Spectral Acceleration (g) Rock Spectral Acceleration (g)
Fig. 8 Amplification factors regressed against rock spectral acceleration for ‘clay’ sites [3]
spectral periods. The natural period of the soil profiles lies in the range of 0.31–0.38 s.
The amplification for the depicted soil type, i.e. sand tends to be highly nonlinear
especially at periods below the average site period. Significant variability is visible for
higher spectral acceleration values at T = 0.2 s, demonstrating even larger variation
at T = 1.5 s. This is mainly due to the large differences in the intensity of the
selected input ground motions. An upward shift was observed at intermediate and
longer period range (T = 1.5 s). The inability of the equivalent linear methodology to
converge to a solution for high-intensity records is the reason behind this upward shift
[17]. Additionally, the equivalent linear method estimates higher amplification at the
smaller period range and overestimates resonant responses when the soil becomes
nonlinear [18]. Further, the amplification reduces drastically for periods after the
resonant vibration period of the sites. Another interesting observation made during
the study is the shift in the natural period depending on the strain induced by various
ground motions.
The amplification observed in clay at T = 0.01 and 0.2 s is less when compared
with sand. However, clay demonstrates sustained amplification at longer periods (T
= 1.5 s). While the majority of the soil deposits have their natural period around
0.4 s, there are a few deposits with a period as high as 0.7–0.9 s displaying a diverse
range. However, the shift in the natural period depending on the induced strain was
noticed similar to the “sand” site. Due to this phenomenon, resonance is attained by
soil profiles at the elongated site period (T = 1.5 s) driving the regression toward a
positive correlation. As a result, the upward shift in the fitted median amplification
curve was observed at these periods. The “clay” site demonstrated consistently higher
AF(f) compared to the other soil categories implying amplification of longer period
ground motions. Overall, “clay” sites demonstrate slower stiffness degradation and
hence, less nonlinear when compared to “sand” sites. The VS(30) is widely accepted
as an index for soil amplification.
However, the present study suggests that the soil characteristics affect the amplifi-
cation to a greater extent. In order to validate this, two soil profiles (Sand & Clay) of
the same Vs(30) were considered for plotting mean amplification in Fig. 9. As evident
from Fig. 9, the sand exhibits higher amplification at lower input values but decreases
gradually with the increase in PGA demonstrating its nonlinear behavior. For the
“clay” site, amplification reduces as the intensity increases and de-amplification can
be noticed for PGA values as small as 0.005 g. However, there is still a very clear
amplification at longer periods as shown in Fig. 8. An important observation made
during the analysis of “clay” deposits is that as the soil plasticity index increases, the
behavior of the soils became less nonlinear.
Fig. 9 Mean site

amplification of the two soil Sand
types at T = 0.01 s [3] 1.5
Clay
1.0
0.5
0.0
0.01 0.1 1
Rock Spectral Acceleration (g)
5 Conclusion
The findings of the study are summarized below.

1. The soil profiles modeled in the study belongs to NEHRP “C” (360–760 ms−1 )
and “D” (180–360 ms−1 ) site categories and the study region belongs to seismic
zone III (moderate level shaking).
2. The average spectral amplification observed 5 for “clay” and 3.5 for “sand” of
the study region.
3. “Sand” sites exhibit nonlinear behavior by undergoing large amplification for
smaller intensity measure but reduce substantially as the spectral acceleration
(Sa ) values exceed 0.1 g.
4. “Clay” sites exhibit amplification even at longer periods but become less
nonlinear with the increase in plasticity index. Hence, the “clay” site plays a
major role in the event of long-period seismic waves.
5. Three soil profiles of VS(30) in the similar range demonstrated distinct amplifica-
tion characteristics. The “sand” site amplifies 29% more than the “clay” site for
lower input Sa values.
6. The amplification characteristics observed in various spectral period frames
suggest that PGA offers an unbiased and better prediction of amplification
function. The same parameters studied at different spectral periods suffer from
resonance (site-specific effect) and demerits of the EQL method (computational
capacity).
References
1. Kanth SR, Iyengar RN (2007) Estimation of seismic spectral acceleration in peninsular India.
J Earth Syst Sci 116(3):199–214
2. Choi Y, Stewart JP (2005) Nonlinear site amplification as function of 30 m shear wave velocity.
Earthq Spectra
3. Shreyasvi C, Venkataramana K, Chopra S (2019) Local site effect incorporation in probabilistic
seismic hazard analysis—a case study from southern peninsular India, an intraplate region. Soil
Dyn Earthq Eng 123:381–398
4. Maheswari RU, Boominathan A, Dodagoudar GR (2010) Use of surface waves in statistical
correlations of shear wave velocity and penetration resistance of Chennai soils. Geotech Geol
Eng 28(2):119–137
5. Hanumantharao C, Ramana GV (2008) Dynamic soil properties for microzonation of Delhi,
India. J Earth Syst Sci 117(2):719–730
6. Dikmen Ü (2009) Statistical correlations of shear wave velocity and penetration resistance for
soils. J Geophys Eng 6(1):61–72
7. Chatterjee K, Choudhury D (2013) Variations in shear wave velocity and soil site class in
Kolkata city using regression and sensitivity analysis. Nat Hazards 69(3):2057–2082
8. Kirar B, Maheshwari BK, Muley P (2016) Correlation between shear wave velocity (vs) and
SPT resistance (N) for Roorkee region. Int J Geosynth Ground Eng 2:9
9. Hasancebi N, Ulusay R (2007) Empirical correlations between shear wave velocity and
penetration resistance for ground shaking assessments. Bull Eng Geol Env 66(2):203–213
10. Anbazhagan P, Kumar A, Sitharam TG (2013) Seismic site classification and correlation
between standard penetration test N value and shear wave velocity for Lucknow City in
Indo-Gangetic Basin. Pure Appl Geophys 170(3):299–318
11. Sil A, Haloi J (2017) Empirical correlations with Standard Penetration Test (SPT)-N for
estimating shear wave velocity applicable to any region. Int J Geosynth Ground Eng 3:22
12. Mhaske SY, Choudhury D (2011) Geospatial contour mapping of shear wave velocity for
Mumbai city. Nat Hazards 59(1):317–327
13. Thokchom S, Rastogi BK, Dogra NN, Pancholi V, Sairam B, Bhattacharya F, Patel V (2017)
Empirical correlation of SPT blow counts versus shear wave velocity for different types of soils
in Dholera, Western India. Nat Hazards 86(3):1291–1306
14. American Society of Civil Engineers (2017) Minimum design loads and associated criteria for
buildings and other structures, ASCE/SEI 7–16. Reston, VA
15. Shreyasvi C, Venkataramana K, Chopra S, Rout MM (2019) Probabilistic seismic hazard
assessment of Mangalore and its adjoining regions, a part of Indian Peninsular: an intraplate
region. Pure Appl Geophys 176(6):2263–2297
16. Stewart JP, Afshari K, Hashash YMA (2014) Guidelines for performing hazard-consistent
one-dimensional ground response analysis for ground motion prediction (October 2014), 1–141
17. Papaspiliou M, Kontoe S, Bommer JJ (2012) An exploration of incorporating site response
into PSHA-part II: sensitivity of hazard estimates to site response approaches. Soil Dyn Earthq
Eng 42:316–330
18. Kim B, Hashash YMA, Stewart JP, Rathje EM, Harmon JA, Musgrove MI, Silva WJ (2016)
Relative differences between nonlinear and equivalent-linear 1-D site response analyses. Earthq
Spectra
19. Seed HB, Idriss IM (1970) Soil moduli and damping factors for dynamic response analysis.
Report no. EERC 70–10. University of California, Berkeley, California
20. Seed HB, Sun JH (1989) Implication of site effects in the Mexico City earthquake of September
19, 1985 for earthquake-resistance-design criteria in the San Francisco Bay Area of California.
Report No. UCB/EERC-89/03. University of California, Berkeley, California
21. Gazetas G, Dakoulas P (1992) Seismic analysis and design of rockfill dams: state-of-the-art.
Soil Dyn Earthq Eng 11(1):27–61
22. Schnabel PB (1973) Effects of local geology and distance from source on earthquake ground
motions (Doctoral dissertation, University of California, Berkeley)
Prediction of Future Surface PGA
in the States of Indo-Gangetic Basin
Considering Site Specific Studies
P. Anbazhagan, Mohammad Rafiq Joo, Meer Mehran Rashid,

and Ketan Bajaj
Abstract Indo-Gangetic Basin (IGB) is the major geographical region of India

extending from Punjab to Bihar including Uttar Pradesh (UP) and Haryana. In this
region, major earthquake damages have been reported in the past. Several attempts
have been made for different site response studies and estimation of earthquake risk
in the region. However, most of them consider only shallow soil information < 50 m
and available ground motions. Present study attempts to analyze the spatial variation
of peak ground acceleration (PGA) at surface for possible future scenario earthquakes
in and around IGB. The earthquakes were identified based on past seismic gaps and
studies whose magnitudes ranged from Mw 7.5 to Mw 9.0. The earthquakes were
simulated for 270 sites with available shear wave velocity data throughout the IGB.
Using proper input parameters of soil column profiles, shear wave velocity, depth
of input motion, suitable shear modulus reduction and damping curves; the detailed
analyses were carried out using DEEPSOIL. This paper arrives at spatial variation
of PGA at surface due to individual earthquakes. The response (bedrock as well as
surface PGA) of different states toward each earthquake has been tabulated. Sites in
Bihar reflect average and maximum surface PGA 0.15 g and 0.68 g, Uttar Pradesh
0.10 g and 1.18 g, Punjab and Haryana 0.12 g and 0.62 g respectively. These values
are indicative of the sensitiveness toward earthquake damages. Maps representing
surface PGA for each scenario earthquake were plotted giving detailed information
about the surface seismic hazard of the area associated with each earthquake.
Keywords PGA · Indo-Gangetic Basin · IGB · Site response · Bed rock
P. Anbazhagan (B) · K. Bajaj

Department of Civil Engineering, Indian Institute of Science, Bangalore 560012, India
e-mail: anbazhagan@iisc.ac.in; anbzhagan2005@gmail.com
M. R. Joo · M. M. Rashid
Department of Civil Engineering, National Institute of Technology Srinagar, Srinagar 190006,
India
52 P. Anbazhagan et al.
1 Introduction and Study Area
Indo-Gangetic Basin, consisting of thick alluvial deposits, lies between longitude

77 E and 88 E and latitude 24 N and 30 N. It covers an area of around 2,50,000 km2
and encompasses densely populated Indian states viz. Bihar, Punjab, Haryana and
Uttar Pradesh as shown in Fig. 1. Its closeness to Himalayan boundary, presence of
thick sedimentary deposits and higher amplification of seismic waves makes the area
seismically vulnerable.
Several researchers in India reported high seismic risk in the region. Ambraseys
[1] reviewed the seismicity of North India during the early instrumentation period
1892–1915 and summarized evidences for 50 events. Khattri [2] also studied three
seismic gaps in the Himalayan Plate boundary which can lead to potential future
earthquakes.
In the recent past, researchers have made several advances in site response studies.
Anbazhagan et al. [3] used borehole data and synthetic ground motions to study
limited sites for 1999 Chamoli earthquake. Researchers [4–8] considered limited
locally recorded ground motions for site response studies and Kumar et al. [9] used
worldwide recorded ground motion for the site-response study irrespective of seismic
background of IGB.
Literature survey reveals that site response studies are mostly limited to 30 m
soil column depth using limited recorded motions. For the first time, our study has
attempted to carry out site response studies for different possible futuristic earth-
quakes with suitable input soil parameters. In the present work, 16 possible earth-
quakes with varying parameters were identified and subsequently used to generate
Fig. 1 Study area and earthquakes considered

Prediction of Future Surface PGA in the States … 53
synthetic ground motions for analysis. Contour maps illustrating variation of esti-
mated surface PGA for each individual earthquake were developed. Wide range of
PGA was observed corresponding to different earthquakes. Further each site exhib-
ited different behavior toward different earthquakes. Sites in Bihar reflect average
and maximum surface PGA 0.15 g and 0.68 g, Uttar Pradesh 0.10 g and 1.18 g,
Punjab and Haryana 0.12 g and 0.62 g respectively. On an average, the percentage
of sites hit by earthquakes in Bihar, Punjab and Haryana and UP are 38%, 35% and
37% respectively.
2 Simulation of Input Motions and Bedrock PGA
Synthetic ground motions are widely used for ground response studies, development
of attenuation relationships, mapping of seismic hazards and amplification studies.
To analyze 270 sites for identified seismic sources, ground motions at every site
corresponding to each scenario earthquake were considered. This was done using
EXSIM [10] taking into account seismotectonic parameters of the study area. The
magnitudes of the identified earthquakes range between Mw 7.5 to 9.0, originating in
and around IGB as depicted in Fig. 1. The detailed parameters of the earthquakes can
be found from Ref. [11] and their locations and magnitudes can also be referred from
Fig. 1. The ground motions used for analysis have varying parameters viz. frequency
content, acceleration and duration. The bedrock PGA ranged between 0.0005 and
0.2996 g in Bihar, 0.0001 and 0.5771 in UP, 0.0003 and 0.6509 g in Punjab and
Haryana. In whole IGB bedrock PGA varied from 0.0001 to 0.6509 g.
3 Input Soil Parameters
To analyze the sites with varying characteristics over IGB, different soil parameters
viz nature of soil, density, shear wave velocity profiles and depth of input motion
have been used as per provisions suggested by researchers. Shear wave velocity (Vs)
as described by [12] while considering representative density from Vs of each layer
[13], reliable depth level of input motion [14] and selected shear modulus reduction
and damping curves as suggested by [15, 16]. The representative curves suggested
have been therefore used to present site response studies of deep sites of IGB. In
case of rock or hard layer, EPRI curve [17] and Zhang curve [18] have been used
depending on whether Vs ≥ 800 m/s or Vs < 800 m/s respectively for deposits of
Quaternary type. For gravel sites with known particle size, Menq curve [19] has been
used otherwise Zhang et al. [18] for deep gravel profiles. Zhang Curve [18] has also
been used for deep sand deposits. For deep clay and silt sites, Darendeli curve [20]
has been used.
4 Site Response Analysis and Surface PGA
For a given motion, PGA is the highest value of horizontal acceleration obtained
from the accelerogram. Generally, horizontal accelerations describe ground motions
because of their natural connection to inertial forces. It has been observed that ground
motions with high peak accelerations are usually more damaging in nature. Site
response analysis establishes the impact on ground surface motion due to the soils
above the bedrock. Site response analysis finds its use in predicting ground surface
motions for development of design response spectra, estimating liquefaction hazards
and to measure earthquake induced forces. It requires the information of different
field parameters like type, thickness and density of soil layers, shear wave velocity
profiles, location of water table, depth of bedrock, shear modulus reduction and
damping ratios. While designing any structure, it becomes important to know about
the behavior of the soil column toward the incoming ground motions. The response
influences the level of shaking which in turn governs the damage caused to the
infrastructure in the area. To assess the site response, we simulated 16 earthquakes
for 58 sites of Bihar, 136 sites of UP and 76 sites for Punjab and Haryana. Only those
input motions which surpassed bedrock PGA 0.005 g were utilized in the study as
the input motions and with PGA below 0.005 g were found ineffective in causing
any significant damage to the infrastructure. 1D (one dimensional) nonlinear site
response studies were carried out using DEEPSOIL V7 [21] and the results obtained
are discussed in the subsequent section.
Seismic waves undergo amplification on traveling from bedrock to ground surface.

Hence, parameters viz. peak ground acceleration, peak spectral acceleration, duration
and frequency content associated with ground motions at surface differ from that at
bedrock level or any hard layer. These modifications in seismic motions are site
dependent and vary considerably from site to site. In present study, spatial variation
of surface PGA corresponding to each earthquake has been shown. Surface PGA
values due to each earthquake for UP, Bihar, Haryana and Punjab are summarized.
These states showed varied levels of PGA and this behavior can be correlated to the
magnitude, depth of fault, hypocentral distance, directivity, nature of soil deposits.
Also, the PGA varied within the states giving rise to different hazard levels. The state
wise analysis for future probable scenario earthquakes is briefly discussed below.
In Bihar, out of sixteen earthquakes considered, only nine have potential to
generate bedrock PGA > 0.005 g and have therefore been used in site response
analysis of 58 sites. The other seven earthquakes may be regarded having no or very
little effect in the state. The highest bedrock PGA observed is 0.30 g in the state.
The bedrock PGA is significantly amplified and goes as high as 0.68 g at surface.
The PGA distribution for individual earthquakes is mapped in Figs. 2, 3, 4, 5, 6, 7,
Fig. 2 Spatial variation of surface PGA for earthquake A
Fig. 3 Spatial variation of surface PGA for earthquake B
8, 9, 10, 11, 12, 13, 14, 15, 16 and 17. The surface PGA ranges from 0.007 to 0.68 g.
Average surface PGA corresponding to each of 9 earthquakes are 0.03, 0.27, 0.1,
0.04, 0.17, 0.06, 0.30, 0.06 and 0.03 g. Earthquakes labeled as C and N affect 100%
of the sites in Bihar and have a potential to generate surface PGA > 0.5 g at certain
sites. Around 77% of analyzed sites in Bihar show average surface PGA > 0.1 g. The
detailed information may be referred from Table 1.
Fig. 4 Spatial variation of surface PGA for earthquake C
Fig. 5 Spatial variation of surface PGA for earthquake D
For Punjab and Haryana, eight earthquakes as listed in Table 2 were found to
produce bedrock PGA > 0.005 g. The bedrock PGA varies between 0.0003 g and
0.6509 g. Ground motions (with PGA > 0.005 g) were inputted at sites in Punjab
and Haryana to study site effects and arrive at distribution of surface PGA. Spatial
variation due to 8 individual earthquakes can be referred from Figs. 2, 3, 4, 5, 6, 7, 8, 9,
10, 11, 12, 13, 14, 15, 16 and 17. Surface PGA also varies significantly for different
Fig. 6 Spatial variation of surface PGA for earthquake E
Fig. 7 Spatial variation of surface PGA for earthquake F
earthquakes and ranged in between 0.0011 g and 0.6238 g of surface PGA. The
average values of surface PGA in the states due to selected earthquakes are 0.1285,
0.1030, 0.2174, 0.0319, 0.1254, 0.0018, 0.0769 and 0.0983 g. Earthquakes labeled
as E, F and H respectively hit 97, 99 and 100% of sites in these states. Average PGA
due to sources E, F and G are 0.1030, 0.2174 and 0.1254 g respectively. Structures in
the area need specific provisions with respect to these seismic sources. About 65%
Fig. 8 Spatial variation of surface PGA for earthquake G
Fig. 9 Spatial variation of surface PGA for earthquake H
of sites analyzed in the region have average surface PGA > 0.1 g. Table 2 shows
summary of results for Punjab and Haryana region.
Based on the same criteria, sites in Uttar Pradesh have been analyzed for thir-
teen seismic sources and surface PGA due to each considered earthquake has been
evaluated. From the data obtained from DEEPSOIL program, spatial variation of
surface PGA for each earthquake has been presented in Figs. 2, 3, 4, 5, 6, 7, 8,
Fig. 10 Spatial variation of surface PGA for earthquake I
Fig. 11 Spatial variation of surface PGA for earthquake J
9, 10, 11, 12, 13, 14, 15, 16 and 17. A wide range of bedrock PGA is observed
varying from 0.0001 g to 0.5771 g. Overall surface PGA for the state varies from
0.0018 g to 1.1827 g. The average values for each earthquake are 0.1096, 0.0420,
0.1504, 0.0205, 0.0669, 0.0556, 0.0578, 0.0924, 0.0260, 0.0724, 0.0208, 0.1912 and
0.1836 g as listed in Table 3. The sites in Uttar Pradesh show average PGA values
lesser compared to Bihar, Punjab and Haryana and thus might be less prone to the
Fig. 12 Spatial variation of surface PGA for earthquake K
Fig. 13 Spatial variation of surface PGA for earthquake L
earthquake sources considered in this study. It is worth mentioning that surface PGA
as high as > 1.0 g is observed for earthquakes 1 and 14 for few sites. Overall, 37%
of analyzed sites show average surface values > 0.1 g.
Average surface PGA observed here is less as compared to PGA with respect
to individual earthquakes. Overall average of 0.12 g is observed for the whole IGB
which is quite different from state wise analysis. Average value of surface PGA
Fig. 14 Spatial variation of surface PGA for earthquake M
Fig. 15 Spatial variation of surface PGA for earthquake N
observed is as low as 0.014 g and as high as 0.32 g. An average value > 0.15 g is
observed at a number of sites in IGB and therefore requires certain consideration in
design of structures. The spatial variation may be referred from Fig. 18. Maximum
surface PGA observed for IGB at any site is as high as 1.18 g and its spatial variation
is shown in Fig. 19. A very smaller number of sites show maximum PGA > 0.7 g.
Fig. 16 Spatial variation of surface PGA for earthquake O
Fig. 17 Spatial variation of surface PGA for earthquake P
Majority of the sites show maximum PGA around 0.25 g. These maximum values
may be referred for design of highly important structures, subjected to public use or
national importance.
Table 1 Analysis of bedrock and surface PGA for Bihar

Earthquake Bedrock PGA(g) Surface PGA(g) Average surface Proportion (%) of
range range PGA(g) sites
A 0.0005–0.0184 0.0093–0.0626 0.0316 21
C 0.0225–0.2996 0.0837–0.6546 0.2739 100
D 0.0021–0.1022 0.0215–0.3112 0.1005 89
I 0.0019–0.0468 0.0126–0.1310 0.0389 58
J 0.0267–0.1910 0.0311–0.3783 0.1671 98
L 0.0023–0.0463 0.0166–0.1266 0.0589 84
N 0.0289–0.3415 0.0599–0.6804 0.2985 100
O 0.0012–0.0428 0.0182–0.1518 0.0553 46
P 0.0008–0.0086 0.0122–0.0440 0.0280 12
Table 2 Analysis of bedrock and surface PGA for Punjab and Haryana
Earthquake Bedrock PGA(g) Surface PGA(g) Average Surface Proportion (%) of
B 0.0026–0.1383 0.0300–0.4319 0.1285 85
E 0.0102–0.0574 0.0297–0.2481 0.1030 97
F 0.0140–0.6509 0.0655–0.6238 0.2174 99
G 0.0004–0.0076 0.0229–0.0441 0.0319 8
H 0.0066–0.3302 0.0269–0.5875 0.1254 100
J 0.0003–0.0020 0.0011–0.0035 0.0018 5
K 0.0056–0.1040 0.0156–0.3087 0.0769 92
M 0.0012–0.1232 0.0208–0.3584 0.0983 67
6 Conclusion
1D nonlinear site response analyses for futuristic scenario earthquakes have been
carried out at 270 sites covering UP, Punjab, Haryana and Bihar. Variation of esti-
mated surface PGA for individual earthquakes has been deliberated using contour
maps. Following are the main remarks concluded from the study:
• Wide range of bedrock PGA ranging from 0.0003 to 0.6509 g for Punjab and
Haryana, 0.0005 to 0.2996 g for Bihar and 0.0001 to 0.5771 g for UP has been
analyzed.
• Sites in Bihar reflect average and maximum surface PGA 0.15 and 0.68 g, Uttar
Pradesh 0.10 and 1.18 g, Punjab and Haryana 0.12 and 0.62 g respectively.
• All the earthquakes hit varied proportion of sites in each state. On an average, the
percentage of sites hit by earthquakes in Bihar, Punjab and Haryana and UP are
38, 35 and 37 respectively.
Table 3 Analysis of bedrock and surface PGA for Uttar Pradesh

Earthquake Bedrock PGA(g) Surface PGA(g) Average surface Proportion (%) of
A 0.0017–0.5771 0.0069–1.1593 0.1096 94
B 0.0001–0.0156 0.0266–0.0789 0.0420 4
C 0.0006–0.2534 0.0178–0.5983 0.1504 74
D 0.0001–0.0067 0.0029–0.0291 0.0205 3
E 0.0005–0.0522 0.0100–0.1919 0.0669 38
F 0.0002–0.0744 0.0148–0.1791 0.0556 19
G 0.0004–0.0466 0.0139–0.1846 0.0578 49
H 0.0010–0.3438 0.0093–0.4008 0.0924 67
I 0.0005–0.0561 0.0018–0.1319 0.0260 68
J 0.0013–0.1373 0.0116–0.2274 0.0724 78
K 0.0002–0.0322 0.0038–0.0684 0.0208 9
L 0.0002–0.0103 0.0137–0.0250 0.0192 8
N 0.0008–0.5105 0.0187–1.1827 0.1836 76
Fig. 18 Spatial variation of average surface PGA
• Earthquakes labeled as C, J and N have relatively higher impact on Bihar as

they hit more than 90% of the sites, Punjab and Haryana show this response for
earthquakes E, F, H and K while UP for A.
• Some of the earthquakes under consideration hit very little area (<10% of sites)
viz. G and J for Punjab and Haryana, B, D, K and L for UP.
Fig. 19 Spatial variation of maximum surface PGA
• Higher surface PGA was observed for nearby seismic sources and have there-
fore higher expected damage level. These should be given due considerations
while designing new structures in the area.
• PGA obtained from the study for futuristic earthquakes is much more than current
seismic code of IS1893 [22].
Acknowledgements The authors are thankful for funding supported by the Science and Engi-
neering Research Board (SERB), Department of Science and Technology [grant numbers
SERB/F/162/2015-2016].
References
1. Ambraseys N (2000) Reappraisal of North-Indian earthquakes at the turn of the 20th century.
Curr Sci 79:101–106
2. Khattri KN (1987) Great earthquakes, seismicity gaps and potential for earthquakes along the
Himalayan plate boundary. Tectonophysics 38:79–92
3. Anbazhagan P, Kumar A, Sitharam TG (2010) Amplification factor from intensity map and
site response analysis for the soil sites during 1999 Chamoli earthquake. In: Proceedings of
the 3rd Indian young geotechnical engineers conference 2010. New Delhi, pp 311–316
4. Mahajan AK, Sporry RJ, Champati PK, Ranjan RR, Slob S, Van WS (2007) Methodology for
site-response studies using multi-channel analysis of surface wave technique in Dehradun city.
Curr Sci 92(7):945–955
5. Govindraju L, Bhattacharya S (2008) Site Response studies for Seismic hazard analysis for
Kolkata city. In: Proceedings of the 12th international conference of the international association
for computer methods and advances in geomechanics 2008, pp 2899–2907
6. Phanikanth VS, Choudhury D, Reddy GR (2011) Equivalent-linear seismic ground response

analysis of some typical sites in Mumbai. Geotech Geol Eng 29(6):1109–1126
7. Kumar A, Anbazhagan P, Sitharam TG (2013) Seismic hazard analysis of Lucknow considering
local and active seismic gaps. Nat Hazards 69:327
8. Jishnu RB, Naik SP, Patra NR, Malik JN (2013) Ground response analysis of Kanpur soil along
the Indo-Gangetic Plains. Soil Dyn Earthquake Eng 51:47–57
9. Kumar A, Baro O, Harinarayan N (2016) Obtaining the surface PGA from site response analyses
based on globally recorded ground motions and matching with the codal values. Nat Hazards
81:543–572
10. Motazedian D, Atkinson GM (2005) Stochastic finite-fault modeling based on a dynamic corner
frequency. Bull Seismol Soc Am 95:995–1010
11. Anbazhagan P, Joo MR, Rashid MM, Al-Arifi NSN (2021) Prediction of different depth amplifi-
cations of deep soil sites for potential scenario earthquakes. Nat Hazards (Accepted, Submission
Id: NHAZ-D-20-00073)
12. Bajaj K, Anbazhagan P (2019) Seismic site classification and correlation between Vs and
SPT-N for deep soil sites in Indo-Gangetic Basin. J Appl Geophys 163:55–72
13. Anbazhagan P, Uday A, Moustafa SSR, Al-Arifi NSN (2016) Correlation of densities with
shear wave velocities & SPT N values. J Geophys Eng 13:320–341
14. Bajaj K, Anbazhagan P (2019) Comprehensive amplification estimation of the Indo Gangetic
Basin deep soil sites in the seismically active area. Soil Dyn Earthquake Eng 127:105855
15. Anbazhagan P, Prabhakaran A, Madhura H, Moustafa SSR, Al-Arifi NSN (2017) Selection
of representative shear modulus reduction and damping curves for rock, gravel and sand sites
from the KiK-Net downhole array. Nat Hazards 88(3):1741–1746
16. Bajaj K, Anbazhagan P (2019) Identification of shear modulus reduction and damping curve
for deep and shallow sites: Kik-net data. J Earthquake Eng.https://doi.org/10.1080/13632469.
2019.1643807
17. Electric Power Research Institute (EPRI) (1993) Guidelines for site specific ground motions.
Palo Alto, California, November, TR-102293
18. Zhang J, Andrus R, Juang CH (2005) Normalized shear modulus and material damping ratio
relationships. J Geotech Geoenviron Eng ASCE 131:453–464
19. Menq FY (2003) Dynamic properties of sandy and gravelly soils. PhD thesis, Department of
Civil Engineering, University of Texas, Austin, TX
20. Darendeli MB (2001) Development of a new family of normalized modulus reduction and
material damping curves. PhD dissertation, University of Texas at Austin, Austin, Texas
21. Hashash YMA, Musgrove MI, Harmon JA, Ilhan O, Groholski DR, Phillips CA, Park D (2017)
DEEPSOIL 7.0, user manual
22. BIS 1893 (Part 1) (2016) Indian standard criteria for earthquake resistant design of structures.
Bureau of Indian Standards, New Delhi, Sixth revision
Synthetic Ground Motion Simulation
for Varanasi City
Manjari Singh, S. K. Duggal, V. P. Singh, and Keshav Kumar Sharma
Abstract Varanasi (latitude 25°28 N and longitude 82°96 E), the cultural capital
of India, is presently clustered with a maze of ancient narrow lanes (Gullies) and
old buildings. Being a sacred city, it is not well planned and structured, due to lack
of adherence to earthquake-resistant building design philosophies and techniques.
Consequently, even a smaller magnitude of an earthquake can cause a considerable
loss. The city is also near to Faizabad ridge, which has been seismically seden-
tary for last 300 years. Due to unavailability of earthquake ground motion (G.M.)
records in this region, it is necessary to simulate G.M. based on regional seismic
data. In this regard, the stochastic approach has been adopted for the synthesis of
G.M. at bedrock level. An EXSIM methodology has been used in this study for
synthesis of strong G.M. for various identified faults (Allahabad Fault, Azamgarh
Fault, Gorakhpur Fault, Deoria Fault, Lucknow Fault, Siwan Fault, Shajhanpur Fault
and Great Boundary Fault) around the city. Various stress drops 70, 100, 125, 150,
175 and 200 bars have been taken for simulations to account for uncertainty in stress
drop. Acceleration time histories due to various faults for Varanasi city has been
simulated and plotted. The maximum PGA estimated was 0.078 g for Azamgarh
Fault at 200 bar among all the faults around the city. Further Response Spectra has
been plotted for stress drop (70–200 bar).
Keywords Varanasi city · Stochastic method · EXSIM · Finite fault · Stress drop
1 Introduction
During the movement of tectonic plates, accrued strain energy gets released in the
form of seismic waves. All of the possible natural hazards, earthquakes are the most
severe calamitous catastrophe for the society with respect to the loss of lives and
M. Singh (B) · S. K. Duggal · V. P. Singh

MNNIT Allahabad, Prayagraj 211004, India
e-mail: majarisingh.2508@gmail.com
K. K. Sharma
NIT Jamshedpur, Jamshedpur 831014, India
68 M. Singh et al.
assets. Earthquakes are ineluctable, but their impact can be reduced appropriately by
considering the soil properties, earthquake type, magnitude and area of impact during
design of infrastructures. The Himalayas spans from J&K to Arunachal pradesh,
Meghalaya, Manipur and Tripura and expanding to Nepal and China, which are
the hub of the most severe earthquakes in the history. The most dreadful Gorkha
earthquake 2015 that triggered in Nepal on April 25, 2015 and took the life of about
9000 people.
Improved structural design has a significant role in minimizing deaths and damage
to property caused by earthquakes. For the design of earthquake-resistant structures,
estimation of the level of ground shaking in terms of G.M. parameters: amplitude
(Peak Ground Acceleration, PGA), frequency content (Fourier acceleration spectra,
response spectra) and duration of motion is required, which reflects Characteristics
of G.M.
Characteristics of G.M. are necessary to comprehend the peril arising from the
event. For a specific site, design G.M. depends upon many factors which include site
location with respect to potential seismic source, nature of rupture and seismicity of
source, travel path between source and site as well as also the local site effects.
If an array of strong-motion networks would exist in the area, then the G.M.
characteristics of the near field can be better understood. However, in those regions
where is a paucity of strong G.M. data, as well as seismicity rates, are also less,
earthquake simulations may then produce synthetic data based on regional seismic
characteristics [1, 2].
G.M. simulation had begun by the pioneering work of Hartzell [3] and Irikura
[4] about three decades ago. Afterwards, various researchers have come up with
different simulation techniques which can be summarized in the following categories:
Parameter-Prediction, Stochastic Process, Physics and Hybrid Methods.
Parameter prediction method requires a simple parameter to simulate G.M. (such
as PGA, PGV and Sa (T)) at multiple periods [5–7]. It offers less computing time, rela-
tively simple, but it is less useful in those regions where G.M. data is rarely available.
Whereas, stochastic process-based, physics-based and hybrid methods simulate G.M.
time history. Housner [8] was first to introduce stochastic process-based model by the
use of white noise to represent earthquake G.M. Physics-based method encompasses
various approaches: Finite fault method, Kinematic method and Dynamic method.
Finite fault method [9, 10] considers an earthquake as the sum of many smaller
earthquakes by first assuming a rupture geometry and then modeling the earthquake
as a distribution of smaller subevents on the rupture plane. Kinematic method [11,
12] consists of two steps one is a simulation of slip distribution, and another one is
wave propagation. In the dynamic method [13, 14], pre-defined stress conditions and
physics laws are simultaneously used to solve slip distribution and wave propagation.
The dynamic method is comparatively more complex, and it is not generally adopted
in the engineering field. On the other hand, hybrid method [15] is a combination
of previously discussed methods which generate more realistic time history and
convenient in the analysis of tall building that is excited at multiple frequencies.
In the present study, Extended Finite-Fault Simulation (EXSIM) software, based
on stochastic finite fault approach to simulate G.M. database for Varanasi City has
Synthetic Ground Motion Simulation for Varanasi City 69
been used. In this city, there is not enough real G.M. data so, EXSIM is used to
develop G.M. simulation for the city.
2 Seismotectonic Background and Seismicity of the Study

Area
The Indo Gangetic Basin (IGB) is the largest newer alluvial sedimentary basin
that comprises approx. 200 million population. IGB comprises the following major
trending fault: NNE-SSW to NE-SW, NNW-SSE to NE-SE and EW. Among them,
EW trending fault lies in Azamgarh-Gorakpur area and describes the MirganjGraben.
The E-W elongated IGB was formed due to the uplift of Himalayas after the collision
of Indian-Asian plates [16, 17]. The area includes several faults/ridges [18]. Several
significant ridges in the IGB increase the seismic activity of the basements. The
Faizabad and Munger ridge, near to Varanasi, is surrounded by many faults which
are the extension of Bundelkhand and Satpura massif. The IGB is comparatively
moderately seismic from the Himalayas [19].
Varanasi, one of the heritage city of India, belongs to zone III in the seismic
map of India [20]. The city inevitably experienced shocks and had even suffered
damages in past when strong earthquakes struck the neighboring vicinities of the
Himalayan territory. Varanasi has also felt shocks of several past earthquakes that
occurred in Uttar Pradesh such as 1925 Raibarielly earthquake M w 6.0, 1956 Buland-
shahar earthquake M w 6.2, 1965 Gorakhpur earthquake M w 5.7 and 1966 Moradabad
earthquake M w 5.8 earthquakes; Countrywide 1916 Dharchulla earthquake M w 7.5,
1945 Uttarakhand earthquake M w 6.5, 1991 Uttarkashi earthquake M w 6.8 and 1997
Jabalpur earthquake M w 6.0 and earthquakes like 1934 Nepal-Bihar earthquake M w
8.4, 2015 Nepal earthquake M w 7.8 from the neighboring country.
The main issue for the designer is the unavailability of systematic data of earlier
earthquakes. Therefore the construction of seismic map for the region, considering
the possibility of occurrence of major earthquake events in recent years, is required.
It significantly includes sources like Main Central (MC) and Main Boundary (MB)
Thrust that is 400 km distant from Varanasi city and are solitary earthquake-prone
areas in the world.
Faizabad ridge, near to the city, is quiescent over more than 300 years as mentioned
by the Disaster Risk Management Programme organized by the Home Affairs
Ministry in collaboration with UN Development Programme (UNDP). Due to the
huge seismic gap, the ridge is substantially stressed and has the capacity to produce
higher magnitude of earthquake in upcoming years [21]. The Indian plate is subducted
underneath the Asian plate by 5.25 m can produce an earthquake of M w 8.0 stated
by Earthquake Mitigation Department of Uttar Pradesh [21].
In order to identify the potential seismic sources, a vast literature survey has been
carried out to observe the presence of linear, aerial and point sources. Only linear sub-
surface faults are observed as a potential seismic source around the study area and
70 M. Singh et al.
Fig. 1 Seismicity around the city
are defined in Seismotectonic Atlas of India. Presence of numerous faults around the
study area increases the seismic hazard of Varanasi city. In the present study, faults
around Varanasi between latitude 25.28° N–29.28° N and longitude 78.96° E–86.96°
E (around 450 km) are plotted using MATLAB and are shown in Fig. 1. Further, the
fault list of the area has been prepared, and various surface and sub-surface faults
are listed and are tabulated in Table 1. The details of these faults like strike, dip, and
mechanism (thrust/reverse, normal, and strike-slip) have been gathered from various
publications like Dasgupta et al. [22], Singh et al. [23], and Kayal [24]. In this study,
only faults having length more than 5.0 km are considered.
3 Methodology
The basic approach in G.M. simulation is to generate reliable synthetic data by

modeling the seismic source mechanisms and properties of regional wave propaga-
tion. Hanks and McGuire [25] developed a basis of stochastic technique that observed
high-frequency strong G.M. can be represented by finite duration band-limited white
Gaussian noise. Boore [26] was the first who defined Band limited white-noise model
in which Gaussian white noise of limited frequency band represents high-frequency
motions of earthquakes.
Further, Beresnev and Atkinson [27] has used this approach to model the finite
faults. The model has been discretized a rectangular fault plane into smaller sub
Table 1 Details of seismic sources
Faults names Faults length (km) Faults position Strike (Deg.) Dip (Deg.) Latitude Longitude Mmax (as per WC
Initial Final Initial Final 1994)
F1 (Allahabad Fault) 57 Subsurface 166 50 25.382 25.873 81.9 81.745 6.3

F2 (Azamgarh Fault) 158 Subsurface 87 50 26 26.074 82.19 83.778 6.9
F3 (Deoria Fault) 106 Subsurface 87 50 26.48 26.53 82.86 83.93 6.6
Synthetic Ground Motion Simulation for Varanasi City
F4 (Gorakhpur Fault) 118 Subsurface 51 50 26.482 27.150 82.8 83.722 6.7

F5 (Lucknow Fault) 126 Subsurface 25 50 26.78 27.808 80.67 81.207 6.8
F6 (Siwan Fault) 94 Subsurface 64 60 26.636 25.945 84.09 84.64 6.6
F8 (Shajhanpur Fault) 130 Subsurface 106 50 28.11 27.788 79.17 80.445 6.8
71
72 M. Singh et al.
faults and considered as a point source and summing the contributions of each sub
fault. Motazedian and Atkinson [10] have used the term dynamic corner frequency
(DCF).
In this concept, the total energy emitted from the fault has been conserved regard-
less of the selected sub fault size. In this paper, the EXSIM program (Extended
Earthquake Fault Simulation Program) has been used for earthquake simulation,
which is based on DCF using stochastic finite fault algorithm [10]. It is an updated
form of Boore’s (1983) point source stochastic model [26].
In this methodology, the Fourier amplitude spectrum of G.M. has been obtained
from point source seismological model. Acceleration Fourier amplitude spectrum
Ai ( f ) of the ith sub fault is expressed as follows (Eq. 1):
√
M0i Hi (2π f )2 2Rθ∅ − πVf RQi
Ai ( f ) = Ge s F( f )e−π f k0 (1)
1 + ( f / f 0i (t))2 4πρVs3
where G denotes the geometric attenuation, Mi is the seismic moment of ith sub
fault, Vs is shear wave velocity, Ri is the distance from sub fault to the site and Q is
the quality factor. Hi is a scaling factor and f 0i (t) defines DCF [10]. F( f ) denotes
site amplification due to earthquake wave propagation from the source region to the
surface. The term e−π f k0 is a high cut filter which takes care of the rapid spectral
decay at high frequencies [28]. Rθ∅ refers to radiation coefficient averaged over
an appropriate range of azimuths and take-off angles and ρ is crust density at focal-
depth. Moment of the ith sub fault is determined from slip distribution is defined as
(Eq. 2):
M0 Di
M0i = N
(2)
i=1 Di
where, M0 -total seismic moment of all sub faults, Di -average final slip acting on the
ith sub fault. The DCF f 0i (t), can be defined from the following equation (Eq. 3)
which relates the terms: the seismic-moment M0 and the stress drop σ :
(1/3)
σ
f 0i (t) = 4.9 × 106 (N R (t))−(1/3) N −1/3 Vs (3)
M0
where, N R (t)-a cumulative number of ruptured sub faults at (t) time. Scaling factor
(Hi ), can be expressed as (Eq. 4).
⎛ 2 ⎞(1/2)
f2
⎜ f 1+( f / f 0 )2 ⎟
Hi = ⎝ N 2 ⎠ (4)
f2
f 1+( f / f 0i (t))2
where, f 0 defines the corner frequency at the end of the rupture, and can be obtained
by putting N R (t) = N in Eq. 3. In Eq. (4), the summation is done from 0 to the
highest frequency exist in the signal.
In this study, the seismic model has been imposed in the time range by using
numerical simulation for which Boore (1983) [26] recommended the following three
steps:
(1) Simulated strong motion duration would be considered equal to the Gaussian
stationary white noise sample of length [1]. In Eq. 5. the terms f c and r denotes
the corner frequency and hypocentral distance, respectively.
T = 1/ f c + 0.05r (5)
(2) A non-stationary modulating function recommended by Saragoni [29] is multi-

plied by this sample. After the simulation, the sample is Fourier transformed
into the frequency domain. Further, it is normalized using rootmean squares
value and then multiplied with Eq. 1. Afterwards, it is converted back into time
range to acquire a sample of sub-surface acceleration time histories.
(3) Summing up all simulated time histories for each sub faults of time delay (t i )
for generating final G.M. acceleration A(t) and expressed as (in Eq. 6):

N
A(t) = Ai (t + ti ) (6)
i=1
Due to the modification between bedrock and A-type sites being a linear problem
as well as 1-dimensional and hence these sites augmentation can be done by using
the quarter wavelength method of Boore [30] directly. Database of kappa factors
have been reported by Chandler et al. [31] and has given an equation (Eq. 7) for
determining the kappa values (k0 ) from the average shear wave in top 30 m of soil
(V s30 ).
0.057
k0 = 0.8
− 0.02 (7)
Vs30
For rock level G.M. simulation, kappa factor calculated as 0.057 for the soft soil
in Varanasi. In this study, Table 2 shows the parameters used in the simulation.
The maximum magnitude of the earthquake Mw for all faults, as shown in Fig. 1
have been estimated by Wells and Coppersmith (WC) [35] relation (Eq. 8). Under
the worldwide dataset, Mark [36] has suggested that the fault rupture length can be
considered as (1/3–1/2) of the total fault length. Though taking such large sub-surface
rupture length generates huge moment magnitude and doesn’t fit with the historical
earthquake records.
Mw = 4.38 + 1.49 ∗ log(R L D)(Allrupturetypes) (8)

74 M. Singh et al.
Table 2 Parameters used for

Parameters Value used in this study
simulation
V s shear-wave velocity and ρ 3500 m/s, 2.9 g/cc [32, 33]
density at source
σ Stress drops 70, 100, 125, 150, 175, 200
bars [24]
Q( f ) quality factor 142 f 1.04 [34]
G geometric attenuation 1/r (for r ≤ 100 km);
√
1/(100r ) (for r > 100 km)
[32]
k0 kappa factor 0.057 for deep profiles (Eq. 7)
σ reference stress drop 100 bars
Pulsing percentage and Sub 50% and 0.25 km × 0.25 km
fault size
Crustal amplification ENA Hard rock amplification
[1]
where Mw = moment magnitude; R L D = length of subsurface rupture (km).
3.1 Validation
2015 Nepal earthquake (also known as Gorkha earthquake) has occurred in the
western part of Nepal with a magnitude M w 7.8 toward North-West of Kathmandu
[37] along the MFT active fault. The main shock of the earthquake was located at
the Barpak village (Longitude 84.708° E and Latitude 28.147° N at shallow Depth
15 km) in the Gorkha district of Nepal. The earthquake epicenter was at Lamjung,
around 80 km north-west of Kathmandu [37]. The characteristics of fault like strike,
dip and depth have been taken from the USGS database [37]. Peak Ground Accel-
eration (PGA), event location and fault distance have been taken from the Center
for Engineering Strong Motion Data CESMD (www.strongmotioncenter.org). The
rupture length and width of the fault have been calculated by using empirical rela-
tionships given by WC [35]. The square grid has been used to categorize sub fault
size configuration. The primary source parameters of Nepal earthquake, shown in
Table 3.
The PGA values are recorded by five strong-motion accelerometers [CESMD]
located at KTP, TVU, PTN, and THM operated by HU-TU (Hokkaido University
and Tribhuvan University, Nepal). Another strong motion accelerometer is located
at KATNP operated by USGS. The recorded PGA at respective stations is provided
in Table 4.
Table 3 Source parameters of Nepal earthquake

Parameters Value
Mw 7.8
Mo (Nm) 6.623 × 1020
Strike (f) 295°
Dip (δ) 10°
Rake (λ) 101°
Focal depth (km) 15
Length of rupture (L) (km) 221
Width of rupture (W) (km) 45
Sub fault length (dl) (km) 5
Sub fault width (dw) (km) 5
Rupture velocity (km/s) 0.8 β (β-fault Shearwave velocity)
Table 4 Recorded peak ground acceleration motion

Station number Station name North latitude (°) East longitude (°) PGA (g)
KATNP (USGS) Kanti path, Kathmandu, 27.712 85.316 0.163
Nepal
KTP (HOKU) Municipality office, 27.682 85.272 0.260
Kirtipur
TVU(HOKU) Dep. Geology, 27.681 85.288 0.234
Tribhuvanuniv., Kirtipur
PTN (HOKU) Pulchowk, 27.681 85.318 0.154
Tribhuvanuniv., Patan
Source www.strongmotioncenter.org
3.1.1 Model Input Parameters
The G.M. has been simulated from Stochastic Finite-Fault Model for A-type site
(hard rock site) and estimated PGA from simulation compared with recorded PGA
at KTP station. Site amplification has been ignored since the simulation is carried
out for the A-type site. The Vs and density of hard rock site is taken as 2.97 km/s
and 2600 kg/m3 respectively [38]. Kappa factor k0 = 0.005s at hard rock level [38].
Stress drop and pulsing area were taken as 160 bar and 50% (www.nset.org.np/eq2
015/intensity_maps.php).
√ Geometrical spreading G is taken as 1/r (for r ≤ 100 km);
1/(100r ) (for r > 100 km) [32]. Q = 500 f 1.04 is used as a quality factor [39]. The
estimated PGA is 0.248 g as shown in Fig. 2 whereas recorded PGA at KTP station
is 0.260 g.
For Azamgarh fault, moment magnitude Mw is determined using the WC [35]
relation (Eq. 8) with sub-surface rupture length of fault as 158 km, it results in a
magnitude of 6.8.
76 M. Singh et al.
Fig. 2 Estimated PGA
Table 5 Simulated peak ground acceleration

Stress Peak ground acceleration (g)
level Allahabad Azamgarh Deoria Gorakhpur Lucknow Shahjahan Siwan
(bar) fault fault fault fault fault fault fault
70 0.0091 0.0254 0.0135 0.0172 0.0043 0.0008 0.0085
100 0.0111 0.0326 0.0157 0.0175 0.0047 0.0009 0.0084
125 0.0160 0.0576 0.0180 0.0157 0.0064 0.0014 0.0112
150 0.0168 0.0737 0.0078 0.0244 0.0059 0.0015 0.0079
175 0.0201 0.0506 0.0231 0.0297 0.0086 0.0021 0.0109
200 0.0211 0.0783 0.0275 0.0288 0.0077 0.0017 0.0136
Acceleration time histories due to all considered faults near Varanasi city have
been simulated, for different stress drops (70–200 bars) and effective simulations
are presented in Table 5. The table presents the maximum PGA values with corre-
sponding stress drop for different faults. It was observed that Azamgarh fault gives
the maximum PGA for all stress drop. The maximum PGA estimated was 0.078 g
for Azamgarh Fault at 200 bar, and the minimum PGA was 0.0008 g at 70 bar for
Shahjahan Fault, among all the faults around the city. Estimated acceleration time
history for Azamgarh fault for all stress drop has been shown in Fig. 3. Also, for
the same critical Azamgarh fault response spectra have been plotted and is shown in
Fig. 4.
5 Conclusions
Varanasi city lies in IGB, which is close to the seismically active Himalayas, which
is vulnerable to great earthquakes, surrounded by numerous inhomogeneities in the
form of faults/ridges. The city is near to Faizabad ridge, which is greatly stressed
due to nonoccurrence of any seismic activity for a prolonged duration. Each seismic
Fig. 3 Simulated acceleration time histories for Azamgarh fault
Fig. 4 Input response spectra for different stress drop (70–200 Bar)
78 M. Singh et al.
source has a capacity to generate massive earthquakes in future. As the absence of

recorded G.M. data of past earthquakes for the city, so G.M. has been simulated for
all the identified faults within 450 km using stochastic finite-fault models for various
stress drops (70–200 bar). Among all the observed faults, Azamgarh fault yielded
maximum PGA (0.078 g), and Shahjahan Fault gives minimum PGA (0.0008 g) at all
stress drops for Varanasi city. Furthermore, simulation procedure has been validated
with recorded PGA of April 25, 2015 Nepal earthquake at KTP station.
References
1. Atkinson GM, Boore DM (2006) Earthquake ground-motion prediction equations for eastern
North America. Bull Seismol Soc Am 96(6):2181–2205
2. Nath SK, Raj A, Thingbaijam KKS, Kumar A (2009) G.M. synthesis and seismic scenario in
Guwahati city—a stochastic approach. Seismolo Res Lett 80(2):233–242
3. Hartzell SH (1978) Earthquake aftershocks as Green’s functions. Geophys Res Lett 5(1):1–4
4. Irikura K (1986) Prediction of strong acceleration motion using empirical Green’s function.
In: Proceedings of 11th Japan earthquake engineering symposium, vol 151, pp 151–156
5. Esteva L, Rosenblueth E (1964) Espectros de temblores a distancias moderadas y grandes.
Boletin Sociedad Mexicana de Ingenieria Sesmica 2:1–18 In Spanish
6. Abrahamson NA, Shedlock KM (1997) Overview. Seismol Res Lett 68(1):9–23
7. Baker JW, Jayaram N (2008) Correlation of spectral acceleration values from NGA G.M.
models. Earthq Spectra 24(1):299–317
8. Housner GW (1947) Characteristics of strong-motion earthquakes. Bull Seismol Soc Am
37(1):19–31
9. Joyner WB, Boore DM (1986) On simulating large earthquakes by Green’s-function addition of
smaller earthquakes. Earthquake source mechanics. American Geophysical Union Monograph
37, pp 269—274
10. Motazedian D, Atkinson GM (2005) Stochastic finite-fault modeling based on a dynamic corner
frequency. Bull Seismol Soc Am 95(3):995–1010
11. Brune JN (1970) Tectonic stress and the spectra of seismic shear waves from earthquakes. J
Geophys Res 75(26):4997–5009
12. Hartzell S, Harmsen S, Frankel A, Larsen S (1999) Calculation of broadband time histories
of G.M.: comparison of methods and validation using strong-G.M. from the 1994 Northridge
earthquake. Bull Seismol Soc Am 89(6):1484–1504
13. Olsen KB, Madariaga R, Archuleta RJ (1997) Three-dimensional dynamic simulation of the
1992 Landers earthquake. Science 278(5339):834–838
14. Kozdon JE, Dunham EM (2013) Rupture to the trench: dynamic rupture simulations of the 11
March 2011 Tohoku earthquake. Bull Seismol Soc Am 103(2B):1275–1289
15. Mai PM, Imperatori W, Olsen KB (2010) Hybrid broadband ground-motion simulations:
combining long-period deterministic synthetics with high-frequency multiple S-to-S backscat-
tering. Bull Seismol Soc Am 100(5A):2124–2142
16. Dewey JF, Bird JM (1970) Mountain belts and the new global tectonics. J Geophys Res
75(14):2625–2647
17. Sinha R, Tandon SK, Gibling MR, Bhattacharjee PS, Dasgupta AS (2005) Late quaternary
geology and alluvial stratigraphy of the Ganga basin. Himalayan Geol 26(1):223–240
18. Valdiya KS (1976) Himalayan transverse faults and their parallelism with subsurface structures
of north Indian plains. Tectonophysics 32:352–386
19. Quittmeyer RC, Jacob KH (1979) Historical and modern seismicity of Pakistan, Afghanistan,
northwestern India, and southeastern Iran. Bull Seismol Soc Am 69(3):773–823
20. IS 1893 (2016) Criteria for earthquake resistant design of structures, part 1—general provisions
and buildings (Fifth Revision). Bureau of Indian Standards, New Delhi
21. Nadeshda TNN (2004) Lucknow is on earthquake list. Times of India. https://timesofindia.ind
iatimes.com/city/lucknow/Lucknow-isonearthquakelist/articleshow/679471.cms. Accessed 17
June 2018
22. Dasgupta et al (2000) Seismotectonic atlas of India and its environs. Geological Survey of
India, Kolkata
23. Singh SK, Garcia D, Pacheco JF, Valenzuela R, Bansal BK, Dattatrayam RS (2004) Q of the
Indian shield. Bull Seismol Soc Am 94(4):1564–1570
24. Kayal JR (2008) Microearthquake seismology and seismotectonics of South Asia, 1st edn.
Springer
25. Hanks TC, McGuire RK (1981) The character of high-frequency strong G.M. Bull Seismol
Soc Am 71(6):2071–2095
26. Boore DM (1983) Stochastic simulation of high-frequency G.M.s based on seismological
models of the radiated spectra. Bull Seismol Soc Am 73(6A):1865–1894
27. Beresnev IA, Atkinson GM (1997) Modeling finite-fault radiation from the ω n spectrum. Bull
Seismol Soc Am 87(1):67–84
28. Anderson JG, Hough SE (1984) A model for the shape of the Fourier amplitude spectrum of
acceleration at high frequencies. Bull Seismol Soc Am 74(5):1969–1993
29. Rodolfo Saragoni G, Hart GC (1973) Simulation of artificial earthquakes. Earthq Eng Struct
Dyn 2(3):249–267
30. Boore DM, Joyner WB, Fumal TE (1997) Equations for estimating horizontal response spectra
and peak acceleration from western North American earthquakes: a summary of recent work.
Seismol Res Lett 68(1):128–153
31. Chandler AM, Lam NTK, Tsang HH (2006) Near-surface attenuation modelling based on rock
shear-wave velocity profile. Soil Dyn Earthq Eng 26(11):1004–1014
32. Singh SK, Dattatrayam RS, Shapiro NM, Mandal P, Pacheco JF, Midha RK (1999) Crustal and
upper mantle structure of Peninsular India and source parameters of the 21 May 1997, Jabalpur
earthquake (Mw=5.8): results from a new regional broadband network. Bull Seismol Soc Am
89(6):1631–1641
33. Mitra S, Priestley K, Bhattacharyya AK, Gaur VK (2005) Crustal structure and earthquake
focal depths beneath northeastern India and southern Tibet. Geophys J Int 160(1):227–248
34. Mohanty WK, Prakash R, Suresh G, Shukla AK, Walling MY, Srivastava JP (2009) Estimation
of coda wave attenuation for the National Capital region, Delhi, India using local earthquakes
Pure. Appl Geophys 166:429–449
35. Wells DL, Coppersmith KJ (1994) New empirical relationships among magnitude, rupture
length, rupture width, rupture area, and surface displacement. Bull Seismol Soc Am 84(4):974–
1002
36. Mark RK (1977) Application of linear statistical models of earthquake magnitude versus fault
length in estimating maximum expectable earthquakes. Geology 5(8):464–466
80 M. Singh et al.
37. U.S. Geological Survey (2018) Earthquake facts and statistics. https://earthquake.usgs.gov/ear
thquakes/eventpage/us20002926#general_region. Accessed 25 Mar 2018
38. Khattri KN, Yu G, Anderson JG, Brune JN, Zeng Y (1994) Seismic hazard estimation using
modelling of earthquake strong G.M.s: a brief analysis of 1991 Uttarkashi earthquake Himalaya
and prognostication for a great earthquake in the region. Curr Sci 343–353
39. Mitra S, Paul H, Kumar A, Singh SK, Dey S, Powali D (2015) The 25 April 2015 Nepal
earthquake and its aftershocks. Curr Sci 108(10):1938–1943
Dynamic Study of Existing Structure
Influenced by Adjacent Deep Excavation
L. Geetha, M. N. Hegde, and M. Mohammed Ayaz
Abstract Rapid growth and expansion in metropolitan cities necessitate the

construction of buildings very close to existing structures. Whenever there is an
excavation made adjacent to structure, the soil below the footings gets disturbed
due to the lateral escape/displacement of soil and subsidence of soil, and hence a
decrease in the bearing capacity of the soil around it. It is therefore required to
assess the impact of construction on the nearby existing structures. The main aim of
the study is to identify the variation in the dynamic properties of an existing frame
structure adjacent to deep excavation and to find the response of the structure under
seismic excitation. The analysis of structure without adjacent excavation is carried
out and is treated as reference for six other cases considered. For three cases, the
distance of excavation from existing frame structure is kept constant (as 0.5 m) and
the depth of excavation is varied (1.5, 3, and 4.5 m). In other three cases, the depth of
excavation is kept constant as 3 m and the distance of excavation from the existing
frame is varied as 1, 2, and 3 m. In all the cases the width of excavation is kept
constant as 3 m. The finite element modeling and analyses are carried out using soft-
ware. The parameters studied are displacement, stress distribution, frequency, mode
shapes, and amplitudes.
Keywords Deep excavation · Adjacent excavation · Soil loosening due to

excavation
L. Geetha (B) · M. N. Hegde

Department of Civil Engineering, Dr. Ambedkar Institute of Technology, Bengaluru 560056, India
e-mail: itsgeethashiv@gmail.com
M. Mohammed Ayaz
Department of Civil Engineering, Dayananda Sagar College of Engineering, Bengaluru 560078,
India
82 L. Geetha et al.
1 Introduction
The constant growth and expansion of city infrastructures across the globe produce a
vast number of engineering problems. In metropolitan areas, construction and reno-
vation projects often occur within close proximity of existing structures due to space
limitations. A major concern during the planning and execution of projects involving
deep excavation is the impact of construction-related ground movements on adja-
cent structures. Excavation support systems are often designed to ensure the structural
stability of adjacent structures. Although this approach produces a support system
that adequately prevents structural failure, it may yield excessive deformations and
ground movements. In response to these ground movements, adjacent structures will
translate, rotate, deform, and possibly sustain damage.
Construction-related ground movements associated with excavation include
movement in response to ground loss during excavation, and short-term and long-
term movements in response: changes in the state of stress in the ground as load is
transferred from the ground to the excavation support system; dewatering activities;
ground treatment or modification; deep foundation installation and vibration caused
by blasting, foundation installation, or general equipment operation. The excavation
for the new structure induces stresses on the existing structure. It is therefore required
to assess the impact of construction on existing structures. In general, the following
should be examined:
1. Variation in the bearing capacity of the soil adjacent to the deep excavation.
2. The extent of displacements that may affect the structural or operational integrity
of adjacent structures.
3. Response of the structure influenced by adjacent deep excavation under seismic
load.
Ni et al. [1] studied numerical calculation model of the dynamic response of a
tall building base foundation system. The structure was modeled and the response
of the structure was found out for both horizontal and vertical seismic action using
Time-History Seismic Analysis, with different elastic modulus for the foundation.
The result showed that the structures on soft foundation take on stronger capacity
of resisting vertical earthquake than that on rigid foundation, and the opposite is
the case of resisting horizontal earthquake. The story drifts on soft foundation under
both horizontal and vertical earthquake action were much less than that on rigid foun-
dation. Che-Ani et al. [2] conducted a study on complaints received by the Local
Authority in Selangor, Malaysia about the impact of adjacent new construction works
on the conditions of existing buildings in the neighboring vicinity. For that a case
study consisting of 28 units of two-storey terrace houses were considered. Inspec-
tion form was used as a main instrument along with a basic surveying tool with a
digital caliper for measuring cracks. Crack-width data were collected and analyzed
using the Statistical Package Social Sciences (SPSS) software. It showed that the
existing building had an impact from the adjacent new construction works. Sebastian
Bryson and Kotheimer [3] presented the results of three-dimensional finite element
Dynamic Study of Existing Structure … 83
analyses of a building adjacent to an excavation, which were used to evaluate the

magnitude of strain that developed in the interior walls in response to the excavation-
related ground movements. The analyses showed that the initial cracking observed in
selected infill wall panels could not have occurred solely in response to excavation-
related deformations. Consequently, it was found that the wall panels cracked as a
result of a combination of strains induced in the structure from self-weight settlement
and excavation-induced displacement. Metwally et al. [4] a thorough analysis of the
pipeline failure influence in different soils on adjacent buildings was investigated.
ANSYS Civil FEM was used to perform numerical simulations. The variable param-
eters like, pipeline settlement, position of settlement, burial depth, soil stiffness,
infiltration of sewage and groundwater were used to simulate the pipeline failure.
Results were presented as vertical and horizontal displacements of ground beneath
building.
A two bay four storey portal frame as shown in Fig. 1 is considered for the study.
The width of beams and columns is taken as 230 mm and depth as 450 mm. M25
grade concrete is used for the structural elements and 30° is the angle of friction
considered for soil. The analysis of structure without adjacent excavation is carried
out and is treated as reference for six other cases considered. For three cases, the
distance of excavation from existing frame structure is kept constant (as 0.5 m) and
the depth of excavation is varied (1.5, 3, and 4.5 m). In other three cases, the depth of
excavation is kept constant as 3 m and the distance of excavation from the existing
frame is varied as 1, 2, and 3 m. In all the cases the width of excavation is kept
constant as 3 m. In ANSYS, the element SOLID65 is used to represent the beams
and columns and SOLID45 is used to represent soil structure, with linear-plastic
properties used in that element. The mesh size used for meshing the frame is 0.2 and
0.6 m for soil. Meshed models of all cases considered are shown in Fig. 2.
Fig. 1 Model of frame

considered
84 L. Geetha et al.
Fig. 2 Meshed models of different cases considered a without adjacent excavation b, c, and d 1.5 m
deep, 3 m deep, and 4.5 m deep adjacent excavation, respectively. e, f, and g excavation 1 m away,
2 m away, and 3 m away from structure, respectively
2 Static Analysis
The displacement and stress variation in the structure under static load condition is
tabulated in Tables 1 and 2. Contour plots of both parameters are from Figs. 3, 4,
5, 6 and 7. Results show that the displacement is more for structure with nearer and
deeper adjacent excavation.
Table 1 Maximum
Type of excavation Maximum Maximum stress
displacement and maximum
displacement (mm) (MPa)
stress for different depth of
excavation Without excavation 3.798 3
1.5 m depth 9.295 3.43
3 m depth 12.354 3.46
4.5 m depth 13.203 3.48
Table 2 Maximum
Type of Maximum Maximum stress
displacement and maximum
excavation displacement (mm) (MPa)
stress for different distances
of excavation Without 3.798 3
excavation
Excavation 1 m 4.064 3.26
away
away
away
Fig. 3 Contour plot of frame without excavation
3 Modal Analysis
With the increase in the depth of the excavation there is a decrease in the natural
frequency of the structure because of the reduced stiffness of the soil near the exca-
vated area as shown in Fig. 8. Similarly nearer the excavation, to the existing structure
86 L. Geetha et al.
Fig. 4 Contour plot of displacement for frames with 1.5, 3, and 4.5 m deep excavation
Fig. 5 Contour plot of stress variation for frames with 1.5, 3, and 4.5 m deep excavation
Fig. 6 Contour plot of displacement for frames with excavation 1, 2, and 3 m away
lesser is the frequency as shown in Fig. 9. This also affects the boundary condition
of the adjacent column.
Fig. 7 Contour plot of stress variation for frames with excavation 1, 2, and 3 m away
10
9
8
Frequency (Hz)
7
6
solid
5
4 1.5m
3 3m
2 4.5m
1
0
0 10 20 30 40 50 60
Mode number
Fig. 8 Mode number versus natural frequency for different depths of excavations
10
9
8
Frequency(Hz)
7
6 solid
5 1m
4
3 2m
2 3m
1
0
0 10 20 30 40 50 60
Mode Number
Fig. 9 Mode number versus natural frequency with excavation 1, 2, and 3 m away
88 L. Geetha et al.
0.025
0.02
Amplitude (m) solid
0.015
1.5 m depth
0.01 3 m depth
4.5 m depth
0.005
0
0 1 2 3 4 5 6
-0.005
Frequency (Hz)
Fig. 10 Amplitude versus frequency for different depths of excavations
0.005
0.004
solid
Amplitude (m)
0.003 1m far
2m far
0.002
3m far
0.001
0
0 1 2 3 4 5 6
-0.001
Frequency (Hz)
Fig. 11 Amplitude versus frequency for excavation 1, 2, and 3 m away
4 Harmonic Analysis
Depth of excavation is directly proportional to the amplitude of response of the

existing structure. Increase in depth of excavation increases the response amplitude
as shown in Fig. 10. The adjacent excavation increases the amplitude of the structure.
If the distance of structure from adjacent excavation increases, the amplitude of the
structure will vary similar to the structure without adjacent excavation as in Fig. 11.
5 Conclusions
1. It can be observed that the increase in depth of the excavation increases the
maximum stress of the existing structure around 14–16% and displacements
can be increased almost 3 times displacement of structure without adjacent

excavation.
2. As the distance of excavation from the existing footing increases the displacement
and stress decreases.
3. The excavation made adjacent to a structure will induce a variation in the dynamic
characteristics of the structure like Natural frequencies, amplitude, and Mode
shapes.
References
1. Ni X, Zhang Y, Zhang S, Huang W (2011) Comparative analysis on seismic response of tall

building on soft and rigid foundation. Adv Mater Res 250–253:2216–2220
2. Che-Ani A, Noor ZM, Pheng LS, Tawil NM, Tahir MM (2011) Building cracks: evidence on
the impact of new construction works on existing buildings. Struct Surv 29(4):337–351
3. Sebastian Bryson L, Kotheimer MJ (2011) Cracking in walls of a building adjacent to a deep
excavation. J Perform Constr Facil 25(6)
4. Metwally KG, Hussein MM, Akl AY (2011) Damage assessment of buildings due to different
parameters of pipeline deterioration. Life Sci J 8(3):278–289
1D and 2D Dynamic Site Response
of Landfill Site Through Numerical
Analysis
Parul Rawat and Supriya Mohanty
Abstract Solid waste management and disposal become a challenge on global level.
Due to this rapid growth of waste, we are losing low-lying areas and valuable lands.
Therefore, it is a peak demand to utilize the waste-filled areas for future use. The
objective of this study is to get the dynamic response of a landfill site by using
already available data in past studies. The response study of landfill site helps to
predict the deformation and amplification behavior of the MSW fill sites. So that
required preventive measures and ground improvement techniques can be imple-
mented according to the response of site. The study area used in this paper includes
landfill sites of Delhi area, which are mostly designed for about 20 m but now they
are overfilled with more than 50 m. The aim is to find the seismic response of the
site under defined dynamic loading. The response of landfill site was numerically
analyzed both in one and two dimensions with the help of Cyclic1D and Plaxis2D
software, respectively. The analysis performed is based on finite element method
under the earthquake of magnitude of 6.5 with duration of 39.9 s and PGA of 0.3 g.
Keywords Landfill · Seismic response · Cyclic1D · Plaxis2D
1 Introduction
India is a fast developing country with neglecting its side effects like generation
of wastes. It generates about 62 million tons of municipal solid waste (MSW) per
annum which is likely to increase by 165 million tons by the end of 2031 [1]. To
fulfill this requirement we have only 59 constructed landfill sites and 376 are under
planning [2]. The major part of landfill site consists of MSW, which is a composite
mixture of household, commercial, industrial, medical waste, etc. This not only leads
P. Rawat (B) · S. Mohanty

Department of Civil Engineering, Indian Institute of Technology (BHU) Varanasi, Varanasi
221005, Uttar Pradesh, India
e-mail: parulrawat.rs.civ18@itbhu.ac.in
S. Mohanty
e-mail: supriya.civ@iitbhu.ac.in
92 P. Rawat and S. Mohanty
to the piling of this waste in landfill sites but also hinders in our economy. The best
solution could be to see its potential use and reduce the waste by converting it into
energy; landfill mining can also be one of the solutions. However, generally sanitary
landfills are designed in such a way that once they achieve designed capacity, they
are covered and can be used for future use such as public parks or construction. For
this particular study Delhi region is considered which comes under seismic zone IV
as per the seismic hazard map of India [3]. The two major landfill sites of Delhi are
Ghazipur and Okhla, which are already filled almost thrice there design capacity.
The Ghazipur dumpsite covers an area of about 28 ha, which is about 680 m long
and 360 m wide. At present, the site receives about 2600 Metric tons of solid waste
per day. Slopes forming the periphery of dump varied between 37° and 75° with
horizontal. The Okhla site spreads in an area of about 16.2 ha and is about 705 m
long and 303 m wide. The site receives about 1100 Metric tons of solid waste per
day. The periphery slope of the dump varied between 36° and 46° [4].
1.1 Past Study on Dynamic Analysis of MSW
There are various past as well as ongoing field and laboratory studies on MSW and
landfill sites have been carried out by researchers. As stability of landfills in active
seismic zone is a very important issue, which may lead to contamination of ground-
water, slope failure as well as other environmental hazards. Therefore, to eliminate
these effects, static and dynamic properties of waste are required to be understood
properly before design of landfills. There are various methods to find the composi-
tion, unit weight, strength and dynamic parameters of landfill sites all over the world,
but as MSW itself is a heterogeneous material, the properties differ from place to
place. It can be seen that unit weight profile is more important parameter than average
unit weight for seismic analysis [5]. Zekkos [6] from various laboratory and field
data analyzed that unit weight is function of waste composition, compaction effort,
depth-dependent confining pressure that is acting on landfill and gave a hyperbolic
equation for unit weight of landfills. It was also investigated from large-scale cyclic
triaxial test that the composition of waste has significant effects on dynamic prop-
erties of MSW [7]. Waste materials are generally soft and show nonlinear behavior
but it shows significant shear resistance due to presence of fibrous material [8]. To
understand the seismic behavior of MSW with single clay liner landfill founded on
sand, a dynamic centrifuge testing was conducted at various earthquake intensities
and frequency [9]. The site-specific one-dimensional seismic response study was
conducted for Delhi Okhla landfill site. For this both field and laboratory data were
used for seismic analysis. The amplification phenomenon observed from the study
was useful for the design of cover system, assessment of permanent displacements
and global stability for the dumpsite during closure [10]. Shear modulus is major
parameter to evaluate the dynamic property of MSW and low strain shear modulus
1D and 2D Dynamic Site Response of Landfill … 93
depends on its mass density and shear wave velocity. The study conducted on Banga-
lore landfill site shows dynamic shear modulus of the waste increases with increasing
confining pressure and shear strains for both dry and dumped wastes [11].
1.2 Objective of Study
The objective of the study is to get the dynamic response of a landfill site by
using already available data in past studies. The response study of landfill site
has been carried out with the help of numerical technique (1D and 2D seismic
analysis) to predict the deformation and amplification behavior of the MSW fill,
so that required preventive measures and ground improvement techniques can be
implemented according to the response of site.
2 Brief Discussion About Numerical Methods
2.1 Modeling Using Cyclic1D
Cyclic1D is a user interface finite element program for execution of the site response,
which includes analysis of liquefaction-induced deformations [12]. The program
operates in time domain with linear and nonlinear analysis for dry as well as satu-
rated strata. In most of the cases, one-dimensional (1D) site response analysis is
performed to assess the effect of soil conditions on ground shaking because verti-
cally propagating and horizontally polarized shear waves dominate the earthquake
ground motion wave field [13]. The number of elements in Cyclic1D can be chosen
from 10 to 400. For the present study in 1D the maximum elements taken after
convergence analysis was 360.
2.2 Modeling Using Plaxis2D
Plaxis2D is a popular two-dimensional (2D) finite element software for the analysis
of deformation, stability and groundwater flow. The finite element model generated
is based upon geometry model, which involves the composition of surface, lines and
points. In FEM, the continuum is divided into number of elements, which consist of
nodes, in Plaxis it can be 6-noded or 15-noded triangular element [14].
Fig. 1 Convergence analysis

for Plaxis2D
2.2.1 Meshing
Once geometry and parameters are defined, the model was analyzed for convergence
and mesh was generated. For this particular study, the model converged at 1989
number of elements with relative element size of 0.7 m (Fig. 1).
2.2.2 Stage Construction Phase
The landfill site was treated as embankment, so it is assumed to be constructed in 5

stages and consolidated for 4 years each, i.e., total of 20 year. The last stage is added
for the input of seismic load.
3 Model Description
The two major landfill sites of Delhi area (Ghazipur and Okhla) were studied and
a generalized profile was considered accordingly. The MSW landfill model was
considered as an embankment with 50 m fill height and side slope of 45°. The model
consists of two layers of subsoil (fine sand and silty sand) of 20 m each with average
groundwater at 12 m below the ground level.
3.1 MSW Parameters
The MSW parameters of two major landfill sites of Delhi were taken from past studies
[4, 10, 15]. Hardening soil model was considered for the study as it is superior to any
linear elastic model and produce more realistic results as well as capable of modeling
modulus reduction with increase in strain. The Young’s modulus of MSW for this
analysis was correlated from SPT data of the site [16].
3.2 Subsoil Parameters
The general site profile of Delhi was considered in study from past data [17, 18]. The
parameters considered are shown in Table 1.
3.3 Input Earthquake Parameters
The input motion selected was Uttarkashi earthquake of magnitude 6.5 and total
duration of 39.9 s (Fig. 2) with peak ground acceleration (PGA) value of 3.04 m/s2 .
Table 1 Model parameters

MSW Fine sand Silty sand
of MSW landfill and subsoil
Model Hardening soil Hardening Hardening
soil soil
Material Drained Drained Drained
behavior
γDry (kN/m3 ) 11 17 18
γSat (kN/m3 ) 14 20 21
eint 0.8 0.5 0.5
Eref 2
50 (kN/m ) 19122.96 23045.62 30890.94
Eref 2
oed (kN/m ) 19122.96 23045.62 30890.94
Eref 2
ur (kN/m ) 40.00E3 48.00E3 62.00E3
ν 0.3 0.3 0.3
c 17 0 20
(°) 32 33 30
Data set Standard Standard Standard
Medium Medium Medium
Fine
kx = ky (m/day) 1.04544E−3 10E−2 10E−4
Damping ratio 15 10 10
(%)
Damping parameters
α 0.628319 0.418879 0.418879
β 0.0318310 0.0212207 0.0212207
Fig. 2 Input motion used for

the analysis (Uttarkashi
Earthquake)
For 1D analysis using Cyclic1D, responses are presented along the depth of the
MSW fill. For 2D analysis in Plaxis2D, seven sections at X = 20, 50, 150, 300, 450,
550 and 580 m all over the landfill are considered along which all the responses are
presented below.
4.1 Displacements
1D Analysis: The results obtained from the analysis done using Cyclic1D show that
vertical displacement is maximum for 50 m MSW landfill and it linearly reduces
to zero at the bottom of the deposit (Fig. 3a). The low density of landfill material
causes compression of the fill. Similarly, the horizontal displacement can also be
seen maximum at the top layers (MSW fill layer) and reduces as the depth increases
(Fig. 3b).
2D Analysis: The results obtained from the analysis conducted using Plaxis2D at
7 sections shows that vertical and horizontal displacements are more toward the end
of the fill near the slopes (at section X = 20, 50, 550 and 580 m) at the end of dynamic
phase (Fig. 4). The displacement reduces as the depth of fill increases. Similar kind of
trends is shown in Cyclic1D results (Fig. 3). However, Cyclic1D analysis was done
for single soil column, which shows less displacement as compared to Plaxis2D
results.
(a) (b)
Fig. 3 a Vertical displacement b Absolute max. horizontal displacement variation with depth in
1D analysis
(a)
(b)
Fig. 4 a Vertical displacement variation with depth b Horizontal displacement variation with depth
for different sections of landfill in 2D analysis
4.2 Stresses
1D Analysis: The drawback of 1D analysis was that only a single soil column can be
considered, so the column was assumed to be at the center of the profile. The shear
stress variation shows increment with depth and maximum shear stress concentration
was noticed in lower layers (Fig. 5a). Similarly, effective confinement also shows
increasing trend with depth (Fig. 5b).
2D Analysis: The vertical and horizontal stress increases with depth in dynamic
phase as overburden pressure increase with depth (Fig. 6). The vertical stresses with
depth at section X = 20 and 580 m are comparatively low, whereas shear stress
increases as we move away from center of landfill toward the slope ends and low in
the center sections due to confinements from both sides (Fig. 7).
(a) (b)
Fig. 5 a Absolute max. shear stress b Absolute max. effective confinement variation with depth in
1D analysis
(a) (b)
Fig. 6 a Vertical stress variation with depth b Horizontal stress variation with depth for different
sections of landfill in 2D analysis
(a)
(b)
Fig. 7 a Shear stress variation with depth at different sections of landfill b Shear stress contour for
2D analysis
4.3 Acceleration
1D Analysis: The horizontal acceleration in 1D analysis can be seen amplified in top

layer of fill and reduced to PGA value again at the bottom (Fig. 8). The amplification
in top layer is because of low confinement, which reduces in middle and again
amplifies at the bottom layer.
2D Analysis: The maximum horizontal acceleration at different sections (Fig. 9)
coincide at bottom is about 2.9 m/s2 . The acceleration near the top surface of landfill
can be seen more for all sections between 1.5 and 2 m/s2 , as at the top layer there is
no vertical overloading of confinement.
Fig. 8 Horizontal
acceleration variation of
landfill with depth in 1D
analysis
Fig. 9 Horizontal
acceleration variation at
different sections of landfill
with depth in 2D analysis
4.4 Excess Pore Pressure
1D Analysis: The water table is considered at 12 m below the ground and in Fig. 10
for Cyclic1D, the water table starts from 62 m as 50 m is MSW landfill height and
water table lies 12 m below that. The excess pore water pressure increases from water
table to the bottom of the profile (Fig. 10).
2D Analysis: The excess pore pressure profile at different sections concludes that
excess pore pressure increases with depth and achieves maximum values at bottom.
The profile (Fig. 11) shows that excess pore pressure is more in the center portion
of landfill. The maximum excess pore pressure can be seen for the middle section,
Fig. 10 Excess pore water

pressure variation of landfill
with depth in 1D analysis

pressure variation at different
sections of landfill with
depth in 2D analysis

pressure ratio variation with
depth in 1D and 2D analysis
i.e., X = 300 m, as that section has more vertical as well as horizontal confinement,
where as the side ends, i.e., section X = 20 and 580 m has lowest Pexcess developed.
In Cyclic1D, the MSW landfill site was analyzed for liquefaction susceptibility
by considering plasticity model into account. In Plaxis2D, hardening soil model was
used which in contrast to elastic perfectly plastic model the yield surface of hardening
plasticity model was not fixed in principal stress space, but it can be expanded due to
plastic straining. The analysis in Cyclic1D was done in only one section or a column,
which can be considered as middle section. Although the trends from both 1D and
2D analysis are same but shear stress values with depth are more for 1D analysis,
if we compare with middle section of 2D analysis. The acceleration shows sudden
change as material in layers changes at 50 and 70 m, whereas that sudden change
can’t be seen in 2D analysis.
Based on pore pressure-related criteria, soil liquefaction is defined as a state
at which excess pore water pressure ratio (ru ) becomes equal to 1. This happens
when pore water pressure increase becomes equal to initial effective overburden
stress. Sometime this is also known as ‘initial liquefaction’ or ‘partial liquefaction’.
Figure 12 shows the variation of ru with depth for 1D and 2D analysis (at section x
= 300 m), which shows no signs of liquefaction of MSW fill as well as foundation
soils.
5 Conclusions
The above results concluded the 1D and 2D seismic response of the Delhi landfill
sites, which shows the amplification of acceleration in loose MSW fill and higher
vertical and horizontal stresses in lower layers. The similar site response for Okhla
Delhi site was shown by Ramaiah [19], variation of maximum horizontal equivalent
acceleration (MHEA) with depth from 1D analysis for local earthquakes (Mw =
6.5) confirms the maximum horizontal acceleration in the top layers. The results
from Cyclic1D are little conservative as, we can see the profile variations at only one
section. The results from 2D analysis from Plaxis2D at middle section give almost
similar results as in 1D analysis by Cyclic1D. For using the landfill sites for future
use, it is required to focus on the slope stability of landfill as it is the zone of maximum
shear stress. Although this modeling is not sufficient for study and design purpose,
it just gives an idea about the site response during any earthquake.
References
1. Planning Commission Report (2014) Reports of the task force on waste to energy, vol I (in the
context of Integrated MSW management)
2. CPCB Status report on municipal solid waste management (2013)
3. IS 1893 (2002) Criteria for earthquake resistant design of structures—Part 1: General provisions
and buildings. Bureau of Indian Standards, New Delhi
4. Ramaiah BJ, Ramana GV, Datta M (2017) Mechanical characterization of municipal solid
waste from two waste dumps at Delhi, India. Waste Manag 68:275–291
5. Kavazanjian Jr E, Matasovic N, Stokoe II KH, Bray JD (1996) Insitushear wave velocity of
solid waste from surface wave measurements. In: Proceedings of the 2nd international congress
on environmental geotechnics, vol 1, pp 97–102, Balkema, Rotterdam, Netherlands
6. Zekkos DP, Bray JD, Kavazanjian E, Matasovic N, Rathje E, Riemer M, Stokoe KH (2005)
Framework for the estimation of MSW unit weight profile. In: 10th international waste
management and landfill symposium, pp 3–7, Santa Margherita di Pula, Cagliari, Italy
7. Zekkos D, Bray JD, Riemer MF (2008) Shear modulus and material damping of municipal
solid waste based on large-scale cyclic triaxial testing. Can Geotech J 45(1):45–58
8. Towhata I, Kawano Y, Yonai Y, Koelsh F (2004) Laboratory tests on dynamic properties of
municipal wastes. In: 11th conference in soil dynamics and earthquake engineering and the
3rd international conference on earthquake geotechnical engineering, vol 1, pp 688–693
9. Thusyanthan NI, Madabhushi SPG, Singh S (2004) Modelling the seismic behaviour of munic-
ipal solid waste. In: 11th international conference on soil dynamics and earthquake engineering,
Berkeley, USA
10. Ramaiah BJ, Ramana GV, Kavazanjian E Jr, Bansal BK (2016) Dynamic properties of municipal
solid waste from a dump site in Delhi, India. Geo-Chicago 2016:121–130
11. Naveen BP, Sitharam TG, Sivapullaiah PV (2014) Evaluating the dynamic characteristics of
municipal solid waste for geotechnical purpose. Curr Adv Civ Eng 2(1):28–34
12. Lu J, Elgamal AWM, Yang Z (2006) Cyclic1D: a computer program for seismic ground
response. Department of Structural Engineering, University of California, San Diego
13. Kramer SL (1996) Geotechnical earthquake engineering. Prentice Hall, Inc., UpperSaddle
River, New Jersey, p 653
14. Plaxis BV (2018) Plaxis 2D reference manual. The Netherlands
15. Ramaiah BJ, Ramana GV (2014) CPTu at a municipal solid waste site in Delhi, India. In: 3rd
international symposium on cone penetration testing, pp 1083–1091, Las Vegas, Nevada, USA
16. Brahma P, Mukherjee S (2010) A realistic way to obtain equivalent Young’s modulus of layered
soil. In: Indian geotechnical conference, pp 305–308, Bombay, India (2010)
17. Inqualabi KQ, Garg RK, Rao KB (2017) Seismic vulnerability of urban bridges due to liquefac-
tion using nonlinear pushover analysis and assessing parameters for damage detection. Procedia
Eng 173:1739–1746
18. Datta M (2012) Geotechnology for environmental control at waste disposal sites. Indian
Geotech J 42(1):1–36
19. Ramaiah BJ, Ramana GV, Bansal BK (2016) Site-specific seismic response analyses of a
municipal solid waste dump site at Delhi, India. In: Proceedings of Geo-China, pp 191–198
A Study on Characteristics of Soil Profile
of Guwahati City Against Different
Ground Motions: 1D NonLinear Ground
Response Analysis
Amar F. Siddique, Anusuya Acharjee, and Binu Sharma
Abstract One-dimensional nonlinear (NL) ground response analysis (GRA) is

carried out for four locations viz. Dharapur and Azara area, VIP and Rani area, Jaluk-
bari area and Assam Engineering College (AEC) area of Guwahati city. The analysis
is done with the help of 23 boreholes using Indo-Burma earthquake, 1997 recorded
at Nongpoh station and Indo-Burma earthquake, 1988 recorded at Nongstoin station
with duration 52.94 s. These two input motions are further scaled up to 0.18 and
0.36 g and used for the GRA analysis. The response of the substrata obtained after
the analysis has been presented in terms of peak spectral acceleration, maximum
shear stress profile, strain distribution and peak ground acceleration (PGA) for all
the locations irrespective of every motions. For most of the locations, soil layers
with high SPT N value, generally stiffer soil, resulted in a lower value of surface
PGA compared to soft soil layers. The PGA and overall energy content of the strong
motion significantly affect the response of a multilayered soil profile, keeping all
other parameters constant. Higher strain amplitudes have been observed at various
sites and at various depths of the soil profiles which may undergo settlement or
earthquake-induced liquefaction. The peak spectral acceleration obtained is much
higher for 0.36 g scaled motion compared with the spectral acceleration of rocky or
hard soil sites (IS: 1893–2002) for both the earthquakes which indicate sensitiveness
of the locations to the induced motions.
Keywords Nonlinear ground response analysis · Peak ground acceleration · Peak

spectral acceleration
A. F. Siddique (B) · A. Acharjee · B. Sharma

Department of Civil Engineering, Assam Engineering College, Guwahati 781013, India
e-mail: amaribnesiddiq98@gmail.com
A. Acharjee
e-mail: anusuyaa798@gmail.com
B. Sharma
e-mail: binusharma78@gmail.com
106 A. F. Siddique et al.
1 Introduction
The Guwahati city is situated beside the mighty river Brahmaputra in the Indian state
of Assam which lies between coordinates of 26.183 N and 91.733E. According to IS
1893 [1], the city falls in the most elevated seismic zonal level (zone V). As per Nath
et al. [4], the city had been hit by quakes of various sizes extending from 5 to 8.7
before. This has featured the need to comprehend the vulnerability of land to seismic
risks caused by future quakes in Guwahati city. Due to the high seismicity level
together with presence of alluvial soil in Guwahati city, one-dimensional (1D) ground
response analysis (GRA) utilizing the nonlinear (NL) procedure was performed for
four locations in Guwahati city viz. Dharapur and Azara, VIP and Rani, Jalukbari
and Assam Engineering College (AEC) area, where bore logs up to 30 m depth are
available with the required physical properties, using the software DEEPSOIL v6.0.
NL analysis is done because it has more potential to simulate soil behavior accurately
and is more realistic in nature.
2 Motivation and Objective(s)
Guwahati city soil consists of alluvial deposits with layers of both coarse and fine-
grained soils. Hence, large modifications in earthquake waves can occur due to change
in variation in soil properties near to the surface of earth. The main objective is there-
fore to study the response of the soil deposits of Guwahati city when subjected to
strong ground motion. The aim is to determine the peak ground acceleration (PGA),
maximum shear stress profile, strain distribution with depth and peak spectral acceler-
ation of 23 boreholes spreading in four locations of Guwahati city. The boreholes(BH)
considered for AEC area are BH01, BH03, BH23, BH136 and BH143; for Dharapur
and Azara area the boreholes considered are BH07, BH08, BH15, BH16, BH17 and
BH18; for VIP and Rani area BH09, BH10, BH11, BH12 and BH13 are considered
and for Jalukbari area BH02, BH04, BH05, BH19, BH21 and BH26. Response spec-
trum is dealt in detail in this work to see the spectral acceleration behavior with respect
to time. The response spectra obtained with the input motion from 1997 Indo-Burma
earthquake and 1988 Indo-Burma earthquake in addition with the response spectra
obtained after scaling the input motion are compared with the response spectra of
rocky or hard soil sites (IS: 1893 [1]). Since response spectra of a site are more
important to design buildings hence these findings will have considerable influence
in building design and estimation of induced effects of earthquakes.
A Study on Characteristics of Soil Profile of Guwahati City … 107
3 Methodology
The present study is accompanied with few steps which are utilized during the anal-
ysis. The steps include: (1) to gather information, (2) to model them for the software
used, (3) to execute the program and (4) to interpret the outcomes. 1D ground response
analysis can be done with the help of three approaches viz. Linear, Equivalent Linear
(EL) and Nonlinear (NL). Linear analysis can be done in both frequency domain and
time domain and it is one of the simplest processes out of all the analysis. But the main
disadvantage in this approach is that the soil behavior is assumed linear in nature and
the soil properties (shear modulus and damping ratio) are constant throughout the
analysis. To overcome the limitation of linear analysis, equivalent linear method was
developed to analyze the nonlinear response of soil using frequency domain with
the aid of linear transfer functions. The nonlinear behavior of the soil is modeled
in terms of equivalent linear properties corresponding to effective shear strain using
iterative procedure (Kramer [3]). However, this technique is computationally helpful
and gives sensible outcomes but it is unfit to speak to the adjustment in soil stiffness
that really happens amid the seismic tremor. Henceforth, nonlinear ground response
analysis is performed to obtain the realistic stress–strain behavior of soil analyzed
in time domain with the help of numerical integration. The current study adopts the
nonlinear analysis with the help of economically accessible program, i.e.DEEPSOIL
v6 software. The modulus reduction curve and damping curves are absent for the
site, therefore, for sandy soil, Seed and Idriss (mean limit) curve and for clay, Vucetic
and Dobry [6] curve is used for the nonlinear GRA analysis. The curves are fitted
with the help of MRDF procedure. Kodner and Zelasko [2] stress–strain model is
used for the analysis including the Mashing criteria. The input parameter shear wave
velocity can be used instead of shear stiffness to represent the stiffness of the soil
layer during the analysis in the DEEPSOIL v6 software. The correlation given in
Eq. (1) by Sharma and Rahman [5] is used to determine the shear wave velocity for
all types of soil.
Vs = 74.639 ∗ N0.3876 (1)
4 Soil Study
The study of soil of a site is considered important to understand the type of soil present
and to know the index properties of the soil, depth of water table and many other
characteristics. For the present study, information of twenty three (23) boreholes up
to 30 m depth were taken from a project sponsored by the Department of Science and
Technology, India. The borehole data consist of SPT N values, depth of water table
and other engineering properties of the soils. Laboratory experiments with the help of
undisturbed and disturbed samples led to determination of engineering properties of
the soil. The fine grained portion for the most part comprises of soils of arrangement
classification CL, CI and CH as indicated by the Indian standard soil classification.
In a couple of areas inorganic soil of characterization ML and CL-ML and nonplastic
inorganic soil were additionally experienced. The coarse grained fraction is for the
most part of grouping SP, SW, SC, SM, SP-SC.
5 Input Motion
The recorded acceleration time histories of Indo-Burma earthquake, 1997 and Indo-
Burma earthquake, 1988 recorded at rocky sites have been selected as the input
motions for the ground response analysis which are further scaled to peak bedrock
acceleration value of 0.18 g and 0.36 g. The motions that are selected have a very low
value of bedrock peak horizontal acceleration (weak motions). These weak motions
do not lead to any harmful effect to the structures. In order to see the vulnerability
and risk of the structures the motions are scaled up to get a strong motion. Table 1
depicts the input data used for the current study. After the input motions are scaled
up, the energy content increases and all other parameters remain constant.
One-dimensional nonlinear ground response analysis (GRA) has been done for
Dharapur and Azara area, VIP and Rani area, Jalukbari area and Assam Engineering
College (AEC) area. The results after the analysis are obtained in terms of PGA
with depth, maximum shear stress profile, strain (%) profile and response spectra.
The PGA value gets amplified at the ground surface. The amplification factor for all
the locations are also reported which is defined as the ratio of surface PGA to the
input bedrock PGA. Table 2 represents the summary of the range of surface PGA,
amplification factor, strain values, maximum shear stress and the maximum spectral
acceleration for all the input motions applied to the four locations. It is found that for
the input motion 0.04 g and scaled up motion 0.18 and 0.36 g recorded at Nongpoh
(NGPH) station for Indo-Burma Earthquake (IND-BM EQR), 1997, the surface PGA
Table 1 Details of input

Earthquake Recording station Description
motion
1997 Indo-Burma Nongpoh Mw 6, depth 90 km,
PGA 0.04 g, duration
47.64 s, site class A
1988 Indo-Burma Nongstoin Mw 7.2, depth 34 km,
PGA 0.05 g, duration
52.94 s, site class A
Table 2 Results of ground response analysis

Earthquake Input Surface Amplification Maximum Stress Spectral
Motion PGA (g) Factor Strain (%) Ratio Acceleration(g)
AEC area
1997 0.04 g 0.10–0.18 2.60–4.52 0.00–0.04 0.10–0.28 1.17
IND-BM 0.18 g 0.25–0.45 1.39–2.50 0.00–0.16 0.25–0.95 3.11
0.36 g 0.47–0.80 1.31–2.24 0.01–0.33 0.46–1.35 4.63
0.05 g 0.08–0.14 1.78–2.98 0.00–0.05 0.08–0.25 0.83
1988 0.18 g 0.28–0.43 1.58–2.43 0.00–0.18 0.36–0.76 2.69
IND-BM 0.36 g 0.45–0.74 1.25–2.08 0.01–7.44 0.49–1.30 4.47
Dharapur and Azara area
1997 0.04 g 0.07–0.15 1.75–3.75 0.00–0.05 0.07–0.15 1.00
IND-BM 0.18 g 0.22–0.37 1.27- 2.11 0.01–0.81 0.17–0.38 2.49
0.36 g 0.37–0.52 1.05–1.44 0.01–1.74 0.23–0.67 3.80
0.05 g 0.09–0.16 1.86–3.32 0.00–0.05 0.09–0.21 0.88
1988 0.18 g 0.22–0.45 1.27–2.50 0.01–4.22 0.23–0.44 2.53
IND-BM 0.36 g 0.39–0.62 1.11–1.72 0.02–30.0 0.18–0.65 3.48
VIP and Rani area
1997 0.04 g 0.08–0.18 2.17–4.65 0.00–0.05 0.08–0.20 0.97
IND-BM 0.18 g 0.31–0.63 1.74–3.52 0.00–0.33 0.32–0.63 3.76
0.36 g 0.37–0.76 1.05–2.11 0.01–1.68 0.38–0.91 4.28
0.05 g 0.08–0.15 1.78–3.12 0.00–0.04 0.09- 0.18 0.66
1988 0.18 g 0.24–0.46 1.37–2.60 0.00–1.69 0.26–0.59 2.05
IND-BM 0.36 g 0.38–0.53 1.08–1.49 0.01–9.19 0.31–0.83 3.28
Jalukbari area
1997 0.04 g 0.07–0.18 1.97–4.65 0.00–0.06 0.08–0.34 1.10
IND-BM 0.18 g 0.22–0.43 1.23–2.41 0.00–0.24 0.18–0.67 2.61
0.36 g 0.38–0.57 1.08–1.59 0.01–0.61 0.28–0.88 3.64
0.05 g 0.08–0.15 1.70–3.08 0.00–0.13 0.11–0.28 0.95
1988 0.18 g 0.20–0.44 1.16–2.48 0.01–9.60 0.22–0.60 2.17
IND-BM 0.36 g 0.43–0.65 1.22–1.72 0.02–48.5 0.30–0.86 3.32
values vary from 0.10 to 0.80 g for AEC area, 0.07–0.52 g for Dharapur and Azara
area, 0.08–0.76 g for VIP and Rani area, 0.07–0.57 g for Jalukbari area. Similarly,
for the input motion 0.05 g and scaled up motion 0.18 g and 0.36 g recorded at
Nongstoin(NGSTN) station for Indo-Burma Earthquake (IND-BM EQR), 1988, the
surface PGA values vary from 0.08 to 0.74 g for AEC area, 0.09–0.62 g for Dharapur
and Azara area, 0.08–0.53 g for VIP and Rani area, 0.08–0.65 g for Jalukbari area.
Also, the amplification factor shows the range correspondingly as per the variation
of the surface PGA. Moreover, as per the input motion strain value and maximum
stress ratio (the ratio of shear stress to the effective vertical stress of the soil) value
also showed variation and it is found that for the former earthquake, the strain value
ranges from 0.00 to 0.33% and stress ratio value ranges from 0.10% to 1.35% for
AEC area, 0.00–1.74% and 0.07–0.67% for Dharapur and Azara area, 0.00–1.68%
and 0.08–0.91% for VIP and Rani area, 0.00–0.61% and 0.08–0.88% for Jalukbari
area. For the second earthquake, the strain value ranges from 0.00% to 7.44% and
maximum stress ratio value ranges from 0.88 to 1.33% for AEC area, 0.00–30% and
0.09–0.65% for Dharapur and Azara area, 0.00–9.19% and 0.09–0.83% for VIP and
Rani area, 0.00–48.5% and 0.11–0.86% for Jalukbari area.
Of the four locations, the result of AEC area is presented in this paper with the help
of graphical representations and similar results are obtained for all other locations.
For AEC area, Fig. 1 shows the comparison of peak horizontal acceleration with
respect to depth for the input motion of Indo-Burma Earthquake, 1997 and Fig. 2
represents the comparison of peak horizontal acceleration with respect to depth for the
input motion of Indo-Burma Earthquake, 1988. It is found that soft soil layer shows
higher surface PGA in comparison to stiff soil layer. Thus soft soil will amplify
the motion because soft soil has a very low SPT N value (N < 10) and low shear
wave velocity than stiff soil. As the value of peak bedrock acceleration (PBRA)
increases after the input motion is scaled up, the value of surface PGA also increases
because the energy content increases keeping the other parameters constant. From
Figs. 1 and 2, it is observed that lower surface PGA and amplification factors are
obtained for Indo-Burma Earthquake, 1988 recorded at Nongstoin station than Indo-
Burma Earthquake, 1997 recorded at Nongpoh station. This is due to the reason
PGA (g) PGA (g)

0 0.05 0.1 0.15 0.2 0 0.1 0.2 0.3 0.4 0.5
0 0
5
5
Depth (m)
10
Depth(m)
10
15
15 20
BH01_NGPH_Mw6
BH03_NGPH_Mw6 25
20 BH01_NGPH_Mw6_0.18g
BH23_NGPH_Mw6
BH136_NGPH_Mw6 30 BH03_NGPH_Mw6_0.18g
25 BH143_NGPH_Mw6 BH23_NGPH_Mw6_0.18g
BH136_NGPH_Mw6_0.18g
30 BH143_NGPH_Mw6_0.18g
PGA (g)
0 0.2 0.4 0.6 0.8 1
0
5
Depth (m)
10
15
20
BH03_NGPH_Mw6_0.36g
30
BH23_NGPH_Mw6_0.36g
Fig. 1 PGA distribution for input motion 0.04, 0.18 and 0.36 g for AEC area
PGA (g) PGA (g)

0 0.25 0.5 0 0.05 0.1 0.15 0.2
0 0
5 5
Depth (m)
10
Depth (m)
10
15
15
20
BH01_NGSTN_Mw7.2_0.18g 20 BH01_NGSTN_Mw7.2
25
BH03_NGSTN_Mw7.2
BH03_NGSTN_Mw7.2_0.18g
30 25 BH23_NGSTN_Mw7.2
BH136_NGSTN_Mw7.2
BH136_NGSTN_Mw7.2_0.18g 30 BH143_NGSTN_Mw7.2
PGA (g)
0 0.2 0.4 0.6 0.8
0
5
Depth (m)
10
15
20
25
30
Fig. 2 PGA distribution for input motion 0.05, 0.18 and 0.36 g for AEC area
that though both the motion has almost same bedrock PGA but the motion recorded
at Nongstoin station has higher hypocentral distance (400.2 km) than the motion
recorded at Nongpoh station (124.1 km).
Similarly, the strain profile of the AEC area is shown in Fig. 3 for the former
earthquake and in Fig. 4 for the later earthquake. It is observed that for few locations,
the layer having very soft soil shows higher strain value. From both the figures, it
can be concluded that when the input motion is scaled up to higher values then the
shear strain value also increases correspondingly with respect to depth and follows
a similar pattern. Also, it is observed that the depth showing the highest strain value
for a particular borehole for a given input motion may vary with the scaled motion.
Because of lower hypocentral distance of motion at Nongpoh station, it is observed
that the strain range is more for motion recorded at Nongpoh station than Nongstoin
station.
The maximum stress ratio distribution is shown in Fig. 5 that represents the effects
of local site and strong motion characteristics for input motion 0.04 g and scaled
motion 0.18 and 0.36 g recorded at Nongpoh station for Indo-Burma Earthquake,
1997 and Fig. 6 shows the distribution of shear stress ratio with respect to depth for
the input motion 0.05 g and scaled up motion 0.18 and 0.36 g recorded at Nongstoin
station for Indo-Burma Earthquake, 1988 for AEC locality. From Figs. 5 and 6 we
find that shear stress ratio also increases as the input motion gets scaled up to higher.
At depths from 3 to 5 m consisting of stiffer soil having higher SPT N value, higher
Strain(%) Strain(%)
0 0.01 0.02 0.03 0.04 0.05 0 0.05 0.1 0.15 0.2
0 0
5
5
Depth (m)
10
Depth (m)
BH03_NGPH_Mw6 15
BH23_NGPH_Mw6 20 BH23_NGPH_Mw6_0.18g
20
BH136_NGPH_Mw6
25 BH136_NGPH_Mw6_0.18g
25 BH143_NGPH_Mw6
30 BH143_NGPH_Mw6_0.18g
30
Strain (%)
0 0.1 0.2 0.3 0.4
0
5
Depth (m)
10
BH03_NGPH_Mw6_0.36g
20
BH23_NGPH_Mw6_0.36g
25
30 BH143_NGPH_Mw6_0.36g
Fig. 3 Shear strain profile of input motion 0.04, 0.18 and 0.36 g for AEC area
Strain (%) Strain (%)

0 0.01 0.02 0.03 0.04 0 0.05 0.1 0.15 0.2
0 0
5 5
Depth (m)
10 BH01_NGSTN_Mw7.2_0.18g
Depth (m)
10 BH01_NGSTN_Mw7.2
BH03_NGSTN_Mw7.2 15
BH23_NGSTN_Mw7.2
20 BH136_NGSTN_Mw7.2_0.18g
20 BH136_NGSTN_Mw7.2
BH143_NGSTN_Mw7.2 25
25
30
30
Strain(%)
0 2 4 6 8
0
5
Depth (m)
15 BH143_NGSTN_Mw7.2_0.36g
20
25
30
Fig. 4 Shear strain profile of input motion 0.05, 0.18 and 0.36 g for AEC area
Stress Ratio (shear/eff. vert.) Stress Ratio (shear/eff. vert.)

0 0.5 1
0 0.1 0.2 0.3 0.4
0
0
5 5
10
Depth (m)
Depth (m)
15 BH03_NGPH_Mw6 15 BH03_NGPH_Mw6_0.18g
BH23_NGPH_Mw6
20 20 BH23_NGPH_Mw6_0.18g
BH136_NGPH_Mw6
25 25
BH143_NGPH_Mw6
30
30
Stress Ratio (shear/eff. vert.)

0 0.5 1 1.5
0
10
Depth (m)
BH01_NGPH_Mw6_0.36g
15
BH03_NGPH_Mw6_0.36g
25
30
Fig. 5 Distribution of stress ratio with depth for input motion 0.04, 0.18 and 0.36 g
maximum stress ratio is obtained for majority of the sites. A higher stress ratio range
is obtained for site due to motion recorded at Nongpoh station than Nongstoin station.
Response spectrum holds a significant role to understand how the structure
responds when subjected to strong ground motions for a particular duration. It is
important to note that response for a particular location will not be same for a specific
duration when the earthquake is different. Therefore, for geotechnical engineers and
structural engineers, the reflection of response spectra obtained from the analysis can
be utilized to design the structures to be built in specific locations, to assure that the
structure can be safe in future when subjected to strong seismic tremor. Response
spectrum is described as spectral acceleration with respect to duration of the earth-
quake for 5% damping. From Table 2, it is seen that in AEC area, for the former
earthquake, the maximum spectral acceleration value is 1.17 for 0.04 g motion while
for scaled up 0.18 and 0.36 g motion the values are 3.11 and 4.63, respectively,
and for the later earthquake the maximum spectral acceleration value is 0.83 for
0.05 g motion while for scaled up 0.18 and 0.36 g motion the values are 2.69 g and
4.47 g, respectively. Similarly, a very high value of spectral acceleration is observed in
0.36 g scaled up input motion for all other locations. It is observed that the maximum
spectral acceleration values for 0.36 g in case of Indo-Burma Earthquake 1997, are
3.80, 4.28 and 3.64 g and for Indo-Burma Earthquake 1988, the maximum values
are 3.48 g, 3.28 g and 3.32 g for Dharapur and Azara area, VIP and Rani area and
Stress Ratio (shear/eff. vert.) Stress Ratio (shear/eff. vert.)

0 0.1 0.2 0.3 0 0.2 0.4 0.6 0.8 1
0 0
5 5
Depth (m)
10 10
Depth (m)
BH01_NGSTN_Mw7.2
15 BH03_NGSTN_Mw7.2 15
BH23_NGSTN_Mw7.2
20 20 BH23_NGSTN_Mw7.2_0.18g
BH136_NGSTN_Mw7.2
25 BH136_NGSTN_Mw7.2_0.18g
BH146_NGSTN_Mw7.2 25
30
30
Stress Ratio (shear/eff. vert.)

0 0.5 1 1.5
0
5
Depth (m)
10
20
25 BH136_NGSTN_Mw7.2_0.36g
30
Fig. 6 Distribution of stress ratio with depth for input motion 0.05, 0.18 and 0.36 g
Jalukbari area, respectively. Figures 7 and 8 show the spectral acceleration variation
with respect to time period for the respective input motions and compared with the
response spectra given by IS 1893 [1] for rock or hard soil in AEC area. It is found
that when the input motion is 0.36 g for both the earthquake of different magnitude,
the response is very high compared to the other two motions and also it crosses the
response spectra given by IS code. Hence, the area will be highly sensitive for that
particular time period when the seismic tremor will have PBRA 0.36 g. The motion
recorded at Nongpoh station gives a higher spectral acceleration value than motion
recorded at Nongstoin station because of lower hypocentral distance of Nongpoh
station.
Figure 9 shows the response spectra for 0.36 g motion of Dharapur and Azara
area, VIP and Rani area and Jalukbari area for the former earthquake taken into
consideration and similar results are obtained for other motions which are not shown
in this paper.
3
Response spectra (IS 3.5 Response spectra (IS
1893:2002, Rock or 1893:2002, Rock or hard soil)
Spectral acceleration (g)
hard soil)

2.5 3 BH01_NGPH_Mw6_0.18g
BH01_NGPH_Mw6
2.5 BH03_NGPH_Mw6_0.18g
2
BH03_NGPH_Mw6 BH23_NGPH_Mw6_0.18g
2
1.5
BH23_NGPH_Mw6 1.5
1
BH136_NGPH_Mw6
0.5
0.5
BH143_NGPH_Mw6
0 0
0.01 0.1 1 10 0.01 0.1 1 10
Period (sec) Period (sec)
5 Response spectra (IS

1893:2002 Rock or hard soil)
4.5
BH01_NGPH_Mw6_0.36g
4
BH03_NGPH_Mw6_0.36g
3.5
BH23_NGPH_Mw6_0.36g
3
2.5 BH136_NGPH_Mw6_0.36g
1.5
1
0.5
0
0.01 0.1 1 10
Period (sec)
Fig. 7 Spectral acceleration for 0.04, 0.18 and 0.36 g input motion recorded at Nongpoh station
for AEC area
7 Conclusions
It is observed that if the soil is soft in nature with low SPT-N value, the PGA value
is magnified in the surface. When input motion value is scaled up i.e. by increasing
the energy content, keeping other parameters constant, it is observed that the surface
PGA value increases for all the sites compared to the original input motion. Response
spectra of the soil for all the sites gives an overview of the response shown by soil
layer when subjected to various ground motion with respect to time period. The peak
spectral acceleration obtained for 0.36 g (scaling) motion for both earthquake shows
much higher value than the spectral acceleration of rocky or hard soil sites (IS: 1893
[1]).
3 Response spectra (IS 3 Response spectra (IS

1893:2002, Rock or
Spectral acceleration (g) 1893:2002, Rock or hard

hard soil)
2.5 BH01_NGSTN_Mw 2.5 soil)
7.2 BH01_NGSTN_Mw7.2_
0.18g
2 2
BH03_NGSTN_Mw
7.2 BH03_NGSTN_Mw7.2_
1.5 0.18g
1.5
BH23_NGSTN_Mw
7.2 BH23_NGSTN_MW7.2
1 1 _0.18g
BH136_NGSTN_M
w7.2 BH136_NGSTN_Mw7.2
0.5 0.5 _0.18g
BH143_NGSTN_M
0 w7.2 0 BH143_NGSTN_Mw7.2
_0.18g
0.01 0.1 1 10 0.01 0.1 1 10
Period (sec) Period(sec)
5 Response spectra (IS
1893:2002, Rock or hard
4.5
soil)
4 BH01_NGSTN_Mw7.2_
0.36g
3.5
BH03_NGSTN_Mw7.2_
3 0.36g
2.5
BH23_NGSTN_Mw7.2_
2 0.36g
1.5 BH136_NGSTN_Mw7.2
1 _0.36g
0.5 BH143_NGSTN_Mw7.2
_0.36g
0
0.01 0.1 1 10
Period (sec)
Fig. 8 Spectral acceleration for 0.05, 0.18 and 0.36 g input motion recorded at Nongstoin station
for AEC area
Response spectra (IS 4.5 Response spectra (IS

4.5 1893:2002, Rock or 1893:2002 , Rock or
hard soil) 4 hard soil)
4 BH14_NGPH_Mw6_
BH07_NGPH_Mw6_0.
3.5 36g 3.5 0.36g
3 BH08_NGPH_Mw6_0. 3 BH13_NGPH_Mw6_
36g 0.36g
2.5
2.5
BH15_NGPH_Mw6_0, BH12_NGPH_Mw6_
2
2 36g 0.36g
1.5
1.5 BH16_NGPH_Mw6_0. BH11_NGPH_Mw6_
36g 1 0.36g
1
BH17_NGPH_Mw6_0. 0.5 BH10_NGPH_Mw6_
0.5 0.36g
36g
0
0 BH09_NGPH_Mw6_
BH18_NGPH_Mw6_0. 0.01 0.1 1 10
0.01 0.1 1 10 0.36g
36g
Period (sec) Period (sec)
4 Response spectra
(IS1893:2002, Rock
Spectral acceleration(g)
3.5 or hard soil)

BH02_NGPH_Mw6_
3 0.36g
2.5 BH04_NGPH_Mw6_
0.36g
2
BH05_NGPH_Mw6_
1.5 0.36g
1 BH19_NGPH_Mw6_
0.5 0.36g
0 BH21_NGPH_Mw6_
0.01 0.1 1 10 0.36g
Period(sec) BH26_NGPH_Mw6_
0.36g
Fig. 9 Spectral acceleration for 0.36 g motion of a Dharapur and Azara area, b VIP and Rani area
and c Jalukbari area for Indo-Burma Earthquake, 1997
References
1. IS: 1893 (Part I) (2002) Indian standard criteria for earthquake resistant design of structures,
fifth revision, part-1. In: Bureau of Indian Standards, New Delhi
2. Konder RL, Zelasko JS (1963) A hyperbolic stress-strain formulation of sands. In: Proceedings
of the 2nd pan American conference on soil mechanics and foundation engineering. Sao Paulo,
Brazil, pp 289–324
3. Kramer SL (1996) Geotechnical earthquake engineering. Prentice Hall, New Jersey (NJ), 653
4. Nath SK, Thingbaijam KKS, Raj A (2008) Earthquake hazard in Northeast India: a seismic
microzonation approach with typical case studies from Sikkim, Himalaya and Guwahati city. J
Earth Syst Sci 117:809–831
5. Sharma B, Rahaman SK ( Use of GIS based maps for preliminary assessment of subsoil of
Guwahati City. J Geosci Environ Prot 2016(4):106–116
6. Vucetic M, Dobry R (1991) The effect of soil plasticity on cyclic response. ASCE Geotech J
117(1):89–107
One-Dimensional Ground Response
Analysis to Arrive at Surface Peak
Ground Acceleration—A Case Study
of Golaghat District in Assam
A. F. Siddique, D. Dutta, and A. Deka
Abstract Ground Response Analysis (GRA) is required to be carried out for

predicting ground surface motions and to evaluate dynamic properties of soil during
an earthquake excitation. In this study one-dimensional equivalent linear (EL) and
nonlinear (NL) GRA was carried out for five boreholes located in Golaghat district of
Assam, India. This region falls under highly seismic zone ‘V’. The soil borehole log
obtained from standard penetration test (SPT) confirmed alluvial deposits with layers
of both coarse and fine-grained soils. The input motions of 2011 Sikkim earthquake
of Mw 6.9 recorded at Gangtok station and 1999 Uttarkashi earthquake of Mw 6.8
recorded at Bhatwari station are considered for the analysis. The results are plotted
in terms of peak ground acceleration (PGA), maximum stress ratio and maximum
strain with depth, and spectral acceleration over range of periods. On comparison
of EL and NL methods, the strain profile follows a similar trend along the depth
of borehole from both methods. However on comparison of both the methods, NL
methods showed maximum shear strain value. It was depicted that the PGA values
for a particular site of interest can be determined directly from the peak horizontal
acceleration (PHA) even in the absence of dynamic soil properties of the site. Such
PGA values can be directly used for earthquake-induced liquefaction analysis and
also for building design purposes.
Keywords Ground response analysis · Equivalent linear method · Nonlinear

method · Surface PGA · Spectral acceleration
A. F. Siddique
Department of Civil Engineering, Assam Engineering College, Guwahati 781013, India
e-mail: amaribnesiddiq98@gmail.com
D. Dutta · A. Deka (B)
Department of Civil Engineering, Central Institute of Technology Kokrajhar, Kokrajhar 783370,
India
e-mail: a.deka@cit.ac.in
D. Dutta
e-mail: dikshadutta1401@gmail.com
1 Introduction
The acceleration time history of a strong ground motion greatly changes when a
motion comes to the ground surface. When the motion travels through different soil
layers it gets amplified to the surface. Ground response analysis (GRA) is done to
know how much changes will occur in amount of acceleration when a motion travels
through different soil layers. In this report 1D equivalent linear (EL) and nonlinear
(NL) ground response analysis has been done for five borehole locations of Golaghat
district of Assam, India. The region is placed on a very high seismic zone, i.e. zone
V [1]. Alluvial deposits with layers of both coarse and fine-grained soils are found in
this region. Hence, large modifications in earthquake waves can occur due to variation
in soil properties near to the surface of earth. The amplification of clay soil is about
1.5 times higher than sandy soil [2]. Hence, the soil characteristics and response
of motion on the surface is very much essential for this region. Based on ground
response analysis, the correlation of amplification factor (AF) value as a function
of PHA is proposed for building design purposes. The advantage of the correlation
is that one can easily determine surface PGA values for the site for any earthquake
motions without undergoing ground response analysis.
2 Methodology
In this study both 1D equivalent linear (EL) and nonlinear (NL) methods are used to
perform ground response analysis. The input motion that is required for the purpose
is obtained from the seismograph station. The acceleration time history of 2011
Sikkim earthquake having bedrock PGA 0.152 g and 1999 Uttarkashi earthquake
having bedrock PGA 0.252 g are considered for the purpose of comparative study
between two methods. The motions are recorded at Gangtok station and Bhatwari
station, India which are rocky sites (site class A; density = 25 kN/m3 and shear wave
velocity = 1500 m/s).
The correlation between PHA and AF is also obtained for the site. For this purpose
an additional three earthquakes from the same tectonic regime viz. 1997 Indo-Burma
EQ recorded at Nongpoh station (PHA = 0.048 g), Loma Gilroy EQ (PHA = 0.357 g)
and Kobe EQ (PHA = 0.82 g) are considered. The input data consist of bulk density,
SPT N value, plasticity indexes and depth of water table. The shear wave velocity
(V s ) at each 1.5 m interval is calculated from the correlation with SPT N value (N)
using the Eq. 1 given by [3].
V s = 97 × N0.314 (1)
As the damping and modulus reduction curves are absent for the site, the curves
for sands [4] and for clay [5] are used for EL method [6]. Similarly the curve proposed
by Seed and Idriss (mean limit) for sandy soil and [5] for clay are considered for NL
One-Dimensional Ground Response Analysis … 121
method and they are subsequently fitted using MRDF procedure. The stress–strain
model [7] is used for performing NL ground response analysis incorporating Masing
criteria.
The results in terms of variation of PGA with depth, maximum stress ratio and
maximum strain over depth and spectral acceleration over range of periods are
resolved. Good comparative studies between both the methods are also done for the
soil profiles of the proposed site. It has been observed that stiffer soil layers usually
the deeper soil layers resulted in similar range of PGA by both the methods however
it get deviated when reaches to the surface. Figures 1 and 2 show the variation of
PGA with depth of five boreholes of Golaghat district of Assam for two different
ground motion 2011 Sikkim and 1999 Uttarkashi earthquake. It has been observed
that the EL surface PGA lie in the range of 0.27 g–0.37 g and 0.40 g–0.67 g against
2011 Sikkim and 1999 Uttarkashi EQ, respectively. Again NL surface PGA lies in the
range of 0.24 g–0.34 g and 0.30 g–0.55 g against 2011 Sikkim and 1999 Uttarkashi
EQ, respectively. From Table 1 it has been found that NL method generally gives
lower surface PGA values than EL method. This is attributed to changing secant
modulus due to dynamic stress–strain characteristics at every time step. Figure 3
shows the comparison of EL and NL methods for BH01 of the site. It has been found
that up to 5 m from the depth the variation of PGA from both the methods are almost
in similar trend but above 5 m it get deviated and NL method shows slightly lesser
value of PGA than EL method.
Fig. 1 Variation of PGA with depth against Sikkim earthquake (EL method)
Fig. 2 Variation of PGA with depth against Uttarkashi earthquake (EL method)
Table 1 Surface PGA values of five borehole sites

Borehole Sikkim 0.152 g Uttarkashi 0.252 g
EL NL EL NL
BH01 0.28 0.24 0.42 0.30
BH02 0.34 0.33 0.40 0.55
BH03 0.27 0.27 0.53 0.42
BH04 0.36 0.34 0.67 0.52
BH05 0.37 0.29 0.64 0.50
Fig. 3 Comparison of PGA profile at BH01 of Golaghat

Fig. 4 Variation of strain (%) with depth at BH01 of Golaghat
Figure 4 shows the shear strain profile at Borehole 01 of Golaghat area. On

comparison of EL and NL methods, the strain profile follows a similar trend along the
depth of borehole from both methods. However on comparison of both the methods,
NL methods show maximum shear strain value. Again the borehole experienced
maximum strain against Uttarkashi motion compared to Sikkim motion due to high
bedrock PHA and high energy content of Uttarkashi motion. The peak of maximum
strain value is obtained at about 5 m because of having soft soil (N = 2). Strains in
soft soil are generally large due to the low stiffness. Table 2 represents the values of
maximum strain at five boreholes site for both EL and NL methods.
A completely opposite picture is seen from the maximum shear stress ratio profile
shown in Fig. 5 where the value of shear stress ratio is obtained higher by EL methods
than NL methods. On comparing both the motions, Sikkim motion shows a lesser
value of shear stress ratio due to lesser energy content and lesser bedrock PGA value
of motion.
The response spectra in terms of spectral acceleration (SA) for 5% damping at
BH01 of the area is shown in Fig. 6. The SA is obtained against 2011 Sikkim and
1999 Uttarkashi earthquake by using both EL and NL analysis. It is found that EL
analysis gives maximum SA value as compared to NL analysis. The SA values for
BH01 against two motions are below the SA of [1], Rock or Hard soil. The spectral
Table 2 Maximum strain (%) values at five borehole sites

EL NL EL NL
BH01 0.497 8.904 1.782 10.070
BH02 0.489 0.523 5.072 2.010
BH03 0.101 0.113 0.323 0.388
BH04 0.177 0.168 0.496 3.201
BH05 0.131 0.136 0.634 0.538
Fig. 5 Variation of maximum stress ratio with depth at BH01 of Golaghat
Fig. 6 Spectral acceleration for BH01 using both EL and NL analysis
acceleration values at five boreholes of Golaghat are shown in Table 3. It is obtained

that for all the boreholes the SA value against Uttarkashi motion are much higher
than Sikkim motion for both EL and NL methods. This is due to high magnitude and
high duration of Uttarkashi earthquake motion.
Figure 7 shows the variation of AF with bedrock PHA for five boreholes of
Golaghat district during different earthquakes. The Indo-Burma earthquake (1997)
having bedrock PHA 0.048 g recorded at Nongpoh station, Sikkim earthquake (2011)
having bedrock PHA 0.152 g recorded at Gangtok station, Uttarkashi earthquake
(1999) having bedrock PHA 0.252 g recorded at Bhatwari station, Loma Gilroy
earthquake having bedrock PHA 0.357 g and Kobe earthquake having bedrock PHA
0.82 g are considered for the analysis. 1D NL GRA was performed for the five
boreholes against the above five different earthquakes and amplification factors (AF)
Table 3 Spectral acceleration (g) values at five borehole sites

EL NL EL NL
BH01 0.94 0.67 1.50 0.95
BH02 1.39 1.44 1.58 2.23
BH03 1.28 1.22 2.04 1.67
BH04 1.76 1.63 2.55 1.75
BH05 1.56 1.32 1.81 1.65
Fig. 7 Plot of AF versus PHA for five boreholes of Golaghat district during different earthquakes
are calculated which is defined as the ratio of surface PGA to bedrock PHA. The
AF’s are plotted with respect to bedrock PHA for all the boreholes sites. Similar
thickness and type of soil deposit might show similar response during an EQ [8].
This might be the reason why the AF variation with respect to PHA values show
similar variation pattern for all the selected sites as shown in Fig. 7. To give a general
equation between AF and PHA, a correlation between AF and PHA is proposed for
Golaghat district as shown in Fig. 7. It has been observed that at PHA < 0.2 g, the
AF changes abruptly, however at PHA > 0.2 g the variation of AF is gradual [9].
Therefore, no linear relation is possible between AF and PHA.
A logarithmic function of the form y = a ln(x) + b, correlates dependent variable
y with an independent variable x, where a and b are two regression constants. For
the present problem a logarithmic function is found to give highest R2 value. For this
a logarithmic function is selected for correlating AF-PHA. The obtained correlation
is shown in the Fig. 7. The main intention of this correlation is that by knowing
this correlation one can simply determine AF and surface PGA values without doing
GRA based on PHA for Golaghat. The surface PGA can be calculated by multiplying
AF with PHA values. Based on this proposed correlation the PGA values for a
particular site of interest can be determined directly from the PHA even in the absence
of dynamic soil properties of the site. Such PGA values can be directly used for
earthquake-induced liquefaction analysis and also used for building design purpose.
4 Conclusions
Study of soil characteristics together with the liquefaction potential study is very
essential for this region. The following points can be summarised by this study:
• The motion is amplified to the surface when it travels through different soil layers.
1D equivalent linear and nonlinear ground surface analysis has been done for 5
boreholes locations of Golaghat district of Assam, India.
• Amplification of clay soil is 1.5 times higher than sandy soil. A similar range of
PGA by both methods was observed, i.e. for stiffer soil layers.
• It was seen that NL method generally gives lower surface PGA values than EL
methods.
• Up to a depth of 5 m the variation of PGA from both the methods are almost in
similar trend but above 5 m it gets deviated and NL method shows slightly lesser
value of PGA than EL methods.
• On comparing both the motion Sikkim motion shows a lesser value of shear stress
ratio due to lesser energy content and lesser bedrock PGA value of motion.
• It has been observed that at PHA < 0.2 g, the AF changes abruptly; however at
PHA > 0.2 g, the variation of AF is gradual.
By these conclusions we can easily determine the AF and surface PGA values
without doing GRA based on PHA for Golaghat. When AF and PHA values are
multiplied we get the surface PGA. Even in the absence of dynamic soil properties
the surface PGA values can be calculated by the prescribed correlation.
References
1. IS 1893–Part-1 (2002) Indian standard criteria for earthquake resistance design of structures.
BIS, New Delhi
2. Boominathan A, Dodagoudar GR, Suganthi A, Maheswari UR (2008) Seismic hazard assessment
of Chennai city considering local site effects. J Earth Syst Sci 117(S2):853–863
3. Imai T, Tonouchi K (1982) Correlation of N value with S-wave velocity and shear modulus. In:
Proceedings of 2nd European symposium on penetration testing, Amsterdam, pp 67–72
4. Seed HB, Idriss IM (1970) Soil moduli and damping factors for dynamic response analyses.
Technical report EERRC-70-10, University of California, Berkeley
5. Vucetic M, Dobry R (1991) The effect of soil plasticity on cyclic response. ASCE Geotech J
117(1):89–107
6. Hashash YMA, Musgrove MI, Harmon JA. Groholski DR, Phillips CA, Park D (2016)
DEEPSOIL version 6.1. User manual, 137 pp
7. Kondner RL, Zelasko JS (1963) A hyperbolic stress-strain formulation of sands. In: Proceedings
of the 2nd Pan American conference on soil mechanics and foundation engineering, Sao Paulo,
Brasil, pp 289–324
8. Kumar A, Harinarayan NH, Baro O (2017) Nonlinear soil response to ground motions during
different earthquakes in Nepal, to arrive at surface response spectra. Nat Hazards 2017(87):13–
33. https://doi.org/10.1007/s11069-017-2751-4
9. Kumar A, Baro O, Harinarayan NH (2016) Obtaining the surface PGA from site response
analyses based on globally recorded ground motions and matching with the codal values. Nat
Hazards 81:543–572. https://doi.org/10.1007/s11069-015-2095-x
Elastic Seismic Design Response Spectra
for Deep and Shallow Basin of the Indian
Subcontinent
Ketan Bajaj and P. Anbazhagan
Abstract In the present study, new elastic design response spectra (EDRS) for the
deep and shallow region of the Indian subcontinent has been proposed for rock and
soil site. For determining the EDRS for bedrock, 50 each rock recorded ground
motions have been used for Intra and Interplate region. For determining the EDRS
for soil sites, 250 and 178 ground motion for deep and shallow sites has been used,
respectively. For SI, only few ground motions are available, hence for developing
EDRS for shallow sites, ground motions recorded at similar tectonics have been used
in the present study. Further, EDRS is derived based on Eurocode, i.e. normalized
elastic design response spectra which is based on one parameter, i.e. effective ground
acceleration at rock.
Keywords Response spectra · Ground motion · Deep and shallow basin · Seismic
site classification
1 Introduction
Local site conditions have great influence on ground surface motion and structural
damage caused by an earthquake event. The Indian Subcontinent (IS) has one of the
most diverse seismotectonic and seismicity. The high level of seismicity is associated
with the Himalaya tectonic province will result in site amplification in the contiguous
deep alluvial deposits named Indo-Gangetic Basin (IGB), due to any major earth-
quake in future. Whereas, low to moderate level of seismicity in the Southern India
(SI) causing high amplification due to shallow thin layers. Evidence from past earth-
quakes suggested that soft soil sites tend to amplify at low frequency, whereas rock
sites have significant intensity at high frequencies. Anbazhagan et al. [1] highlighted
that local site effect is the major factor that causes the damage due to an earthquake.
K. Bajaj (B) · P. Anbazhagan

Department of Civil Engineering, Indian Institute of Science, Bengaluru, India
e-mail: ketanbajaj@iisc.ac.in
P. Anbazhagan
e-mail: anabazhagan@iisc.ac.in
130 K. Bajaj and P. Anbazhagan
2001 Bhuj (7.7 Mw ), 1999 Chamoli (6.8 Mw ), 2011 Sikkim (6.9 Mw ), and 2015
Nepal (7.8 Mw ) earthquakes are the recent examples that explained the effect of
thick deposits on site-specific damage in the IS. Various researchers [e.g. 2–4] have
studied the local site effect and estimated amplification factors using the 1D site
response study but most of these studies are limited to soil column of 30 m depth.
Moreover, in the previous site response studies, the input ground motions were either
selected randomly from global database or simulated based on the occurred earth-
quake scenario. In India very limited attempts have been made to drive the design
spectrum with different damping level considering regional recorded data.
In most of the modern seismic design, the estimation of seismic force of a typical
structure is based on the 5% damped design response spectrum of recorded data in
the region. In India limited attempts have been made to the drive design spectrum
with different damping level considering regional recorded data, some attempt has
been made by Anbazhagan et al. [5]. Generally, the design spectra of a given site
are obtained by modifying the uniform hazard spectrum by considering site factors
corresponding to a particular seismic site. Conventionally, the design force is speci-
fied via. response spectrum amplitude. However, with the increased complexity of the
modern structure and understanding the seismic performance of structure demands,
it is now essential to define the amplitude and shape of the design spectra.
This paper aims at the development of new elastic design response spectra for Intra
and Interplate region of the Indian subcontinent considering region-specific seismic
data. Further, EDRS is derived based on Eurocode, i.e. normalized elastic design
response spectra which is based on one parameter, i.e. effective ground acceleration
at rock.
2 Study Area
The Indo-Gangetic Basin is the foredeep depression that is situated between the
Indian Peninsular shield and the Himalayan region. IGB lies roughly between longi-
tude 74o E, and 88o E and latitude 24o N and 32o N (Fig. 1a). The sediment depth
varies from few tens of meter in the south part of the IGB and progressively increasing
up to ~5–6 kms in the northernmost part. High neotectonic activity and reactivation
of tectonic features and lineaments are acknowledged by various researchers [e.g. 6
etc.]. IGB is contiguous to the most seismically active Himalayan region and experi-
encing the strong compressional stress conditions. Any large to moderate earthquake
in the Himalayan region may result in massive destruction in the IGB due to site
amplification and liquefaction.
Southern India is considered as one of the oldest geologically evolved and tectoni-
cally stable continental crust of the IS. The seismotectonic of the SI is majorly consist
of various faults, ridges, shear zones, and tectonic lineaments. Various researches
[e.g. 7, etc.] defined the tectonic feature of the SI and many authors reported the
reactivation of fault along the western part of the Peninsular India. Additionally, SI
Elastic Seismic Design Response Spectra for Deep … 131
Fig. 1 Study area used in the present study
is having an irregular seismicity. As micro-seismicity is reported in the South Gran-

ulite Terrain, Eastern Dharwar craton is surrounded by intermediate seismicity, and
Koyna-Warna region and Deccan Volcanic Province has high seismicity. The SI is
also marked in Fig. 1b.
3 Methodology
Malhotra [8] concluded that the response of the structure derived using acceleration
time history does not correspond to the velocity and displacement time histories.
Response of the flexible structures (long period) can be contradictory if computed
only using processed acceleration time history [8]. Based on that, Malhotra [8]
proposed a methodology to compute elastic response spectra for incompatible accel-
eration, velocity, and displacement time histories. The following is the procedure
recommended by Malhotra [8] for deriving the normalized response spectra and used
in the present study. For determining the EDRS for bedrock, 50 each rock recorded
ground motions have been used for Intra and Interplate region. For determining the
EDRS for soil sites, 250 and 178 ground motion for deep and shallow sites has been
used, respectively. For SI, only few ground motions are available, hence for devel-
oping EDRS for shallow sites, ground motions recorded at similar tectonics have been
used in the present study. After smoothening the recorded ground motion, the elastic
design response spectra have been defined considering the following equations

Sa (T ) T
0 ≤ T ≤ TB : = s. 1 + .(β − 1) (1)
P G Ar ock TB
Sa (T )
TB ≤ T ≤ TC : = s.β (2)
P G Ar ock
Sa (T ) TC
TC ≤ T ≤ TD : = s.β (3)
P G Ar ock T
Sa (T ) TD
TD ≤ T : = s.β.TC 2 (4)
P G Ar ock T
Here, P G Ar ock is the design ground acceleration at rock-site conditions, S and β

are the soil amplification and spectral amplification factors. TB and TC are the limits
of constant acceleration branch and TD is the beginning of the constant displacement
range of the spectrum. Distributing the spectra into two regions helps in better repre-
sentation of the shape of the response spectra. β, TB , TC , and TD are determined
based on the shape of the normalized response spectra. S and β have been deter-
mined considering the methodology proposed by Pitilakis et al. [9]. Typical EDRS
is given as Fig. 2. The amplification factor corresponding to acceleration, velocity,
and displacement is denoted as α A , αV , and α D . T1 , T2 , T3 , T4 , T5 , and T6 in Fig. 7.2
is denoted as control periods (where the straight-line segments meet).
Fig. 2 Typical EDRS derived in this study
4 Result and Discussion
Pitilakis et al. [9] defined different approaches in determining the soil amplification
(S). In this study, it is calculated considering the period-independent amplification
factor with respect to rock site. The period-independent amplification factor for soil
site and different magnitude at an interval of 0.5 with respect to rock is calculated as
S = (Isoil /Ir ock ).(1/S R) (5)
here, S R is the spectral ratio, Isoil and Ir ock are the spectral intensities for soil and
rock, respectively. Housner [10] defined the response spectrum intensity as
2.5
I = P SV (ξ, T )dT (6)
0.05
i.e. the area under the pseudo velocity response spectrum between periods 0.05 to
2.5 s. To compute the Ir ock for IGB and SI, the area under the PSV curve has been
determined. Similarly, the area under the PSV curve for soil site is computed and
Isoil /Ir ock has been determined. S R reflects the difference in shape of soil spectra
with respect to rock spectra, when all the spectra are normalized and have the same
ordinate at starting period. The spectral amplification factor (β) is analogues to α A
(see Fig. 3) is also determined for both the regions and given as Table 1. From Table
1, it has been observed that S is 1.22 in case of IGB and 1.13 in case of SI for soil
site. Similarly, β is 2.91 in case of IGB and 2.21 in case of SI, for soil.
The parameters β, TB , TC , and TD have been determined using the normalized
spectrum. These factors are the result of fitting the smooth spectrum to the median
Fig. 3 Typical smooth medium response spectrum considering
Table 1 Parameters of the proposed design response spectrum for 5% damping

Parameters Soil Rock
IGB SI IGB SI
TB 0.16 0.12 0.15 0.08
TC 0.52 0.29 0.38 0.28
TD 1.90 1.41 2.33 1.52
S 1.22 1.13 1.00 1.00
β 2.91 2.21 2.29 2.81
normalized spectrum (see Fig. 3). The values are given in Table 1. The derived
spectrum for the IGB and SI is given as Fig. 4a, b, respectively. The proposed spectra
in this case depend on the effective ground acceleration at rock sites (i.e. PGA) and
soil amplification factors (S, β), whereas control periods remain constant for the
corresponding site class. The derived EDRS is also compared with design spectrum
given in Indian standard code (IS:1893) and given in Fig. 4.
5 Conclusion
In the present study, new elastic design response spectra for the deep and shallow soil
sites in Inter and Intraplate regions of the Indian subcontinent have been proposed.
EDRS for bedrock is determined by considering 50 each rock recorded ground
motions for Intra and Interplate region. For determining the EDRS for soil sites,
250 and 178 ground motion for deep and shallow sites has been used, respectively.
Fig. 4 Design response spectra for deep and shallow basin of India for bedrock and soil sites
EDRS is derived based on Eurocode, i.e. normalized elastic design response spectra
which is based on one parameter, i.e. effective ground acceleration at rock. Response
spectra for IS:1893 are underestimating the spectral values at long and short periods.
References
1. Anbazhagan P, Kumar A, Sitharam TG (2010) Site response of Deep soil sites in Indo Gangetic
plain for different historic earthquakes. In: Proceedings of the 5th international conference on
recent advances in geotechnical earthquake engineering and soil dynamics, vol 3.21b. San
Diego, California, p 12
2. Boominathan A, Dodagoudar GR, Suganthi A, Maheshwari RU (2008) Seismic hazard
assessment of Chennai city considering local site effects. J Earth Syst Sci. 117(S2):853–863
3. Anbazhagan P, Sitharam TG (2008) Site characterization and site response studies using shear
wave velocity. J Seism Earthq Eng. 10(2):53–67
4. Naik N, Choudhury D (2013) Site specific ground response analysis for typical sites in Panjim
city, Goa. In: Proceedings of Indian geotechnical conference, Roorkee, India
5. Anbazhagan P, Bajaj K, Reddy GR, Phanikanth VS, Yadav DN (2016) Quantitative assessment
of shear wave velocity correlations in the shallow bedrock sites. Indian Geotech J 46(4):381–397
6. Singh IB (1996) Geological evolution of Ganga plain – an overview. J Paleont Soc India
41:99–137
7. Ramasamy SM (2006) Remote sensing and active tectonics of south India. Int J Rem Sens
27(20):4397–4431
8. Malhotra PK (2001) Response spectrum of incompatible acceleration, velocity and displace-
ment histories. Earthquake Eng Struct Dyn 30(2):279–286
9. Pitilakis K, Riga E, Anastasiadis A et al (2013) New code site classification, amplification
factors and normalized response spectra based on a worldwide ground motion database. Bull
Earthq Eng 11(4):925–966
10. Housner GW (1952) Spectrum intensities of strong-motion earthquakes. In: Proceedings of
symposium on earthquakes and blast effects on structures, earthquake engineering research
institute, pp 20–36
Case Studies on Preshaking
and Reliquefaction Potential
for Different Earthquakes in Japan
Gowtham Padmanabhan and B. K. Maheshwari
Abstract The paper examines the effect of liquefaction potential of the site due to
normal earthquake and sequential earthquakes occurred in Japan. Two earthquakes
have been considered (a) 2016 Kumamoto sequential earthquakes with mainshock
(Mw = 7.3) and a couple of foreshock in the previous days (Mw = 6.2 and Mw = 6.5);
and (b) 2005 Fukuoka-ken Seiho-oki earthquake (Mw = 7.0), the earthquake has only
mainshock. With the help of available literatures, damages associated with the lique-
faction was examined with reference to effect of preshaking, liquefaction history and
earthquake pattern. It is evident that, liquefaction effects of the 2005 earthquake are
larger than those associated with Kumamoto sequential earthquakes. It is inferred
that liquefaction resistance and reliquefaction potential of the site was highly influ-
enced by the seismic preshaking, liquefaction history and earthquake pattern. This
is the most probable explanation for the higher liquefaction resistance exhibited by
the Kumamoto earthquake. Preshaking effect is highly associated with earlier earth-
quake history or significant foreshocks as in the 2016 Kumamoto earthquakes. The
occurrence of sequential earthquakes induces reliquefaction phenomenon in the site.
It is concluded that, liquefaction history and preshaking effect of the site contributed
to the enhanced liquefaction resistance of the site.
Keywords Sequential earthquakes · Liquefaction history · Preshaking ·

Reliquefaction
1 Introduction
Earthquakes are known for its complex nature, as its return period and frequency
cannot be predicted accurately. It has been known to occur successively at the same
site within a short period of time, as shown by the examples from Christchurch
G. Padmanabhan (B) · B. K. Maheshwari

Department of Earthquake Engineering, IIT Roorkee, Roorkee, India
e-mail: gowtham@eq.iitr.ac.in
B. K. Maheshwari
e-mail: bkmahfeq@iitr.ac.in
138 G. Padmanabhan and B. K. Maheshwari
earthquake 2010–2011, Japan earthquake 2011, Nepal earthquake 2015, Kumamoto

earthquake 2016, Indonesia earthquake 2018 and recent Philippines earthquake 2019
[1, 2, 8, 10, 16, 20].
Soil liquefaction is an intricate phenomenon occurs due to the generation of excess
pore water pressure due to seismic shaking under undrained soil conditions. Lique-
faction of saturated loose sands during earthquakes has been the cause of severe
damage to various buildings, embankments and retaining structures. The liquefaction
was first reported in 1964 Niigata earthquake, Japan and in Good Friday earthquake,
Alaska in the same year and has been extensively studied since [12, 15, 23]. Some
recent earthquakes with the major liquefaction damages are the 2010 Christchurch,
2011 Japan, 2015 Nepal and the recent 2018 Indonesia earthquakes.
Sequential earthquakes are in the form: (1) couple of strong earthquakes in short
interval (2) significant foreshocks followed by mainshock (3) mainshock followed by
couple of aftershock (4) combination of both foreshock and aftershock. Sequential
earthquakes resulted in the preshaking of the site, that alters the geological proper-
ties of the soil. From the experimental studies, it is concluded that preshaking was
responsible for the densification of loose sand deposits resulted in enhanced lique-
faction resistance [3–7]. Seismic preshaking of a zone results in a decreasing risk of
future liquefaction [3, 4]. Each previous instance of liquefaction allows soil to build
up resistance to future liquefaction in terms of densification or change in orientation
of soil deposits. Reliquefaction phenomenon was clearly observed during the 2011
Japan sequential earthquake, in which many man-made and natural sites relique-
fied. Based on above literature, it can be concluded that, very limited work has been
reported for the effect of sequential earthquakes on the liquefaction potential. This
has been examined in the present study as discussed in the scope.
2 Scope of the Paper
The paper is an attempt toward understanding the influence of preshaking and lique-
faction history in developing liquefaction resistance. The potential of sequential
earthquakes in inducing the reliquefaction and the resistance associated with this
phenomenon is presented. Case studies considering the seismic preshaking and
earthquake and liquefaction history was rarely reported.
The existing reports focused on earthquake parameters and characteristics,
damages associated with earthquakes. This paper is a first of its kind in comparing
the two earthquakes based on the earthquake pattern, preshaking history, liquefaction
history and the liquefaction damages associated with the site.
Case Studies on Preshaking and Reliquefaction … 139
3 Case Study I—Kumamoto Sequential Earthquakes
3.1 Sequential Earthquakes, Location and Characteristics
A sequential earthquake is a sequence of seismic events occurring in a local area

within a relatively short period of time. The successive earthquakes were recorded
for a period of two months. From USGS data it is evident that, several foreshocks
and aftershocks hit the same region along with mainshock. Black dots represent
microearthquakes with magnitude <5 and yellow circles represent medium earth-
quakes with magnitude greater than or equal to 5 and lesser than 6, Pink circles
represent the strong events which are accounted in the study with magnitude ≥6 [17].
Table 1 shows the detailed list of the significant events which contributed for the
preshaking and reliquefaction. Earthquakes with magnitude ≥6 are considered as
major events that have potential to induce preshaking and reliquefaction in the site.
The Kumamoto earthquake was observed in the year 2016 with a sequence of
two foreshocks of moment magnitude 6.5 on 14th April at 21.26 local time, the
hypocentral parameters include 32.74 N 130.81 E at a depth of 11 km and 6.2 on
15th April at 00:03 local time and the location of the focus was 32.70 N 130.78 E
at a hypocenter depth of seven kilometers (USGS 2016). A mainshock of moment
magnitude 7.0 at 01:25 local time, at a location of 32.75 N 130.76E and a depth of
12 km was occurred on 16th April [17, 18, 19]. The three major earthquake events
each had high-intensity ground motions with recorded PGAs ranging from 0.2 to
1.2 g. The 2016 Kumamoto earthquake ground motions were sufficient enough to
induce required Cyclic Stress Ratio (CSR) intensity and duration for liquefaction in
the Kumamoto plain areas.
Though major events alone are considered in the present study, the events with
magnitude ≤6 are also acknowledged for the contribution in preshaking effect and
resulted in the development of liquefaction resistance and reliquefaction. Table 2
provides the details of earthquake events based on magnitude scale, which was
recorded in the Kumamoto region for a period of two months from April 01 to
May 31, 2016. A total of 1157 events were observed in the specified duration within
the Kumamoto region.
Table 1 Details of
Date and time Moment magnitude
Kumamoto earthquakes,
2016 [17]. Location: 14 April 2016 21:26 JST (12:26 UTC) 6.5
Kumamoto Chiho of 15 April 2016 00:03 JST (14 April 2016 6.2
Kumamoto Prefecture 15:03 UTC)
16 April 2016 01:25 JST (15 April 2016 7.3
16:25 UTC)
Table 2 Number of seismic

Moment magnitude Number of earthquakes
observations in Kumamoto
region [18, 19] 7 1
6 2
5 12
4 80
3 224
2 461
1 377
3.2 Seismic Preshaking History of Site
For better understanding of seismic preshaking history of the site, sequential

occurrence of earthquakes or foreshock/aftershock events alone are not sufficient.
The earthquake history of the corresponding site should also be considered for
accurate prediction of preshaking effect in minimizing damages associated with
earthquake-induced liquefaction [1].
Table 3 shows the earthquake history of the Kumamoto region. Kumamoto region
is identified as seismically active zone in Japan. 15 major events with moment magni-
tude ≥6 were recorded in the region since sixteenth century witness the seismic
potential of the site [13, 14].
Table 3 Earthquake history of the Kumamoto region

S. No Date dd/mm/yyyy Region and name Moment magnitude
1 01/05/1619 Higo and Yatsushiro 6.0
2 21/06/1625 Kumamoto 6.1
3 28/10/1707 Houei earthquake 8.4
4 19/12/1723 Higo, Bungo and Chikugo 6.5
5 29/08/1769 Higo, Bungo and Chikugo 7.8
6 21/05/1792 Unzendake 6.4
7 24/12/1854 Ansei Nankai earthquake 8.4
8 28/07/1889 Kumamoto 6.3
9 19/11/1941 Hyuga-nada sea 7.2
10 21/12/1946 Nankai earthquake 8.0
11 23/01/1975 Northern edge of Mount Aso 6.1
12 17/01/1995 Southern Hyogo Prefecture earthquake 7.3
13 14/04/2016 Kumamoto 6.5
14 15/04/2016 Kumamoto 6.2
15 16/04/2016 Kumamoto 7.0
3.3 Liquefaction History of Site
Earlier to the 2016 earthquake, Kumamoto city was indicated as being a moderate
to significant liquefaction vulnerable zone [10].
The liquefaction data of the Kumamoto region was shown in Table 3. From the
table, it is clear that, Kumamoto city and the surrounding regions are highly suscep-
tible to liquefaction. Around 11 cities are considered in present study to evaluate
the liquefaction damages associated with the sequential earthquakes. The earlier
earthquakes recorded in the corresponding region shown in Table 3 induced higher
liquefaction susceptibility.
3.4 Liquefaction Damages
Liquefaction seems to have been a concern in the 2016 sequential Kumamoto

earthquake; given the high-intensity ground motions.
Despite higher liquefaction susceptibility and high-intensity ground motions, the
damages associated with the liquefaction are limited when compared to earlier earth-
quakes. The factors hindered the liquefaction damages are (a) Occurrence of relique-
faction, which develops the liquefaction resistance of the sand deposits by the densi-
fication effect (b) Effect of seismic preshaking has a significant role in increasing the
liquefaction resistance due to the reorientation of sand particles.
4 Case Study II—Fukuoka-ken Seiho-oki Earthquake,

2005
4.1 Earthquake Location and Characteristics
The strong earthquake occurred at 10:53 am on March 20, 2005 in the Kyushu region,
about 70 km west of shimonoseki city in Yamaguchi prefecture. The hypocentral
parameters include 33.90 N 130.20 E at a depth of 9 km [14]. The depth of the
earthquake was extremely shallow, with a presumed Richter scale of 7.0. Fukuoka
is not as seismically active as many other parts of Japan, and was known prior to
the earthquake as one of Japan’s safest locations in terms of natural disasters; the
previous earthquake, a magnitude 5, had occurred over a hundred years ago and it
had been centuries since the city had experienced a serious earthquake [21, 22]. A
quake measuring about seven on the Richter scale struck three hundred years ago,
but no major quake has struck after that event [14].
4.2 Liquefaction Damages
Liquefaction was observed mainly in the reclaimed lands of the Hakata bay area.
Severe damages were identified in the structures and pavements due to liquefaction.
The earthquake-induced liquefaction damages were found to be very severe and
resulted in loss of human life and economic resources [14].
The considered region is free from seismic preshaking, reduced liquefaction
susceptibility, absence of foreshocks/aftershocks. Still, liquefaction damages associ-
ated with the Fukuoka-ken Seiho-oki Earthquake found to be significant. This shows
that, liquefaction resistance of the sand deposits depends on several factors.
5 Discussions
1. In the present study, two sites with different geological compositions, seismic
history, liquefaction history and earthquake pattern were considered. The extent
and severity of the liquefaction damages were found to be higher in 2005
Fukuoka-ken Seiho-oki earthquake, when compared to the 2016 Kumamoto
earthquakes. It is due to the reorientation of the sand particles and effect of
densification observed in the site.
2. The liquefaction resistance of the sand deposits increases with increase in number
of earthquakes recorded in the site. This pattern was observed in the 2016
Kumamoto sequential earthquakes. Three earthquakes with magnitude >6 hit the
same region, seismic preshaking takes place and resulted in the enhanced lique-
faction resistance. Whereas in case of 2005 Fukuoka-ken Seiho-oki Earthquake,
foreshocks/aftershocks are not observed.
3. The reliquefaction phenomenon is observed in the 2016 Kumamoto earthquake
as a result of successive foreshocks and mainshock. Densification mechanism
is associated with the occurrence of reliquefaction. As a result, liquefaction
resistance of sand deposits increased to a greater extent.
4. Earthquake shaking was so strong in 2016 Kumamoto region comparatively
to 2005 Fukuoka-ken Seiho-oki earthquake, still the damages associated with
liquefaction are lesser. It is strongly suggested that, there must be other factors
that hindered liquefaction potential in Kumamoto region.
5. All of these discussions are valid only for the specific site and magnitude of
earthquakes considered in the study. Further research is needed to investigate the
combined influence of seismic preshaking and occurrence of reliquefaction in
reduced liquefaction damages in the sand deposits.
6. The work presented in the paper may be helpful for analyzing and comparing
liquefaction history and preshaking effect in liquefaction resistance and relique-
faction potential of sandy soils.
6 Conclusions
From this study, following conclusions can be drawn.

1. The liquefaction resistance of the sand deposits was highly influenced by the
intense seismic preshaking history of the site.
2. The reliquefaction potential of the sites strongly depend on the characteristics of
the earthquake motions.
3. Successive earthquakes or combination of foreshocks/aftershocks events are
responsible for occurrence of reliquefaction phenomenon.
4. A site liquefied by earthquake loading can also be reliquefied by a subsequent
earthquake even when the magnitude caused by a later was less than that caused
by previous earthquake.
5. The liquefaction history of the site also contributed to liquefaction resistance,
higher the liquefaction history; higher will be the liquefaction resistance.
References
1. Anderson DJ (2019) Understanding soil liquefaction of the 2016 Kumamoto earthquake

2. Cubrinovski M et al (2011) Soil liquefaction effects in the central business district during the
February 2011 Christchurch earthquake. Seismol Res Lett 82(6):893–904
3. El-Sekelly WE (2014) The effect of seismic preshaking history on the liquefaction resistance
of granular soil deposits. Rensselaer Polytechnic Institute
4. El-Sekelly W, Abdoun T, Dobry R (2015) Liquefaction resistance of a silty sand deposit
subjected to preshaking followed by extensive liquefaction. J Geotech Geoenviron Eng
142(4):04015101
5. El-Sekelly W et al (2018) Experimental simulation of the effect of preshaking on liquefaction
of sandy soils. In: Physical modelling in geotechnics, vol 2. CRC Press, pp 949–953
6. Goda K et al (2014) Sensitivity of tsunami wave profiles and inundation simulations to
earthquake slip and fault geometry for the 2011 Tohoku earthquake. Earth Planets Space
66(1):105
7. Goda K et al (2015) The 2015 Gorkha Nepal earthquake: insights from earthquake damage
survey. Front Built Environ 1:8
8. JMA. https://www.jma.go.jp/jma/indexe.html. Accessed 12 Dec 2019
9. Kayen RE et al (2017) Geotechnical aspects of the 2016 MW 6.2, MW 6.0, and MW 7.0
Kumamoto earthquakes. Geotechnical Extreme Events Reconnaissance Association
10. Kiyota T et al (2017) Geotechnical damage caused by the 2016 Kumamoto earthquake, Japan.
ISSMGE Int J Geoeng Case Histor 4(2):78–95
11. Kokusho T et al (2015) Energy-based liquefaction potential evaluation and its application to
a case history. In: Proceedings of 6th international conference on earthquake geotechnical
engineering, Christchurch, NZ, Paper
12. Kramer S (1996) Geotechnical earthquake engineering. Prentice-Hall, Inc., New Jersey, pp
348–422
13. Mukunoki T et al (2016) Reconnaissance report on geotechnical damage caused by an
earthquake with JMA seismic intensity 7 twice in 28 h, Kumamoto, Japan. Soils Found
56(6):947–964
14. Nagase H et al (2005) Zoning for liquefaction and actual damage during the 2005 Fukuokaken-
Seiho-Oki earthquake. In: Geotechnical engineering for disaster mitigation and rehabilitation:
(With CD-ROM), pp 293–298
15. Prakash S (1981) Soil dynamics. McGraw-Hill Companies
16. Towhata I, Goto H, Kazama M, Kiyota T, Nakamura S, Wakamatsu K, Wakai A, Yoshida N
(2011) On gigantic Tohoku Pacific earthquake in Japan. Earthquake News, Bull Int Soc Soil
Mech Geotech Eng 5(2)
17. USGS. https://earthquake.usgs.gov/earthquakes/eventpage/usp000djku/executive. Accessed 5
Dec 2019
18. USGS. https://earthquake.usgs.gov/earthquakes/eventpage/official20110311054624120_30/
executive. Accessed 1 Dec 2019
19. USGS. https://earthquake.usgs.gov/earthquakes/eventpage/us20005hzn/executive. Accessed
12 Dec 2019
20. Wakamatsu K (2000) Liquefaction history, 416–1997, Japan. In: Proceedings of 12th world
conference on earthquake engineering, Paper no 2270
21. Yasuda S, Matsumura S (1991) Microzonation for liquefaction, slope failure and ground
response during earthquakes in Fukuoka City. Proc 4th Int Conf Seis Zonat 3
22. Youd TL (1984) Recurrence of liquefaction at the same site. Proc 8th World Conf Earthq Eng
3. Prentice-Hall Inc
23. Youd TL, Izzat MI (2001) Liquefaction resistance of soils: summary report from the 1996
NCEER and 1998 NCEER/NSF workshops on evaluation of liquefaction resistance of soils. J
Geotech Geo-environ Eng 127(4):297–313
Seismic Response of Basal Geogrid
Reinforced Embankments Supported
on a Group of Vertical and Batter Piles
Radhika M. Patel, B. R. Jayalekshmi, R. Shivashankar, and N. R. Surya
Abstract Basal geogrid reinforced embankments supported on vertical piles are

proven to be a feasible and effective solution for constructing embankments over
thick soft clay deposits and bridge approaching embankments. These solutions mini-
mize the lateral displacements, total and differential settlements of embankment crest
and toe by transmitting embankment loads into the deeper stratum through pile foun-
dations and arching action of geogrid. Basal geogrid reinforcements provide good
restraint against lateral spreading of the toe. Providing batter piles near the toe will
further enhance this restraint against lateral spreading. Not many studies are available
in literature on performance of batter piles below embankment toe, especially under
seismic excitations. The present study aims to find the advantages of providing batter
piles below embankment toe under seismic excitations. A 6 m high basal geogrid
reinforced embankment having 1 V:1.5H side slope constructed over 28 m thick
soft clay is considered for the 3-Dimensional finite element analysis. The soft clay
is stabilized with 22 m long 300 mm diameter vertical and batter piles spaced at
three times the pile diameter. Embankment crest vertical displacements, toe hori-
zontal displacements, maximum differential settlements at the crest and crest lateral
accelerations are analysed for different batter angles of 0°, 5°, 10°, 15°. Analysis
of results reveals that larger the batter angle more is the reduction of toe horizontal
displacements.
Keywords Pile foundations · Embankments · Seismic analysis · Finite element

method
1 Introduction
Inclined (or batter) piles are often used to resist lateral thrust in offshore construc-
tions. In the pile-supported embankments in relatively very soft grounds, these batter
piles are used below embankment toe to prevent lateral spreading of toe. The addition
R. M. Patel (B) · B. R. Jayalekshmi · R. Shivashankar · N. R. Surya

Department of Civil Engineering, NITK, Surathkal, Mangalore 575025, Karnataka, India
e-mail: radikagmpatel@gmail.com
146 R. M. Patel et al.
of basal geogrid in the embankment eliminates the necessity of batter piles below
embankment toe (BS 8006-2010) [3] under static loading conditions. Most of the
studies from literature on pile-supported embankments under static loading condi-
tions considered only vertical piles [4, 16, 17, 19, 20, 23]. Numerical analysis on pile-
supported embankments considered unit cell models instead of full 3-Dimensional
models [2, 5, 6, 10], thereby the presence or absence of batter pile is ignored. Thach
et al. [21, 22] and Han et al. [11] analysed geogrid reinforced pile-supported embank-
ments subjected to cyclic loading. To investigate the pile-pinning effects for the
embankment supported over piles, dynamic centrifuge model tests were performed
by Armstrong et al. [1]. Wang and Mei [24] studied the seismic performance of
micropile-supported embankment. Patel et al. [18] analysed the seismic behaviour
of end bearing and floating pile-supported embankment. Many studies reported the
beneficial and detrimental effects of these batter piles subjected to dynamic loading
conditions [8, 9, 25]. The presence of batter piles below embankment toe in a pile-
supported embankment under dynamic loading conditions is beneficial or detrimental
need to be studied yet.
Hence the present study aims to find the seismic behaviour of geogrid rein-
forced embankment supported over vertical and batter piles. The results are analysed
based on the toe horizontal displacements, crest vertical displacements, differential
settlements and lateral accelerations at crest.
2 Methodology
2.1 Numerical Analysis
6 m high embankment made of pulverized fuel ash (PFA) having 38 m base width
constructed over 28 m thick soft marine clay is considered for the time-history anal-
ysis. Hard stratum exists below the clay layer. The embankment geometry is shown
in Fig. 1.
2.2 Idealization of Soil
The properties of PFA, surface fill [16], soft marine clay which exists in Cochin
region [12, 14] and hard soil considered for the analysis are listed in Table 1. PFA
and surface fill were modelled using mohr–coulomb material model with drained
condition. The soft marine clay and hard soil were modelled using mohr–coulomb
material model including consolidation.
Seismic Response of Basal Geogrid Reinforced Embankments Supported … 147
Fig. 1 Pile-supported embankment geometry
Table 1 Soil properties considered

Soil Unit Young’s Poisson’s Cohesion Friction Permeability vp (m/s) vs (m/s)
type weight Modulus ratio (kN/m2 ) angle (m/s)
(kN/m3 ) (MN/m2 )
PFA 18.5 20 0.3 10 30° – 119.5 63.9
Surface 18.5 7 0.3 15 28° – 70.7 37.8
fill
Soft 14 4 0.45 12.5 2° 5 × 10–10 103.7 31.3
Marine
clay
Hard 21 250 0.3 50 40° 5 × 10–8 400.3 213.9
soil
2.3 Idealization of Pile Foundations
Piles of length (L) 22 m having 300 mm diameter (D) and arranged in a 3D spaced
square grid pattern were considered. Below embankment toe, a batter pile with a
batter angle of 0°, 5°, 10° and 15° are considered. Piles were modelled as linear
elastic isotropic material with a modulus of elasticity corresponding to M20 grade
concrete, unit weight of 25 kN/m3 and Poisson’s ratio of 0.15.
2.4 Idealization of Geogrid Reinforcement
Geogrid with tensile modulus of 4000 kN/m and Poisson’s ratio of 0.3 with linear
elastic isotropic material property was used as basal geogrid.
2.5 Modelling
3-dimensional finite element modelling was performed using general-purpose finite

element software ANSYS. PFA, surface fill and piles were modelled using SOLID185
element. SOLID185 is an eight noded element with three degrees of freedom at each
node: translations in the nodal x, y and z directions. Soft marine clay and hard soil
were modelled using CPT215 element. CPT215 is a coupled pore-pressure mechan-
ical solid element. It is defined by eight nodes having four degrees of freedom at each
node: translations in the nodal x, y and z directions, and one pore-pressure degree of
freedom at each corner node. Geogrid was modelled using SHELL181 element with
membrane effect. It is a four noded element with 3 translational degrees of freedom
at each node. A slice of 3-Dimensional finite element model of embankment, pile,
geogrid and viscous boundaries are shown in Fig. 2.
Fig. 2 a 3-Dimensional finite element model of embankment. b Enlarged view of geogrid reinforced
pile-supported embankment. c Enlarged view of viscous boundaries
2.6 Boundary Conditions
For time-history analysis, Ghosh and Wilson [7] suggested that the lateral boundary
should be at a distance of four times the base width of the embankment so that
the waves propagated from the soil cannot reflect back. To simulate the infinite soil
medium, viscous boundaries were applied for the lateral boundaries using spring-
damper element given by Kianoush and Ghaemmaghami [15]. The equation of
motion with additional damping matrix C* can be written as follows when the viscous
boundaries are taken into account.

[M]{ü(t)} + [C]{u̇(t)} + C∗ {u̇(t)} + [K]{u(t)} = −[M] üg (t) (1)
where, [M] is the structural mass matrix, [C] is the structural damping matrix, [K]
is the structural stiffness matrix,{üg (t)} is the ground acceleration vector,{ü(t)} is
?
the nodal acceleration vector, {u(t)}is the nodal velocity vector, {u(t)} is the nodal
displacement vector and [C*] is the special damping matrix that is considered as
follows,
⎡ ⎤
∗ An ρvp 0 0
C = ⎣ 0 At1 ρvs 0 ⎦
0 0 At2 ρvs
where, vp and vs are the dilatational and shear wave velocity of the considered medium
(Table 1), ρ is the density of soil medium, An , At1 and At2 are the fields controlling the
viscous dampers and the subscripts n and t represent normal and tangential directions
in the boundary.
2.7 Loading
The present study considers embankments situated at sites classified as earthquake

zone III under seismic zoning of India. Hence the embankment is subjected to time-
history of accelerations corresponding to Indian Standard (IS-1893) [13] code spec-
trum for Zone III scaled to 0.35 g. The time duration corresponding to bracketed
duration was used for the analysis. The acceleration time-history record of IS code
spectrum for Zone III scaled to 0.35 g is shown in Fig. 3. For time-history loading
the macro developed by Sahin, 2010; was used in ANSYS software.
Fig. 3 Acceleration time-history record of IS Code Zone III ground motion scaled to 0.35 g
3.1 Crest Vertical Displacements
Figure 4 shows the time history of embankment vertical displacements at crest centre
and crest edge for embankments supported over vertical and batter piles. Table 2
shows the maximum vertical displacements at embankment crest due to IS code zone
III ground motion. For embankment toe supported on 5° batter piles, the crest centre
and crest edge vertical displacements are about 2% lesser when compared with the toe
supported over other batter piles. It is also observed from Fig. 4 that, for embankment
toe supported over 0° and 5° batter piles crest centre vertical displacements are less
than the crest edge vertical displacements. But in 10° and 15° batter pile-supported
embankment, crest centre vertical displacements are more than the crest edge vertical
displacements.
3.2 Differential Settlements at Crest
From Fig. 4 and Table 2 it is noticed that embankment supported over 5° batter piles
experiences less differential settlements. It is also observed from Table 2 that increase
in batter angle beyond 5° increases the differential settlements at crest.
Fig. 4 Time history of embankment vertical displacements at crest centre and edge for embank-
ments supported over vertical and batter piles
Table 2 Maximum vertical displacement and differential settlement at embankment crest

Batter angle Maximum vertical displacement (mm) Maximum differential
At crest centre At crest edge settlement (mm)
0° 209.12 211.65 2.53

5° 203.4 205.18 2.06
10° 208.8 209.4 9.03
15° 209.73 212.98 8.99
3.3 Toe Lateral Displacements
Figure 5 shows the time history of toe lateral displacements for embankments
supported over vertical and batter piles. From Fig. 5 it is observed that increase
in batter angle reduces the toe lateral displacements. About 4.2% reduction in toe
lateral displacements is seen by increasing the batter angle from 0° to 15°. Toe lateral
displacements for 15° batter pile-supported embankment without basal geogrid are
2.5% higher than the embankment supported only with vertical piles and basal
geogrid.
Fig. 5 Time history of toe lateral displacements for embankments supported over vertical and
batter piles
Fig. 6 Time history of crest lateral accelerations for embankments supported over vertical and
batter piles
3.4 Lateral Accelerations at Crest
Figure 6 shows the time history of crest lateral accelerations for embankments
supported over vertical and batter piles. From Fig. 6 it is noticed that increase in
batter angle reduces the crest centre lateral accelerations. About 3% reduction in
crest lateral accelerations were observed by increasing the batter angle from 0° to
5° and about 5% reduction in crest lateral accelerations were observed by increasing
the batter angle from 0° to 15°.
4 Conclusions
3-Dimensional finite element analysis was conducted for the embankment supported
over soft marine clay improved with vertical and batter piles along with basal geogrid
subjected to IS code zone III ground motions.
From the analysis it is revealed that, higher batter angles (10°, 15°) lead to higher
vertical displacements at crest centre than at crest edge. This also increases the
differential settlements at crest.
Toe lateral displacements reduce with increase in batter angle. About 4.2% reduc-
tion in toe lateral displacements is seen by increasing the batter angle from 0° to
15°.
Toe lateral displacements for embankment supported over vertical piles and basal
geogrid are less when compared with the embankment supported over vertical and
15° batter piles without basal geogrid.
Crest centre lateral accelerations reduce with increase in batter angle. More reduc-
tion in crest lateral accelerations occurs for 5° batter pile-supported embankment.
References
1. Armstrong RJ, Boulanger RW, Beaty MH (2013) Liquefaction effects on piled bridge
abutments: centrifuge tests and numerical analyses. J Geotech Geoenviron Eng ASCE
139(3):433–443
2. Ariyarathne P, Liyanapathirana DS (2015) Review of existing design methods for geosynthetic-
reinforced pile-supported embankments. J Soils Found 55:17–34
3. BS 8006: code of practice for strengthened/reinforced soils and other fills (2010) British
Standard Institution, UK
4. Briançon L, Simon B (2011) Performance of pile-supported embankment over soft soil: full-
scale experiment. J Geotech Geoenviron Eng 138(4):551–561
5. Bhasi A, Rajagopal K (2014) Geosynthetic-reinforced piled embankments: comparison of
numerical and analytical methods. Int J Geomech ASCE 15(5):04014074
6. Bhasi A, Rajagopal K (2015) Numerical study of basal reinforced embankments supported on
floating/end bearing piles considering pile-soil interaction. Geotext Geomembr 43:524–536
7. Ghosh S, Wilson EL (1969) Analysis of axi-symmetric structures under arbitrary loading (No.
69–10) EERC report
8. Gerolymos N, Giannakou A, Anastasopoulos I (2008) Evidence of beneficial role of inclined

piles: Observations and summary of numerical analyses. Bull Earthq Eng 6(4):705–722
9. Giannakou A, Gerolymos N, Gazetas G, Tazoh T, Anastasopoulos I (2010) Seismic behavior
of batter piles: elastic response. J Geotech Geoenviron Eng 136(9):1187–1199
10. Han J Gabr MA (2002) A numerical study of load transfer mechanisms in geosynthetic rein-
forced and pile supported embankments over soft soil. J Geotech Geoenviron Eng ASCE
128(1):44–53
11. Han GX, Gong QM, Zhou SH (2014) Soil arching in a piled embankment under dynamic load.
Int J Geomech 15(6):04014094
12. IRC:113-guidelines for the design and construction of geosynthetic reinforced embankments
on soft subsoils (2013)
13. IS 1893(Part 1): criteria for earthquake resistant design of structures (2016)
14. Jose BT, Sridharan A, Abraham BM (1988) A study of geotechnical properties of Cochin
marine clays. Mar Georesour Geotechnol 7(3):189–209
15. Kianoush MR, Ghaemmaghami AR (2011) The effect of earthquake frequency content on
the seismic behavior of concrete rectangular liquid tanks using the finite element method
incorporating soil–structure interaction. Eng Struct 33(7):2186–2200
16. Liu HL, Charles W, Ng W, Fei K (2007) Performance of a geogrid-reinforced and pile-
supported highway embankment over soft clay: case study. J Geotech Geoenviron Eng ASCE
133(12):1483–1493
17. Liu KW, Rowe RK, Su Q, Liu B, Yang Z (2017) Long-term reinforcement strains for column
supported embankments with viscous reinforcement by FEM. Geotext Geomembr 45(4):307–
319
18. Patel RM, Jayalekshmi BR, Shivashankar R (2019) Seismic response of Basal geogrid rein-
forced embankments supported over floating and end bearing piles. In: Silvestri, Moraci
(eds) Earthquake geotechnical engineering for protection and development of environment
and constructions, pp 4620–4628. Rome, Italy. ISBN 978-0-367-14328-2
19. Smith M, Filz G (2007) Axisymmetric numerical modeling of a unit cell in geosynthetic-
reinforced, column-supported embankments. Geosyn Int 14(1):13–22
20. Shen P, Xu C, Han J (2017) Model tests investigating spatial tensile behavior of simulated
geosynthetic reinforcement material over rigid supports. J Mater Civ Eng 30(2):04017288
21. Thach PN, Liu HL, Kong GQ (2013) Evaluation of PCC pile method in mitigating embankment
vibrations from a high-speed train. J Geotech Geoenviron Eng 139(12):2225–2228
22. Thach PN, Liu HL, Kong GQ (2013) Vibration analysis of pile-supported embankments under
high-speed train passage. Soil Dyn Earthquake Eng 55:92–99
23. Wachman GS, Biolzi L, Labuz JF (2009) Structural behavior of a pile-supported embankment.
J Geotech Geoenviron Eng 136(1):26–34
24. Wang Z, Mei G (2012) Numerical analysis of seismic performance of embankment supported
by micropiles. Mar Georesour Geotechnol 30(1):52–62
25. Wang Y, Orense RP (2019) Numerical analysis of inclined pile group performance in
liquefiable sands. Silvestri, Moraci (eds) Earthquake geotechnical engineering for protec-
tion and development of environment and constructions. Rome, Italy, pp 5630–5637. ISBN
978-0-367-14328-2
Seismic Response of Low Height
Embankment over Soft Foundation Clay
Debabrata Ghosh, Narayan Roy, and R. B. Sahu
Abstract Low height embankments are widely used for construction of highways
and rural roads in India. Sometimes, it becomes necessary to construct such types of
embankment on soft foundation soil under unavoidable circumstances. In addition to
the stability under static loading, seismic excitation very often plays a major role in
the stability of such an embankment. So, to study the behaviour of such embankment
under seismic loading becomes very important as the failures may lead to huge
loss of property and life. The present study analyses the response of a low height
embankment overlying soft foundation soil under seismic loading with the help of
finite difference analysis using FLAC. The Mohr-Coulomb failure criterion is used
to define the soil constitutive model for both embankment and foundation soil. A
stiffer half-space has been considered below the soft foundation soil and the analysis
has been performed varying the shear wave velocity of the stiffer half-space. Low
to moderate level of shaking has been applied at the base of the stiffer half-space by
scaling the original input motion of 1987 Loma Prieta earthquake.
Keywords PGA · Soft foundation soil · Embankment · Seismic excitation ·

Excess pore pressure · Mohr-Coulomb
1 Introduction
Earthen embankment is a common civil engineering structure that serves a wide

range of purpose, like, roadways, railways, earthen dams, etc. The sudden failure
of an embankment as in the case of seismic event causes huge loss of property and
life. Sometimes low height embankments are built on soft foundation clayey soil
for construction of highways and rural roads without having an alternate choice. So,
even if these kinds of embankments are safe against static failure but under seismic
excitation they may undergo substantial deformations due to strength degradation.
Ansal and Erken [1] reported that degradation of underlying soil does not occur if
D. Ghosh · N. Roy (B) · R. B. Sahu

Jadavpur University, Kolkata, India
e-mail: narayan.roy04@gmail.com
156 D. Ghosh et al.
the number of equivalent load cycles is less even though the stress level exceeds the
critical level. They indicated that performance of an embankment depends upon the
frequency of the dynamic loading. Egawa et al. [2] conducted an extensive experi-
mental study on the behaviour of embankment on soft peaty ground under seismic
excitation generated by centrifuge and found that the deformation increases as the
loading frequency approaches the natural frequency. Okamura and Tamamura [3]
investigated the seismic stability of embankment on soft clay and peat deposits. The
deformation of the embankment was shown to be concave type with maximum settle-
ment at the crest due to the higher excess pore pressure during shaking. Rahardjo
[4] discussed some case histories on embankment failures in Indonesia and showed
that the main cause of bearing capacity failure of the foundation soil was excess
pore pressure generation during the cyclic loading. Gordan et al. [5] investigated the
dynamic behaviour of short embankment on soft soil using FE method simulating
the earthquake loading for specified PGA and duration and the response was shown
to be a function of modular ratio of embankment soil and foundation clay. Most of
above studies are for embankment overlying a deep soft clay deposit. However, in
general, in alluvial deposit, the top thick soft clay layer is generally underlain by a
deep deposit of stiff/very stiff/hard silty clay/clayey silt/dense/very dense sand layer.
When earthquake waves propagate upward through the stiff soil layer to the soft
foundation soil, the motion that will finally reach to embankment may substantially
get modified and affect the response of the embankment. With this background the
present study highlights the seismic response of a 5 m high embankment built on
10 m thick soft foundation clay overlying a stiffer half-space. To study the influence
of different bed-rock conditions, three different half-space shear wave velocities of
stiffer layer have been considered. Also in order to study the effect of level of shaking,
input motions with three different level of shaking have been considered by scaling
the original input motion for each case of half-space velocity.
2 Methodology
Numerical modelling and analysis for a 5 m high embankment underlain by a soft

foundation soil of 10 m thick have been performed using the finite difference software
FLAC (version: 7.0). The soft foundation soil is again underlain by a half-space of
stiff clay layer considered down to 40 m depth and the input motion has been applied at
the base of this stiff half-space. Figure 1 depicts the embankment with foundation soil
considered in the study. The engineering properties considered for the embankment
and foundation soil layers are given in the Table 1.
For the stiffer half-space, three different shear wave velocities, 300, 500 and
800 m/s, have been considered for the analysis so as to find out their influence on
the response of the embankment. Again for each half-space case, three different
acceleration time histories with varying PGA values have been considered. For this
purpose, the original time history has been scaled to three different PGA values. So,
Seismic Response of Low Height Embankment … 157
10m
RL+5.0
Embankment
(Compacted Clay) 1V: 2H
RL 0.0
Soft clay
(10 m thick)
RL -10.0
Stiff clay (half space) (40 m thick)

RL -50.0
Fig. 1 Schematic of the considered embankment with foundation soil
Table 1 Properties of embankment and foundation soil

Properties Unit Embankment Foundation soil Half-space
Unit weight kg/cum 1900 1700 2000
Cohesion (c) kPa 40 32 70/80/100
Friction angle Degree 20 0 0
Poison’s ratio 0.35 0.45 0.35
Shear modulus (G) MPa 42.75 17 180/500/1280
Porosity 0.35 0.4 0.3
Permeability (k) m/s 10−7 10−6 10−7
Shear velocity (Vs) m/s 150 100 300/500/800
Table 2 Combinations of study considered in the analysis

Sl. No. Model name Embankment Vs Foundation Vs Half-space Vs Input motion
(m/s) (m/s) (m/s) PGA
1 Model 1 150 100 300 0.07g
2 Model 2 150 100 300 0.15g
3 Model 3 150 100 300 0.25g
4 Model 4 150 100 500 0.07g
5 Model 5 150 100 500 0.15g
6 Model 6 150 100 500 0.25g
7 Model 7 150 100 800 0.07g
8 Model 8 150 100 800 0.15g
9 Model 9 150 100 800 0.25g
158 D. Ghosh et al.
Fig. 2 A sample
embankment model with
generated mesh considered
in the study
total nine combinations of models have been analysed in the current study. Table 2
presents the combinations of models analysed in the study.
At first, the static equilibrium of the embankment is checked. Figure 2 presents
a sample discretized model using FLAC. The Mohr-Coulomb soil model is adopted
for the analysis of the problem. The entire model is divided into the small rectan-
gular/triangular elements for analysis with finite difference. The size of the element
is decided on the basis of maximum considered frequency of the input motion and
the minimum shear velocity of foundation soil. The element size has been decided
considering the combination that causes the worst effect from the stability point of
view. Now from acceleration spectrum analysis, the maximum considered frequency
is obtained approximately as 20 Hz. Kuhlemeyer and Lysmer [6] showed that for an
accurate representation of the wave transmission through the soil model, the spatial
element size, l, must be smaller than approximately one-tenth to one-eighth of the
wavelength associated with the highest significant frequency.
shear velocity 100

wavelength (λ) = = m = 5m
significant maximum frequency 20
Maximum linear dimension of mesh element l ≤ (λ/10) = (5/10) m = 0.5 m.

Hence, the rectangular mesh with element size of 0.5 m × 0.5 m has been consid-
ered in the modelling. For static analysis, the usual boundary condition with fixity
at a distance of 20 m from the toe and 50 m from the base of embankment has been
considered. For nonlinear dynamic analysis, the vertical boundaries are modelled in
such a way that the reflection of the wave does not happen and the waves can radiate
out to minimize the geometry effect. The water table is considered at the bottom of
embankment, i.e. at the ground level (at RL 0.0).
Now the static analysis has been performed and the stability of the embankment is
evaluated in terms of factor of safety, and a pattern of potential base failure is found
for all cases as the base foundation soil is soft (Fig. 3). The displacement contours
Fig. 3 Failure pattern after the static analysis of Model 1
show the maximum static displacements in between 20 and 23 mm at the crest of

the embankment for all the models. The static factor of safety for all the considered
models has been found greater than 1.5.
Now, once the static equilibrium is checked, the dynamic analysis is performed
using three different acceleration time histories with PGA values 0.07g, 0.15g and
0.25g. The acceleration time history is of 40 s duration and the predominant frequency
of the motion is 1.0 Hz. The input motion can be given mainly in three different ways:
as acceleration time history, velocity time history and stress wave history. The input
motion is given at level RL-50.0 m, i.e. throughout the bottom edge of half-space in
form of a stress wave which corresponds to varying particle velocity with respect to
the time period of 40 s. The expression for stress wave is given below.
σ = 2ρCvs (1)
σ = stress generated, ρ = mass density, C = shear wave velocity, vs = particle

velocity.
Initially, the model is analysed without considering any kind of damping and then
considering hysteresis damping. A small amount of Rayleigh damping (0.2%) is
considered combined with hysteresis damping. At first, all the models are analysed
without the embankment portion. Then the embankment loading is applied after
which the static analysis is done. It is assumed that the seismic loading acts on
the model after the completion of consolidation of the soft foundation clay due to
embankment.
After static analysis, the stress wave history corresponding to the PGA magnitudes
of 0.07g, 0.15g and 0.25g is applied at the base of the models. The acceleration time
history is shown in Fig. 4.
After the completion of analysis for dynamic loading, different parameters like
displacement vectors, displacement contours, stresses, etc., are evaluated for deter-
mining the response of the models under dynamic loading. A vertical displacement
contour of Model 9 is shown in Fig. 5 that shows the maximum vertical displacement
occurs near the top edge of the embankment. This model is subjected to input motion
with PGA 0.25g and Half-space velocity 300 m/s.
160 D. Ghosh et al.
(a) 4
Acceleration (m/s2) (b) 4 (c) 3
Acceleration (m/s2)
Acceleration (m/s2)
2
2 2
1
0 0 0
-1
-2 -2
-2
-4 -4 -3
0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50
Time (s) Time (s) Time (s)
Fig. 4 Input motions considered in the study with PGA a 0.07g b 0.15g c 0.25g
Fig. 5 Vertical displacement contour of embankment with Model 9
In the static analysis, the factor of safety is calculated for all the models and it ranges
between 1.55 and 1.56. The failure pattern shows a probable base failure or the
foundation soil would fail first and it appears to be upward movement of the base
soil near the toe of the embankment. The base failure envelope is tangential to the
interface of the foundation soil and stiff half-space. In Fig. 3 the shear strain vector
of the Model 1 is shown after static analysis and factor of safety is found to be 1.556.
The base failure pattern is clear from the Fig. 3. After ensuring the static stability the
dynamic analysis is performed.
3.1 Variation of Amplification
After the dynamic analysis, the variations of PGA at different depth have been
obtained and presented as amplification ratio with depth. Amplification ratio presents
the ratio between PGA at a particular depth to the bed-rock PGA. Figures 6 and 7
present the variation of horizontal acceleration and PGA amplification ratio with
depth for half-space velocity 800 m/s and 500 m/s, respectively, along the mid-
width from bottom of model to crest of the embankment considering hysteresis
damping condition. The figures clearly depict the increase in amplification ratio
towards the surface. On the other hand, weaker motion exhibits higher amplifica-
tion in comparison with strong motion, as in the case of strong motion the effect of
(a) (b)
1.2 1.2
1 1
Embankment
Height Ratio(h/H)
Height Ratio(h/H)
0.8 soft foundation soil 0.8 soft foundation soil
0.6 0.6 stiff half-space

0.07g 0.07g
0.4 0.4
0.15g 0.15g
0.2 0.25g 0.25g
0.2
0
0
0 1 2 3 4
0 2 4
horizontal acceleration (m/s2) Amplification Ratio
Fig. 6 a Horizontal acceleration versus height ratio b Amplification ratio versus height ratio from
RL-50 for half-space Vs = 800 m/s) (with hysteresis damping) (Horizontal acceleration is in m/s2 )
(a) 1.2 (b) 1.2
1 1
Height Ratio(h/H)
Height Ratio(h/H)
0.8 soft layer 0.8 soft layer
0.6 0.6
0.4 0.4 0.07g

0.07g
0.2 0.15g 0.2 0.15g
0.25g 0.25g
0 0
0 2 4 0 2 4
horizontal acceleration (m/s2) Amplification Ratio
Fig. 7 a Horizontal acceleration versus height ratio b Amplification ratio versus height ratio from
RL-50 with half-space Vs = 500 m/s with hysteresis damping (horizontal acceleration is in m/s2 )
162 D. Ghosh et al.
Table 3 Vertical displacement at the middle of the crest after loading period
Sl. No. Model name Disp. static (mm) Disp. (undamped) Disp. (hysteresis)
dynamic (mm) dynamic (mm)
1 Model 1 23 118 130
2 Model 2 23 156 176
3 Model 3 23 185 225
4 Model 4 21 88 83
5 Model 5 21 170 190
6 Model 6 21 230 260
7 Model 7 20 78 68
8 Model 8 20 189 215
9 Model 9 20 258 270
induced soil non-linearity in the soil layer is more prominent. The results exhibit that
the soft foundation layer modifies the motion significantly in comparison with stiffer
half-space for all the considered cases. So, the seismic response of an embankment
constructed on soft foundation soil might be quite critical as the motion will get
amplified substantially by the soft soil layer.
3.2 Variation of Vertical Crest Displacement
The variations in vertical crest displacements, at the middle of the embankment, have
been presented in Table 3. Displacements have been tabulated from three different
types of analysis, static, undamped and hysteresis damping analysis. The results
show that displacements increase from static to undamped and it further increases in
hysteresis analysis. Further, the analysis points that as the level of shaking increases
the crest displacement of the embankment increases considerably; like for Model 7,
8 and 9 in case of hysteresis damping the vertical crest displacement for PGA 0.07g,
0.15g and 0.25g the displacement are 68 mm, 215 mm and 270 mm, respectively. The
results also show that for low level of shaking, i.e. PGA 0.07g, when the half-space
velocity increases the crest displacement decreases. For stronger shaking (PGA 0.15
and 0.25g), with the increase in half-space velocity the crest displacement is found
to increase for a particular input motion.
3.3 Variation of Excess Pore Pressure
The comparison of excess pore pressure is shown in Fig. 8a, b and c for Model 7,
8 and 9, respectively, with input ground motion of PGAs 0.07g, 0.15g and 0.25g,
respectively, for half-space velocity of 800 m/s. From the figures it is evident that
Fig. 8 Excess pore pressure contour after the dynamic analysis considering hysteresis damping
for: a Model 7 b Model 8 and c Model 9
the excess pore pressure is maximum near the point of application of input ground
motion and it gradually decreases while travelling upwards. Again, in the soft soil
layer, a little built up of excess pore pressure is observed and this built up of excess
pore water pressure is concentrated mainly in both the sides, away from the middle of
the embankment, which tends to reduce the effective stress of the portion and could
lead to instability.
Figure 9 depicts the variation of pore water pressure with time at the bottom,
middle and top of the soft foundation soil layer. From the figure it is observed that
Fig. 9 Pore pressure variation at bottom, mid and top of soft foundation soil layer of Model 8 with
time under hysteresis damping condition; X-axis: time (s); Y-axis: Pore pressure (Pa)
164 D. Ghosh et al.
built up of excess pore water pressure is maximum at the top of the soft layer; whereas
minimum excess pore water pressure is observed at the bottom of the soft layer. The
reason probably is the substantial amplification of ground motion while it reaches at
the top of the soft soil layer. From the Fig. 9, it can be found that at middle of soft
layer, the static pore water pressure is around 67 kPa and after application of input
motion (with PGA 0.15g) it rises to around 85 kPa, which is almost 25% increase in
pore pressure. On the other hand, the increase in pore pressure is around 75% (from
35 to 61 kPa) at the top of the soft layer. But in case bottom of soft layer the increase
is negligible.
4 Concluding Remarks
From the above analysis, following conclusions can be drawn:

(i) The analysis clearly shows a probable mode of failure is base failure due to the
soft foundation soil and the crest settlement is in between 20 and 23 mm under
static condition for all the models.
(ii) Analysis clearly depicts the increase in amplification ratio towards the surface.
Weaker motion exhibits higher amplification (3–3.5) in comparison with strong
motion (1–2). This decrease in amplification in case of strong motion is due to
the effect of soil non-linearity which is appearing more prominent for higher
level of shaking.
(iii) The vertical displacement of crest is found to increases for same property
model with the increase in level of shaking as the input motion with higher
shaking induces higher magnitude of stress wave with higher PGA. With the
introduction of hysteresis damping the crest settlement is found to increase.
(iv) The excess pore pressure is found to decreases from bottom of stiff soil layer
towards top of the model except in the soft foundation soil layer where little
built up of excess pore water pressure is observed. In the soft soil layer, at the top
and middle the pore pressure increment are around 75% and 25%, respectively,
whereas, the pressure increment at the bottom is negligible.
References
1. Ansal AM, Erken A (1989) Undrained behaviour of clay under cyclic stresses. J Geotech Eng
115
2. Egawa T, Nishimoto S, Tomisawa K (2004) An experimental study on the seismic behaviour of
embankments on peaty, soft ground through centrifuge model test. In: 13th world conference
on earthquake engineering, Vancouver, B.C, Canada, Paper 36
3. Okamura M, Tamamura S (2011) Seismic stability of embankment on soft soil deposit. Int J
Phys Model Geotech 11(2):50–57
4. Rahardjo PP (2014) Geotechnical failures case histories of construction on soft soils forensic
investigations and counter measures in Indonesia. Int J Integr Eng 6(2):11–23
5. Gordan B, Adnan A, Aida MAK. Soil saturated simulation in embankment during strong earth-
quake by effect of elasticity modulus, vol 2014, Article ID 191460, 7 pp. Hindawi Publishing
Corporation Modelling and Simulation in Engineering. http://dx.doi.org/10.1155/2014/191460
6. Kuhlemeyer RL, Lysmer J (1973) Finite element method accuracy for wave propagation
problems. J Soil Mech Found Div 99(5):421–427
Influence of Toe Cutting on Seismic
Response of a Typical Hill Slope
in North-East India
Rubi Chakrabory and Arindam Dey
Abstract In hilly regions of India, excavation of slope is a common practice for

construction of roadways. Toe cutting of slopes is repeatedly carried out for construc-
tion of new roads or widening of existing roads with rapid growth in population
needing urbanization. In earthquake prone hilly regions such as North Eastern part
of India, seismic activity can play the role of a major triggering agent for catastrophic
slope failure due to non-engineered and impromptu toe excavation. In majority of
the cases, effect of toe cutting on hill slope stability under earthquake condition is
studied based on Limit Equilibrium (LE) pseudo-static method, wherein the safety is
measured in terms of a time independent single factor of safety (FoS) value. However,
pseudo-static approach, in which earthquake is represented by a constant inertia force
acting on the slope, cannot simulate the earthquake condition accurately and may
overestimate the earthquake force. Hence, in this paper rigorous dynamic analysis is
performed for safe and economical excavation of toe regions of the slope before road
construction. The study reveals that the conventional pseudo-static analyses provide
conservative results, which when used as a basis of mitigation measures, will lead to
uneconomical stabilization technique.
Keywords Toe cutting · Limit equilibrium · Pseudo-static method · Factor of

safety · Dynamic analysis
1 Introduction
The primary means of transportation and communication in hilly regions of North-

East India are the highways constructed by toe excavation of the hill slopes. There-
fore, road cuts are often carried out for widening or construction activities along the
hill slopes. But due to inaccessibility in hilly terrain, lack of detailed site investigation
R. Chakrabory
Department of Civil Engineering, NIT Meghalaya, Shillong 793003, India
A. Dey (B)
Department of Civil Engineering, IIT Guwahati, Guwahati 781039, Assam, India
e-mail: arindam.dey@iitg.ac.in
168 R. Chakrabory and A. Dey
may lead to improper mitigation measures. Therefore, failure of non-engineered cut

slopes is a common phenomenon in this part of the country, causing a high monetary
loss and safety issues very often. The primary reason of slope failure is the reduction
in confining stress in natural slope after excavation. Geometry of slope structure, soil
shear strength and position of ground water table are the prime design parameters for
slope excavation. For excavation of slopes with cohesionless soil, slope stability does
not depend on height of cut and therefore slope angle is the most crucial parameter
of design. For slopes with cohesive soils, the height of the excavation is the primary
design parameter. For c-ϕ and saturated soils, both height and angle of slope are
important in slope excavation design.
Development of roadway along hill slopes is subjected to very high risk taking
into consideration the various past hazards reported around the world [7, 8]. Stark
et al., in 2005 [9] investigated the cause of distress to a single-family residence
located adjacent to a major highway cut slope. It was reported that the distress
was caused due to a landslide induced by excavation made to widen an existing
highway. Umrao et al., in 2011 [11] investigated the stability of road cut in rock
slopes at five different locations along NH-109 in Himalayan territory of India, by
the means of continuous slope mass rating (CSMR) technique. Kainthola et al., in
2015 [3] studied a high basaltic and lateritic road cut hill slope in their natural state at
Mahabaleshwar, India. They investigated several critical parameters from the study
i.e., mode of failure, factor of safety, shear strain rate, displacement magnitudes,
etc. Mahanta et al., in 2016 [5] investigated stability of slopes in areas susceptible
to failure determined from hazard zonation studies in the vicinity of Luhri village
(NH-305), India. Finite element method (FEM) was used to investigate the defor-
mation mechanism associated with such slope failures. Chakraborty and Dey, in
2018 [1] investigated the effect of vertical toe excavation on hill slope stability by
means of deterministic LEM for a wide range of geotechnical, hydraulic, and seismic
parameters. Based on the study recommendations were given regarding the critical
horizontal extent of vertical toe cutting in hill slopes depending on the type of slope
and different geotechnical parameters. In all the studies so far, effect of toe cutting
on hill slope stability under earthquake condition is studied based on Limit Equilib-
rium (LE) pseudo-static method [1, 2, 4], wherein the safety is measured in terms of
time independent single factor of safety (FoS) parameter. In pseudo-static method
earthquake is represented by a constant time independent inertia force acting on the
slope. Pseudo-static method overestimates the earthquake force and cannot simulate
the earthquake condition accurately. Hence, rigorous dynamic analyses need to be
performed for more accurate assessment of seismic response of slope in different
stages of toe excavation and corresponding safety parameters, leading to risk-free
and economical excavation of toe regions of the slope before road construction.
In this paper, a dynamic analysis is performed and compared with the outcomes
from pseudo-static analysis, to assess the slope stability more precisely for various
horizontal extent of vertical toe cutting under earthquake conditions. Firstly, the
static and pseudo-static LE analyses of the slope, for different horizontal extent of
vertical toe cutting (bt ) are performed using GeoStudio Slope/W v2007. Dynamic
analysis is conducted using GeoStudio Quake/W v2007 and the stresses generated
Influence of Toe Cutting on Seismic Response … 169
are incorporated in Slope/W module for evaluating the finite element FoS and its
variation for the entire duration of the assigned earthquake time history. It has been
observed that the pseudo-static analyses provide conservative results, which when
used as a basis of mitigation measures, will lead to large structural dimensions of the
chosen stabilization technique.
2 Static and Pseudo-static LEM Analysis for Slope Toe

Excavation
In this section of the paper, a typical hill slope (Slope height = 40 m, Slope inclination
= 30°, Cohesion = 50 kPa, Angle of internal friction = 20°, Unit weight = 18 kN/m3 )
is analyzed under static and pseudo-static condition using LEM by Slope/W v2007.
Defining the location of critical slip surface is of high importance in LEM analysis.
There are various methods to specify the locations of trial slip surfaces in SLOPE/W
2007, namely Entry and Exit method, Grid and Radius method, Fully specified slip
surfaces. In this study, Entry and Exit method is used. In this technique, the location
of the trial slip surfaces can be specified with their likely entry and exit ranges, thus
facilitating a large range of slip surfaces to be analyzed for assessing the critical
stability of the slope. Among the different limit equilibrium methods available in
SLOPE/W 2007, Morgenstern-Price method [6] is used and shear strength of soil
materials is simplistically represented by Mohr-Coulomb’s failure criterion.
In geotechnical earthquake engineering pseudo-static condition refers analyzing
the seismic response of any geotechnical structure by simply adding a permanent time
independent inertia force representing the earthquake force to a static LEM analysis.
For present pseudo-static analysis, horizontal acceleration considered is 0.12g, which
is equal to the peak horizontal acceleration (PHA) of the earthquake time history
assigned for dynamic analysis in Sect. 3.
The results obtained for static and pseudo-static analysis are given in Table 1.
From Table 1 it is noticed that in static condition, the FoS value is above unity (i.e.,
slope is safe) for a high range of horizontal extent of toe excavation (approximately
Table 1 Results of static and pseudo-static analyses

Horizontal extent of vertical toe cutting (bt ) (m) Static FoS Pseudo-static FoS
0 1.43 1.14
5 1.39 1.11
10 1.33 1.07
15 1.25 1
20 1.14 0.93
25 1.02 0.85
30 0.92 0.77
bt = 25 m). Whereas, pseudo-static analysis reports a safe toe cutting up to 15 m.

Therefore, it is understood that earthquake force can be a crucial agent to trigger
failure in unplanned excavated slope for road construction or widening to such an
extent (in this case beyond 15 m) without proper mitigation measures.
Among different techniques available in literature for slope stability study under
dynamic loading, LEM pseudo-static is the simplest approach as it is based on
simple mathematical calculations leading to less computational time. However, there
are some inherent limitations associated with LEM based pseudo-static method. In
LEM the failure mechanism is predetermined, it does not allow the failure surface to
develop through the actual weakest path. Moreover, pseudo-static approach highly
overestimates the earthquake loading acting on a geotechnical structure. As high risk
and economy is associated with highway construction planning and design, relying
completely on LEM based pseudo-static analysis for safety is not justifiable. To over-
come the above-mentioned limitations, in next section of the paper a dynamic FEM
based slope stability study is carried out where the test slope structure is assigned
an earthquake time history for more accurate assessment of seismic response of the
slope structure subjected to toe excavation in various stages.
3 Dynamic Analysis for Slope Toe Excavation
In this section of the paper, the same test slope (Elastic Modulus = 5000 GPa,
Poisson’s ratio = 0.3, Damping ratio = 0.1) as described in Sect. 2 is analyzed
under equivalent linear dynamic condition by Quake/W v2007. For finite element
based FoS calculation for the entire duration of the assigned earthquake time history
(Fig. 1), Quake/W v2007 is coupled with Slope/W v2007. El-Centro seismic motion
is used in the present study, which is scaled to a PGA similar to that experienced in
the Northeastern region of India. Further, although the actual recorded motion was
of longer durations, however, only 10 s motion is used in the present study that is
obtained as the bracketed duration of the original recorded motion. The bracketed
duration is obtained by applying the proposition by Trifunac and Brady [10]. The
Fig. 1 Earthquake time

history
stresses generated due to the earthquake motion assigned at the bottom of the slope
structure in Quake/W analysis are incorporated in Slope/W for evaluation of time-
dependent variation of FoS at different stages of toe cutting operation. Determining
the location of critical failure surface having the lowest FoS is one of the key issues in
a slope stability analysis. In LEM the location of the slip surfaces is predetermined,
whereas FEM is capable to seek out the actual weakest path through the soil domain.
Hence, FEM represents the failure mechanism more precisely.
The equivalent linear dynamic model is an approximation of the actual non-
linear behavior of soil. With the Equivalent Linear model, a dynamic analysis is
conducted in Quake/W with the specified Gmax value. Quake/W steps through the
entire earthquake record and find the peak shear strain value for each iteration. Then
the shear modulus value is modified according to the specified G reduction function
as shown in Fig. 2a, and the process is repeated. This iterative process is continued
until the required G modifications are within a specified tolerance. Figure 2b shows
that G is constant during one iteration through the earthquake record and the change
in slope shows the reduction in G value in next iterations. The cyclic shear strain
is estimated from the FE analysis. The estimated shear strain together with the G
reduction function and the specified Gmax value are used to evaluate new G values
for each iteration. The damping ratio in QUAKE/W also can be specified as a constant
or as a function. Similar to the G reduction function, the damping ratio is a function
of the cyclic shear strain. Damping ratio is considered as constant in present analysis.
The FoS variation with time estimated from FEM based dynamic analysis is
shown in Fig. 3. Dynamic analysis (Fig. 3) shows that up to 20 m of toe cutting
FoS does not fall below unity for the assigned earthquake motion having a peak
horizontal acceleration of 0.12g. Beyond 20 m of excavation, FoS falls below unity
several times during the earthquake. It is understood that even if the FoS reaches
the stability limit for certain instances of time, it does not lead to the failure of the
slope, owing to the stress reversals in the following time instances. Hence, dynamic
analysis reports that a toe excavation of 20 m may be carried out and beyond 20 m
proper safety measures are required. Whereas, the same slope structure shows a safe
Fig. 2 a G reduction function, b change of G value for each iteration through the earthquake
(adopted from Quake/W Engineering Book, 2007)
Fig. 3 FoS variation with

time for dynamic analysis
horizontal excavation extent of 15 m under pseudo-static earthquake condition having

horizontal acceleration equal to the peak horizontal acceleration of the assigned
earthquake time history in dynamic analysis. Hence, pseudo-static analysis provides
conservative results leading to uneconomical safety measures. It is worth mentioning
that different other seismic motions could have been chosen for better understanding
of the effect of seismicity on the stated problem. However, as the present study
focuses primarily on the difference in the outcomes as obtained by LEM and FEM
methods, the application of several other motions is not included in the present study.
4 Conclusions
In this paper, a report is provided against the safe excavation of the toe of a typical hill
slope that is carried out based on pseudo-static LEM and FEM based dynamic earth-
quake analyses. The results from both the analyses are compared. As limit equilibrium
methods do not exhibit the evolution of failure mechanism due to toe cutting, finite
element (FE) studies have been conducted to delineate the development of failure
mechanism with progressive toe cutting of the hill slope. A rigorous dynamic FE
analysis illustrated the time-dependent factor of safety (FoS). It is understood that
even if the FoS reaches the stability limit for certain instances of time, it does not
lead to the failure of the slope, owing to the stress reversals in the following time
instances. Under such conditions, it has been observed that the pseudo-static analyses
provide conservative results, which when used as a basis of mitigation measures, will
lead to large structural dimensions of the chosen stabilization technique.
References
1. Chakraborty R, Dey A (2018) Effect of toe cutting on hillslope stability. In: Anirudhan IV,
Maji V (eds) Geotechnical applications. Lecture notes in civil engineering, vol 13. Springer,
Singapore
2. Choudhury D, Basu S, Bray JD (2007) Behavior of slopes under static and seismic conditions
by limit equilibrium method. Geo-Denver, Denver, CO
3. Kainthola A, Singh PK, Singh TN (2015) Stability investigation of road cut slope in basaltic
rockmass, Mahabaleshwar, India. Geosci Front 6(6):837–845
4. Ling HI, Leshchinsky D, Mohri Y (1997) Soil slopes under combined horizontal and vertical
seismic accelerations. Earthq Eng Struct Dyn 26:1231–1241
5. Mahanta B, Singh HO, Singh PK, Kainthola A, Singh TN (2016) Stability analysis of potential
failure zones along NH-305, India. Nat Hazards 83(3):1341–1357
6. Morgenstern NR, Price VE (1965) The analysis of the stability of general slip surfaces.
Geotechnique 15(1):79–93
7. Singh PK, Kainthola A, Singh TN (2014) Influence of rock mass parameters on the stability
of high hill slopes. In: Indorock 2014, New Delhi, pp 577–587
8. Singh R, Umrao RK, Singh TN (2014) Stability evaluation of road-cut slopes in the Lesser
Himalaya of Uttarakhand, India: conventional and numerical approaches. Bull Eng Geol
Environ 73:845–857
9. Stark TD, Arellano WD, Hillman RP, Hughes RM, Joyal N, Hillebrandt D (2005) Effect of toe
excavation on a deep bedrock landslide. Managing 244–255
10. Trifunac MD, Brady AG (1975) A study on the duration of strong earthquake ground motion.
Bull Seismol Soc Am 65(3):581–626
11. Umrao RK, Singh R, Ahmad M, Singh TN (2011) Stability analysis of cut slopes using
continuous slope mass rating and kinematic analysis in Rudraprayag district, Uttarakhand.
Geomaterials 1(3):79–87
August, 2019 Landslide Events
in Kinnaur, H.P.—An Assessment
of Earthquake and Landslide
Consequences Using Satellite Data
Madan A. Mohan, Vidya Sagar Khanduri, and Amit Srivastava
Abstract On August 18, 2019, the Kinnaur region in the Indian state of Himachal
Pradesh experienced many landslides due to heavy rains. In H.P.’s Kinnaur district,
National Highway 5 was blocked, heavy rainfall triggered many landslides in isolated
regions of Himachal Pradesh, blocking large number of roads, it was blocked near
Ribba village of Kinnaur district causing heavy traffic jam. According to reports,
this event resulted in the death of 22 people and a significant loss of property. This
study presents the results of the Kinnaur landslide damage assessment derived from
the analysis of very high-resolution images (VHR) received from different satellites.
These datasets were obtained through a coordinated effort by the government organ-
isation major disasters. The damage is mainly attributed to rock slides from the area
of the slope that later became debris flows by sanding the material along the exit
zone.
Keywords Landslides · Heavy rainfall · VHR · Earthquake · Damage assessment
1 Introduction
Topographical, meteorological, physical and anthropogenic factors are the major

causes of slope failures. Topographical factors include terrain, curvature, drainage,
slope, elevation, landslide spatial distribution and frequency control. Meteorolog-
ical factors consist of rainfall intensity, rainfall magnitude and climatic conditions.
Physical factors consist of type of soil, structure, texture, roughness and its under-
lying geology. Anthropogenic factors include human activities like over grazing
of animals, cutting of trees, forest fires, excavation of slopes for construction of
M. A. Mohan (B) · V. S. Khanduri · A. Srivastava

School of Civil Engineering, Lovely Professional University, Jalandhar, Punjab, India
e-mail: chinnu2madan@gmail.com
V. S. Khanduri
e-mail: vidya.18579@lpu.co.in
A. Srivastava
e-mail: amit.22986@lpu.co.in
176 M. A. Mohan et al.
roads, mineral mining, laying of pipelines, etc. [11]. There are many records kept
regarding occurrence of earthquake and rainfall data causing heavy destruction in
Kinnaur district. Due to the development in urbanisation, construction of roads and
dams, environmental changes and global warming, the terrain condition is changing
leading to landslides. So, to take preventive measures and reduce risk Landslide
Hazard Zonation is prepared with the help of very high-resolution images (VHR)
taken from different satellites. Landslide Hazard Zonation is the process in which
an area is selected and it is divided into different subclasses based on the landslide
susceptibility which consists of different parameters and doing the assessment of
earthquake induced landslides. Since it consists of lot of factors to be considered, the
analysis of the Landslide Hazard Zonation becomes a complex work and requires
skilled and experienced person. Landslide Hazard Zonation can be done in both qual-
itative and quantitative approach [14]. The quantitative approach includes logistic
regression, linear regression and discriminant regression and qualitative approach
includes parameter weighted method, site visitation and recording, landslide hazard
mapping, soil testing.
The objective of the study is
• To investigate and understand the impact of landslides in Ribba region which lead
to economic and physical loss by using landslide susceptible mapping.
• To evaluate the hazard associated with Ribba landslide by using very high-
resolution images, rainfall and earthquake data.
• To determine the geotechnical properties of the soil sample collected from Ribba
landslide region.
• To suggest slope stability analysis and its frequency by using kinematic and finite
element analysis.
2 Area and Location
Kinnaur is a Trans-Himalayan region which is located in the bank of Sutlej and

tributary of Baspa. It is the east frontier district of Himachal Pradesh. The total
area of Kinnaur is about 6401 km2 . Ribba is a village panchayat located in Kinnaur
district, Himachal Pradesh, India. The coordinates of Ribba are longitude 78.364
and latitude 31.584. In eastern part of Kinnaur lies Ngari region located in Tibet.
In southern region, Douladhar ranges which separates Shimla district of Himachal
Pradesh and Uttarkashi district of Uttar Pradesh. Srikhand Dhar makes boundary
between Kuku and Kinnaur in the western region. In norther region Spiti is located
which shares Indo-Tibetan border (2017) (Fig. 1).
The elevation of Kinnaur varies between 1220 and 3050 m. But in mountainous
region the average height of the peaks is 5500 m which consist of ice and snow
throughout the year. Kinnaur district consists of rivers, glaciers, deep valleys gorges
and very high mountains.
The mountains of Kinnaur are full of craggy rocks of massive heights perilous
gradients in spurs. The mountains are covered with huge boulders and cliffs which
August, 2019 Landslide Events in Kinnaur, H.P. … 177
Fig. 1 Physical and satellite map of Kinnaur district [8]
makes a dangerous location during landslides. Rainfall occurs throughout the year in
this region. As the district is located in the higher region, it has temperate climate with
longer winters from October to May compared to remaining parts of India (2017).
From 2011 census records of India, Kinnaur district population is 84,121, with
population growth rate of 7.61%. The district consists of 65 panchayats with 12,503
households. Due to tourism and verity of culture tourist from different parts of the
world are attracted to this region (2017).
2.1 Zone of Landslide in Ribba Village
The landslide of Ribba village is located at 31°59 20 N and 78°36 49 E, along which
another landslide zone is located called Pawari landslide zone where the landslides
are common. These two zones are located near the bank of Sutlej River. Recently
on August 18, 2019 due to heavy rainfall a massive landslide was occurred in Ribba
landslide zone due to which National Highway 5 was blocked. The width and length
of landslide zone are 2.58 km and 859 m, respectively, and has an area of about 2.21
km2 . [9] (Fig. 2).
The landslide occurred at MSL of 1993 m and at an altitude of 2850 m. The
surrounding villages are Rarang, Akpa, Panah and Purbani. In this landslide zone
the northern part is a cliff with debris flow and rock boulders. In the southern zone
there is a valley of Sutlej river. By geomorphic characteristics, this zone has steep
angle of slope with a difference of 50° and 65° in the northern region and 40° and
55° in the southern region [1].
3 Frequency of Earthquake and Rainfall
The rainfall data can be obtained from meteorological centre, Shimla which can be
used to observe the pattern of rainfall occurring in Kinnaur district. With this data
Fig. 2 Ribba village landslide zone [9]
the affected area of landslide is accessed. The earthquake data is compiled from ‘The
tribune’ [13].
The largest earthquake occurred in Kinnaur district is ‘1975 Kinnaur Earthquake’
with a magnitude of 6.8 and causing extensive damage in whole Himachal Pradesh.
Its epicentre was in Kinnaur district in the southern east of Himachal Pradesh. Earth-
quakes are common in Himachal Pradesh which can be varied from low intensity to
high intensity. Since the state of Himachal Pradesh lies toward the southern boundary
of Himalayan region belt, the collision between Eurasian plate and Indian plate causes
ground movements. There are a lot of incidents of earthquake recorded in this region
by Indian meteorological department.
The district receives rainfall from the month of June till September. The marginal
shift in the monsoon pattern has been noticed over the period of years. The average
annual rainfall in the district is 816 mm. However much of the rainfall is received in
parts of lower Kinnaur. There is a progressive decrease in rainfall as one goes from
west to east. The parts of upper Kinnaur receive more snowfall than the rainfall. The
district receives heavy snowfall from November end or early December till March
or sometimes April. During this period, the area remains totally cut off from rest
of the area. The higher peaks of Kinnaur district are completely covered with snow
throughout the year (Tables 1 and 2).
4 Data Preparation for Landslide Hazard Zonation
The required data of landslide can be obtained from Geological Survey of India,
National Remote Sensing Centre which gives the satellite images. By using GPS and
Google Earth the present situation of landslide can be obtained for mapping. With
Table 1 Earthquake data of past 10 years [7]

S.no Date Intensity Epicenter Casuality Remarks
1 28, May 2010 4.8 Shimla dist No
2 02, Oct 2012 4.5 Border of districts of No
Chamba, Lahaul and
Spiti
3 12, Nov 2012 4.1 Baijnath in Douladhar No
ranges
4 05, Jun 2013 4.8 Border of the districts No The region falls in
of Chamba and seismic zone-IV
Lahaul-Spiti
5 26, Apr 2015 7.9 Nepal Killed-36
Injuries-200
6 27, Aug 2016 4.6 Kullu Region No
7 27, Oct 2017 4.4 Mandi Region No
8 25, May 2018 3.6 Kinnaur Dist No
9 03, Feb 2019 3.5 Kangra Dist No
10 03, May 2019 4.2 North east of Mandi No
Table 2 Rainfall data of Kinnaur district of past six years [4]

Year 2014 2015 2016 2017 2018 2019
Jan 78.1 67.2 16.1 135 1.8 81.8
Feb 140.7 166.7 44.3 38.6 36.8 154.6
Mar 69.5 228.2 79.2 69.5 74.6 102.2
Apr 52.9 32.7 105.2 121.1 40.1 9.4
May 25.6 44.1 15.2 17.4 36.0 69.4
Jun 7.3 44.1 10.1 39.4 24.9 38.2
July 25.9 35 29.8 41.9 51.9 19
Aug 8.6 23.3 49.2 27.5 24.5 53.7
Sep 10.2 37.8 27.2 51.6 78.8 9.5
Oct 1.4 4.5 1.7 0.1 0.0 0
Nov 0.2 8.9 0.0 21.5 35.7 74
Dec 39.3 18.0 5.0 29.9 13.3 0
Total 426.5 710.5 383 593.5 645.1 611.8
the collected information, by using a software called ARC GIS 10 a thematic layer
of the landslide can be developed by considering the factors of land cover, slope
angle, type of soil, curvature, density of drainage, geological features and relief. By
using previous records of landslides occurred in the Kinnaur district with different
properties like landslide pattern, geological characteristics, runoff distribution and
earthquake pattern an inventory is prepared. In Kinnaur District a total of 6322
landslides have been occurred in the entire district. Using the image processing
techniques and field investigations in detail gives the type of landslide, its mechanism,
failure pattern and its causes. There are two main aspects of hazard analysis, Aspect
and Slope [6]. As the distance of change in slope increases the slope value varies,
if the slope value is more then it represents the steeper terrain and vice versa. The
slope value can be calculated by

Slope (In degrees) = 57.3 tan−1 (dz/dx)2 + (dz/dy)2
The slopes which have been modified is classified into five zones with different
land use pattern. Now the steepest slope direction gives the aspect and can be
calculated by
Aspect = 57.3 tan−1 2[(dx/dx) + (dz/dy)]
x and y are the direction of slope which are represented by dz/dx and dz/dy [6].
5 Stability Analysis
Since the landslide zone of Ribba village consist of rock mass and boulders the
stability analysis is done to determine the slope deformation and rock failure mode.
Now the rock failure mode can be determined by using kinematic analysis based
on stereonet and slope deformation is obtained by using shear strength reduction
technique and plain strain (FEM).
5.1 Rock Failure Mode by Kinematic Analysis
With the help of stereographical projection method, the rock exposure, joint orien-
tation data and stereo plot can be obtained in the northern part. The kinematic anal-
ysis gives the possible failure mode and direction in a rock mass by using angular
relationship data between slope surface and discontinuities [14].
5.2 Finite Element Analysis
In the finite element analysis, a two slope model is used to determine the failure
criteria of slope. Generalised Hooke brown model gives the failure criteria for slope
1 by using the following formula

σ1 = σ3 + σci [m b (σ3 σci ) + s] (a)
σ1 Maximum principal stress.

σ3 Minimum principal stress.
σ ci Intact compressive strength (mpa).
mb , s and a Hoek–Brown parameters.
Now Generalised Hooke brown model is dependent on a parameter called Geolog-

ical Strength Index (GSI) of rock mass. The GSI can be obtained by correlating
discontinuity condition and rock mass blocky nature which has a standard chart
proposed by Cai [3].
After the finding out the failure criteria of slope 1, Mohr coulomb criteria is used
to determine debris flow in slope 2 with the formula
τ = c + σ tan θ
τ shear stress or shear strength of material.

c cohesion of material.
θ angle of internal friction.
By using these two failure criteria, a model is prepared in CartoSat-1 Digital

Elevation Model which is used for execution of the slope parameters for finite element
analysis.
6 Soil Properties
By performing soil experiments on the soil sample collected from the Ribba landslide
it is found that the soil is poorly graded with coarse and very fine grade category.
• The average specific gravity of the soil is 2.47.
• By performing sieve analysis, the following graph is plotted in which the value
of coefficient of uniformity is 16.84 and coefficient of curvature is 0.31.
• In Atterberg limits, Liquid Limit (LL) is 23.339%, Plastic Limit (PL) is 19.67%
and shrinkage limit (SL) is 17.05%.
• From standard proctor test, a compaction curve is plotted to obtained optimum
moisture content (OMC) of 25.1% and maximum dry density (MDD) of 1.744
gm/cc (Figs. 3, 4 and 5).
100
90
80
70
% Finer by mass
60
50
40
30
20
10
0
0.01 0.1 1 10
Grain Size (mm)
Fig. 3 Particle size distribution
Atterberg Limit Calculation

35
30
25
Water Content
20
15
10
0
0 5 10 15 20 25 30 35
Number of Blows
Fig. 4 Atterberg limit calculation
7 Landslide Susceptibility Mapping
There are different factors of mapping which are responsible for landslide suscep-
tibility. The previous landslides data and information are accessed so that these can
be used for future analysis of landslides to be occurred. A landslide inventory is
used to prepare a whole map showing the old and currently active landslides and
Dry Density-Moisture Content Relationship

1.79
1.74
1.69
Dry Density
1.64
1.59
1.54
1.49
15 20 25 30 35
Water Content
Fig. 5 Determination of optimum moisture content and maximum dry density
to perform further investigation in those regions. Thematic layers of these map are
prepared by overlaying the different factors showing geological, morphological and
land use conditions [5]. Finally, a model is prepared on the landslide susceptibility
for assessment which can be qualitative or quantitative approach. Landslide vulner-
ability includes different aspects based on physical, land use and land cover, people
perception, comprehensive vulnerability (Fig. 6).
7.1 Morphometric Properties
Due to heavy rainfall in this region, a high amount of volume of water is discharged
with greater velocity which leads to erosion of soil. As the slope has steeper gradient
the flow of water along with debris is high and this flow washes away the pavement
constructed and finally into the Sutlej river. The bifurcation ratio is defined as the
ratio of number of streams of given order to number of streams to next order. The
bifurcation ratio gives us the physical loss of the soil mass (Fig. 7).
Here the Sutlej River has a low bifurcation value which says that it has a controlled
bedrock. Many streams are formed from higher ridge, and due to continuous flow,
the boulders and soil gets washed away into Sutlej basin.
Fig. 6 Landslide susceptibility mapping
Fig. 7 Flow of rainwater

along with debris and soil
blocking NH 5
Fig. 8 Positive curvature of

landslide
7.2 Curvature of Landslide
The deceleration and acceleration of the flow due to profile curvature affect the nature
of the landslide. If the curvature has a negative value then it shows that the surface
is upward convex and the flow will be decelerated. If positive then the surface is
upward concave with accelerated flow (Fig. 8).
Here the landslide occurred at Ribba has a positive curvature due to which high
amount of movement of soil mass, boulders and debris took place.
7.3 Lithology of the Location
The lithology of the location is obtained from the geological survey of India and
google maps. It is found that the area of Kinnaur district is mostly covered with type
of peat with the mixture of limestone, phosphate, quartzite, greenish grey sandstone
with area of 29%. The affected area was about 1.2% which consists of boulders,
peat, sediments of sandstone and clay. The area of forest covers with limestone,
shale which are highly vulnerable to floods, soil erosion and landslides [1].
7.4 Hydroelectric Projects
Construction of hydroelectric project is a major issue which finally leads to slope

failures and landslides. If surrounding of the Sutlej is considered it is observed that
slope instability is common causing sliding of soil mass and boulders. There are four
hydroelectric projects constructed in Sutlej basin and are different in type which
were under investigation, clearance, one was commissioned and another was under
construction. It was found that the maximum number of landslides were occurred due
to under construction type and were causing risk. As the continuity of construction
Fig. 9 Hydroelectric project

[10]
activities, the chances of landslides in the surrounding region are increasing. The
wildlife sanctuary, forestland and topsoil and many villages were threatened under
hydroelectric projects [2] (Fig. 9).
When these hydroelectric projects are constructed blasting technique and
tunnelling is done which will cause ground disturbances and finally causing
landslides.
7.5 Occurrences of Earthquake
Land movements due to earthquake in Kinnaur region have an intensity between

4.0 and 6.1. The average magnitude recorded is 4.5 and low intensity earthquakes
commonly occur in this region. By comparing with the Mercalli intensity scale
Kinnaur district falls under IV to V. Every time when a low intensity earthquake is
occurred, the tremors can be felt in the surrounding region, with cracks developed in
buildings, pavement failure, spalling, settlement, utility disturbances. People living in
the surrounding also explained about how they experienced those tremors (Fig. 10).
Even due to blasting operations and tunnelling construction they felt those
vibrations. Due to these conditions every village in Kinnaur district comes under
earthquake vulnerability zone.
8 Questionnaire Survey
A questionnaire survey was constructed in Ribba landslide zone to gather the infor-
mation about the landslide occurred based on different parameters. Out of 70 people
50 people told that the vibrations which they observe are due to blasting operations
Fig. 10 Small range seismic waves due to blasting and tunnelling operations
and tunnelling activities. When heavy rainfall occurs and a huge amount of water is
released through dam causing heavy flow of river finally leads to slope failure told by
43 people. All 70 people told that the loss of property and life were common in this
region. 27 people told that most of the landslides are both overburden and boulders
sliding. Some of them also stated that there are no dumping site and the waste mate-
rials were dumped near river or beside roads. Some told that due to deforestation
activities, the soil in this region is getting eroded.
9 Conclusion and Discussions
From the study it was found that the Ribba village zone is highly vulnerable to land-
slide because of its clumsy geophysical nature. Most of the landslides in this region
are occurring due to blasting operations as the slopes are steeper so failure occurs
easily. By considering morphological factors it can be stated that this region comes
under highly vulnerable region. The use of hydropower plant was causing distur-
bances in soil and also ecological imbalance in Sutlej valley. Further construction
activities near the steeper region should be avoided. The recent landslide occurred
in Ribba village lead to 22 persons death, and many households got destroyed. The
Ribba landslide zone comes under high vulnerability with an impact of 70% failures
if any disaster occurs. For further studies it is necessary to consider detailed study
of slope stabilisation and rock mechanics and these research works should be under-
taken by experienced institutes and departments. The following pie chart shows the
research work done or are to be carried out in different areas by considering previous
research papers or review papers. Total of eight papers have been found related to
Kinnaur district landslides in which these have been divided into six categories in
which further investigation can be carried out (Fig. 11).
It can be seen that most of the work has been done in assessment of impact of
landslides causing economic and physical loss and investigation of slope failure,
ground movements due to blasting operations. Even a little work was also carried
in vulnerability assessment in Kinnaur district but still the proper data has been not
Research Work Areas
Vulnerability assessment in Kinnaur

district
Hazard evaluation of Ribba landslide

3% zone by using VHR images
20%
Usage of stability analysis in Ribba
30%
landslide zone with PLAXIS 2D or
3D, VisLOG, GEO5
10%
Assessment of impact of landslide
0% causing economic and physical loss
Investigation of slope failure and

ground movement due to blasting
37%
operations
Detailed study of slope stabilization
and rock mechanics
Fig. 11 Pie chart showing research work done or to be carried out in ribba landslide zone [1–3, 5,
6, 10, 12, 14]
provided in any of their research work regarding Ribba landslide zone. So, the main
area of work can be carried out in slope stabilisation and rock mechanics analysis.
Moreover, no stability analysis has been done in Ribba landslide zone by using
PLAXIS 2D or 3D, VisLOG and GEO5 software.
References
1. Amit Jamwal NK (2019) Use of geographic information system for the vulnerability assessment
of landscape in upper Satluj basin of district Kinnaur, Himachal Pradesh, India. Geol Ecol
Landscapes
2. Asher M (2019) Hydropower projects ruining the lives in the himalayas
3. Cai M (2004) Estimation of rock mass deformation modulus. J Rock Mech Min Sci
4. centre M (2019) https://www.weathershimla.nic.in
5. Kundan Digambar Rangari RS (2017) Macro scale (1:50,000) landslide susceptibility mapping
in parts of Kullu, Mandi, Chamba, Kangra and Lahaul & Spiti districts of Himachal Pradesh.
Spec Publ J Eng Geol
6. Laxmi Devi VRS (2019) Comparative evaluation of GIS based landslide hazard. J Geol Soc
India
7. List of earthquakes in India (2019) Wikipedia. Retrieved from https://en.wikipedia.org/wiki/
List_of_earthquakes_in_India
8. maphill (n.d.) Maps of kinnaur
9. Maps G (2019) Google Maps
10. Negi BB (2014) Environmental impect of making hydropower projects in kinnaur district of
himachal pradesh
11. Prakash S (2013) Earthquake related landslides in the Indian Himalaya. Springer, pp 327–334
12. Suneet Naithani AS (2018) Mapping of natural hazards and expected incidences in Great
Himalayan National Park Conservation Area, Himachal Pradesh. Indian J Ecol
13. The Tribune (2019). Retrieved from The tribune
14. Vipin Kumar VG (2018) Hazard evaluation of progressive Pawari landslide zone, Satluj valley,
Himachal Pradesh, India. Springer
Geotechnical Investigation of Landslide
of Atharamura and Baramura Hill,
Tripura
Kousik Adak and Sujit Kumar Pal
Abstract Landslide is a natural disaster that occurs when rock, earth or debris flows
to downward slope due to gravity after being detached from slope underneath. It is
essential to know the geotechnical properties to analyse the causes of landslide. For
the present study geotechnical investigation has been carried out on the slopes of
Atharamura and Baramura Hill, Tripura. Results obtained from laboratory test and
field investigation revealed that the underlying causes of the landslide could be (a)
the geological formation of those hills (b) permeability through the different layers
of soil (c) shear strength of the soil (d) cutting of hills slope for reconstruction
and widening of the road. Landslide triggered due to heavy precipitation during the
monsoon season.
Keywords Landslide · Debris flow · Geotechnical property · Shear strength ·

Precipitation
1 Introduction
In India about 10% of total population was affected by natural disaster in 2019.
Among this, landslides also play an important role in affecting the people of India.
Geological Survey of India says that 15% of the land in our country is hazardous
to landslide. North-east portion of the country is vulnerable to the landslide hazard.
It is a very common phenomenon in the hilly region of Tripura especially in North
and Dhalai districts. Almost every monsoon due to heavy rainfall total hilly region is
mostly affected by the landslide. Due to its geological instability, non-tectonic activity
and extreme rainfall infiltration during monsoon season [1, 2, 3]. As the North-east
portion is the developing region there is new road and railways construction along the
unstable slope of the hilly region. This road construction and various infrastructural
K. Adak · S. K. Pal (B)

Department of Civil Engineering, NIT Agartala, Tripura, Jirania 799046, India
e-mail: skpal1963@gmail.com
K. Adak
e-mail: kousik7258@gmail.com
192 K. Adak and S. K. Pal
works repeatedly increase the number of landslides in those areas [4]. In the time of
rigorous field survey most of the landslide occurred alongside of the NH-44 which
connecting Triupra state with Assam. Due to landslide mainly in monsoon time when
rainfall is moderate to heavy (where moderate is defined as 2.6–7.6 mm per hour ≥
7.6 mm) blocking and damaging of NH-44 is a common occurrence.
As landslide is a complex phenomenon worldwide it is very difficult to say the
exact causes of it. Most of the publications are based on landslide hazard zonation
mapping deals with landslide susceptibility mapping [5]. But field investigations also
helpful to know the type and analysis of landslide as we have seen that there is very
heterogeneity in geotechnical properties of soil in a very small area.
It also helps us to find the further remedial of landslide. Based on field investigation
and laboratory similar case studies have been carried out [6–8, 9]. Our present study
is the outcome of field investigation and the laboratory testing of soil collected from
the landslide areas which help us to analyse the causes of landslide.
2 Experimental Test Programme and Set-Up
Experimental test programme of present study is based on the laboratory experiment.

Our study is mainly to know about the geotechnical property of the landslide area
and to find out the basic causes of landslide. Mainly we have visited those landslide
prone areas in the monsoon season. The soil sample is collected from four different
sites from Atharamura hills and Barramura hills. Using cylindrical sampler of size
10 cm diameter and 30 cm long we have taken undisturbed soil and also collected
some disturbed sample using plastic bag. On July 9, locations 1 and 2 201,923.810
N,91.580 E and 23.880 N,91.780 E near Kulai Rain Forest besides NH 44 Shillong-
Agartala-Sabrum road; and some soil samples are collected in November 2019,
locations 3 & 4 from Baramura Hill, of North Tripura district. After collecting the
samples we have done laboratory test programme given below (Figs. 1 and 2):
(a) Specific Gravity Grain Size Analysis (b) Atterberg limit determination (c)
Direct shear test (d) Compaction test (e) Permeability test.
3 Test Procedure
The tests were carried out systematically and the following steps were followed:
Specific Gravity
The specific gravity test is conducted as per IS: 2720 (Part 3/Sec-1) 1980. The specific
gravity of soil shall be calculated from the following formula:-
G = [{W2 − W1 /(W2 − W1 )} − (W3 − W4 )}]

Geotechnical Investigation of Landslide of Atharamura and Baramura Hill, Tripura 193
Fig. 1 Soil sample

collection
Fig. 2 Landslide affected

area
W1 = Weight of the density bottle (g)

W2 = Weight of the density bottle + soil (g)
W3 = Weight of density bottle + soil + water (g)
W4 = Weight of the density bottle full with water (g)
Grain Size Analysis
Grain size analysis has been done to find out the classifications and the group of the
soil on the basis of ASTM D6913. The particle size distribution can be explained
as a plot of percentage finer to the particle size plot. This is determined to know
the exact type of the soil. This is followed the following three steps. i. Dry sieve
analysis; ii. Wet sieve analysis; iii. Hydrometer analysis. However for sandy soil dry
sieve analysis is enough to have the properties. For dry sieve analysis the dry soil
has been taken and sieving is carried out by the standard IS sieves. And the plot is
generated. Then the wet sieving is done with the soil solution. And the retained part
from 0.075 mm sieve is taken for the hydrometer analysis. In case of hydrometer
analysis the soil particles are allowed to settle with time and using stokes law the
particle size distribution is generated to have the full profile of grain size distribution
of soil.
Atterberg Limits
The Atterberg limits are a basic measure of the critical water contents of a fine grained
soil, such as its liquid limit, plastic limit and shrinkage limit. Depending on the water
content of the soil, it may appear in four states: solid, semi-solid, plastic and liquid.
• Liquid limit (LL): The boundary between the liquid and plastic states;
• Plastic limit (PL): The boundary between the plastic and semi-solid states;
• Shrinkage limit (SL): The boundary between the semi-solid and solid states;
Direct Shear Test

This test is also conducted to find out the ‘c’ and ‘F’ of the soil having more angle
soil is first subjected to some amount of normal stress. Then the shearing stress is
given until the soil sample fails or deforms too much. This one set of experiments
will give a point in the plot of normal stress (i.e. ‘o’) to the failure shearing stress
(i.e. ‘ı’).
τ = φ+
In the laboratory the direct shear test is conducted with the soil sample of skin
friction (i.e. silt, sand, etc.) (Fig. 3).
Permeability Test
The permeability has been determined on the basis of ASTM D 2434. The property
of the soil which permits water (fluid) to percolate through its interconnecting voids
is called permeability. The coefficient of permeability may be determined both in the
laboratory and field by direct test. I did falling head permeability test in laboratory.
In laboratory we determine the permeability of soil by using falling head
permeability test.
Compaction Test
The compaction test is conducted as per ASTM D698 & D1557. The test was
conducted on soil sample. Water was added to the mixtures for determining the
maximum dry density (MDD) and optimum moisture content (OMC) of the sample
on the basis of light compaction (Fig. 4).
Fig. 3 Direct shear test apparatus
Fig. 4 Standard proctor test

apparatus
4 Test Results and Discussion
After performing all the above tests on the collected sample from four different sites
we are able to find out the geotechnical properties of those samples. We present the
properties of the soil in tabular format.
Physical Properties of the Soil Samples (Table 1)
Engineering Properties of the Soil Samples (Table 2)
5 Test Results and Discussion
The results of this present study fully depend on the geotechnical and geological
properties of soil which effect the causes of landslide. So the causes of landslides
are discussed below:
Table 1 Physical properties of the soil sample

Physical prop. of soil Location 1 Location 2 Location 3 Location 3
Specific gravity 2.65 2.62 2.62 2.65
Sand (%) 88.75 88.64 85.29 86.28
Still (%) 6.68 6.65 6.6 6.5
Clay (%) 4.57 4.71 8.11 7.22
Cu & Cc 6.2 & 1.68 6.5 & 1.65 6.2 & 1.5 7.13 & 2.5
Classification Silty sand Silty sand Silty sand Silty sand
L.L (%) & P.L (%) 25.25 & 22.08 28.28 & 21.09 30.12 & 25.03 29.23 & 26.65
Table 2 Engineering properties of the samples

Engineering properties Sample 1 Sample 2 Sample 3 Sample 4
C&F c-23.58 KN/m2 c-20.62 KN/m2 c-19.52 KN/m2 c-20.58 KN/m2
F −30.520 F −28.230 F −25.690 F −31.530
Maximum dry density 1.54 gm/cc 1.68 gm/cc 1.59 gm/cc 1.62 gm/cc
(gm/cc)
Optimum moisture content 18 20.61 19.23 23.25
(%)
Effect of Soil Characteristics on Landslide

The rock information we get from Bhuban teratory information. The soil informa-
tion we get from the laboratory test the soil is unconsolidated sandy materials with
occasional intercalation of moderate to poorly sorted silt or clay layers. After site
investigation we saw that the whole slope is composed of highly weathered and
fragile shale and mudstone. In some parts of the slope, coarse grained sand with mud
balls was also found. The whole landslide scar is mainly composed of huge thickness
of laminated siltstone, shale with narrow bands of sandstone; occasionally lenticular
structure of medium to coarse sandstone with mudstone. Basically the whole area is
composed of mainly silty sand soil that’s why the area is prone to landslide (Fig. 5).
Effects of Cohesion and Shear Strength
From the results obtained from the direct shear strength test we have seen that cohe-
sion value is in the range of 19–24 kN/m2 . So the value of cohesion is very less. It is
one of the causes of instability of slope. As the cohesion is the main component for
resistance of soil to sleep then the angle of internal friction also triggers the landslide
(Fig. 6).
Effect of Permeability on Landslide
Collected samples from the study area we have done falling head permeability test
for different layer of soil. We have seen that the permeability of upper layer soil is less
than the permeability of lower layer soil. For this after rainfall the water gets logged
30
25
20
% Finer
locaƟon 1
15
locaƟon 2
10
locaƟon 3
5
locaƟon 4
0
1 0.1 0.01
ParƟcle size(mm)
Fig. 5 Particle size distribution curve of silty sand
1.4
1.2
Shear Stress(kg/cm 2)
1
0.8
0.6
0.4
0.2
0
0 0.5 1 1.5 2
direct stress(kg/cm2)
Fig. 6 Shear stress versus normal stress curve
in the upper layer soil naturally pore water pressure is developed in this region. The
upper layer of the soil gets sheared off due to the pore water pressure in this region.
We also have seen that permeability is decreasing day by day in the soil. Because
of this after few more days of rainfall there will be a chance of shear failure to be
started (Fig. 7).
Effects of Vegetation Cover on Landslide
Vegetation cover has an effective impact on resistance of slope failure. Due to
increasing population infrastructural works going on the slope cutting trees. During
the time of site investigation we have seen that due to less vegetation landslides are
more on those areas (Fig. 8).
1.6
1.4
1.2
Permeablity
1
locaon 1
0.8
locaon 2
0.6
locaon 3
0.4
locaon 4
0.2
0
0 2 4 6 8
Day
Fig. 7 Permeability in 10–4 mm versus no of days curve
Fig. 8 Landslide area with less vegetation cover
Effects of Vertical Slope Cutting on Landslide

Basically in hilly areas, roads and other infrastructural works are done by cutting the
slope without precautions. Due to less skill sometimes we have seen during our site
investigation they have cut the slope vertically for the widening of pre-existing road.
Table 3 Stability analysis of different locations

Location 1 Location 2 Location 3 Location 4
C&F c-23.58 KN/m2 c-20.62 KN/m2 c-11.52 KN/m2 c-10.58 KN/m2
F −30.52 F −28.23 F −25.69 F −31.53
SLOPE ANGLE (β) 70 65 85 85
Height in m 6.5 7.5 6 7.2
Unit weight () 15.56 17.52 18.32 15.53
tan φ 0.395 0.29 0.21 0.15
tan ∅ 0.38 0.45 0.59 0.7
FOS 1.55 1.19 0.81 0.876
6 Slope Stability Analysis
We used Taylor’s stability method to analyse the stability of slopes by the data
obtained from field survey and experiment conducted on laboratory. From this we
have seen that locations 3 & 4 have the factor of safety less than one that means those
slopes are unstable and prone to landslides (Table 3).
7 Conclusion
Landslide is a complex phenomenon and very deep knowledge is required to analyse

the causes of landslide. This has mainly happened when there is a loosely deposited
siltstone and sandstone and occasionally sandstone with limestone. Rainfall is also
a triggering factor for occurrence of this hazard. As it is a complex phenomenon
a deep knowledge is required to take some mitigation steps. It is very difficult to
predict the causes of landslide only by checking the geotechnical properties. Grain
size and shear strength have the main effects on the occurrence of landslides. The
result obtained from the present study is very helpful for slope management thus its
helps to do infrastructural works on those landslide prone areas.
References
1. Anbarasu K, Sengupta A, Gupta S & Sharma S P (2010) Mechanism of activation of the Lanta
Khola landslide in Sikkim Himalayas. Landslides, 7(2), 135–147.
2. Dahal R K, Hasegawa S, Nonomura A, Yamanaka M, Masuda T & Nishino K (2008) GIS-based
weights-of-evidence modelling of rainfall-induced landslides in small catchments for landslide
susceptibility mapping. Environmental Geology, 54(2), 311–324.
3. Wei Y, Wu X, Cai C, Wang J, Xia J, Wang J & Yuan Z (2019) Impact of erosion-induced
land degradation on rainfall infiltration in different types of soils under field simulation. Land
Degradation Develop 30(14):1751–1764
4. Ghosh S, van Westen CJ, Carranza EJM, Jetten VG (2012) Integrating spatial, temporal, and
magnitude probabilities for medium-scale landslide risk analysis in Darjeeling Himalayas,
India. Landslide 9(3):371–384
5. Mandal B, Mandal S (2018) Analytical hierarchy process (AHP) based landslide susceptibility
mapping of Lish river basin of eastern Darjeeling Himalaya, India. Adv Space Res 62(11):3114–
3132
6. De SK, Jamatia M, Bandyopadhay S (2008) Geo-technical investigation of Mirik Landslide,
Darjiling Himalayas. Geomorphology in India. Prayag Pustak Bhawan, Allahabad, pp 207–216
7. Dutta KK (1966) Landslips in Darjeeling and neighbouring hill slopes in June 1950. Bull GSI,
Ser B 15(1):7–30
8. Paul DK (1973) Report on geo-technical investigation of hill slope stability and alignment of
border road under project Swastik in Darjiling district, West Bengal and Sikkim. Unpublished
Geological Survey of India Report
9. Basu SR (1985) Some considerations on recent landslides at Tindharia and their control. Indian
J Power River ValY Dev 88–194
10. Varnes DJ (1958) Landslide types and processes. In: Landslides and engineering practice, vol
24, pp 20–47
Stability Analysis of Kattipara
and Meppady Regions of Hill Soil Slope
Abstract Kerala has been witnessing a large number of slope failure issues nowa-
days. After 2018 and 2019 floods, due to the landslide activities occurred in Kerala hill
regions, people are getting aware of the seriousness of the slope instability problems.
Factors such as geomorphology, slope angle, terrain curvature, slope length and steep-
ness, soil type and land use or land cover are considered to affecting slope stability.
The present study concentrates on the stability of Kattipara and Meppady regions
of hill soil slopes. 2018 Karincholamala landslide comes under Kattippara region,
and 2019 Puthumala landslide occurred at Meppadi region. The study area faces
several slope failure issues. Rainfall is the primary triggering factor of slope failures
in the region. Human-made activities of deforestation, constructing high rise build-
ings at slopes and mining explorations take part in making slopes more vulnerable to
failure. Disturbances of soil due to vehicle motions, quarrying, construction activi-
ties, and earthquakes can affect the dynamic stability of slopes. Before performing the
stability analysis under static and dynamic loading conditions, strength properties of
soils collected from those regions are determined after conducting the experiments.
SLOPE/W software using Limit equilibrium and pseudo static methods are used to
obtain the factor of safety of slopes. Results are indicating that the soil slopes are
about to fail under static as well as pseudo static conditions if the soils along the
slopes are submerged below the piezometric line.
Keywords Soil strength · Stability of slope · Statics and dynamics
1 Introduction
Kerala is highly vulnerable to climatic variations and natural disasters, including land
slides and geostructural failures. Landslides are becoming the most common natural
hazard in recent days of Kerala state. Massive damages to soil structures occurred
due to the landslides triggered in the monsoon seasons of 2018 and 2019 years. The
P. Aswathi · K. Rangaswamy (B)

Department of Civil Engineering, NIT Calicut, Kozhikode 673601, Kerala, India
e-mail: ranga@nitc.ac.in
202 P. Aswathi and K. Rangaswamy
infrastructures are entirely damaged, those including roads, railways, bridges, power
supply utilities, and communication networks located nearby the steep slopes. Many
people have died or injured and also lost the shelter during the past occurred landslide
events. Kerala experiences several types of landslides include rock falls, rock slips,
slumps, creeps, debris flows and in a few cases, rotational kinds of slides. Among
those, the most prevalent and disastrous type of mass movements observed in Kerala
is the ‘debris flows.’ Majority of downslope slides that have occurred in the Kerala
state is due to the strength loss of saturated soil mass. The saturated overburden soil
mass containing debris material is ranging in size from soil particles to huge boulders
destroying and carrying with it everything that is lying in its path [1].
Recent literature on landslide investigations reveals the significance of analyzing
and stabilization of soil slopes in order to mitigate the failures. Arca and Lorenzo
[2] present the spatial distribution of Factor of Safety (FS) using Limit Equilibrium
Method (LEM) and GIS techniques to improve the landslide mapping at Philip-
pines mountain slopes. The accuracy of the model was validated through actual field
observations and also by overlying the landslide inventory. It was found that 78% of
inventoried landslides are in low to very low stability zones. This technique provides
better insight on the landslide susceptibility throughout the study area and can be
used directly for engineering design purposes. Wang et al. [3] analyzes the stability
of the Duanjiagou landslide on the Bazhong to Guangan Expressway K134–K135
segment in China. The soil samples were collected from bore holes of the slided
zone and carried out direct shear and residual shear tests. Stability was analyzed by
the landslide force transmission and numerical simulation methods using Geostudio.
The results showed that the safety factor obtained was in good agreement with the
field observation situation.
Montrasio et al. [4] examines the efficacy of SLIP model used to derive a simpli-
fied method to estimate multiple seasonal cycles of the mean degree of saturation of
soil and to carry out the time varying stability analysis of a test site slope. It uses
available climatic data, such as air temperature and rainfall depth, and is validated
through the comparison with long-term field measurements on a slope in Canneto
Pavese, northern Italy. This model used to obtain the factor of safeties of slopes
those subjected to excess rain fall. Gutiérrez-Martín et al. [5] developed an innova-
tive analytical model using the limit equilibrium method supported by a geographic
information system (GIS). This model is especially useful for predicting the risk
of landslides in scenarios of heavy unpredictable rainfall. To validate the model, it
is applied to a real life example of a landslide which resulted in human and material
losses. Pandit et al. [6] performed a slope stability analyses on Himalayan landslide
using limit equilibrium and finite element methods. It was resulting that the slope is
just marginally stable, and decided to improve the slope by the application of soil nails
and/or pile reinforcement at critical locations along the slope. It also describes the
design methodology and the significance of pile reinforcements as a novel technique
for slope stabilization.
Present study concentrates on analyzing the stability of Kattipara and Meppady
regions of hill soil slopes in Kerala. 2018 Karincholamala landslide comes under
Kattippara region in Kozhikode district, and 2019 Puthumala landslide occurred
Stability Analysis of Kattipara and Meppady Regions … 203
at Meppadi region in Wayanadu district. The areas like Wayanad and Kozhikode
districts are prone to deep-seated landslides. However, Idukki and Kottayam areas
are susceptible to shallow landslides, which is mainly due to the excessive amount
of rainfall [7]. Soil slopes in this study area failed due to excessive rain and other
human-made factors like deforestation, mining, construction of high rise buildings
along the hill slopes, etc. Rainfall is the primary triggering factor of slope failure in
this region. The daily or the threshold rainfall alone cannot be used as the determinant
for landslide because antecedent rainfall increases the permeability of the soil, and
subsequent storm events may trigger an avalanche or landslide [8]. Due to the excess
rainfall, the soil becomes fully saturated and water level reaches the ground surface.
Here, the main triggering factor from rainfall is pore pressure build-up which causes
to reduce the effective stress. Hence in the numerical model, pore pressure coefficient
is considered to one and accordingly the pore pressure heads along the failure surface
were calculated to obtain the effective stresses. Factor of safety against sliding is
calculated based on effective stresses. The present study used the limit equilibrium
numerical method of Morgenstern–price analysis model which one inbuilt in the
Slope/W software.
2 Stability Analysis Methods
2.1 Limit Equilibrium Method
For calculating slope stability, various numerical modeling techniques are available
now. But this softwares require various geotechnical parameters for stability compu-
tations. Hence it’s important to study the geotechnical characteristics of soil slopes
such as shear strength, grain size distribution, specific gravity, and Atterberg’s limits.
Various methods used for slope stability analysis include Limit Equilibrium Methods,
Numerical methods, Artificial Neural Methods, and Limit Analysis. Among all, the
limit equilibrium method is most commonly used due to its reliability and conve-
nience. It considered the principle of static forces and moments. It requires the input
of soil shear strength parameters such as cohesion and angle of internal friction
including density and slope geometry. Present study used SLOPE/W software using
Limit equilibrium method to obtain factor of safety of slopes under static loading.
Factor of safety is calculated as
Mr cl + (W cos α) tan φ

FS = = (1)
Mo W sin α
2.2 Pseudo Static Method
If the soil slopes are susceptible to continuous disturbances due to heavy traffic, quar-
rying, construction activities, machine induced vibrations and earthquakes, the slopes
are needed to be analysed for dynamic stability. Dynamic slope stability could be
carried out by using either time history or pseudo static analyses. Present study used
pseudo static analysis to get critical acceleration i.e., defined as the ground motion
necessary to begin the process of slope failure. In this method, the pseudo static
forces in horizontal (F h ) and vertical (F v ) directions are to be applied in different
magnitudes to examine the stability of slopes. Pseudo static analysis involves simply
adding a permanent body forces i.e., pseudo static coefficients (k h and k v ) repre-
senting the earthquake shaking to a static limit equilibrium analysis [9]. Here the
factor of safety can be found out by the equation of
clab + ((W − Fv ) cos α − Fh sin α) tan φ

FS D = (2)
(W − Fv ) sin α + Fh cos α
where Fh = Wg ah = kh W and Fv = Wg av = kv W .
A common approach to utilizing pseudo static analysis is simply to run the limit
equilibrium analysis using different values of pseudo static coefficients. The pseudo
static coefficient is called as yield coefficient if the factor of safety becomes unity.
In the simplest form, any ground acceleration due to inertia that exceeds the product
of gravitational force and pseudo static factor is causing to failure of slope under
dynamic loads.
3 Study Area
Two study areas namely Kattippara and Meppady hill slopes are chosen for the
present study. Kattippara is a small Panchayath (21.29 km2 ) in Kozhikkode district
consisting of steep hill slopes. It is located about 10 km from Thamarassery town
and lies between North 11°21 40 and East 76°0 35 . The slopes are covered almost
with earth material. Kattippara has a generally cool humid climate with a very hot
season extending from March to May. The average annual rainfall is more than
3500 mm and it is the highest rainfall in this region which is greater than average
rainfall of 3266 mm in Kozhikode city region. Another study area Meppady is a
Panchayath (198.65 km2 ) in Wayanad district and it shares its southern boundary
with Kozhikode district. It’s an extremely scenic hill station on the State Highway
between Kozhikode and Ooty. The nearest city is Kalpetta and is the headquarters
of the Wayanad revenue district. Meppady lies in the coordinates of North 11°34 0
and East 76°9 0 . It’s about 7000 ft above average mean sea level.
3.1 Geology and Geomorphology of the Study Area
Kozhikode district has divided in to three geological belts viz., (i) linear NW–
SE trending gneissic belt, along the middle extending from north to south, (ii) a
charnockite belt occupying areas in the northeast and south, extending to the adjacent
districts and also occurring as pockets within the gneissic terrain, and (iii) a narrow
coastal belt. In which Kattippara region consists of Hornblende-biotite gneiss of the
Migmatite complex of Archaean period extending from north to south and is well foli-
ated. These medium grained, foliated, banded rock types occur within the migmatites
and associated retrograded charnockites. It consists of alternate layers rich in horn-
blende or biotite. Rocks are having bands of coarse to medium grained light grey to
pink granites. The rocks are with highly deformed admixtures of contorted bands and
enclaves of pyroxene granulite, calcium granulite, and hornblende-biotite granulite.
NW–SE trending dolerite dykes are also present in this region and these dykes are
about 10–20 m wide.
Kattippara region mainly consists of denudation hills and valley flats. They are
exposed as a group of massive hills with resistant rock bodies and rounded summits
and are formed due to differential erosion and weathering. Denudation hills are
identified in the satellite imagery by their massive size and dome to elliptical shape.
They appear as dark green in color in the satellite imagery. These hills are covered
with big boulders and sparse vegetation in contrast to structural hills. This landform,
in general, act as high runoff zone, due to its moderate to steep slope (5–25°). Rubber
and coconut plantation are the main cultivations in the productive slopes of the region
(District survey Report of Minor Minerals, Kozhikode District, 2016). Wayanad
district was broadly divided into four geological domains viz., (i) Peninsula Gneissic
complex in the north and central part, (ii) Migmatite complex in the south central part,
(iii) Charnockite group in the south, and (iv) Wayanad Group in the north. Meppady
region comes under charnockite group which comprises the charnockite forming
hilly terrain in the south and southeast. Pyroxene granulite and banded magnetite
quartzite are occurring as narrow bands within charnockite. Geomorphology of the
region consists of structural cum denudation hills along with dissected plateau in the
northern part and erosional remnants in the southern part (District survey Report of
Minor Minerals, Wayanad District, 2016).
4 Sampling and Geometry of Soil Slope
A sufficient number of undisturbed soil samples were collected from the top and
middle part of the soil slopes in Kattippara and Meppady region of hill soil slopes.
The soil samples were sealed in the metal core cutters by using paraffin wax and
transported to the laboratory without disturbing the structure of soil. Location details
of soil collection and slope geometrical features are tabulated in Tables 1 and 2.
Table 1 Location details and slope geometry of Kattippara hill soil slopes
Soil slope at Soil Location of soil slope Slope details
collected at Latitude Longitude height Angle (°)
(m)
Karinchola Top N11°28.046 E075°55.849 20 25
Middle N11°28.558 E075°55.836
Calvary Top N11°28.046 E075°55.256 15 20
Middle N11°28.010 E075°55.246
Poovanchola Top N11°28.787 E075°55.813 10 16
Middle N11°28.756 E075°55.806
Kualamala-Triveni Top N11°28.893 E075°55.789 10 20
Middle N11°28.891 E075°55.774
Kualamala-Triveni Top N11°28.947 E075°55.894 20 25
(near water fall) Middle N11°28.936 E075°55.888
Theyyathumpara Top N11°27.612 E075°55.811 20 27
Middle N11°27.585 E075°55.797
Mavullapoyil Top N11°29.362 E75°54.979 15 18
Middle N11°29.375 E075°54.973
Cherachankdy Top N11°28.816 E075°55.037 25 20
Middle N11°28.876 E075°55.041
Kelanmoola Top N11°28.386 E075°56.202 30 20
Middle N11°28.371 E075°56.210
Table 2 Location details and slope geometry of Meppady hill soil slopes
Soil slope at Soil Location of soil slope Slope details
collected at Latitude Longitude height Angle (°)
(m)
Mummykunnu Top N11°30.896 E076°08.039 8 40
Middle N11o 30.900 E076°08.032
Kalladi-Puthumala Top N11°30.611 E076°08.240 6 45
Roadside Middle N11°30.622 E076°08.202
Panchami-Kalpatta Top N11°33.788 E076°07.752 6 70
Roadside Middle N11°33.797 E076°07.715
Poothakolli Top N11°34.670 E076°06.484 5 60
Middle N11°33.660 E076°08.596
Thazhe Arrapett Top N11°33.238 E076°09.039 6 75
Middle N11°32.821 E076°09.443
5 Experimental Program
Initially the undisturbed soil samples collected from study area are tested for natural
moisture content and bulk unit weight of soil mass. Further the experiments were
carried out to determine the particle size distribution, Atterberg’s limits in the fine
grained soils, and also shear strength parameters. Shear strength parameters of cohe-
sion and friction angles of all the soil samples were determined after conducting the
direct shear box tests. All the experimental tests were followed by the test proce-
dures according to Indian standard codes. All the basic properties of soil samples
determined from the tests are summarized in Tables 3 and 4.
Results are indicating that the Kattippara soils have the water content in the range
of 15–31% with average unit weight of 15–18 kN/m3 . Meppady soils exhibit the water
content in the range of 30–40% with average density of 12–17 kN/m3 . However, these
water contents are increasing toward saturated values during heavy rains and causing
the slopes under failure due to occurring of undrained condition along the failure
surfaces. Based on particle size distribution, it is understood that the Kattippara soils
are purely cohesionless and Meppady soils are predominantly cohesive soils. Soils
are belonging to low to medium plastic consistency. The cohesion strength values
are in the range of 0 to 28.5 kPa for Kattippara soil samples and 0 to 40 kPa for
Meppady soil samples. Cohesion helps to bind the soil particles tightly and there is
an inverse correlation between friction angle and cohesion with landslide potentiality
[10]. Friction angles of soil samples were found in-between 18 and 34°. Bonding
of the soil particles due to cohesion is influenced and loosened by the presence of
lubricating agent (water and ice particles) and ensure the soil mass to collapse.
6 Stability Analysis of Soil Slopes
Slope stability analysis is performed by using numerical program called SLOPE/W

feature built in Geostudio 2019 software. Input parameters required for a slope
stability analysis includes (i) geometry of slope—description of the stratigraphy
and shapes of potential slip surfaces, (ii) soil strength parameters used to describe
the capacity of soil, and (iii) unit weight of soil including piezometric surface or pore-
water pressures. SLOPE/W uses the analytical methods in solving the stability of soil
slopes are: Ordinary (Fellenius) method, Bishop simplified method, Spencer method,
Janbu simplified method, Morgenstern–price method, generalized limit equilibrium
and finite element methods. Except Morgenstern–price and generalized limit equi-
librium methods, all the other methods do not consider interslice shear forces in their
analysis. Morgenstern-Price and General limit equilibrium methods use both normal
and shear forces between the slices and satisfy the equilibrium condition of both the
forces and moments [11].
6.1 Factor of Safety with Static Loading Conditions
In the present study, factor of safety of soil slopes in the regions of Kattippara and
Meppady were found out at two extreme conditions of fully dry and saturated states.
Typical contours on static stability analysis of two soil slopes in Kattippara and
Table 3 Basic properties of soil samples collected from Kattippara soil slopes
208
Soil sample at Natural state of soil Grain size distribution Atterberg’s limits Shear strength parameters
Density Water Gravel (%) Sand (%) Silt + LL (%) PL (%) PI C (kPa) F (degrees)
(kN/m3 ) content (%) clay (%)
Karinchola
Top 16 22.8 3 78 19 NA NA 0 10 27.3
Middle 15.1 21.5 2 62 36 31 21 10 0 30
Calvary
Top 16.42 18.6 2 68 30 25 16.2 8.8 3 24.8
Middle 16.02 15.4 2 70 28 NA NA 0 0 21.7
Poovanchola
Top 17.16 18 3 53 44 54 33.5 20.5 18 29.6
Middle 15 22 3 54 43 26 18 8 10 6.5
Kualamala-Triveni
Top 15.8 23.43 1 62 37 38 23 15 20 21.2
Middle 15.26 19.45 2 73 25 34 19.8 14.2 17 18.1
Kualamala-Triveni (near water fall)
Top 15.46 27.6 1 47 52 68 35.7 32.3 15.8 26.4
Middle 15.16 21.51 1 56 43 29 19 10 2 25.7
Theyyathumpara
Top 17.69 18.74 3 88 9 NA NA 0 0 32.5
Middle 17.27 15.64 6 82 12 NA NA 0 28.5 13.35
Mavullapoyil
Top 16.44 31 14 53 33 53 35.8 17.2 12 31
(continued)
Table 3 (continued)
Soil sample at Natural state of soil Grain size distribution Atterberg’s limits Shear strength parameters
Density Water Gravel (%) Sand (%) Silt + LL (%) PL (%) PI C (kPa) F (degrees)
Middle 15 25.8 10 46 44 61 37.4 23.5 11 34
Cherachankandy
Top 17.45 28.41 3 59 38 46 32 14 20 28.4
Middle 17.37 21.98 0 47 53 50 33.6 16.4 22 32.6
Kelanmoola
Top 16.88 26.7 2 56 42 50 28.4 21.6 20 32.6
Middle 16.11 19.3 4 64 32 37 22.2 14.8 11 26.6
Stability Analysis of Kattipara and Meppady Regions …
209
210
Table 4 Basic properties of soil samples collected from Meppady hill soil slopes
Soil sample at Natural state of soil Grain size distribution Atterberg’s Limits Shear strength parameters
Density Water Gravel (%) Sand Silt + LL (%) PL (%) PI C (kPa) (degrees)
(%)
Kalladi-Puthumala roadside
Top 14.42 30.6 6 46 48 67 46.4 20.6 17 30
Middle 14.40 36.8 14 41 45 58 42.75 15.25 0 31
Mummykunnu
Top 16.9 31.52 0 25.5 74.5 64 40 24 30 22
Middle 17.3 36.48 3 31 66 58 38.46 19.54 25 20
Panchami-Kalpetta roadside
Top 12.15 30.8 5 28 67 60 35.8 24.2 26 15
Middle 12.50 32.2 4 28 68 52 30.6 21.4 15 27
Poothakolli
Top 15.11 32.5 0 14 86 66 45 21 40 23
Middle 15.21 35 0 16 84 59 41.2 17.8 21 22
Thazhe Arrapetta
Top 11.77 28.71 0 20 80 63 43.5 19.5 42 26
Middle 12.89 30.66 0 22 78 56 37.6 18.4 15 26.5
Meppady regions are shown in Figs. 1 and 2. The analysis is extended to all other
soil slopes and the factor of safety against slope stability for all the soil slopes under
static loading with dry and saturated water conditions was reported in Table 5. Results
are indicating that the existing soil slopes are to be failed during rainy season due
to excess rainfall and development of continuous pore pressures along the failure
surfaces of slopes. The factor of safeties of soil slopes determined is in the range of
0.56–1.5 under saturation during rainfall seasons. Even the slopes are safe against
sliding (FSL > 1.5) during dry season without water content, the slopes become
unstable with excessive rainfall and pore pressure developments. A few slopes are
under margin of stable with factor of safety of 1.5. It is observed that the slide at
Karinchola of Kattippara and the slide at Kalladi-Puthumala of roadside of Meppady
regions are having lowest factor of safety of 0.45 and 0.28, respectively. Poothakolli
slide in Meppady region is safe under static loading with factor of safety around
2.3. However, even the slopes are stable under only rainfall conditions, but they may
fail with man-made activities of mine explorations, continuous heavy traffic and
Fig. 1 Typical contours on stability analysis of Calvary hill soil slopes in Kattippara. a Dry condition
and b saturated condition
Fig. 2 Typical contours on stability analysis of Mummykunnu soil slopes in Meppady. a Dry
condition and b saturated condition
Table 5 Factor of safety against stability of soil slopes under static loading
Soil slopes in Factor of safety against Soil slopes in Factor of safety against
Kattippara region at stability Meppadi region at stability
Dry Saturated Dry Saturated
condition condition condition condition
Calvary 1.481 0.582 Mummykunnu 2.036 1.456
Karinchola 1.438 0.459 Kalladi-Puthumala 1.401 0.28
Roadside
Poovanchola 2.978 1.549 Panchami-Kalpatta 2.003 1.211
Roadside
Kualamala-Triveni 2.335 1.502 Thazhe Arrapetta 2.346 1.46
Kualamala-Triveni 1.567 0.627 Poothakolli 2.823 2.233
(Water fall)
Theyyathumpara 1.464 0.569
Mavullapoyil 3.046 1.555
Cherachankandy 2.736 1.578
Kelanmoola 1.922 0.866
industrial vibrations, and seismic events. So the slopes are to be examined for the
stability against various pseudo static forces.
6.2 Factor of Safety with Pseudo Static Loading Conditions
Slope stability analysis is carried out for all the slopes by varying the pseudo static
coefficients, kh values, in the range of 0.1–0.8. Vertical pseudo static coefficient is
considered to zero since it reduces both the driving and resisting forces. A typical
contour on pseudo static slope stability of Karinchola, and Kalladi-Puthumala hill soil
slopes during excess rainfall and subjected to pseudo static coefficient of 0.2 under
dynamic loading is shown in Fig. 3. It is extended to all the other soil slopes and
the reduction of factor of safety of Kattippara and Meppady soil slopes by varying
pseudo static coefficients in dry and excess rainfall conditions and are shown in
Figs. 4, 5, 6 and 7.
Factor of safety of Karinchola and Kalladi-Puthumala hill soil slopes subjected
pseudo static loading at kh = 0.2 are about 0.25 and 0.16 as shown in Fig. 3 and are
still lesser when compared with static loading. Figures 3, 4, 5, 6 and 7 are indicating
that the factors of safeties of soil slopes are decreasing with increase in the amplitudes
of horizontal pseudo static coefficients. It is concluding that the soil slopes even under
dry conditions fail if the intensity of pseudo static coefficients exceeds above 0.3–0.4.
The saturated soil slopes under excess rainfall conditions are completely unstable if
the slopes are susceptible to dynamic loading even at pseudo static coefficient of 0.2.
Fig. 3 Typical contours on pseudo static stability analysis with kh = 0.2: a Karinchola, and
b Kalladi-Puthumala hill soil slopes during excess rainfall
2.2
Mavullapoyyil
2.0 Cherachankandy
Poovanchola
1.8 Kualamala
Kelanmoola
Factor of safety
1.6
Kualamala Waterfall
1.4 Theyyathumpara
Karincholamala
1.2 Calvary
1.0
0.8
0.6
0.4
0.0 0.2 0.4 0.6 0.8 1.0

Horizontal Pseudostatic coefficient,Kh
Fig. 4 Reduction of factor of safety of Kattippara soil slopes by varying pseudo static coefficients—
dry weather condition
7 Conclusions and Recommendations
A series of experiments were conducted on collected soil samples from the study area
of soil slopes and the parameters required to give as input into the numerical program
are determined. SLOPE/W software using limit equilibrium static and pseudo static
methods are performed to obtain the factor of safety of soil slopes. Present study
reveals that from the geometrical features of the slope height and slope angle affect
the factor of safety adversely. Meppady soil slopes are having steeper slope angles
and show less factor of safety compared to flatter slopes of same height and soil
properties. Kattippara soil slopes are flatter when compared with Meppady soil slopes
and Kattipara soil slopes are having slightly higher factor of safeties. The major
observations are: (i) The existing soil slopes fail during rainy season due to excess
rainfall and development of continuous pore pressures along the failure surfaces of
Poothakolli
Thazhe Arrapetta
2.2 Panchami-Kalpetta road
2.0 Mummykunnu
Kalladi-Puthumala road
1.8
Factor of safety
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.0 0.2 0.4 0.6 0.8

Fig. 5 Reduction of factor of safety of Meppady soil slopes by varying pseudo static coefficients—
dry weather condition
Cherachankandy
Kualamala
1.2
Mavullapoyyil
Poovanchola
1.0 Kelanmoola
Kualamala Waterfall
Calvary
Factor of Safety
0.8
Theyyathumpara
Karincholamala
0.6
0.4
0.2
0.0
0.0 0.2 0.4 0.6
Fig. 6 Reduction of factor of safety of Kattippara soil slopes by varying pseudo static coefficients—
excess rainfall condition
slopes. Even a few slopes are under margin of stable with factor of safety of 1.5,
but they fail completely under the dynamic influences with increasing the pseudo
static coefficients. Soil slopes are susceptible to sliding under dynamic influences
even at dry weather conditions. (ii) Slides at Karinchola of Kattippara and the slide
at Kalladi-Puthumala of roadside of Meppady regions are having lowest factor of
safety of 0.45 and 0.28 under static loading. These factor of safeties are reduced to
still lower values 0.25 and 0.16, respectively, with the pseudostatic loading coefficient
of 0.2., and (iii) Soil slopes even under dry conditions fail if the intensity of pseudo
static coefficients exceeds above 0.3–0.4. The saturated soil slopes under excess
Poothakolli
1.8 Mummykunnu
Thazhe Arrapetta
1.6 Panchami-Kalpetta roadside
Kalladi-Puthumala roadside
1.4
Factor of safety
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0.0 0.2 0.4 0.6 0.8
Fig. 7 Reduction of factor of safety of Meppady soil slopes by varying pseudo static coefficients—
excess rainfall condition
rainfall conditions are completely unstable if the slopes are susceptible to dynamic
loading even at pseudo static coefficient of 0.2. Hence the existing soil slopes are
to be stabilized by using various techniques including lime column intrusions, soil
nailing, piling or using any innovative chemical grouting methods.
References
1. Thampi PK, Mathai J, Sankar G, and Sidharthan S (1998) Evaluation study in terms of landslide
mitigation in parts of Western Ghats, Kerala. Research report submitted to the Ministry of
Agriculture, Government of India. Centre for Earth Science Studies, Government of Kerala,
Thiruvananthapuram, India
2. Arca MCQ, Lorenzo GA (2018) Landslide hazard mapping using limit equilibrium method
with GIS application of roadway traversing mountain slopes: the case of Kitaotao Bukidnon,
Philippines. J Nepal Geol Soc 55:93–101
3. Wang X, Xia T, Zhang L, Gao M, Cheng K (2020) Field geological investigations and stability
analysis of Duanjiagou landslide. ISPRS Int J Geo-Inf 9:23. https://doi.org/10.3390/ijgi90
10023
4. Montrasio L, Valentino R, Meisina C (2018) Soil saturation and stability analysis of a test
site slope using the shallow landslide instability prediction (SLIP) Model. Geotech Geol Eng
2018(36):2331–2342. https://doi.org/10.1007/s10706-018-0465-3
5. Gutiérrez-Martín A, Herrada MÁ, Yenes JI, Castedo R (2019) Development and validation of
the terrain stability model for assessing landslide instability during heavy rain infiltration. Nat
Hazards Earth Syst Sci 19(721–736):2019. https://doi.org/10.5194/nhess-19-721-2019
6. Pandit K, Sarkar S, Samanta M, Sharma M (2016) Stability analysis and design of slope
reinforcement techniques for a Himalayan landslide. In: Conference proceedings on recent
advances in rock engineering. https://doi.org/10.2991/rare-16.2016.16
7. Kuriakose SL, Sankar G, Muraleedharan C (2009) History of landslide susceptibility and a
chorology of landslide-prone areas in the Western Ghats of Kerala in India. Environ Geol
57:1553–1568
8. Rahardjo H, Li XL, Toll DG, Leong EC (2001) The effect of antecedent rainfall on slope
stability. Geotech Geol Eng 19:371–399
9. Jibson RW (2011) Methods for assessing the stability of slopes during earthquakes—a
retrospective. Eng Geol 122: 43–50
10. Mandal S (2015) Assessing cohesion, friction angle and slope instability in the Shivkhola
watershed of Darjiling Himalaya. Res J Earth Sci 3
11. Karhn J (2004) Stability modeling with slope/W: an engineering methodology, 1st edn, 17 pp
Numerical Studies on the Behavior
of Slope Reinforced with Soil Nails
Akhil Pandey, Sagar Jaiswal, and Vinay Bhushan Chauhan
Abstract Soil nailing is commonly used to stabilize cut slopes and earth retaining
structures, embankments and sometimes to reduce the lateral earth pressure on
retaining wall. This is an in-situ reinforcement technique that uses passive rigid
bars usually made up of steel that can withstand tensile forces, shearing forces,
and bending moments. In the present study, an attempt has been made to analyze the
behavior of soil slope reinforced with soil nails by numerical simulation under seismic
loading using a finite element numerical modeling tool, OptumG2. Two-dimensional
models of the slope are considered (with and without nails) and subjected to static
and dynamic loading to investigate the possible modes of failure (base failure, slope
failure, and toe failure). The corresponding factor of safety values for unreinforced
soil slope is calculated using the strength reduction method (SRM) at different slope
angles (β) at 30°, 40°, 50°. Based upon the obtained critical slope surface in the
case of the unreinforced slope, soil nails are provisioned in the slope and the same is
analyzed under gravity and seismic loading. A detailed investigation of reinforced
soil slope has been conducted considering the key factors governing the factor of
safety of reinforced slope namely length of the soil nails, the number of soil nails,
spacing of soil nails (s), the inclination of soil nails (θ ) in different slope angles
(β) under seismic loading with a peak horizontal ground acceleration = 0.12 g. The
present study also investigates the possible failure patterns under static and seismic
conditions, along with the analysis of the development of internal reactions in soil
nails such as bending moment and shear force.
Keywords Soil nailing · Finite element method · OptumG2 · Seismic loading
A. Pandey · S. Jaiswal · V. B. Chauhan (B)

Civil Engineering Department, Madan Mohan Malaviya University of Technology, Gorakhpur
273010, India
e-mail: chauhan.vinaybhushan@gmail.com
A. Pandey
e-mail: akhilpandey812@gmail.com
S. Jaiswal
e-mail: sagarjaiswal.work@gmail.com
218 A. Pandey et al.
1 Introduction
The downward falling or sliding of a soil mass from a steep slope may occur in
regions like hilly areas, offshore and onshore environment, zones of heavy rainfall,
etc. The slip of slope can cause natural damages and sometime cause human losses
also [5]. It can be controlled by the use of proper slope stabilization techniques such
as soil nailing, etc. Soil nailing is one of the recent in-situ methods for soil slope
stabilization and can be used for the future expansion of construction work in the
sloping region applicable to hard to stiff fine-grained soils, silty clays, sandy clays
and very dense granular soils [1–3]. Soil slope stability analysis is an important
component in the design and construction of various geotechnical structures such as
rail or road embankment, road widening under existing bridge end, and temporary
and permanent excavation in those problematic regions [7]. This technique consists of
the passive reinforcement of existing ground by installing closely spaced solid steel
bars (i.e., nails) although hallow-system bars are also available. They are passive
inclusions, which improve the shearing resistance of the soil. The holes with chosen
diameters are drilled at the desired position to install the solid bars and grout or
shotcrete is used to avoid caving in the drilled holes before installing the nails [4].
Centralizers, made of synthetic materials are installed to ensure a minimum thickness
of grout which completely covers the nail bar. The nail-head which is a square-shaped
concrete structure provides bearing strength and transfer bearing loads from the soil
mass to soil nails. Temporary and permanent facing is also provided to ensure the
support to the exposed soil and it also acts as a bearing surface for the bearing plate
[4]. The nails can be arranged in many ways such as steeply inclined nail arrangement,
sub-horizontal nail arrangement, and hybrid nail arrangement. Steeply inclined nail
arrangement generally satisfies the design objectives of sustaining the earth’s pressure
upon the liquefaction of the loose fill. Sub-horizontal nail arrangement is effective
in sliding failure in cut slopes.
Two types of regions are generated in the soil nail system, an active region which
consists of the soil mass above the failure surface which shows the movement of soil
masses and passive region which consists of embedded part of soil nail behind the
failure surface which doesn’t show any movement of soil masses. During the slope
failure, the active region tends to deform which results in axial displacement along
with soil nails that are placed across the slip plane. This results in the development
of tensile forces in soil nail in the passive zone which resists the deformation of the
active zone. This tension force results in an increment of the normal force coming
on a slip plane and reduces the driving shear force. The soil nails are embedded in
the passive region through which it resists the pull-out of the nail from slope through
friction between nails and soil [4].
However, shear force and bending moment carrying capacity of the soil nails are
conventionally unnoticed in the design procedure [13]. Hence, an attempt is made
to examine the generation of these internal stresses, and bending moment and shear
force resistance generated in the soil nails are also evaluated for all the cases of the
reinforced soil slope in the present study. Detailed analysis of the performance of soil
Numerical Studies on the Behavior of Slope Reinforced … 219
slope reinforced with soil nails under seismic loading is still quite fewer in number.
Most of the literature on soil nailed system emphasizes only the static condition.
Few studies have been performed for the slope stability of soil nails reinforced slope
and its failure mechanism under seismic condition [16]. In the present work, a two-
dimensional model of soil slope is considered with and without nails, subjected to
gravity as well as dynamic loading and the corresponding minimum factor of safety
values for soil slope is calculated by the strength reduction method at different slope
angles (β) at 30°, 40°, 50°. Variation of development of maximum bending moment
and shear force is also analyzed with the different failure mechanisms.
2 Methodology
In this present work, the finite element method is adopted for numerical simulation
of soil slope reinforced with and without soil nails under static and seismic condi-
tions. Once the soil parameter is adopted, soil slope is modeled using the Optum G2
numerical tool [10]. The shear reduction method is used to calculate the factor of
safety in all the condition [9] and then the analysis is performed by using soil nails
of constant diameter at different nail inclination angle (θ ) and spacing (s). Different
failure mechanism and effect of bending moment are also analyzed for the different
conditions for finding some beneficial conclusion for slope stability of soil using
nails.
3 Numerical Modeling
Analysis of soil slope by wedge method is complicated as well as time-consuming,

moreover detailed analysis of various soil surface interactions cannot be done
through it as this analysis involves many assumptions related to inter-slice force and
slip surface, which are not true.
Hence numerical modeling using finite element analysis is preferred because of its
simple and efficient approach. A finite element method is a widely adopted method
for the analysis of slope stability because no assumptions of inter-slice force and
slip surface are required [7]. In this study numerical modeling of soil reinforced
with nails under seismic condition is conducted for various parameters to evaluate its
performance by finite element analysis. All the variation in the considered parameters
for the analysis of reinforced slope is as shown in Table 1.
The Mohr–Coulomb constitutive model is considered for the soil which can simu-
late perfectly plastic material conditions with the development of irreversible
strains.
The strength reduction method using the finite element method is used to obtain
the factor of safety for slopes [8]. Designing of soil nails is performed by using
FHWA guidelines [6] and the calculation of the diameter of soil nail is performed
Table 1 Slope and soil nail

Spacing of soil nail Soil slope angle (β) Soil nail inclination
parameters
(s) (θ)
0.5 m, 1 m, 2 m 30° 10°, 20°, 30°, 40°
40° 10°, 20°, 30°, 40°
50° 10°, 20°, 30°, 40°
by the adopted horizontal and vertical spacing of soil nails [6] and subsequently, the
length of soil nails is calculated for the slope.
The soil slopes are reinforced using high yield strength deformed (HYSD) bars
nails at four different nail inclinations of 10°, 20°, 30°, and 40°. The boundary
conditions are equivalent to the fixed bottom, and the roller at the vertical end of the
soil mass is provisioned.
A diagrammatic representation including complete details of the geometric
configuration of soil slope with nailed element is shown in Fig. 1.
The accuracy of numerical simulation is governed by the optimum number of
total elements considered in the mesh, therefore a sensitivity analysis is carried out
to predict the optimum number of total elements in the domain of the numerical
mesh considered in the present study [11, 12, 15, 16]. Variation of the total number
of elements from 1000 to 4000 with the increment of 1000 elements in each iteration
is done, it is noted that a further increase in the total number of elements beyond
2000 elements in the mesh does not improve the factor of safety of the nail soil slope.
The material properties adopted for the numerical simulation are shown in Table
2.
Fig. 1 Numerical Model of soil slope with soil nailing under the seismic condition
Table 2 Material properties

Properties Sand Soil nail
[14]
Young modulus (E), MPa 20 2.6 × 104
Poisson ratio (υ) 0.2 –
Cohesion (c) (kPa) 1.4 –
Friction angle (∅), degree 30 –
Dry unit weight (γd ), kN/m3 17 –
Lateral strength (FL max ), kN/m - 42
The soil slope models are analyzed under static and seismic loading conditions. In the
case of the seismic condition, a horizontal seismic coefficient of 0.12 g and the factor
of safety is determined for the different slopes without soil nails having inclination
ranging from 30° to 50°. The factor of safety obtained from this analysis is shown
in Fig. 2. The obtained factors of safety are calculated for the lower bound (LB)
and upper bound (UB) analysis and presented with the average of factors of safety
and error bars in Fig. 2, whose upper and lower tips show the factors of safety in
the lower bound and upper bound analysis. A very insignificant difference in UB
and LB factors of safety strengthens the fact that proper selection of the number of
elements is chosen for carrying out the present study. It is indicated that with the
Fig. 2 Variation of the factor of safety under the static and seismic condition
increase in the soil slope, the factor of safety decreases, which is self-explanatory
as steeper slopes are unstable. Furthermore, with the application of seismic forces,
FOS of a given slope is found to be lower compared to the static case (under gravity
loading), and this reduction in the FOS was found to be 11.4, 9.5, and 8.6 for slope
having an inclination of 30°, 40°, and 50°.
Furthermore, analysis is also performed for the slope stabilized with soil nails
with similar designing parameters and the factors of safety are examined with the
variation in spacing and inclination of the soil nails, and slope inclination from the
horizontal.
At 0.5 m spacing of soil nails, factor of safety of the soil slope increases with the
increase in soil slope inclination of 30° and 40° and remains constant for 50° slope
angle.
In case of soil nails spacing of 1 m, the factor of safety increases with the variation
of inclination of soil nails at 30° soil slope, for 40° slope angle factor of safety increase
up to θ = 30° but reduce at θ = 40°, and for 50° soil slope angle factor of safety
decreases with increase in soil nail inclination up to 30° and decrease at 40° soil nail
inclination. At 2 m spacing of soil nails factor of safety increase with the increment
in soil nail inclination for 30° and 50° soil slope angle but for 40° soil slope angle
factor of safety decreases up to 30° soil nails inclination and show a little increment
at θ = 40°.
The variation of a factor of safety with the variation of soil slope angle, spacing
of soil nails, and soil nails inclination is shown in Figs. 3, 4, 5.
Analysis of failure patterns observed from the analysis is performed to understand
the generation of failure mechanism under seismic loading and the variation of the
same due to change in the inclination of the soil slope, soil nails’ inclination and
spacing. From the observed failure pattern from each analysis, a significant failure
pattern is shown in Figs. 6, 7, and 8 that the failure surface obtained from all the
simulations performed the present study for a wide range of different slope angles
shows two distinct failure surfaces. Failure patterns for different cases are shown
in. For a mild slope of 30°, the same failure patterns are obtained with the variation
and spacing of soil nails slip surface originating from the top of the slope crest and
progresses toward the bottom of the slope with huge soil mass except for 2 m spacing
and 40° soil nail inclination. In this case, the top surface of soil slope fails with the
small amount of soil mass. For 40° soil slope, failure surface originates from the top
of the soil slope surface and moves toward the bottom for 0.5 m and 1 m spacing
with the variation of soil nails inclination from 10° to 40° but for 2 m spacing, failure
surface shows a different pattern which involves the movement of the soil surface
from the surface of soil slope with small soil mass. Failure pattern shows a vivid
behavior for β=50°, the only top surface near about the soil nails fails with a small
amount of soil mass along with them for the variation of spacing and inclination of
soil nails.
Internal stresses developed in the soil nails i.e., bending moment and shear force in
nails also aid in the stability of the reinforced slope. In view of the above, maximum
bending moment and shear force developed in the nails for all the cases (variation
of slope angle, spacing, and inclination of soil nails) of the reinforced slope are
Fig. 3 Variation of a factor of safety with soil nail inclination (θ)attheslope(β) = 30°
Fig. 6 Failure pattern of soil slope under the seismic condition for slope angle (β) = 30° at a θ=10°
and s = 0.5 m; b θ=30°and s = 1.0 m; and c θ=40° and s = 2 m
analyzed and summarized in Table 3. In Table 3, Observed magnitude of bending

moment and shear force for a given slope with a provided inclination and spacing
of nails is shown, where first numeric value signifies the magnitude of maximum
bending moment/shear force and the alphabet (U, M, and L) in the parentheses
denote the location of the nail for the mentioned internal reaction, where L, M, and
U signify the lower (z = 0–2 m), middle (z = 2–6 m) and upper (z = 6–8 m) section
of soil slope, z specifies the distance measured from the top horizontal surface of the
slope.
Fig. 7 Failure pattern of soil slope under the seismic condition for slope angle (β) = 40° at a θ =
10° and s = 0.5 m; b θ = 30° and s = 1.0 m; and c θ = 40° and s = 2 m
Fig. 8 Failure pattern of soil slope under the seismic condition for slope angle (β) = 50° at a θ =
10° and s = 0.5 m; b θ = 30° and s = 1.0 m; and c θ= 40° and s = 2 m
It is found that spacing and inclination of soil nails plays a vital role in slope
stability. At 30° slope angle, maximum bending moment and shear force are generated
at θ = 10° for 0.5 m, 1 m, and 2 m soil nail spacing and minimum bending moment
and shear force are obtained at θ = 40° and 20° for 0.5 m, 1 m, and 2 m spacing
of soil nails, respectively. At 40° slope angle, maximum bending moment and shear
force are obtained at θ = 10° and 20° for 0.5 m, 1 m, and 2 m spacing of soil nails
and minimum bending moment and shear force are obtained at θ = 40° and 30° for
0.5 m, 1 m, and 2 m spacing of soil nails, respectively. At 50° slope angle, maximum
bending moment and shear force are obtained at θ = 40° and 10° for 0.5 m, 1 m, and
2 m spacing of soil nails and minimum bending moment and shear force are obtained
at θ = 20° and 40° for 0.5 m, 1 m, and 2 m spacing of soil nails, respectively, as shown
in Table 3. It is also noticed that bending moment and shear force developed in the
soil nails show a different variation for different slope angles with the variation of
spacing and inclination angle of soil nails. In the case of 30° slope bending moment
and shear force go on decreasing with the increment in the inclination of soil nails
for 10° to 40° and spacing from 0.5 m to 2 m. For 40°, decrement in bending moment
and shear force is noticed with the increment of soil nail inclination at 0.5 m and 1 m
but at 2 m spacing increase in internal stresses is noticed. At 50° soil slope bending
Table 3 Maximum bending moment and shear force developed in the soil nails for different
reinforced slopes
Soil slope (β) → 30° 40° 50°
Spacing Inclination Maximum Maximum Maximum Maximum Maximum Maximum
of soil of soil nail bending shear bending shear bending shear
nails (s) (θ) moment force (kN) moment force (kN) moment force (kN)
(kN m) (kN m) (kN m)
0.5 (m) 10° 30.15 (M) 23.49 (M) 33.51 (L) 23.26 (L) 11.95 (U) 8.58 (U)
20° 23.89 (L) 15.69 (L) 19.00 (L) 18.41 (L) 12.42 (U) 8.69 (U)
30° 21.98 (L) 14.65 (L) 9.91 (M) 10.18 (L) 15.87 (U) 9.98 (U)
40° 18.80 (L) 13.92 (L) 11.73 (U) 9.18 (U) 22.74 (U) 14.37 (U)
1 (m) 10° 38.72 (M) 29.55 (M) 39.24 (L) 25.87 (L) 30.28 (M) 22.88 (U)
20° 32.97 (U) 17.95 (U) 24.01 (L) 16.28 (L) 32.03 (U) 24.24 (U)
30° 28.38 (U) 17.39 (U) 24.87 (M) 16.78 (M) 34.78 (M) 26.67 (M)
40° 15.81 (L) 14.84 (L) 23.65 (M) 14.93 (M) 48.45 (M) 30.78 (M)
2 (m) 10° 37.76 (M) 27.53 (M) 36.51 (M) 24.19 (M) 53.49 (M) 38.58 (M)
20° 26.44 (L) 24.15 (L) 44.56 (M) 27.59 (M) 64.01 (M) 45.97 (M)
30° 36.16 (L) 22.30 (L) 37.26 (M) 20.51 (M) 52.80 (M) 42.33 (M)
40° 33.17 (L) 21.51 (L) 43.30 (M) 32.19 (M) 38.01 (M) 27.16 (M)
moment and shear force increase with an increase in soil nail inclination at 0.5 m
and 1 m soil nail spacing, at 2 m spacing of soil nails bending moment and shear
increases up to 20° soil nail inclination but later on decreases with increase in soil
nail inclination.
5 Conclusion
A comprehensive finite element numerical analysis for soil slope reinforced with
soil nails at a various inclination (θ ) and spacing (s) of soil nails is performed. It is
noticed that the inclination (θ ) of soil nails and spacing (s) of soil nails play a vital
role in the slope stability of sand. On the basis of present work following conclusions
are drawn:
1. It is found that keeping the diameter of soil nails fixed, the inclination of 30°
gave the maximum factor of safety with respect to 10°, 20° and 40° for the slope
of 30° under horizontal seismic coefficient of 0.12 g at the spacing(s) of 0.5 m.
The increment of 39.5% in the factor of safety would occur by using soil nails.
2. For the 40° slope, the maximum factor of safety can be obtained at soil nail
inclination (θ ) of 30° at 0.5 m spacing(s). The increment of 50.27% in the factor
of safety can be obtained by using soil nails of given parameters under k h =
0.12 g
3. In the case of the slope angle of 50°, the increment of 42.9% can be obtained
under k h = 0.12 g at soil nail inclination (θ ) of 30° at 0.5 m spacing(s).
4. Two different failure patterns are observed during this study, for soil slope 30°
and 40° slip surface found from model testing is in a form of circular slip surface
that originates from the slope crest and progresses towards the bottom of the slope
face and for 50° soil slope only top surface of soil slope fail with the movement
of the small amount of soil mass.
5. The maximum bending moment is obtained at 1 m spacing and 10° soil nail
inclination for 30° slope angle. For 40° and 50° slope angle maximum bending
and shear force is obtained 2 m spacing and 20° soil nail inclination.
References
1. Alsubal S, Indra SH, Babanigida NM (2017) A typical design of soil nailing system for stabi-
lizing a soil slope: case study. Indian J Sci Technol 10(4). https://doi.org/10.17485/ijst/2017/
v10i4/110891
2. Azzam W. Sobhey MA (2019) Utilization of soil nailing technique to improve sand slopes
under seismic loading. In: International conference on advances in structural and geotechnical
engineering. Hurghada, Egypt, pp 0–10
3. Babu GLS, Singh VP (2008) Numerical analysis of performance of soil nail walls in seismic
conditions. ISET J Earthquake Technol 31–40
4. Budania R, Arora RP (2016) Soil nailing for slope stabilization : an overview. Int J Eng Sci
Comput 6(12):3877–3882
5. Fawaz A, Farah E, Hagechehade F (2014) Slope stability analysis using numerical modelling.
Am J Civil Eng 2(3):60–67. https://doi.org/10.11648/j.ajce.20140203.11
6. Federal Highway Administration (FHWA) (2003) Soil nail walls, geotechnical engineering
circular no. 7, report no. FHWA0-IF-03–017. Federal Highway Administration
7. Garzón-Roca J, Capa V, Torrijo FJ, Company J (2019) Designing soil-nailed walls using the
amherst wall considering problematic issues during execution and service life. Am Soc Civil
Eng 19:1–14. https://doi.org/10.1061/(ASCE)GM.1943-5622.0001453
8. Kannaujiya P, Chauhan VB (2020) Behavior of anchored sheet pile wall. In: Shehata H, Brandl
H, Bouassida M, Sorour T (eds) Sustainable thoughts in ground improvement and soil stability,
GeoMEast 2019. Springer, Cham, pp 184–195. https://doi.org/10.1007/978-3-030-34184-8_12
9. Ojha R, Chauhan VB (2020) Performance of geosynthetic reinforced segmental retaining walls.
In: Shehata H, Brandl H, Bouassida M, Sorour T (eds) Sustainable thoughts in ground improve-
ment and soil stability, GeoMEast 2019. Springer, Cham, pp 196–206. https://doi.org/10.1007/
978-3-030-34184-8_13
10. Optum G2 (2017) Optum computational engineering. Copenhagen NV, Denmark
11. Pandey A, Chauhan VB (2020) Evaluation of pull-out capacity of helical anchors in clay using
finite element analysis. Geotechnical Special Publication 2020. GSP 317, pp 60–68. https://
doi.org/10.1061/9780784482803.007
12. Pandey A, Chauhan VB (2020) Numerical analysis for the evaluation of pull-out capacity of
helical anchors in sand. In: Shehata H, Brandl H, Bouassida M, Sorour T (eds) Sustainable
thoughts in ground improvement and soil stability, GeoMEast 2019. Springer, Cham, pp 207–
218. https://doi.org/10.1007/978-3-030-34184-8_14
13. Rawat P, Chatterjee K (2018) Seismic stability analysis of soil slopes using soil nails.
Geotechnical Special Publication, pp 79–87. https://doi.org/10.1061/9780784481486.009
14. Rawat S, Gupta AK (2016) Analysis of a nailed soil slope using limit equilibrium and finite
element methods. Int J Geosynth Ground Eng. https://doi.org/10.1007/s40891-016-0076-0
15. Seyed-Kolbadi SM, Sadoghi-Yazdi J, Hariri-Ardebili MA (2019) An improved strength

reduction-based slope stability analysis. Geosciences 1–11. https://doi.org/10.3390/geoscienc
es9010055
16. Sengupta A, Giri D (2011) Dynamic analysis of soil-nailed slope. Ground Improve 164:225–
234. https://doi.org/10.1680/grim.8.00023
Seismic Stability of Non-homogenous
Cohesive Soil by Using Calculus
of Variation
Sourav Sarkar and Manash Chakraborty
Abstract In this article, the factor of safety of a cohesive soil slope subjected to
seismic load is determined by employing the variational method and pseudostatic
analysis. Unlike the conventional limit equilibrium method, there is no requirement to
consider any kinematical or static assumption in the variational method. The factor
of safety (F) is defined as a functional which is minimized by using the Euler–
Lagrangian equation. The (i) transversality and boundary conditions are imposed at
the intersection of slip surface, and the slope surface, and (ii) continuity and natural
boundary states are forced at the intermediate point of the slip surface. The soil is
considered to be completely saturated and loaded under undrained conditions. The
cohesion of the soil is assumed to increase linearly with depth. The critical slip
surface and consequently critical factor of safety, F s is being obtained by varying
the slope geometry, soil properties, and seismic loadings. The available solutions
compare quite well with the convenient solution for the pseudostatic slope stability
analysis. The proposed design charts will be quite useful to practicing engineers.
Keywords Variational calculus · Slope stability · Seismic coefficient · Cohesive

soil
1 Introduction
Slope stability analysis is one of the most exciting and challenging problems in
the geotechnical engineering field. Since the past eight decades, several researchers
[4, 9, 11, 17, 28, 34–36] had studied this classical problem thoroughly. Most of the
previous studies related to slope stability were accomplished with the help of the limit
equilibrium method (LEM). However, in LEM, few assumptions are being considered
to assess the stability of slopes. They are as follows: the nature of the interslice forces,
the shape of the slip surface, and distribution of the normal stress along the slip
surface. Moreover, strain and displacement compatibility are not being considered
S. Sarkar · M. Chakraborty (B)

Indian Institute of Technology (BHU), Varanasi 221005, India
e-mail: manashchakra.civ@itbhu.ac.in
230 S. Sarkar and M. Chakraborty
in LEM, and hence, the method suffers severe limitations as discussed by Duncan
[10] and Krahn [22]. Several other researchers [3, 12, 18, 23, 27, 37, 38] had applied
various other analytical and numerical methods for analyzing the homogenous slopes.
Zaki [38] and Lane and Griffiths [23] applied the strength reduction method (SRM)
in the framework of the conventional displacement based finite-element method for
analyzing the homogenous slope. Yu et al. [37] and Kim et al. [18] used finite-element
limit analysis to analyze the homogenous slopes with and without considering the
effect of pore water pressure. Michalowski [27] provided stability charts for uniform
slopes subjected to pore water pressure and horizontal seismic force by using the
kinematic approach of limit analysis. Baker et al. [3] proposed stability charts for
homogeneous slopes by applying the variational method and the strength reduction
technique considering pseudostatic analysis.
The stability of non-homogeneous slopes was also studied rigorously especially
in the past few years [5, 6, 8, 13–16, 20, 21, 24–26, 29]. Gibson and Morgenstern
[13] analyzed non-homogeneous slopes considering linearly increasing strength with
cohesion at the ground surface (cu0 ) as zero by using φ = 0 limit equilibrium anal-
ysis for the static case. By following his work, Hunter and Schuster [16] extend the
problem for cu0 > 0 and provide solutions. Booker and Davis [5] and Chen et al.
[8] used the kinematical approach of limit analysis (LA) to obtain the upper-bound
solution for non-homogeneous slope problems. Koppula [20] used the limit equilib-
rium method to analyze the slope consisting of linearly increasing strength for the
static case. Koppula [21] employed the equivalent static force (ESF) method for the
seismic case. Leshchinsky and Smith [24], Low [26] evaluated the factor of safety of
embankments constructed on the soft clay by adopting LEM. Chai et al. [6] applied
the finite-element method (FEM) to assess the failure of an embankment on soft
clayey soil-based modified cam clay model. Griffiths and Yu [14] employed Matlab
optimization programs to evaluate the critical slip surface and corresponding critical
factor of safety of the slope with linearly increasing undrained cohesion with depth
and firm stratum at the bottom of the slope. Li et al. [25] applied the variational
method to assess the stability of the slope consisting of undrained clay with linearly
increasing strength for both static and seismic cases.
After a thorough literature study, it is noticed that very few research works on non-
homogeneous soil slopes under seismic forces by using the variational approach had
been carried out earlier. The use of the variational approach is more advantageous
than the limit equilibrium method as no prior assumption regarding the critical failure
surface and the normal stress distributions along the failure surface are needed. The
variational method was first introduced by Kopacsy [19] in solving the slope stability
problem. Later on, some researchers [1–3, 7, 30–33] carried on a further study on
this method and applied it to solve various problems on slope stability.
In this present article, the variational method is implemented to evaluate the critical
slip surface and corresponding critical factor of safety (F s ) for any non-homogeneous
soil slope subjected to seismic loading. The formulation recommended by Revilla
and Castillo [31] for analyzing the homogenous cohesive soil slopes is extended
by making generalized for any purely undrained cohesive soil (φ u = 0) slopes with
linearly varying cohesion under seismic conditions. In the present analysis, the effect
Seismic Stability of Non-homogenous Cohesive Soil … 231
of pore water pressure is not considered (the provision of pore water pressure is
kept for future scope). The critical factor of safety (F s ) is obtained for a different
combination of slope angle (β) and cohesion coefficient (λ) of soil corresponding to
varying horizontal (k h ) and vertical (k v ) seismic coefficients. The obtained critical
slip surfaces are also presented for a few cases.
2 Problem Statement and Mathematical Background
Figure 1 shows a problem containing a rectilinear slope of angle β with non-

homogeneous undrained cohesive soil (φ u = 0) subjected to horizontal and vertical
seismic forces within a chosen domain where the cohesion of the soil (cu ) is linearly
increasing with depth. The objective is to determine the critical factor of safety (F s )
and the critical slip surface for the given slope. The domain size is kept adequately
high so that the slip surface is well accommodated. However, it is to be noted that
domain size does not influence the computed solution.
The definition of factor safety is the ratio of the available resisting moment to the
actual driving moment. Both these parameters are the function of the slip surface
and the slope surface; consequently, they can be represented as functional. The opti-
mization of these functionals is essential for obtaining the critical slip surface of a
slope corresponding to the critical factor of safety. The Euler–Lagrangian equation
is being applied to optimize the safety functional, as recommended in the theory of
calculus of variation. Pseudostatic analysis is being implemented to incorporate the
seismic forces.
Fig. 1 Schematic diagram of a slope along with slip surface and considered slices
3 Analytical Formulation of the Problem
The linear relationship between undrained shear strength of purely cohesive soil with
the depth can be represented as follows:
cu = cu0 + kz (1)
where cu is the undrained shear strength of soil at a certain depth z below the ground
surface, cu0 is the undrained shear strength of soil at the ground surface, and k is the
gradient at which undrained shear strength of soil increases with depth z.
In this problem, the variational method is applied in the framework of the ordinary
slice method [11] subjected to seismic forces in both horizontal and vertical direc-
tions. For any slice, i, as shown in Fig. 1, the equation of static equilibrium along
both horizontal and vertical directions are satisfied. Hence, the factor of safety can
be formulated as
n
i=1 [cu li ]
F = n (2)
i=1 (Wi sinθi + Fh cos θi − Fv sin θi )
where (i) cu is the shear strength of soil at a specific depth z.

(ii) W i , θ i , x i , l i (=x i /cosθ i ) are the weight, inclination angle, width, and
length of the ith slice, respectively; n indicates the total number of slices.
(iii) F h (=k h W i ) and F v (=k v W i ) are the horizontal and vertical seismic forces,
respectively.
Plugging the expressions of F h , F v , and l i in Eq. (2), the factor of safety can be
further expressed as
n
cu xi sec2 θi
F = n i=1
(3)
i=1 Wi [(1 − kv ) tan θi + k h ]
This formulation is being expressed as the continuous slope surface by following

the study of Revilla and Castillo [31] and Baker and Garber [2]
x3 x3
cu 1 + yi 2 d x P(x, y, y)d x
x0 x0
F= = (4)
x4 x3
γ ( f i − yi )[(1 − kv )yi + kh ]d x R(x, y, y)d x
x0 x0
where (i) yi = tan θi , (ii) Wi = γ ( f i − yi )xi ; γ is the unit weight of soil, and f i
and yi represent the slope and the slip surface, and (iii) x 0 and x 3 are the abscissas of
two endpoints where the slip surface intersect with the slope surface.
The rectilinear slope as shown in Fig. 1 can be expressed mathematically with the
following three functions within the range of x 0 and x 3 :
⎫
f 1 (x) = 0 (x0 ≤ x ≤ 0) ⎪
⎬
f 2 (x) = hh1 x (0 ≤ x ≤ h 1 ) (5)
⎪
⎭
f 3 (x) = h (h 1 ≤ x ≤ x3 )
Corresponding to each f i (x) there would be a distinct yi (x) in the interval of (x i-1 ,
x i ). The expression of the factor of safety in Eq. (4) takes the following form:
x1 x2 x3

P1 (x, y, y)d x + P2 (x, y, y)d x + P3 (x, y, y)d x
x0 x1 x2
F= (6)
x1 x2 x3
R1 (x, y, y)d x + R2 (x, y, y)d x + R3 (x, y, y)d x
x0 x1 x2
To obtain critical slip surface and corresponding critical factor of safety, F s the
above functionals are required to be minimized. This is being executed by using the
Euler–Lagrangian equation as follows:
∂P ∂P
∂ yi
− d
dx ∂i 2cu yi
Fs = = ; (i = 1, 2, 3) (7)
∂R
− d ∂R γ [(1 − kv ) f i + kh ]
∂ yi dx ∂ yi
From Eq. 7, the slip surface in the three different ranges, as depicted in Eq. 5,
can be obtained. Figure 2 representing all three surfaces is constructed with the
assumption that the factor of safety remains to be the same along the slip surface.
The three surfaces are
kh
y1 = x 2 + B1 x + D1 (8a)
4(1 + λ)N h
Fig. 2 Slip surface by x2 f3=h x3

variational method
f2=hx/h1 h
y3
f1=0
x0 x1 h1
y1
y2

(1 − kv ) hh1 + kh
y2 = x 2 + B2 x + D2 (8b)
4(1 + λ)N h
kh
y3 = x 2 + B3 x + D3 (8c)
4(1 + λ)N h
Here, N (=cu0 /(F s γ h)) and λ(=kh/cu0 ) indicate the stability number and the cohe-
sion coefficient, respectively. The equation of each surface contains two integration
constants. Therefore, the total number of unknown parameters becomes nine—(i) x 0
and x 3 , (ii) F s, and (iii) six integration constants (B1 , B2 , B3, D1 , D2 , and D3 ). These
unknown parameters can be determined by following nine equations.
(a) Two transversality conditions: Specified at two endpoints (x 0 and x 3 )
[Py − Fs R y ][ f (x) − y(x)] + P − Fs R|x=xi = 0; where (i = 0, 3) (9)
(b) Two continuity conditions: Specified at the intermediate points x 1 and x 2 (i.e.,
the intersections of the slip surfaces)
yi (xi ) = yi+1 (xi ) (i = 1, 2) (10)
(c) Two natural boundary conditions: Specified at two intermediate points x 1 and
x2
yi (xi ) = yi+1 (xi ) (i = 1, 2) (11)
(d) Two boundary conditions: Specified at two end-points (x 0 and x 3 )
y1 (x0 ) = f 1 (x0 ); y3 (x3 ) = f 3 (x3 ) (12)
(e) The equation for Factor of Safety, as provided in Eq. (6).

By using these nine equations, nine unknown parameters can be evaluated. The
expressions for the unknown parameters are as follows:
kh x0
B1 = B2 = −1 −
2(1 + λ)N h
kh x3
B3 = 1 −
2(1 + λ)N h
kh x02
D1 = D2 = x 0 +
4(1 + λ)N h
(1 − kv )h 1 kh h 1 (x3 − x0 ) kh x02
D3 = − 2h 1 + x0 + +
4(1 + λ)N 2(1 + λ)N h 4(1 + λ)N h
h(1 − kv ) + kh (x3 − x 0 )
N=
4(1 + λ)h
According to the literature [14, 25] non-homogeneous soil slope under seismic
condition will fail by developing toe failure surface (x 0 = 0). Hence in the present
analysis, the value of x 0 is considered equal to zero.
(1 − kv )(h 1 + h)
x3 =
[(1 − kv ) − kh ]
In the present study, some typical non-homogeneous slopes are analyzed by using
the formulation derived in the previous segment. The solutions are presented in the
form of the critical factor of safety (F s ). The height of the slope (h) and the unit
weight of soil (γ ) are to be maintained identically for all cases. The critical factor
of safety is evaluated for different combinations of λ and slope inclination angle (β)
corresponding to varying k h and k v . The value of β has been varied within the range
of 20°–90° at an interval of 5°. Three different values of λ, namely, 0, 1.5, and, 3.0
are chosen for each specific value of β. The magnitude of k h is considered as 0.0,
0.1, 0.2, and 0.3. For these values of k h , corresponding magnitudes of k v are 0, 0.5k h ,
and k h .
Table 1 depicts the typical result of F s of non-homogeneous soil slopes for a
different combination of λ, β subjected to the different magnitude of k h and k v . The
Table 1 F s for different slopes corresponding to different seismic coefficients

kh kv λ
β = 45° β = 55° β = 65° β = 75°
0.0 1.5 3.0 0.0 1.5 3.0 0.0 1.5 3.0 0.0 1.5 3.0
0 0 0.50 1.25 2.0 0.50 1.25 2.0 0.50 1.25 2.0 0.50 1.25 2.0
0.00 0.48 1.20 1.75 0.47 1.17 1.88 0.46 1.14 1.83 0.44 1.11 1.78
0.10 0.05 0.46 1.14 1.83 0.45 1.12 1.80 0.44 1.09 1.75 0.43 1.06 1.70
0.10 0.44 1.09 1.92 0.43 1.07 1.72 0.42 1.05 1.68 0.41 1.02 1.64
0.00 0.43 1.10 1.76 0.42 1.05 1.68 0.40 1.00 1.60 0.37 0.94 1.50
0.20 0.10 0.41 1.02 1.63 0.39 0.98 1.57 0.37 0.93 1.49 0.35 0.88 1.41
0.20 0.38 0.95 1.52 0.37 0.91 1.46 0.35 0.88 1.40 0.33 0.83 1.33
0.00 0.37 0.91 1.46 0.34 0.85 1.36 0.31 0.78 1.26 0.29 0.71 1.14
0.30 0.15 0.35 0.87 1.39 0.33 0.82 1.31 0.30 0.76 1.22 0.28 0.70 1.12
0.30 0.32 0.81 1.29 0.31 0.77 1.23 0.29 0.72 1.16 0.27 0.67 1.08
Fig. 3 Variation of critical factor of safety (F s ) with k h for a β = 20π; b β = 30π; c β = 40π; d β
= 50π
results are presented for four different values of β, namely, 45°, 55°, 65°, and 75°.
Figure 3 shows the variation of F s with k h for different values of k v /k h . The graphs
represent the variation of F s values for different combination λ values corresponding
to four distinct values of β, namely, 20°, 30°, 40°, and 50°. It is clearly noticed from
the obtained solutions that the magnitude F s decreases with the increase in k h and
k v values. For example, for β = 20° and λ= 3, when the magnitude of k h increases
from 0 to 0.1, F s reduce by 10.50%, and for the same slope, the reduction of F s is
66.67% when the magnitude of k h increases from 0 to 0.3. It is also observed that F s
reduce with (i) increase in β and (ii) decrease in λ. For k h = 0.3, k v = 0.5k h and β =
20° F s improves by 75.21% with the increase in λ from 0 to 3. The stability of the
slope further improves by reducing β. F s increases by 32.14% with the decrease of
β from 50° to 20° for λ = 3 and k h = k v = 0.3.
Figure 4 illustrates the shape of the critical slip surface for β= 20°, 40°, 60°,
and 80° with a different combination of k h . The magnitude of λ and k v are kept 3
and 0, respectively. The shape of the critical slip surface changes markedly with the
Fig. 4 Form of the critical slip surfaces for a β = 20°, λ = 3, k v = 0; b β = 40°, λ = 3, k v = 0;

c β = 60°, λ = 3, k v = 0 d β= 80°, λ = 3, k v = 0
geometric profile of the slope. The figure also gives the impression that the volume
of soil within the critical slip surface increases with the increment of the magnitude
of seismic forces.
5 Comparison of Results
Table 2 shows the comparison of the present solution evaluated from the variational
method with the available solution provided by Koppula [20] and Li et al. [25] for the
static case. The comparison is carried out for different values of cohesion coefficient
(λ) corresponding to various β for both static and pseudostatic case. The present
solution is compared with the available solution for a particular slope consisting of
non-homogeneous purely cohesive soil with slope height, h = 10 m, γ = 16 kN/m3 ,
and cu0 = 20 kPa. Table 3 depicts the comparison between the present solution and
the result presented by Li et al. [25] for the pseudostatic case. The present analytical
solutions are quite agreeable with the available solutions (Table 3).
Table 2 A comparison of F s values obtained from the present study with the solution of Koppula
[20] and Li et al. [25] considering k h = k v = 0 for β = 45°
λ Present study Koppula [20] Li et al. [25]
0.5 0.75 1.00 1.00
1 1.00 1.27 1.27
2 1.50 1.79 1.79
3 2.00 2.31 2.31
4 2.50 2.83 2.83
5 3.00 3.34 3.34
10 5.50 5.92 5.92
Table 3 A comparison of F s values obtained from the present study with the solution of Li et al.
[25] considering k v = 0
β = 60° β = 75°
kh λ Present study Li et al. [25] Present study Li et al. [25]
0.1 0 0.44 0.57 0.43 0.50
0.5 0.66 0.74 0.64 0.66
1 0.88 0.96 0.85 0.78
1.5 1.10 1.14 1.06 0.89
2 1.31 1.25 1.28 1.04
2.5 1.53 1.39 1.49 1.25
3 1.75 1.79 1.70 1.39
0 0.38 0.54 0.36 0.45
0.2 0.5 0.57 0.66 0.54 0.57
1 0.76 0.83 0.72 0.69
1.5 0.95 0.96 0.90 0.83
2 1.14 1.14 1.08 0.96
2.5 1.33 1.25 1.26 1.04
3 1.52 1.39 1.43 1.25
6 Conclusions
In the present article, pseudostatic analysis of non-homogeneous soil slope is

executed by using the variational approach without considering any prior assump-
tion on the shape of the slip surface. Design charts are being presented in the form
of the critical factor of safety for different slope geometry (β), soil properties (λ),
and seismic forces (k h and k v ). The factor of safety reduces as the magnitude of k h
and k v increases, and soil strength decreases. It is also noted that the critical factor
of safety decreases significantly with an increment of slope angle for pseudostatic
cases. Critical slip surfaces are presented here for a few cases. The present solu-
tion has also been well compared with the previous one that instills a further belief
in applying the calculus of variation to solve other stability problems. The method
can be comfortably used in solving multilayer slope stability problems provided
analytical formulation to be changed only.
Acknowledgements The corresponding author acknowledges the support of the “Depart-

ment of Science and Technology (DST), Government of India” under grant number
DST/INSPIRE/04/2016/001692.
References
1. Baker R (2003) Sufficient conditions for existence of physically significant solutions in limiting
equilibrium slope stability analysis. Int J Solids Struct 40:3717–3735
2. Baker R, Garber M (1978) Theoretical analysis of the stability of slopes. Geotechnique
28(4):395–411
3. Baker R, Shukha R, Operstein V, Frydman S (2006) Stability charts for pseudostatic slope
stability analysis. Soil Dyn Earthq Eng 26(9):813–823
4. Bishop AW (1955) The use of the slip circle in the stability analysis of slopes. Geotechnique
5(1):7–17
5. Booker JR, Davis EH (1972) A note on a plasticity solution to the stability of slopes in
inhomogenous clays. Géotechnique 22(3):509–513
6. Chai J, Carter JP (2009) Simulation of the progressive failure of an embankment on soft soil.
Comput Geotech 36(6):1024–1038
7. Chen J, Yang Z, Hu R, Zhang H (2016) Study on the seismic active earth pressure by variational
limit equilibrium method. Hindawi Publishing Corporation Shock and Vibration, pp 1–11
8. Chen WF, Snitbhan N, Fang HY (1975) Stability of slopes in anisotropic, non-homogeneous
soils. Can Geotech J 12(1):146–152
9. Chen ZY, Morgenstern NR (1983) Extensions to the generalized method of slices for stability
analysis. Can Geotech J 20(1):104–119
10. Duncan JM (1996) State of the art: limit equilibrium and finite-element analysis of slopes. J
Geotech Eng 122(7):577–596
11. Fellenius W (1936) Calculation of stability of earth dam. In: Transactions 2nd congress large
dams, Washington, DC, vol 4, pp 445–462
12. Gens A, Hutchinson JN, Cavounidis S (1988) Three-dimensional analysis of slides in cohesive
soils. Geotechnique 38(1):1–23
13. Gibson RE, Morgenstern N (1962) A note on the stability of cuttings in normally consolidated
clays. Geotechnique 12(3):212–216
14. Griffiths DV, Yu X (2015) Another look at the stability of slopes with linearly increasing
undrained strength. Géotechnique 65(10):824–830
15. Hossley A, Leshchinsky B (2019) Stability and failure geometry of slopes with spatially varying
undrained shear strength. J Geotech Geoenviron Eng 145(5):06019002
16. Hunter JH, Schuster RL (1968) Stability of simple cuttings in normally consolidated clays.
17. Janbu N (1954) Application of composite slip surface for stability analysis. Proc Eur Conf Stab
Earth Slopes, Sweden 3:43–49
18. Kim J, Salgado R, Yu HS (1999) Limit analysis of soil slopes subjected to pore-water pressures.
J Geotech Geoenviron Eng 125(1):49–58
19. Kopacsy J (1961) Distribution des contraintes a la rupture forme de la surface de glissement
et hauteur theorique des talus. In: Proceedings of the 5th international conference on soil
mechanics and foundation engineering paris, France, pp 641–650
20. Koppula SD (1984) On stability of slopes in clays with linearly increasing strength. Can Geotech
J 21(3):577–581
21. Koppula SD (1984) Pseudo-static analysis of clay slopes subjected to earthquakes. Geotech-
nique 34(1):71–79
22. Krahn J (2003) The 2001 RM hardy lecture: the limits of limit equilibrium analyses. Can
Geotech J 40(3):643–660 (2003)
23. Lane PA, Griffiths DV (2000) Assessment of stability of slopes under drawdown conditions. J
Geotech Geoenviron Eng 126(5):443–450
24. Leshchinsky D, Smith DS (1989) Deep-seated failure of a granular embankment over clay.
Stability analysis. Soils Found 29(3):105–114
25. Li B, Zhang F, Wang D (2018) Impact of crack on stability of slope with linearly increasing
undrained strength. Math Prob Eng 1–11. https://doi.org/10.1155/2018/1096513
26. Low BK (1989) Stability analysis of embankments on soft ground. J Geotech Eng 115(2):211–
227 (1989)
27. Michalowski RL (2002) Stability charts for uniform slopes. J Geotech Geoenviron Eng
128(4):351–355
28. Morgenstern NU, Price VE (1965) The analysis of the stability of general slip surfaces.
29. Nakase A (1970) Stability of low embankment on cohesive soil stratum. Soils Found 10(4):39–
64
30. Narayan CGP (1975) Variational methods in stability analysis of slopes. PhD dissertation
submitted to Indian Institute of Technology, Delhi, India
31. Revilla J, Castillo E (1977) The calculus of variations applied to stability of slopes.
32. Sarkar S, Chakraborty M (2019) Pseudostatic slope stability analysis in two-layered soil
by using variational method. In: Earthquake geotechnical engineering for protection and
development of environment and constructions (7th ICEGE, Rome), pp 4857–4864
33. Sarkar S, Chakraborty M (2020) Pseudo-static slope stability analysis for cohesive-frictional
soil by using variational method. In: Advances in computer methods and geomechanics.
Springer, Singapore , pp 159–171
34. Sarma SK (1973) Stability analysis of embankments and slopes Geotechnique 23(3):423–433
35. Spencer E (1967) A method of analysis of the stability of embankments assuming parallel
inter-slice forces. Geotechnique 17(1):11–26
36. Taylor DW (1937) Stability of earth slopes. J Boston Soc. Civ Eng 24(3):197–247
37. Yu HS, Salgado R, Sloan SW, Kim JM (1998) Limit analysis versus limit equilibrium for slope
stability. J Geotech Geoenviron Eng 124(1):1–11
38. Zaki A (1999) Slope stability analysis overview. University of Toronto (1999)
Site Characterization Using Satellite
Data and Estimation of Seismic Hazard
at Ground Surface
Naveen James , Sreevalsa Kolathayar , and T. G. Sitharam
Abstract This paper presents the seismic site characterization carried out for
Karnataka (state level) as well as for India (country level) using topographic slope
map derived from Digital Elevation Model (DEM) data. Two DEM data, SRTM, and
ASTER were used to derive the slope maps. For Karnataka (state level), the slope
map was generated from ASTER DEM considering a grid size of 5 × 5 km and for
India (country level), the slope map was generated from SRTM DEM considering
the grid size of 10 × 10 km. Based on the slope value, every grid point was char-
acterized into various NEHRP site classes, and spatial variation of average shear
wave velocity for top 30 m (Vs30) value throughout the study areas is presented
in this paper. Peak horizontal acceleration (PHA) at bedrock level was evaluated
for the same grid points using deterministic as well as probabilistic methodologies.
The amplification factor for every grid point was obtained from the site coefficients
corresponding to NEHRP site class. The surface level peak horizontal acceleration
(PHA) was then evaluated for every grid point by multiplying bedrock level PHA
with the corresponding amplification factor. Spatial variation of seismic hazard at
the surface for the state of Karnataka as well as for entire India is presented in this
paper.
Keywords Hazard · PSHA · DSHA · DEM · PHA
N. James (B)
Department of Civil Engineering, Indian Institute of Technology Ropar, Punjab, India
e-mail: naveen.james@iitrpr.ac.in
S. Kolathayar
Department of Civil Engineering, National Institute of Technology Karnataka, Surathkal, India
e-mail: sreevalsa@nitk.edu.in
T. G. Sitharam
Director, Indian Institute of Technology Guwahati, Guwahati, Assam, India
e-mail: sitharam@iisc.ac.in
242 N. James et al.
1 Introduction
It is a well-known fact that the characteristics of seismic waves will undergo severe
alteration due to impedance contrast while traveling from bedrock to ground surface.
Hence, the role seismic site characterization is important in evaluating the intensity
of ground shaking at a site due to any future earthquake. Usually, geotechnical or
geophysical tests are carried out to characterize the site. For the seismic site character-
ization of vast regions, topographic gradient data can be very useful. Studies already
proved that topographic slope data derived from digital elevation model (DEM) is
an index for the seismic site conditions. Several studies have proved that apart from
source parameters, the characteristics of surface-level earthquake motion are mainly
influenced by engineering properties of overburden soil such as thickness, density,
sub-surface geometry. Seismic site characterization aims to categorize the overburden
soil strata based on either local geology [1–5] or geotechnical/geophysical test data.
The most important parameter needed for the site characterization is the average
shear wave velocity for top 30 m depth estimated using geophysical tests such as
SASW test, MASW test, suspension logging test, cross-hole test, seismic cone pene-
tration test. The top 30 m is the depth corresponding to the one-quarter wavelength
of the period of interest as suggested by Joyner et al. [6]. Many studies have been
carried out to correlated geotechnical test data, such as SPT and CPT values with the
shear wave velocity [7–9]. Most of the building codes such as [10–12] recommend
site characterization based on V s30 where
n
di
Vs30 = 30/ (1)
V
i=1 i
where di is the depth of each soil layer,Vi is the corresponding shear wave velocity
for each layer and n is the number of soil layers. Soil strata with Vs30 > 1500 m/s
are categorized as site class A. Soil strata with Vs30 ranging from 760 to 1500 m/s,
360–760 m/s and 180–360 m/s are categorized as site class B, C, and D respectively.
The strata where the Vs30 < 180 m/s are categorized as site class E.
Wald and Allen [13] presented a technique to derive first-order site-condition
maps from topographic gradient using global 30 arc, sec topographic data and
VS30 measurements. They found that maps derived from the topographic data were
correlating well with other independently derived, regional site-condition maps.
Allen and Wald [14] which have examined the use of high-resolution DEMs can
be used to recover improved estimates of VS30 beyond that which can be delivered
using the original correlations of [13]. James and Sitharam [15] have compared the
Vs30 map of Bangalore and Chennai cities, derived from 1 arc-second DEM data
with the Vs30 map developed by Sitharam and Anbazhagan [16] and [17] using
geotechnical and geophysical tests data. Studies by [15] have found that for both
Bangalore and Chennai, the Vs30 maps derived from topographic slope data are
comparable with the Vs30 maps developed based on geotechnical and geophysical
tests data. Yong et al. [18] have used the geomorphometry for the site characterization
Site Characterization Using Satellite Data … 243
using the time averaged shear wave velocity to a depth of 30 m. Grelle et al. (2018)
have generated a 3D-topographic surface of East Mountain area (Utah, USA), using
the digital elevation model (DEM) for computing Topographic Aggravation Factors
(TAF). This paper presents the case studies on site characterization of Karnataka
(state level) as well as India (country level) using topographic slope map derived
from DEM data, based on the VS30 values. The slope maps of the two regions were
derived separately from two different types of DEM (ASTER and SRTM) having
different resolutions.
1.1 Digital Elevation Model (DEM)
DEMs are widely used in a large variety of engineering disciplines such as in

Hydrology, landslide hazard assessment, and transportation. Two types of DEM
are freely available currently, Shuttle Radar Topography Mission (SRTM) and the
Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER).
Shuttle Radar Topography Mission (SRTM). The Shuttle Radar Topography
Mission (SRTM) is a joint venture between the National Geospatial-Intelligence
Agency (NGA) and the National Aeronautics and Space Administration (NASA),
producing digital topographic data land areas between 60° north and 56° south lati-
tude, with 3-arc second (approximately 90 m) resolution. The data were downloaded
from various sources such as Cartosat DEM from NRSC Bhuvan, SRTM from USGS
EROS, interpolated seamless SRTM data v4.1 from CGIAR CSI, and void filled
data in mountainous terrain from viewfinder panoramas. The absolute vertical accu-
racy of the elevation data will be 16 m (at 90% confidence). The Advanced Space-
borne Thermal Emission and Reflection Radiometer (ASTER) Global Digital Eleva-
tion Model (ASTER GDEM) is a freely available DEM, developed by Ministry of
Economy, Trade, and Industry (METI) of Japan in collaboration with the National
Aeronautics and Space Administration (NASA) in the United States. The ASTER
GDEM covers land surfaces between 83° N and 83° S with a resolution of 1 arc-
second (30 m). The estimated vertical accuracy for ASTER GDEM was 20 m at 95%
confidence. Since ASTER DEM (Fig. 1a) has higher resolution, it is used to generate
the slope map for the state of Karnataka and similarly lower resolution SRTM DEM
(Fig. 1b) is used to generate the slope map for entire India. These ASTER DEM and
SRTM DEM maps can be freely downloadable from websites.
1.2 Generation of Slope Map
The slope maps of the two study areas were developed in ArcGIS software. The
DEMs were first projected to the UTM coordinate system and then reassembled to
5 × 5 km for Karnataka and 10 × 10 km for the entire India. Further, using the spatial
analyst tool, the slope map for the two study areas was derived from the respective
244 N. James et al.
a) b)
Fig. 1 Digital elevation model a Karnataka (ASTER) b India (SRTM)
DEM and presented in Fig. 2. Slope maps thus obtained were converted to vector
data (points) and then exported in a table form. The geographic co-ordinates and
slope values were computed for each grid point considered.
a) b)
Fig. 2 Slope map of Karnataka and India (gradient values m/m)

a) b)
Fig. 3 Spatial distribution of various site classes in Karnataka and India
2 Site Characterization Based on Topographic Slopes
Once slope maps are obtained, site characterization of the two study areas was carried
out using methodology suggested by Wald and Allen (2007). Based on the correlation
studies conducted for active tectonic and stable continental regions, Wald and Allen
(2007) have proposed slope ranges corresponding to each NEHRP site class. Based
on the slope range given by Wald and Allen (2007), site class for each grid point was
determined. Maps showing the spatial variation in site classes throughout Karnataka
and India were developed and presented in Fig. 3. The site classes obtained from
various locations in the study areas were then compared with the existing site char-
acterization map for the same location generated for the microzonation purpose by
various researchers. Comparing with the studies of [19–22], related to microzonation
of Bengaluru, Chennai, New Delhi, and Imphal, the present work identifies similar
site class for the above study areas.
3 Evaluation of Seismic Hazard
The conventional seismic hazard analysis provides the peak horizontal acceleration
(PHA) at bedrock level. As it is well known that there are lots of variations observed
in depth of bedrock, even in a small area of landmass and in many cases, most of the
foundations rest at shallow depth and do not extend up to bedrock.
Hence, applying the PHA value at bedrock level obtained from seismic hazard
analysis for the earthquake resistant design of shallow founded structures is not the
right procedure. So, the PHA at the ground surface has to be evaluated after the
characterization of overburden soil mass. As a first step, the PHA at bedrock level
246 N. James et al.
for Karnataka as well as entire India for all the grid points where the slope value was
evaluated. Both deterministic and probabilistic methodologies were used to evaluate
seismic hazard at bedrock.
Deterministic Hazard Analysis. The deterministic seismic hazard approach
(DSHA) considers only the critical scenario by assuming the occurrence of the
largest magnitude earthquake at the closest proximity of site, thereby giving single-
valued ground motion at that site. The deterministic seismic hazard analysis analogies
presented by [23] and [24] were used in the present study for Karnataka and India,
respectively. In the present study, DSHA was carried out at all grid points (whose
slope values are available) in the two study areas using linear source model.
Probabilistic Hazard Analysis. Unlike DSHA, the probabilistic seismic hazard
analysis (PSHA) considers various uncertainties on magnitude and distance which
are tallied judiciously to get a comprehensive picture of seismic hazard [25]. Similar
to DSHA, all the known earthquake events and sources within 300 km from the
boundary of study areas (Karnataka as well as India) were compiled and seis-
micity parameters were evaluated [26]. The PSHA analogy for Karnataka and India
presented by [23] and [27], respectively, were used in the present study. Figure 4
describes a general framework for DSHA and PSHA.
Evaluation of PHA at Ground Surface. Non-linear site amplification technique
proposed by [27] was adopted in the present study for determining surface-level
seismic hazard for the two study areas. The PHA at the ground surface Ys is obtained
by multiplying suitable amplification factor Fs to the bedrock level PHA Ybr as per
Eq. 2.
Ys = Ybr × Fs (2)
Fig. 4 General methodology for the hazard analysis for DSHA and PSHA
a) b)
Fig. 5 Spatial variation in PHA value at surface level throughout Karnataka and India (DSHA)
The value of the amplification factor Fs varies with PHA values, and it accounts for
the non-linear behavior of soil in amplifying the seismic signals. [28] have developed
a regression relationship as per Eq. 3 for determining amplification factor Fs .
ln(Fs ) = a1 Ybr + a2 + ln δs (3)
where a1 and a2 are regression coefficients, Ybr is the spectral acceleration at rock
level, and δs is the error term. The values of the regression coefficients a1 and a2
will vary for different site classes and for different time periods. These values were
derived based on the statistical simulation of ground motions [28], and they take
into account the non-linear site response of soils. For site classes C and D, the
amplification factor was found non-linear dependency on bedrock level PHA when
compared with site classes A and B. The method adopted for evaluation of the
amplification factor (Fs ) values consider this effect and the value of Fs varies with
the rock level PHA values. Figures 5 and 6 present the spatial variation in PHA value
at bedrock throughout Karnataka as well as India from deterministic and probabilistic
methodologies, respectively.
4 Conclusions
The seismic hazard at the ground surface is a critical parameter to perform earthquake
resistant design of buildings. It is also required to carry out liquefaction analysis for
given soil strata. In this paper, the surface level PHA values were evaluated for
Karnataka (state level) and India (country level) based on NEHRP site classes by
248 N. James et al.
a) b)
Fig. 6 Spatial variation in PHA value at surface level throughout Karnataka and India (corre-
sponding to 475 years, PSHA)
considering the local site effects. Site characterization study shows that the interior
regions of the Karnataka fall in the category of site class D. However, the Western
Ghat region of the Karnataka state was identified as site class C to the B-type terrain.
Similarly, for India, the site characterization study shows that the majority of central
India falls in site class D. The northern and north-eastern regions of India fall in site
class C. In the absence of Vs30 values, the site classes can be identified based on
local geological conditions. In this paper, the spatial variation of Vs30 throughout
Karnataka and India was evaluated from the topographic slope maps. The PHA value
was brought to the surface level using appropriate amplification factors. Thus, this
method provides a simplified methodology for evaluating the surface-level PHA
values. The PHA values obtained based on the DSHA analysis can be taken as the
upper bound PHA values.
Thus, this method provides a simplified methodology for evaluating the surface
level PHA values. The PHA values obtained based on the DSHA analysis can be
taken as the upper bound PHA values. For the state of Karnataka, the very high
site amplification (PGA ≤ 0.3 g) value is observed for regions close to Bidar in
north Karnataka, Mangalore–Udupi in the southwest and Bangalore–Mysore in the
southeast. For India, very high site amplification (PGA ≤ 0.3 g) is observed for the
regions in the north, the northeast, Kutch in the west, and Indo-Gangetic belt in the
central to north-central India. The seismic site characterization based on topographic
slope maps is ideal for large landmasses like a state or a country where geotech-
nical/geophysical test-based site characterization is not physically and economically
viable. It is recommended to use the topographic slope data to perform the first
level of seismic zonation for identifying various classes. Effect of topography, basin
geometry, and soil non-linearity on earthquake motion at bedrock were ignored in
the present study. These macro-level surface-level seismic hazard maps are excel-
lent tools for engineers and designers to identify potential hazardous zones while
planning to set up various infrastructures.
References
1. Medvedev J (1962) Engineering seismology. Academia Nauk Press Moscow, p 260

2. Borcherdt RD, Gibbs JF (1976) Effects of local geological conditions in the San Francisco Bay
region on ground motions and the intensities of the 1906 earthquake. Bull Seismol Soc Am
66(2):467–500
3. Shima E (1978) Seismic microzoning map of Tokyo. Proc Second Inter Conf Microzonation
1:433–443
4. Evernden JF, Thomson JM (1985) Predicting seismic intensities. Evaluating earthquake hazards
in the Los Angeles Region an earth-science perspective. US Geol Surv Prof Pap 1360:151–202
5. Midorikawa S (1987) Prediction of isoseismal map in the Kanto plain due to hypothetical
earthquake. J Struct Eng 33B:43–48
6. Joyner WB, Warrick RE, Fumal TE (1981) The effect of Quaternary alluvium on strong ground
motion in the Coyote Lake, California, earthquake of 1979. Bull Seismol Soc Am 71(4):1333–
1349
7. Imai T (1982) Correlation of N value with S wave velocity and shear modulus. Proc 2nd ESOPT,
pp 57–72
8. Anbazhagan P, Sitharam TG (2008) Mapping of average shear wave velocity for Bangalore
region: a case study. J Environ Eng Geophys 13(2):69–84
9. James N, Sitharam TG, Padmanabhan G, Pillai CS (2014) Seismic microzonation of a nuclear
power plant site with detailed geotechnical, geophysical and site effect studies. Nat Hazards
71(1):419–462
10. IBC (2009) International building code. International Code Council, Washington
11. Eurocode-8 (2003) BS-EN 1998-1, Design of structures for earthquake resistance part1:
General rules, seismic actions and rules for buildings. European Committee for Standardization,
Brussels
12. BSSC (2003) NEHRP recommended provisions for seismic regulations for new buildings and
other structures (FEMA 450), Part 1: Provisions. Building Seismic Safety Council for the
Federal Emergency Management Agency, Washington, D.C., USA
13. Wald DJ, Allen TI (2007) Topographic slope as a proxy for seismic site conditions and
amplification. Bull Seismol Soc Am 97(5):1379–1395
14. Allen TI, Wald DJ (2009) On the use of high-resolution topographic data as a proxy for seismic
site conditions (VS30). Bull Seismol Soc Am 99(2A):935–943
15. James N, Sitharam TG (2016) Seismic zonations at micro and macro-level for regions in the
Peninsular India. Int J Geotech Earthquake Eng (IJGEE) 7(2):35–63
16. Uma Maheswari R, Boominathan A, Dodagoudar GR (2010) Seismic site classification and
site period mapping of chennai city using geophysical and geotechnical data. J Appl Geophys
72(3):152–168
17. Yong A, Hough SE, Iwahashi J, Braverman A (2012) A terrain-based site-conditions map
of California with implications for the contiguous United States. Bull Seismol Soc Am
102(1):114–128
18. Anbazhagan P, Sitharam TG (2008) Site characterization and site response studies using shear
wave velocity. J Seismol Earthquake Eng 10(2):53–67
19. Boominathan A, Dodagoudar GR, Suganthi A, Uma Maheswari R (2008) Seismic hazard
assessment of Chennai city considering local site effects. J Earth Syst Sci 117:853–863
20. Hanumantharao C, Ramana GV (2008) Dynamic soil properties for microzonation of Delhi,
India. J Earth Syst Sci 117:719–730
250 N. James et al.
21. Pallav K, Raghukanth STG, Singh KD (2015) Estimation of seismic site coefficient and
seismic microzonation of Imphal City, India. Using the probabilistic approach. Acta Geophysica
63(5):1339–1367
22. Sitharam TG, Naveen James KS, Vipin K, Ganesha Raj (2012) A study on seismicity and
seismic hazard for Karnataka State. J Earth Syst Sci 121(2):475–490
23. Kolathayar S, Sitharam TG, Vipin KS (2012) Deterministic seismic hazard macrozonation of
24. Kramer SL (1996) Geotechnical earthquake engineering. Pearson Education Pvt. Ltd, Reprinted
2003, Delhi, India
25. Gutenberg B, Richter CF (1944) Frequency of earthquakes in California. Bull Seismol Soc Am
34(4):185–188
26. Sitharam TG, Kolathayar S, James N (2015) Probabilistic assessment of surface level seismic
hazard in India using topographic gradient as a proxy for site condition. Geosci Front 6(6):847–
859
27. Raghu Kanth STG, Iyengar RN (2007) Estimation of seismic spectral acceleration in Peninsular
GIS-Based Landslide Hazard Zonation
and Risk Studies Using MCDM
Ankit Tyagi, Reet Kamal Tiwari, and Naveen James
Abstract In India, landslides are the most frequently occurring disaster in the
regions of the Himalayas and the Western Ghats. They are mainly triggered either by
rainfall or earthquake or the combination of both, causing severe damage to human
life and infrastructure. This study presents a comprehensive use of the multi-criteria
decision-making (MCDM) method in landslide risk assessment for the Tehri area in
the state of Uttarakhand, India. The Tehri area is situated in the Lesser Himalaya of
Garhwal hills which lies in zone IV of seismic zoning map of India. Because of the
large-scale slope instability in the area, it has received the special attention of the
researchers. In the recent past,—many landslide hazards and risk zonation is carried
out for different regions in the Uttarakhand state. However, limited work is done
considering temporal factors such as seismic ground shaking, rainfall, and seismic
amplification at surface level. The DEM data is used to produce topographic char-
acteristics such as slope, aspect, and relative relief. DEM data is also used for the
detailed drainage analysis which includes topographic wetness index (TWI), stream
power index (SPI), drainage buffer, and reservoir buffer. Seismic hazard analysis is
performed using the deterministic methodology to estimate the peak horizontal accel-
eration. The amplification factor is calculated using the non-linear site amplification
method. In this study, the analytical hierarchy process (AHP) is used to evaluate the
landslide hazard index which is used to generate landslide hazard zonation (LHZ)
map. Further, the landslide vulnerability assessment is done for the study area. The
vulnerability map of the study area is derived in terms of landuse/landcover (LULC)
using remote sensing data of Landsat 8 which can provide useful information that
helps people to understand the risk of living in an area.
Keywords PHA · AHP · GIS · Landslide
A. Tyagi (B) · R. K. Tiwari · N. James

Department of Civil Engineering, Indian Institute of Technology Ropar, Rupnagar 140001, India
e-mail: annkittyagi@gmail.com
R. K. Tiwari
e-mail: reetkamal@iitrppr.ac.in
N. James
e-mail: naveen.james@iitrpr.ac.in
252 A. Tyagi et al.
1 Introduction
Landslide is defined as a displacement of slopes, forming earth material by fall,

topple, slide, or flow under the gravity [1]. Landslides are a major threat to both lives
and property and damages natural resources in the Himalayas. Naithani [2] reported
that every year more than 200 people die, and property damage is more than US$ 1
billion in the Himalayas because of landslides which is approximately 30% of world-
wide landslide losses. The Himalayas are highly susceptible to landslides due to their
varying slope, complex geology, and are triggered by heavy rainfall, earthquakes,
floods, etc. There is a need to carry out Landslide Hazard zonation (LHZ) to identify
these landslide-prone areas. Landslide Hazard Zonation is a method of classifying
different parts of an area according to their landslide potential [3]. In the last few
decades, landslide hazard and risk assessment have become a topic of great interest
for engineering professionals. It started with field-based studies of hazard zonation
with manual data integration [4] which have some disadvantages, such as small
area cover, poor integration capacity, and manual overlay. Recently remote sensing
and geographic information systems (GIS) technology have attracted great attention
with their brilliant spatial data processing capacity. With their help, it is possible
to efficiently gather, visualize, operate, and combine different types of data such as
aspect, geology, LULC, slope, etc. of a selected zone that is used to generate LHZ
[5] map. In this study, the LHZ map is generated for the Tehri Garhwal region of
Uttarakhand with the help of satellite data and GIS technology. The vulnerability
map is derived from the LULC classification of multispectral data. Using the AHP
technique, vulnerable regions and hazard zones were combined to obtain the risk
zone of the area.
2 Study Area
The Tehri area is located at latitude 30° 22 40 N and longitude 78° 28 50 E in the
Lesser Himalaya of Garhwal hills in Uttarakhand, India. The area of study covers
about 1500 km2 area which falls in a part of Bhagirathi and Bhilangna river basins.
These two rivers confluence to form a huge reservoir at Tehri Dam where landslides
are very common. A total of 195 landslides were recorded in this area from field
observations, image interpretation, and historical data (Fig. 1).
3 Data Used
For deriving the causative factors of a landslide, various types of remote sensing
data, earthquake data, and ancillary data are used as listed in Table 1.
GIS-Based Landslide Hazard Zonation and Risk Studies … 253
(A) (B)
Fig. 1 a Location and landslide inventory map, b geological Map of the study area
Table 1 Data used

Data type Sensors Resolution Derivative
Multispectral Landsat 8 30 m LULC
data Landslide
inventory
DEM Cartosat DEM 30 m Slope
Aspect
Relative relief
Curvature
TWI
SPI
Drainage
density
Ancillary data Published 1:250,000 Digitized
geology map. geology map
A report on 1:250,000 Digitized soil
Uttarakhand map
soils.
Toposheet 1:50,000 Digitized base
53 J/7 NW map
254 A. Tyagi et al.
4 Terrain Parameters
4.1 Geological Setting
The geology of the Tehri region defined here is mostly based on the work of Valdiya
[6]. The rocks here are of Rautgara formation, Deoban formation and Berinag forma-
tion of the Inner Lesser Himalayas, and the rocks Chandpur formation, Nagthat
formation and Blaini formation of outer Lesser Himalayas are found here.
4.2 Soil Type
Soil categories of the reservoir area consist of alluvial soil, sandy soil, clayey soil
[7]. Soils of different categories have different influences on landslides. Clayey soil
is comparatively more stable because of thick green vegetation cover. Recently
deposited alluvium soil and boulders are unstable as they are less compacted and
have more moisture. As compared to alluvial soil, sandy loamy soil is more stable.
4.3 Primary Topographic Attributes
Topographic map and DEM data are used to derive primary factors of topography.
Slope angle is varying from 0° to 75°. The slope angle is classified into five classes
based on its influence to cause landslide (Figs. 2, 3). It is widely accepted that area
having low slope angle is less prone, whereas areas with a higher slope angle have
more chance of landslides [8].
The aspect of a slope governs the Sun ray’s effect which is related to temperature
and other climatic conditions (Fig. 4). Aspect as a great influence on the terrain can
be seen in the Himalayan region as south direction slopes are forested and moist,
whereas north direction slopes are glaciated and dry. The Himalayan terrain slopes
that are facing south have more chances of landslides [9]. Relative relief is another
important DEM derivative, which is the difference in maximum and minimum points
of elevation in an area. Slope curvature is another important factor as water flowing
in and out of the area is controlled by it [10]. Calculation of curvature is done parallel
to the direction of the maximum slope in which, a positive profile specifies that the
terrain is concave and a negative value specifies that the terrain is convex and a value
of zero specifies that the terrain is level [11]. Acceleration of flow across the terrain is
affected by curvature. Therefore, a curvature map is created, indicating both profiles
of concave and convex.
(C) (D)
(E) (F)
Fig. 2 c Soil map, d slope map, e aspect map, and f relative relief map of the Tehri area
4.4 Secondary Topographic Attributes
For the LHZ mapping, two secondary topographic units, i.e., TWI and SPI are used.
TWI indicates the accumulation of flow of the terrain TWI [12]. The following
formula is used to calculate
CA
TWI = ln (1)
tan slp
256 A. Tyagi et al.
(G) (H)
(I) (J)
Fig. 3 g Curvature map, h TWI map, i SPI map, and j Drainage density map of the Tehri area
here CA represents the catchment area and slp represents the slope gradient of the
area. As the area increases and slope decreases, soil moisture content and TWI
increases. SPI is another important factor which corresponds to the erosive power of
the stream. The following formula is used to calculate
SPI = ln(CA × tan slp) (2)

(K) (J)
Fig. 4 k Drainage buffer map and j reservoir buffer map of the Tehri area
As the area and slope increases, erosion and SPI increases. Many landslides are
triggered during the monsoon season because of the high stream density of the Tehri
reservoir area. A high value of SPI indicates more possibility of a landslide. SPI
values have been divided into five categories by applying a natural break classifier.
4.5 Distance/Buffer Layers
The Arc hydro tool of ESRI GIS package is used to derive a drainage map from
the DEM. As the drainage density increases, the possibility of landslide occurrence
also increases. In hilly regions, unstable slopes are created because of the continuous
erosion of its banks by streams that are very prone to failures. Drainage buffer map,
drainage density map, and reservoir map are generated as drainage pattern has a great
influence on landslides.
4.6 Seismicity
Uttarakhand has been repeatedly shaken by earthquakes in the past. Uttarkashi

suffered one of India’s deadliest earthquakes in 1991 that killed nearly 730 people
and affected over three lakhs. Another major earthquake hit the Chamoli district of
Uttarakhand in 1999 where 103 people died. Hence in this study, within 500 km of
radius from the study area, events of earthquakes were taken from the USGS website
258 A. Tyagi et al.
(K) (L)
Fig. 5 k Decluttered earthquake event and l peak ground acceleration map of the Tehri area
(Fig. 5). About 950 seismic events were found in the 500 km vicinity of the study
area from 1950 to 2019 of magnitude 4 and above. Hence, it can be stated that the
Uttarakhand in the Himalayas is situated in a highly seismic zone. A deterministic
method is used in this study for seismic hazard assessment. A relationship for the
Himalayas in India for peak ground accelerations which was developed by Sharma
[13] is used to calculate PHA. Following is the general formula which shows the
relationship between ground motion and parameters like magnitude, distance, etc.

log(y) = b1 + b2 M + b3 log 2
R 2J B + b42 + b5 S + b6 H (3)
here, y is acceleration, S is kept as 1 as the study area is rocky, H varies from 0 to 1, M

is the magnitude of the earthquake, RJB is Joyner distance, and b1 , b2 , b3 , b4 , b5 , and
b6 are the regression coefficients. For the study area, a map of PHA is prepared at the
bedrock level (Fig. 5). To calculate PHA at the surface level, amplification technique
is used here [14]. The PHA at the surface is calculated for slopes having more than or
equal to 10°. The surface with a slope angle less than 10° are neglected considering
them to be flat [15]. The surface with slope angle 10° and above comes under B-type
site class [16], for which the velocity of shear wave is more than 760 m/s for top
30 m overburden (Vs30 ) [17]. The following relationship is used to calculate PHA at
the surface level.
Ys = Ybr + Fs (4)
Here, Y s is surface-level PHA, F s is the amplification factor, and Y br is PHA at

bedrock, where amplification factor is calculated using the following relationship
proposed by Raghu Kanth and Iyengar in 2007.
ln FS = a1 Ybr + a2 + ln(δs ) (5)
Here, a1 and a2 are site class regression coefficients, Y br is the spectral acceleration
at the bedrock level, and δs is the error term [18].
5 Methodology
An inventory map for the landslide is prepared with the help of satellite data, field
observation and historical information available. Many landslides were found in the
close vicinity of the reservoir. In this study, 12 causative factors are used to prepare
LHZ map. According to Yalcin [19], it is necessary to be sure that any selected factor
is operational, complete, uniform, measurable, and non-redundant in GIS Study. For
synthesizing weights of these causative factors, the AHP method is used. It is a
method to derive ratio by paired comparisons. AHP has many applications in the
selection of the sites and doing LHZ [20]. AHP is an MCDM technique invented by
Saaty [21] where both objective as well as subjective factors can be considered while
taking decisions [22]. AHP breakdowns difficult decision-making problems into a
pyramid of factors and alternatives. Weights are assigned to factors and alternatives on
a nine-point scale (Table 2) by pair-wise comparison between them. In this study, the
comparison is done between all twelve causative factors and their respective classes.
Matrix is formed with different causative factors and their respective classes [23].
For comparison, weights are provided which are subjective. In this study, weights are
given based on a field visit and past works of literature. From the matrix, Eigenvalues
and Eigenvectors are obtained which represent the influence of that factor or class
for causing a landslide.
As these weights are given subjective, hence it is important to check the consis-
tency of the weight provided. To check consistency, Saaty (1980) has defined two
Table 2 The priorities scale

Preference Degree
between two causative factors
in AHP according to Saaty 1 Equal
3 Moderate
5 Strong
7 Very strong
9 Extreme
2, 4, 6, 8 In-between
Reciprocals Reverse
260 A. Tyagi et al.
Table 3 Random consistency index prepared by Saaty

N 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
RI 0 0 0.57 0.90 1.11 1.23 1.31 1.42 1.44 1.48 1.52 1.54 1.55 1.58 1.58
terms, i.e., consistency index (CI) and consistency ratio (CR). The following formula
is used to calculate the consistency index:
λmax − N
CI = (6)
N −1
Here, N is the number of factors or classes and λmax is the maximum eigenvalue.
After calculating the CI, Random index is obtained from the table prepared by Saaty.
From the ratio of CI and RI, the consistency ratio is calculated.
Table 3 of the Random index was prepared by Satty in 1980 by doing random
sampling. CR value greater than 0.1 shows inconsistency. While CR value 0 shows
that it is perfectly consistent. The following are the matrix prepared for all the
causative factors and their respective classes (Table 4).
The next step was the calculation of LHI. It was computed by multiplying weights
with respective classes and then using the following formula:

N

LHI = weight of factor W j × weight of factor classes Wi j (7)
j=1
where W ij denotes weight of ith class of causative factor J. LHI map generated is
again divided into five classes as very low, low, moderate, high, and very high.
6 Landslide Vulnerability Map
Landslide vulnerability map in the form of LULC is prepared from multispectral

Landsat 8 data. LULC map is generated by a supervised classification technique.
The maximum likelihood classification algorithm uses the spectral signature of the
training class to classify the whole image into five classes, i.e., reservoir, settlement,
dense forest, scrub forest, and agriculture land. Among them settlement or built-up
areas are more vulnerable than other areas.
Table 4 Scores obtained by performing AHP

Factors 1 2 3 4 5 6 7 8 9 10 11 12 13 Normalized
and eigen
classes
Factors
Geology 1 0.05455
LULC 1/2 1 0.03570
Soil 1/3 1 1 0.03637
Slope 4 5 4 1 0.13450
Aspect 1/4 1/3 1/3 1/5 1 0.02222
Relative 3 4 4 1/3 3 1 0.10070
relief
Curvature 1/4 1/4 1/3 1/6 1/2 1/3 1 0.01762
TWI 1/2 2 1 1/4 2 1/3 4 1 0.04065
SPI 1/3 1/2 ½ 1/5 2 ¼ 2 1/3 1 0.02454
Drainage 4 4 4 1 5 3 5 4 5 1 0.13061
density
Drainage 4 4 4 1 4 1/3 5 4 5 1 1 0.11822
buffer
Reservoir 4 5 4 3 5 4 5 5 5 3 3 1 0.20538
buffer
PHA 3 4 4 1/3 3 ½ 5 3 4 1/3 1/4 1/3 1 0.07889
CR = 0.0785, maximum eigenvalue = 14.4615
Geology
Blaini formation 1 0.07565
Nagthat formation 3 1 0.24704
Chandpur formation 4 2 1 0.33496
Berinag formation 1 1/4 1/5 1 0.07176
Deoban formation 1 1/4 1/5 1 1 0.06555
Mandhali formation 1/3 1/5 1/5 1/3 1/2 1 0.03877
Rautgara formation 4 1/3 1/3 3 3 4 1 0.16623
LULC
Dense forest 1 0.07173
Scrub forest 3 1 0.15323
Agriculture 4 2 1 0.24412
Settlement 5 4 3 1 0.53092
(continued)
262 A. Tyagi et al.
Table 4 (continued)
LULC
Soil cover
Alluvial soil 1 0.638
Forest soil 1/5 1 0.101
Sandy soil 1/3 3 1 0.262
Slope
0°–13° 1 0.04308
14°–23° 2 1 0.06682
24°–32° 4 3 1 0.14245
33°–42° 6 5 3 1 0.29466
43°–75° 8 6 4 2 1 0.45299
Aspect
North 1 0.04161
Northwest 3 1 0.05852
West 3 2 1 0.08241
Southwest 4 3 3 1 0.15122
South 5 4 4 3 1 0.30804
Southeast 4 3 3 3 1/3 1 0.21129
East 2 2 1 1/3 1/4 1/3 1 0.06652
Northeast 2 1 ½ 1/3 1/4 1/4 1/2 1 0.05234
Flat 1/3 1/3 1/3 1/4 1/5 1/5 1/3 1/3 1 0.02712
Relative relief
13–270 m 1 0.04674
280–400 m 2 1 0.07483
410–510 m 3 2 1 0.11732
520–620 m 5 4 3 1 0.25099
630–840 m 8 6 5 3 1 0.51012
Profile curvature
Concave 1 0.66666
Convex 1/2 1 0.33333
(continued)
Table 4 (continued)
Profile curvature
CR = 0, maximum eigenvalue = 2.00
TWI
2–4 1 0.07600
5–6 3 1 0.14751
7–8 4 3 1 0.27356
>9 5 4 2 1 0.50293
SPI
<3 1 0.06232
3–4 2 1 0.09846
5–6 3 2 1 0.16083
7–8 4 3 2 1 0.26192
>8 5 4 3 2 1 0.41646
Drainage density
0–300 1 0.06232
301–600 2 1 0.09846
601–900 3 2 1 0.16083
901–1200 4 3 2 1 0.26192
1201-1500 5 4 3 2 1 0.41646
Drainage buffer
0–50 m 1 0.48637
51–100 m 1/3 1 0.23394
101–150 m 1/4 1/3 1 0.15493
151–200 m 1/5 ¼ 1/2 1 0.08045
>200 m 1/7 1/6 1/4 1/3 1 0.04432
Reservoir buffering
0–100 m 1 0.38348
101–200 m 1/3 1 0.25485
201–300 m 1/4 1/2 1 0.16406
301–400 m 1/5 1/3 1/2 1 0.10303
401–500 m 1/6 1/4 1/3 ½ 1 0.06494
>500 m 1/9 1/5 1/4 1/3 1/2 1 0.02964
(continued)
264 A. Tyagi et al.
Table 4 (continued)
Reservoir buffering
PHA
0.11–0.12 g 1 0.06232
0.12–0.13 g 2 1 0.09846
0.13–0.14 g 3 2 1 0.16083
0.14–0.16 g 4 3 2 1 0.26192
0.16–0.19 g 5 4 3 2 1 0.41646
7 Results
AHP technique is used to give weightage to different causative factors for preparing
an LHZ map for the study area. Each factor of the raster map is given weight value.
Weighted raster maps were integrated resulting in an LHI map, where the high value
of LHI means that the grid is lying on a more critical landslide zone and the low value
of LHI means that the grid is lying on a less critical landslide zone. Hence, the LHI
map is classified into the following five categories: Very low, low, moderate, high,
and very high (Fig. 6). Entire region is divided into five zones of susceptibility, i.e.,
1% of the entire area is found in very low class, 59% in low class, 30% in moderate
class, 9% in high class, and 2% in very high. Vulnerable areas in the form of built-up
(M) (N)
Fig. 6 m Vulnerability Map and n landslide hazard zonation map of the Tehri area
Fig. 7 Landslide Risk map

of the Tehri area
areas are found out using LULC map (Fig. 6) and a landslide risk map (Fig. 7) is
prepared from LHZ map and landslide vulnerability map.
8 Conclusions
The present study uses the ability of AHP in predicting areas prone to landslides.
This method is effective in allocating weights to different factors and their respective
classes based on the relative importance of different factors, which depends on the
choice of expert and skilled knowledge. Weights are used to classify study areas into
zones and a hazard map is prepared. Further, LULC classification is applied to the
study area and vulnerable areas are identified. A risk map is generated by overlaying
hazard and vulnerability maps. Recognizing zones with a high level of risk can help
in increasing security in urban planning by applying safety and maintenance actions.
References
1. Varnes DJ (1978) Slope movement types and processes. In: Schuster RL, Krlzek RJ (eds)
landslides: analysis and control. Transportation Research Board, Nat. Acad. Scl., Washington,
D.C., pp 11–33
2. Naithani AK (1999) The Himalayan landslides. Employment News, vol. XXIII no. 47, 20–26
February, 1–2
266 A. Tyagi et al.
3. Varnes DJ (1984) Landslide hazard zonation: a review of principles and practice. UNESCO,
Paris, pp 1–63
4. Anbalagan R (1992) Landslide hazard evaluation and zonation mapping in mountainous terrain.
Eng Geol 32:269–277
5. Gupta RP, Joshi BC (1990) Landslide Hazard Zonation using the GIS approach—a case study
from the Ramganga Catchment, Himalayas. Eng Geol 28:119–131
6. Valdiya KS (1980) Geology of Kumaun lesser Himalaya. Wadia Institute of Himalayan
Geology, Dehradun. Interim. Report, 291
7. Kumar R, Anbalagan R (2015) Landslide susceptibility zonation of Tehri reservoir rim region
using binary logistic regression model. Curr Sci 1662–1672
8. Gupta RP, Saha AK, Arora MK, Kumar A (1999) Landslide hazard zonation in a part of
Bhagirathy Valley, Garhwal Himalayas, using integrated Remote Sensing & GIS. J Him Geol
20:71–85
9. Nagarajan R, Mukherjee A, Roy A, Khire MV (1998) Temporal remote sensing data and GIS
application in landslide hazard zonation of part of Western Ghat, India. Int J Remote Sens
19:573–585
10. Ayalew L, Yamagishi H (2005) The application of GIS-based logistic regression for land-
slide susceptibility mapping in the Kakuda–Yahiko Mountains, Central Japan. Geomorphology
65:15–31
11. ESRI FAQ (2012) What is the Jenks optimization method? http://support.esri.com/en/knowle
dgebase/techarticles/detail/26442
12. Wilson JP, Gallant JC (2000) Terrain analysis: principles and applications. Wiley & Sons Ltd.,
New York, USA, pp 1–27
13. Sharma ML (1998) Attenuation relationship for estimation of peak ground horizontal acceler-
ation using data from strong-motion arrays in India. Bull Seismol Soc Am 88(4):1063–1069
14. Stewart JP, Liu AH, Choi Y (2003) Amplification factors for spectral acceleration in tectonically
active regions. Bull Seismol Soc Am 93:332–352
15. James N, Sitharam TG (2017) Multicriteria based landslide hazard assessment and vulnerability
studies for the state of Sikkim, India. In: 16th world conference on earthquake, Santiago Chile
16. Wald DJ, Allen TI (2007) Topographic slope as a proxy for seismic site conditions and
amplification. Bull Seismol Soc Am 97:1379–1395
17. BSSC (2003) NEHRP recommended provisions for seismic regulations for new buildings and
other structures (FEMA 450), Part 1: Provisions. Building Seismic Safety Council for the
Federal Emergency Management Agency, Washington, D.C., USA
18. Raghu Kanth STG, Iyengar RN (2007) Estimation of seismic spectral acceleration in peninsular
India. J Earth Syst Sci 116:199–214
19. Yalcin A (2008) GIS-based landslide susceptibility mapping using analytical hierarchy process
and bivariate statistics in Ardesen (Turkey): comparisons of results and confirmations. Catena
72:1–12
20. Ayalew L, Yamagishi H, Marui H, Kanno T (2005) Landslides in Sado Island of Japan:
Part II. GIS-based susceptibility mapping with comparisons of results from two methods and
verifications. Eng Geol 81:432–445
21. Saaty TL (1980) The analytic hierarchy process: planning, priority setting, resource allocation.
McGraw-Hill International Book Company, New York
22. Feizizadeh B, Blaschke T (2013) GIS-multicriteria decision analysis for landslide susceptibility
mapping: comparing three methods for the Urmia lake basin, Iran. Nat Hazards 65:2105–2128
23. Saaty TL (1977) A scaling method for priorities in hierarchical structures. J Math Psycho
15:234–281
A Comparative Study on Landslide
Susceptibility Mapping Using AHP
and Frequency Ratio Approach
Malemnganbi Lourembam Chanu and O. Bakimchandra
Abstract Landslide is a naturally occurring phenomenon in most of the moun-

tainous regions of the world. Manipur, being a landlocked hilly state, is continuously
facing the problems of landslides in the rainy seasons and the economic conditions
are highly affected due to blockages of the highways of the region which served
as the lifelines. So it becomes very important to check the problems caused due to
this natural disaster. In this particular study, an attempt is being made for devel-
oping the landslide susceptibility mapping of the region using two GIS-based land-
slide susceptibility approaches––Analytic hierarchy approach and Frequency ratio
approach. Eight causative factors Land Use Land Cover (LULC), Normalized Differ-
ence Vegetation Index (NDVI), slope, aspect, curvature, elevation, rainfall, and soil
types are considered in the study. The output landslide susceptibility maps devel-
oped by the two different approaches have been compared and validation of both
the models have been done using landslide locations of the region. Both the models
show good accuracy but the Frequency Ratio shows higher accuracy when compared
to the AHP approach.
Keywords Landslide · Analytic hierarchy process (AHP) · Frequency ratio ·

Geographical information system (GIS) · Success rate curve
1 Introduction
Landslides are one of the natural phenomena prevalent in the mountainous region of
the world. If there is any downward movement caused by the influence of gravity, it
is termed a landslide [1]. Landslides are one of the major natural geological hazards
prevalent in Manipur, a border state of India. Sometimes landslide is triggered due to
M. L. Chanu (B) · O. Bakimchandra

Department of Civil Engineering, National Institute of Technology, Imphal 795004, Manipur,
India
e-mail: malemnganbilc@gmail.com
O. Bakimchandra
e-mail: bakim143@gmail.com
268 M. L. Chanu and O. Bakimchandra
anthropogenic activities such as large-scale deforestation, agricultural practices on

slopes, wildfires, etc. ([2, 3]). Each and every year, many casualties, death, and huge
economic loss have been reported from various parts of the world and nearly 17% of
fatalities from natural hazards are from landslides worldwide [4]. Generally, land-
slides are caused by the combination of geo-environmental factors [1, 5]. Landslide
occurrences depend on the complex interactions among a large number of causative
factors. These factors can be categorized into static and dynamic factors (rainfall and
earthquakes) [6]. Static factors or primary factors are basic surface-related charac-
teristics that are related to sliding [7]. Proper understanding of the causative factors
is necessary in order to develop reliable landslide susceptibility maps. The identifi-
cation of susceptible landslide zones is a must for proper planning and construction
purposes and also for the safety of the inhabitants.
Through proper surveying and analysis by scientific studies, it is possible to assess
and predict the landslide susceptible regions and the damage can be reduced through
proper planning purposes and using the prediction models. Landslide susceptibility
is defined as the probability of landslide occurrence based on the conditions of the
local terrain [8]. A landslide susceptibility map shows the landslide potential areas
of the future by considering the factors that contributed to past landslides [9]. The
spatial prediction of landslide hazards is one important field of geoscientific research
in which statistical classification rules have been applied [10]. Multi-temporal infor-
mation is particularly important since future landslide hazards partly depend on the
scarps of past events [11] and should therefore be modeled conditionally.
In this particular study, an effort is being constructed to produce the landslide
susceptibility map of the region using the Frequency Ratio Approach and Analytic
Hierarchy Process. Eight causative factors are taken into consideration as topographic
data––elevation, slope, aspect, curvature, soil information, Normalized difference
Vegetation Index (NDVI), Tropical Rainfall Measuring Mission (TRMM) precipi-
tation data. Landslide location data, collected from field survey and various depart-
ments, has been employed to develop the spatial relationship with the various
causative factors. The landslide causative factors are integrated into the GIS plat-
form using the frequency ratio approach and AHP to generate the final Landslide
Susceptibility Map. The maps were validated and the accuracy of the models has
been checked.
2 Study Area
Manipur is a state lying at the northeastern border of India with a total area of 22,327
km2 . The region is geographically divided into two broad regions, viz., the hill and
the valley regions. The average elevation of the valley is about 790 m above the sea
level and that of the hill’s ranges from 1500 to 1800 m. Around 90% of the total area
constitutes the hilly regions and the remaining 10% constitutes the valley region.
Specially, the hilly districts are highly affected in the rainy months. The state has
A Comparative Study on Landslide Susceptibility Mapping Using AHP … 269
Fig. 1 Study area, Manipur,

India
to face huge losses in economy due to blockages of Highways- 37, 2, 102, and 202
during the monsoons and rainy seasons every year (Fig. 1).
In general, the causes of landslides in Manipur are mainly due to weak lithology
consisting of complex structural features like highly jointed rock formation; unstable
slope and high intensity rainfall during the rainy seasons. Landslides occurring in
the areas are both natural and anthropogenic. Landslides are quite frequent and past
records show that landslides mostly occur during the rainy seasons in the month of
June–September in the region. The damage caused by landslides is huge and the state
has to face huge economic losses.
3 Data Description and Methodology
Eight factors have been considered for the preparation of the landslide susceptibility
map of the region. The acquisition of the conditioning factors involves a wide variety
of data sources. LANDSAT 8 imageries downloaded from the USGS have been used
to generate the LULC map of the region, using the algorithm of SAM (Spectral Angle
Mapper) and is also used to generate the vegetation index, Normalized difference
Vegetation Index (NDVI). SRTM (Shuttle Radar Topographic Mission) DEM data of
30 m resolution has been used to generate various topographic layers––slope, aspect
and curvature of the area. And soil map from the National Bureau of Soil Survey and
Land Use Planning (NBSSLUP) has been used to obtain the soil type (Fig. 2).
Historical landslides locations of past field visits collected from Earth Science
Department, Manipur University and Manipur Science and Technology Council
(MASTEC) and also landslides locations taken during reconnaissance field survey
have been used for model validation. Rainfall data from (Tropical Rainfall Measuring
Fig. 2 Landslide conditioning factors a Aspect, b Elevation, c Soil type, d LULC, e NDVI,
f Rainfall, g Slope, h Curvature
Mission) TRMM is used to generate the annual rainfall layer, for the year 2017, using
the Geostatistical-Kriging interpolation technique (Fig. 3).
Different data have been derived and prepared from various sources. LANDSAT
8 images are obtained to derive the LULC and the spectral index NDVI. SRTM DEM
data is used for the derivation of various topographic parameters like slope, aspect,
curvature, elevation map, and rainfall map is prepared using Kriging Interpolation
technique from TRMM data.
4 Assigning Weights for Each Factor and Classes Using

AHP
The Analytic Hierarchy Process (AHP) is a semi-quantitative multi-criteria decision-

making process of measurement through pair-wise comparisons of causative factors
and to derive priority scales. It has four levels, viz., defining problem, determination
of goals and alternatives, construction of pair-wise comparison matrix, determining
weights and obtaining overall priority. AHP converts complex decision-making
problem into a hierarchy of factors and alternatives. Weights of each factor and alter-
natives are assigned on a nine-point ordinal scale (Table 1) by pair-wise comparison
between them. Both subjective and objective factors are to be considered in the
decision-making process [12].
DATA PREPARATION
LANDSAT 8 TRMM SOIL MAP SRTM DEM

SATELLITE DATA
NDVI RAINFALL SOIL TYPE SLOPE

PARAMETER MAPS
LULC ELEVATION
CURVATURE
ASPECT
ANALYTIC HIERARCHY PROCESS FREQUENCY RATIO ANALYSIS
LANDSLIDE SUSCEPTIBILITY MAP LANDSLIDE SUSCEPTIBILITY MAP
VALIDATION USING VALIDATION USING

LANDSLIDE LOCATIONS LANDSLIDE LOCATIONS
COMPARISION
FINAL LANDSLIDE SUSCEPTIBILITY

MAP OF MANIPUR
Fig. 3 Methodological framework
Table 1 Ordinal scale representing preference of judgment [13]

Importance scale Degree of preference Definition
1 Equal Equal importance
3 Moderate One factor moderately favored over other
5 Strongly Judgment strongly favors one over other
7 Very strongly One factor very strongly favored over other
9 Extreme importance One factor favored over other in highest degree
2, 4, 6, 8 Intermediate Compensation between weights 1,3, 5,7, and 9
Reciprocals Opposite Refers inverse comparison
In the present study, the relative value of each pair of the factors and factor classes
was determined by the frequency of landslides on each class. Matrix calculation
is used to derive the weights of factor/class weights in terms of the Eigen vector.
One important aspect of the AHP is the calculation of Consistency Index (CI) and
Consistency Ratio (CR).
Table 2 Random consistency index (RI) [14]

N 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
RI 0 0 0.58 0.9 1.12 1.24 1.32 1.41 1.45 1.49 1.51 1.53 1.56 1.57 1.59
CR is calculated as
CI
CR = (1)
RI
where RI stands for random index (Table 2) (Satty 1980) based on the number of
random samples. CR value of 0.1 is the maximum threshold of consistency of the
matrix. CR value >0.1 is thought to be inconsistent whereas value 0 indicates a
perfectly consistent comparison result.
AHP method has its advantages in weighting/rating of factors and their classes
along with some deficiencies. Pair-wise comparison can provide a simple and accept-
able decision rule. Some landslide conditioning factors have a certain degree of
dependency in influencing landslides but AHP considers each factor in the hierarchy
as an independent entity. Overall AHP multi-criteria decision-making provides a
very simple and flexible decision-making.
Landslide susceptibility index (LSI) was computed as given below

N

LSI = Weight of factor (Wj) × Weight of factor classes Wij (2)
j=1
where Wij = weight of ith class of factor j.

Lastly, the LSI map was classified into five classes as very low, low, moderate,
high, and very high susceptibility classes of susceptibility using Natural Jenk’s break
classification.
4.1 Frequency Ratio Approach for Landslide Susceptibility
Frequency ratio approach is a bi-variate statistical approach of landslide suscepti-

bility assessment based on observed relationships between landslide distribution and
landslide causative factors. Generally, it is necessary to assume that landslide occur-
rence is determined by landslide-related factors to predict landslides, and that future
landslides will occur under the same conditions as past landslides. Using this assump-
tion, the relationship between landslides occurring in an area and the landslide-related
factors can be distinguished from the relationship between landslides not occurring
in an area and the landslide-related factors.
The frequency ratio is used to represent the distinction quantitatively. The

frequency ratio is the ratio of the area where landslides occurred to the total study
area, and also, is the ratio of the probabilities of a landslide occurrence to a non-
occurrence for a given factor’s attribute. Therefore, the greater the ratio above unity,
the stronger the relationship between landslide occurrence and the given factor’s
attribute, and the lower the ratio below unity, the lesser the relationship between
landslide occurrence and the given factor’s attribute.
%target occur ence in each sub categor y ( points in f actor class/total points)
Frequency ratio, F R = =
%categor y o f an independent f actor ( f actor class ar ea/total area)
(3)
PR = |RFmax − RFmin|/|RFmax − RFmin|min (4)
Table 5 has been used to give weights to each and every factor and the Relative
Frequency (RF) from Table 3 has been used to give weights to each class of every
factor.
5 Results
The relative frequency obtained from Table 3 has been used to assign weights to the
factor classes of the eight conditioning factors and the prediction rate from Table 4
has been used to assign weights to the conditioning factors (Table 5).
The final Landslide Susceptibility Index Map (LSI) for Frequency Ratio Approach
is found out by summing up all the weighted factors.

LSI = Fij ∗ Wij (5)
where Fij = Weighted factors and Wij = weights obtained from Prediction rate.
Landslide Susceptibility Index (LSI) for Analytic Hierarchy Process (AHP) [15]
was computed using Eq. 2.
Lastly, the LSI maps have been classified into five classes as—no landslide, low,
moderate, high, and very high susceptibility classes of susceptibility using Natural
Jenk’s break classification, Jenkins optimization technique. In the above landslide
susceptibility maps produced by Frequency Ratio and AHP Approach, the patterns
of the landslide classes are almost similar. To understand the pattern more, the area
of each class has been found out which is shown in Table 6.
The above map shows the status of the combination of the two resulting maps
obtained from frequency ratio and AHP approach and Table 7 depicts the area falling
under each status class (Figs. 4 and 5).
Table 3 Landslide area falling under each class of the factors and their relationship with landslides
274
Factors Factor classes No. of points % of points Class area % of class area Ratio (+) Relative frequency
Slope angle 1 6300 6.730769231 7,799,085 31.448641 0.21402417 0.04
2 26,100 27.88461538 8,859,387 35.724150 0.78055364 0.14
3 51,300 54.80769231 6,203,538 25.014837 2.19100737 0.40
4 9000 9.615384615 1,658,073 6.685931 1.43815198 0.26
5 900 0.961538462 279,351 1.126441 0.85360745 0.16
Total 93,600 100 24,799,434 100 5.47734461 1
Aspect 1 0 0 847,866 3.539918 0 0.00
2 9900 10.57692308 2,751,132 11.486229 0.92083514 0.12
3 2700 2.884615385 2,769,487 11.562863 0.24947243 0.03
4 13,500 14.42307692 3,149,691 13.150249 1.09679115 0.14
5 15,300 16.34615385 3,235,370 13.507967 1.21011208 0.15
6 17,100 18.26923077 2,861,654 11.947668 1.52910429 0.19
7 10,800 11.53846154 3,059,620 12.774194 0.90326334 0.11
8 12,600 13.46153846 3,180,422 13.278554 1.01378046 0.13
9 11,700 12.5 2,944,193 12.292276 1.01689873 0.13
Total 93,600 100 23,951,569 100.000000 7.94025762 1
Elevation 1 14,400 15.38461538 4,444,324 17.921070 0.85846525 0.17
2 15,300 16.34615385 8,367,688 33.741446 0.48445327 0.10
(continued)
M. L. Chanu and O. Bakimchandra
Table 3 (continued)
3 41,400 44.23076923 6,488,959 26.165753 1.69040687 0.34
4 18,900 20.19230769 4,063,427 16.385160 1.23235343 0.25
5 3600 3.846153846 1,435,037 5.786571 0.66466887 0.13
Total 93,600 100 24,799,435 100 4.93034769 1
Curvature 1 61,200 65.38461538 11,328,410 45.680117 1.43135832 0.56
2 4500 4.807692308 2,556,218 10.307566 0.46642361 0.18
3 27,900 29.80769231 10,914,805 44.012317 0.67725797 0.26
Total 93,600 100 24,799,433 100 2.5750399 1
LULC 1 0 0 115,634 0.468443 0 0.00
2 0 0 312,288 1.265105 0 0.00
3 7200 7.692307692 16,649,192 67.447265 0.11404922 0.03
4 1800 1.923076923 1,633,547 6.617635 0.29059882 0.07
5 84,600 90.38461538 5,974,094 24.201553 3.73466184 0.90
Total 93,600 100 24,684,755 100 4.13930987 1
Soil 1 4500 4.807692308 6,632,774 26.745720 0.17975558 0.01
2 5400 5.769230769 2,325,911 9.378906 0.61512833 0.04
3 5400 5.769230769 258,565 1.042627 5.53336202 0.37
A Comparative Study on Landslide Susceptibility Mapping Using AHP …
4 15,300 16.34615385 5,170,820 20.850598 0.78396572 0.05

5 10,800 11.53846154 3,426,371 13.816355 0.83513067 0.06
6 23,400 25 3,696,266 14.904668 1.67732686 0.11
(continued)
275
Table 3 (continued)
276
7 11,700 12.5 2,080,577 8.389631 1.48993434 0.10
8 17,100 18.26923077 1,208,101 4.871496 3.75023022 0.25
Total 93,600 100 24,799,385 100 14.8648337 1
NDVI 1 0 0 107,702 0.436190 0 0.00
2 27,900 29.80769231 3,497,782 14.165912 2.10418447 0.39
3 39,600 42.30769231 4,530,842 18.349774 2.30562469 0.43
4 19,800 21.15384615 6,585,861 26.672539 0.79309457 0.15
5 6300 6.730769231 10,077,056 40.811774 0.16492224 0.03
Total 93,600 100 24,691,541 100.4361899 5.36782597 1
Rain 1 9900 10.57692308 4,693,059 18.92442624 0.55890324 0.15
2 51,300 54.80769231 8,786,654 35.43155658 1.54686098 0.42
3 26,100 27.88461538 6,912,463 27.87401483 1.0003803 0.27
4 6300 6.730769231 2,855,297 11.51378183 0.5845837 0.16
5 0 0 1,551,477 6.256220525 0 0.00
Total 93,600 100 24,798,950 100 3.69072822 1
M. L. Chanu and O. Bakimchandra
Table 4 Prediction rate calculation using the above equation and the information from Table 3
Class PR
Slope angle 1.874262
Aspect 1
Elevation 1.270137
Curvature 1.945859
LULC 4.685121
Soil 1.870182
NDVI 2.230425
Rain 2.17639
Table 5 Weights obtained from frequency ratio and pair-wise comparison

Factors Prediction rate using FR Prediction rate using pairwise comparison
Aspect 100.000 46.914
Elevatiion 127.014 59.588
Soil 187.018 87.738
Slope angle 187.426 87.930
Curvature 194.586 91.289
Rain 217.639 102.104
NDVI 223.043 104.639
LULC 468.512 219.799
Table 6 Area falling under each landslide susceptibility classes

Class Frequency ratio approach Analytic hierarchy approach
Area (Km2 ) % Area Area (Km2 ) % Area
No landslide 5189.56 23.36 7317.70 32.94
Low 7513.96 33.83 6308.47 28.40
Moderate 4135.54 18.62 4975.35 22.40
High 2547.22 11.47 2358.88 10.62
Very high 2825.97 12.72 1251.85 5.64
Table 7 The status of combined maps and the area for each status
Value Count FR AHP Status Area (km2 )
1 2,223,461 3 3 Moderate 2001.115
2 900,192 4 4 High 810.1728
3 1,592,210 5 4 VH-H 1432.989
4 3,937,685 2 2 Low 3543.917
5 1,931,552 3 2 M-L 1738.397
6 1,139,685 1 2 NL-L 1025.717
7 3,208,101 2 1 L-NL 2887.291
8 1,199,578 2 3 L-M 1079.62
9 1,243,185 5 5 Very high 1118.867
10 4,607,890 1 1 No landslide 4147.101
11 147,620 4 5 H-VH 132.858
12 1,781,954 4 3 H-M 1603.759
13 18,600 1 3 NL-M 16.74
14 125,099 3 4 M-H 112.5891
15 304,576 5 3 VH-M 274.1184
16 314,792 3 1 M-NL 283.3128
17 3479 2 4 L-H 3.1311
18 138 3 5 M-VH 0.1242
19 3 1 4 NL-H 0.0027
20 483 4 2 H-l 0.4347
Fig. 4 Landslide susceptibility maps obtained from frequency ratio and AHP approach
6 Validation
Validation is important in landslide susceptibility analysis. Without validation, no

interpretation is possible and no support of the method or the input information
can be provided [16]. For validation of landslide susceptibility calculation models,
it is assumed that landslides are related to spatial information such as topography,
LULC, and future landslides will be triggered by specific factors such as rainfall.
Two types of plots have frequently been used in landslide susceptibility modeling:
the success rate and prediction rate curves [17]. The success rate curve is obtained
by varying the decision threshold and plotting the respective sensitivities against
the total proportions of the data set classified as landslide. This may be done on
independent test data sets (Fig. 6).
Some authors use either test data from the training area but for a different time
period for evaluation [17, 18], or data from an adjacent test area [17]. But in this
particular study, 75% of the total landslide has been used to train the model which
will give the success rate of the model and remaining 25% has been used to test
the model which will give the prediction rate of the model and thus accuracy of the
model can be checked.
The Area Under Curve (AUC) value for finding accuracy was calculated by using
simple trapezium method. AUC value of 0.7512 is obtained from the success rate
curve and 0.7314 from the prediction rate curve for AHP approach. So, the model
gave an accuracy of 75.12 and 73.14% for success and prediction rate, respectively.
Whereas an accuracy of 75.26 and 74.21% have been obtained from the success rate
and prediction rate curve for Frequency Ratio Approach.
7 Conclusions
From the present study, we can conclude that both AHP and Frequency Ratio
Approach can be conventionally used for the landslide susceptibility analysis for
regional scale by combining only some easily available data such as topographic
slope, aspect, curvature, elevation which can be derived from DEM and ancillary
data such as soil map and rainfall data were also considered for the better results. It
can be concluded from the validation that both the model shows acceptable agree-
ment in accordance with the landslide occurrences with the frequency ratio showing
a slightly higher accuracy i.e. 75.12% as compared to AHP approach, i.e., 75.26%.
Fig. 5 Combined status map of the similarity and dissimilarity
AUC for AHP AUC for FREQUENCY RATIO
SUCCESS RATE PREDICTION RATE SUCCESS RATE PREDICTION RATE
120 120
Cumulave percentage of landslides
Cumulave percentage of landslides
100 100
80 80
60 60
40 40
20 20
0 0
100
10
19
28
37
46
55
64
73
82
91
1
1 10 19 28 37 46 55 64 73 82 91 100
LSI index in percentage (decreasing) LSI index in percentage (decreasing)
Fig. 6 Cumulative frequency diagrams for AHP approach and frequency ratio, respectively
References
1. Varnes DJ (1984) Landslide hazard zonation: a review of principles and practice, natural hazards
3. Communication landslides IAEG, UNESCO, Paris, France, p 63
2. Sidle RC, Pearce AJ, O’Loughlin CL (1985) Hillslope stability and land use. Water Resources
Monograph Series 11, American Geophysical Union, Washington, DC
3. Kuruppuarachchi T, Wyrwoll KH (1992) The role of vegetation clearing in the mass failure of
hillslopes: Moresby Ranges, Western Australia. Catena 19, 193–208
4. CRED (2006) Retrieved from international disaster database. https://www.cred.be/emdat
5. Hutchinson, J.N. (1995) Keynote paper: landslide hazard assessment. In: Bell (ed) Landslides,
Balkema, Rotterdam, pp 1805–1841
6. Dai FC, Lee CF (2002) Landslide characteristics and slope instability modeling using GIS,
Lantau Island, Hong Kong. Geomorphology 42, 213–238
7. Sidle RC, Ochiai H (2006) Landslide processes, prediction, land use. American Geophysical
Union, Washington, DC, 1–312
8. Brabb E (1984) Innovative approaches for landslide hazard evaluation. In: IV International
symposium landslides, Toronto, pp 307–323
9. Santacana N, Baeza B, Corominas J, de Paz A, Marturia J (2003) A GIS-based multivariate
statistical analysis for shallow landslide susceptibility mapping in La Pobla de Lillet Area
(Eastern Pyrenees Spain). Nat Hazards 30:281–295
10. Brenning A (2005) Spatial prediction models for landslide hazards: review, comparison and
evaluation. Nat Hazards Earth Syst Sci 5:853–862
11. Casadei M, Dietrich WE, Miller NL (2003) Testing a model for predicting the timing and
location of shallow landslide initiation in soil-mantled landscapes. Earth Surf Proc Land
28:925–950
12. Yalcin A (2008) GIS-based landslide susceptibility mapping using analytical hierarchy process
and bivariate statistics in Ardesen (Turkey): comparisons of results and confirmations. Catena,
72, 1–12
13. Saaty TL (1977) A scaling method for priorities in hierarchical structures. J Math Psycho, 15,
234–281.
14. Saaty TL (1980) The analytic hierarchy process: planning, priority setting, resource allocation.
McGraw-Hill International Book Company, New York
15. Saaty T (2008) Decision making with the analytical hierarchy process. Int J Serv Sci 1(1):83–98
16. Chung CJF, Fabbri A (2003) Validation of Spatial Prediction Models for Landslide Hazard
Mapping. Nat. Hazards, 30, 451–472
17. Chung CF, Fabbri A (2003) Validation of spatial prediction models for landslide hazard
mapping. Nat Hazards 30:451–472
18. Chung CF, Kojima H, Fabbri AG (2002) Stability analysis of prediction models for landslide
hazard mapping, in Allison 3–19
Support Vector Machine for Evaluation
of Liquefaction Potential Using SPT Data
Dev Kumar Pradhan, Suvendu Kumar Sasmal, Vamsi Alla,

and Rabi Narayan Behera
Abstract The geotechnical engineers usually have difficulties in resolving compli-

cated problems that involve a number of influencing parameters. Sometimes, it is
also complicated to describe a problem mathematically. This study aims for an alter-
native algorithm known as Support Vector Machine (SVM), which can be used to
solve the classification-type problem. In this study, liquefaction potential is anal-
ysed using influencing parameters viz. the Standard Penetration Test (SPT) values,
different soil parameters, and depth of water table taken from a borehole database
of a certain depth. This borehole database was prepared by collecting borehole data
of different sites located in different parts of the country. A deterministic approach
has been used to evaluate the liquefaction potential and it is expressed in the form of
factor of safety (FS). The SVM model developed using the dataset of the liquefac-
tion potential evaluated from a deterministic approach showed an overall accuracy
of 96.8%.
Keywords Liquefaction · Standard penetration test · Deterministic approach ·

Classification · Support vector machines
1 Introduction
If a mass of sand in a saturated condition with a greater void ratio than its critical
void ratio is subjected to a suddenly applied shearing stress, as from an earthquake,
D. K. Pradhan · S. K. Sasmal · V. Alla · R. N. Behera (B)

Department of Civil Engineering, National Institute of Technology Rourkela, Rourkela 769008,
India
e-mail: rnbehera82@gmail.com
D. K. Pradhan
e-mail: devpradhan.nit@gmail.com
S. K. Sasmal
e-mail: suvendukumarsasmal@gmail.com
V. Alla
e-mail: vamsi.royalcivil@gmail.com
284 D. K. Pradhan et al.
heavy blasting, pipe driving, or any other dynamic force, the sand tends to decrease
in volume. As a result, the pore water is subjected to a suddenly applied hydrostatic
excess pressure, and a portion of the weight of overlying material is transferred
from inter-granular pressure to pore water pressure. The effective stress in the soil
is thus reduced. Since the shearing strength of granular soil depends on the effective
stress, this transfer of pressure causes a sudden decrease in shearing strength. If it
is reduced to a value below the applied shearing stress, the mass will fail in shear.
This type of failure is sometimes referred to as liquefaction. As such, there is no
single major method to study the liquefaction problem. In the present work, this
problem is addressed using soft computing technique. One of the recent trends in
soft computing across engineering applications is the use of Support Vector Machine
(SVM). In recent times, SVM has been used in the field of geotechnical engineers by
various researchers. Zhang et al. [12] analyzed the soil concrete interfacial strength
using LS-SVM. Zhang et al. [11] used the SVM method for the analysis of geometry
factors in rock slope. Ji et al. [4] analyzed the stability of slopes using this technique.
Kurnaz and Kaya [5] studied the compression index of clay using the Support Vector
Machine. Borthakur and Dey [1] observed the behavior of micropile in clay using the
SVM technique. One of the key criteria to forecast the occurrence of the liquefaction
is to observe the SPT N values of the corresponding soil. The Standard Penetra-
tion Tests is generally conducted as per IS-2131:1981 [3]. This test is performed to
determine different soil parameters like relative density or density index, angle of
internal friction, unconfined compressive strength of clay, ultimate pile load-carrying
capacity, ultimate bearing capacity as per shear criteria, and allowable bearing pres-
sure on the basis of settlement criteria. In this test, a split sampler is placed over the
soil in a borehole of 55–150 mm diameter and this sampler is driven by the dynamic
mechanism of hammer. The weight of hammer is 65 kg and the height is 760 mm.
The test is conducted at every 2–5 m interval or at a change of soil strata. The SPT
number is defined as the number of blows of hammer required for 300 mm penetra-
tion. Usually, test is performed in 3 stages 150 mm each. SPT number is taken as
the number of blows required for the last 300 mm of penetration, which means the
number of blows required for the first 150 mm penetration is ignored. In this study,
a number of SVM classification techniques are presented to predict the liquefaction
potential in terms of SPT data. Outcomes corresponding to different kernels are also
presented. A comparative analysis among the different considered methods is also
discussed, which will be helpful in deciding the method selection for liquefaction
analysis.
1.1 Support Vector Machine (SVM)
SVM is one of the most reliable soft computing techniques to predict the dependent
variable in terms of the independent variable. SVM was introduced as a classifier
by Cortes and Vapnik [2], which was initially popularized in the Neural Information
Processing Systems community.
Support Vector Machine for Evaluation … 285
In a linearly separable case, for a data set, a linear hyperplane that can separate
two classes can be defined by Eq. (1),
f (x) = w · x + b = 0 (1)
where w Rn , b R = bias. Separating hyperplane for two classes can also be

represented as
w · xi + b ≥ 1 for yi = 1 (2)
w · xi + b ≤ −1 for yi = − 1 (3)
The above two equations can be written as
yi (w · xi + b) ≥ 1 (4)
Sometimes, slack variables (ξ i > 0) are used because of the effects of misclassi-
fication during the selection of training data of mixture of classes. So Eq. (4) can be
written as
yi (w · xi + b) ≥ 1 − ξi (5)
|1+b|
w = perpendicular distance from w · xi + b = −1 from the origin.
Similarly, |b−1|
w = perpendicular distance from w · xi + b = 1 from the origin. The
margin (ρ(w, b)) between the planes is simply (Fig. 1)
2
ρ(w, b) = (6)
w
For nonlinear classification kernel function is used, which calculates the dot
product of two vectors.
SVM Classifier: The classifiers used in this study are as follows:
C-SVM: In this type of SVM,
minimization of error function:
1 T N
w w+C ξi (7)
2 i=1
subjected to

yi wT ∅(xi ) + b ≥ 1 − ξi and ξi ≥ 0, i = 1, . . . , N (8)
Fig. 1 Support vector with maximum margin [7]
where C = capacity constant, w = vector of coefficients, b = constant, and ξ i

= parameters for handling non-separable data (inputs). The index i labels the N
training cases. The kernel φ is used for the purpose of transform data from the
input (independent) to the feature space. The problem with the parameter C is that
it can take any positive value and the other one is that it has no direct interpretation.
Therefore, it is hard to choose correctly and one has to resort to cross-validation or
direct experimentation to find a suitable value.
nu-SVM: Schölkopf et al. [8] reformulated SVM to take a new regularization
parameter nu. This parameter is that it is bounded between 0 and 1 and also it has a
direct interpretation.
The classification model involves minimization of error function:
1
N
1 T
w w − νρ + ξi (9)
2 N i=1
subjected to the constraints

yi wT ∅(xi ) + b ≥ ρ − ξi , ξi ≥ 0, i = 1, . . . , N and ρ ≥ 0. (10)
Bound constraint c-SVM: In this type of SVM offset b is introduced which

simplifies the dual solution with only bound constraints remaining.
Minimize the error function:
C
N
1 T 1
w w + b2 + ξi (11)
2 2 N i=1
subjected to

yi wT ∅(xi ) + b ≥ 1 − ξi , ξi ≥ 0 and i = 1, . . . , N (12)
Weston and Watkins multi-class SVM [10]:

Minimization of error function:
1 T
k l
wm wm + C ξi,t (13)
2 m=1 i=1 t=y i
subjected to

yi wT ∅(xi ) + b ≥ t wT ∅(xi ) + b + 2 − ξi,t ,
ξi,t ≥ 0, i = 1, . . . , l , and t ∈ {1, .., k}\yi (14)
the resulting decision function is

arg max fm (x) = arg max wmT ∅(x) + bm (15)
m m
Crammer and Singer’s multi-class SVM:

Minimization of error function:
1 T
k l
wm wm + C ξi (16)
2 m=1 i=1
subjected to
wyTi ∅(xi ) − wtT ∅(xi ) ≥ 1 − δyi ,t − ξi , I = 1, . . . , l and t ∈ {1, . . . , k} (17)
where, δi,j is the Kronecker delta (defined as 1 for I = j and as 0 otherwise). The
resulting decision function is
arg max fm (x) = arg max wmT ∅(x) (18)

m m
2 Methodology
2.1 Deterministic Approach
The cyclic stress ratio [9]:

amax σv
CSR = 0.65 × × × rd (19)
g σv
where amax = maximum horizontal acceleration, σ vo , σ ’vo = total and effective

overburden stress, r d is the stress reduction coefficient.
The expressions to estimate the average values of stress reduction factor (r d ) have
already been reported in Liao and Whiteman [6].
2.2 SVM Approach
In this technique, the seven input variables considered were the depth of interest,
SPT value, bulk density at the depth of interest, saturated density at the same depth,
groundwater table, fine contents, and the earthquake moment magnitude (M w ). The
output considered for the classification were liquefied soil and non-liquefied soil in
the deterministic approach case. The types of SVM classifiers used in this study are
C-SVM, nu-SVM, bound constraint c-SVM, Weston Watkins multi-class SVM, and
Crammer–Singer multi-class SVM along with different kernel functions to obtain an
optimal model.
The kernel functions are used for nonlinear SVM classification. The kernel
functions are as follows:
• The linear kernel function is the simplest of kernel functions.

k x, x = x, x (20)
• The Gaussian Radial Basis Function (RBF) kernel

2
k x, x = exp −σ x − x (21)
• The polynomial kernel

degree
k x, x = scale · x, x + offset (22)
• The hyperbolic tangent kernel


k x, x = tanh scale · x, x + offset (23)
• The Bessel function kernel

Besseln(ν+1) σ x − x
k x, x = (24)
(x − x )−n(ν+1)
• The Laplace Radial Basis function kernel

k x, x = exp −σ x − x (25)
• The ANOVA Radial Basis function kernel

n 2
d
k x, x = exp −σ xk − xk (26)

k=1
• The linear spline kernel in one dimension

3
x + x 2 min x, x
k x, x = 1 + xx min x, x − x, x + (27)
2 3
and for multidimensional case
n

k x, x = k xk , xk (28)
k=1
The optimal model was chosen which gave the least number of misclassification
or better accuracy. Training and testing of SVM for classification problem have been
carried out using R Studio.
Training and testing performance can be calculated from the following formula:
Training performance (%) or testing performance (%)

No of data predicted accurately by SVM
= × 100. (29)
Total data
3 Result and Discussions
In this study, the database of liquefaction potential based on SPT values is prepared
by using deterministic approaches. The database consists of 532 cases. The data
represents the SPT value data of 5 sites situated in different locations of India. For
the study, data were taken from various boreholes of certain depths from the ground
surface.
The database of liquefaction potential based on SPT values is prepared by evalu-
ating the factor of safety against liquefaction using 532 SPT values obtained from 5
sites. The factor of safety is the ratio of cyclic resistance ratio (CRR) to cyclic stress
ratio (CSR). Liquefaction is predicted to occur if the factor of safety is less than or
equal to 1. After evaluation, it is observed that there are 276 liquefied cases and 256
non-liquefied cases.
In this study, SVM is performed in R Studio with the help of the e1071 package
and kernlab package. The optimal model is found using different SVM classifiers
with the help of different kernels. For finding the optimal model determination of
certain parameters which were obtained using trial and error procedure. The details
are shown in Table 1.
Out of 532 cases, 372 cases were considered for training and 160 cases were
considered for testing the SVM model. In the training data, 196 cases were liquefi-
able cases and 176 cases were non-liquefiable cases. In testing data, 80 cases were
liquefiable cases and 80 cases were non-liquefiable cases. The optimal model is
chosen which gave the least number of misclassification or better accuracy.
It can be seen that the optimal model is obtained using c-SVM classifier with
radial kernel. In training process, parameters obtained cost = 35 and gamma =
0.49 gave accuracy of 99.19%. When the same parameters were used in the testing
process, it showed an accuracy of 91.25%. In the training process, out of 372 cases
only 3 cases were misclassified whereas in the testing process it misclassified 14
cases out of 160 cases, achieving an overall accuracy of 96.8%. The summary of the
results is presented in Table 2. The Bound constraint c-SVM, Wetson and Watkin
SVM and Crammer–Singer SVM have very good accuracy both for training and
testing, indicating the fact that these methods have good generalization capability.
Irrespective of the different techniques used in the present analysis for liquefaction
analysis, the radial and polynomial kernels yield very good results in terms of the
accuracy of the model.
4 Conclusions
The liquefaction data, based on SPT data obtained from the deterministic approach,
is analyzed to develop the SVM model.
• The optimal model is achieved using c-SVM classifier and radial kernel function
using cost parameter = 35 and gamma parameter = 0.49. The trained SVM model
correctly classified 369 out of 372 cases of training data (99.19% accuracy) and
146 out of 160 cases of testing data (91.25% accuracy).
• Radial and Polynomial Kernels are found to be more useful in terms of the accuracy
of the model, irrespective of the classifier considered.
Table 1 Parameters used to develop SVM model using liquefaction data

Classifier Kernels Parameter used No. of support vectors
C-SVM Radial cost = 35 143
gamma = 0.49
Polynomial cost = 2 140
gamma = 0.47
degree = 3
Linear cost = 1 170
gamma = 0.142
Sigmoid cost = 24 179
gamma = 0.01
coeff = 0
nu-SVM Radial nu = 0.16 153
gamma = 0.5
Polynomial nu = 0.16 127
gamma = 0.08
degree = 5
Linear nu = 0.46 174
gamma = 0.01
Sigmoid nu = 0.46 175
gamma = 0.01
coeff = 0
Bound constraint c-SVM Polynomial cost = 1 169
degree = 1
scale = 1
offset = 1
Linear cost = 1 169
Radial cost = 1 190
sigma = 0.153
Hyperbolic tangent cost = 1 159
scale = 1
offset = 1
Laplace cost = 1 218
sigma = 0.190
Bessel cost = 1 186
sigma = 1
order = 1
degree = 1
Anova cost = 1 167
sigma = 1
degree = 1
Weston–Watkins multi-class Polynomial cost = 1 169
SVM degree = 1
scale = 1
offset = 1
(continued)
Table 1 (continued)
Classifier Kernels Parameter used No. of support vectors
Linear cost = 1 169
Radial cost = 1 190
sigma = 0.153
scale = 1
offset = 1
sigma = 0.190
Bessel cost = 1 186
sigma = 1
order = 1
degree = 1
Anova cost = 1 167
sigma = 1
degree = 1
Crammer–Singer multi-class Polynomial cost = 1 167
SVM degree = 1
scale = 1
offset = 1
Linear cost = 1 166
Radial cost = 1 180
sigma = 0.153
scale = 1
offset = 1
sigma = 0.190
Bessel cost = 1 175
sigma = 1
order = 1
degree = 1
Anova cost = 1 160
sigma = 1
degree = 1
• The Bound constraint c-SVM, Wetson and Watkin SVM and Crammer–Singer
SVM have very good generalization ability.
Table 2 Summary of SVM classification results

Classifier Kernels Training Testing accuracy Overall accuracy
accuracy (%) (%) (%)
c-SVM Radial 99.19 91.25 96.8
Polynomial 92.47 85.625 90.41
Linear 86.56 87.5 86.84
Sigmoid 83.33 88.75 84.96
nu-SVM Radial 98.65 91.25 96.42
Polynomial 93.27 83.125 90.22
Linear 82.8 88.75 84.58
Sigmoid 83.06 88.75 84.77
Bound Constraint Linear 53.76 51.875 53.19
c-SVM Polynomial 83.33 87.5 84.58
Radial 87.09 90 87.97
Laplace 87.36 89.375 87.97
Bessel 86.56 90 87.6
Anova 85.48 88.125 87.6
Hyperbolic 58.87 59.375 59.02
tangent
Weston Watkin Linear 83.33 87.5 84.58
SVM Polynomial 83.33 87.5 84.58
Radial 87.09 90 87.97
Laplace 87.365 89.375 87.97
Bessel 86.56 90 87.6
Anova 85.48 88.125 86.27
Hyperbolic 59.14 56.87 58.45
tangent
Crammer Singer Linear 84.67 86.875 85.33
SVM Polynomial 83.87 86.875 84.77
Radial 87.63 90 88.35
Laplace 89.78 88.75 89.47
Bessel 86.3 88.125 86.84
Anova 86.56 88.75 87.22
Hyperbolic 68.55 71.875 69.54
tangent
References
1. Borthakur N, Dey AK (2020) Evaluation of group capacity of micropile in soft clayey soil from
experimental analysis using SVM-based prediction model. Int J Geomech 20(3):04020008
2. Cortes C, Vapnik V (1995) Support-vector network. Mach Learn 20
3. IS: 2131:1981(Reaffirmed 2002) Method for standard penetration test for soils (First Revision)
4. Ji J, Zhang C, Gui Y, Lü Q, Kodikara J (2017) New observations on the application of LS-SVM
in slope system reliability analysis. J Comput Civil Eng 31(2):06016002
5. Kurnaz TF, Kaya Y (2018) The comparison of the performance of ELM, BRNN, and SVM
methods for the prediction of compression index of clays. Arab J Geosci 11(24):770
6. Liao SSC, Whitman RV (1986) Catalogue of liquefaction and non-liquefaction occurrences
during earthquakes. Department of Civil Engineering, Massachusetts Institute of Technology
7. Samui P (2013) Liquefaction prediction using support vector machine model based on cone
penetration data. Front Struct Civil Eng 7:72–82
8. Schölkopf B, Smola A, Williamson R, Bartlett PL (1998) New support vector algorithms.
NeuroCOLT technical report NC-TR-98-031
9. Seed HB, Idriss IM (1971) Simplified procedure for evaluating soil liquefaction potential. J
Soil Mech Found Div 97(9):1249–1273
10. Weston J, Watkins C (1999) Multi-class support vector machines. In: Proceedings of ESANN-
99
11. Zhang Y, Dai M, Ju Z (2016) Preliminary discussion regarding SVM kernel function selection
in the twofold rock slope prediction model. J Comput Civil Eng 30(3):04015031
12. Zhang C, Ji J, Gui Y, Kodikara J, Yang SQ, He L (2016) Evaluation of soil-concrete interface
shear strength based on LS-SVM. Geomech Eng 11(3):361–372
Assessment of Seismic Liquefaction
of Soils Using Swarm-Assisted
Optimization Algorithm
T. Vamsi Nagaraju, Ch. Durga Prasad, Babloo Chaudhary, and B. M. Sunil
Abstract Assessment of liquefaction potential of soils due to the earthquake has

been carried out in this research using the nature-inspired Metaheuristic swarm-
assisted algorithm (PSO). An assessment has been made on the basis of actual field
data from the previous research. The field data consists of 59 sets having variables
of total stress of soil (⌐o ), effective stress of the soil (⌐ o ), percentage fines, mean
size of soil particles (D50 ), standard penetration value (SPT), the equivalent dynamic
shear stress (Tav /⌐ o ), maximum horizontal acceleration at ground surface (a/g) and
the earthquake magnitude (M). PSO-based models were developed for both single
variable and multivariable linear approaches. The results revealed that for the assess-
ment of liquefaction of soils, the developed PSO models perform good estimations
in terms of the errors and convergent solution. And also, with a damping coefficient
and varying input variables, there is a significant improvement in the best solution.
These developed models can be useful for practicing engineers in the field.
Keywords Liquefaction · SPT · Dynamic shear stress · PSO algorithm
1 Introduction
Civil engineering infrastructure construction in the earthquake influence zone has

a major concern about the liquefaction potential of soils in the designing process
[1]. Liquefaction of the soil depends on many factors such as soil type, soil state,
water table and earthquake intensity [2–4]. In recent times and past many civil
infrastructure damages occurred during earthquakes mostly by the liquefaction of
soils. The efficient way of determination of the liquefaction potential of soils in the
field is Standard Penetration Test (SPT) and Cone Penetration Test (CPT), and in
T. V. Nagaraju (B) · Ch. D. Prasad

Department of Civil Engineering, S.R.K.R. Engineering College, Bhimavaram 534204, India
e-mail: Varshith.varma@gmail.com; tvnraju@srkrec.edu.in
B. Chaudhary · B. M. Sunil
Department of Civil Engineering, National Institute of Technology Karnataka, Mangalore
575025, India
296 T. V. Nagaraju et al.
the laboratory is Cyclic Stress Ratio (CSR) for better understanding the liquefac-
tion of soils [5, 6]. Nowadays, many researchers were interested to develop predic-
tion tools to estimate the many geotechnical parameters including the liquefaction
potential of soils. From the very past, many researchers have been contributed
correlations for liquefaction of soils based on the field SPT and CPT data sets
[7, 6]. Later, correlations were developed by using laboratory data. However, these
correlations not provided stable and accurate predictions because of having many
constraints. To overcome the constraints with promising predictions, artificial intelli-
gence techniques have emerged with complex structures. The gaining importance in
artificial intelligence stems from their ability to find the best solutions in an effective
and efficient way. In artificial intelligence techniques, few of them are inspired by
neurobiological studies and others from nature-inspired algorithms. The prediction
models were developed based on sufficient input–output data sets [8, 9]. Further,
it keeps generating and regenerating new data. To use artificial intelligence to its
fullest potential in a civil engineering discipline, it can optimize material utiliza-
tion, reduce cumbersome testing, reduce time-consuming tasks, assess properties of
materials, and reduce construction costs [8, 9]. The developed prediction models
can estimate the behaviour of materials by considering the required data information
obtained from the experimental results. In this regard, some research was carried out
to develop models for the prediction of liquefaction potential of soils. Over time, only
mathematical methods are used for developing correlations and equations for lique-
faction potential. These methods have no convergent findings due to the complex
structure. This has recently motivated researchers for the application of optimization
techniques for predicting the liquefaction potential of soils, and a brief account of
results is presented below.
Young-Su and Byung-Tak [10] considered undrained cyclic triaxial testing results
of sand were used as input data sets for developing prediction models to predict
Cyclic Resistance Ratio (CRR) of soils using Artificial Neural Network (ANN).
Hanna et al. [11] adapted General Regression Neural Network (GRNN) models that
were developed between the seismic parameters and soil properties. The obtained
results were convergent with 94%.
Bagheripour et al. [12] proposed a model to estimate the liquefaction potential
using the Genetic Algorithm (GA). The main advantage of this model is that it relates
both the Factor Of Safety (FOS) and the probability of liquefaction, which helps in the
estimation of liquefaction potential when deterministic approaches are considered.
Samuli and Sitharam [13] developed prediction models for liquefaction susceptibility
using machine learning techniques such as ANN and Support Vector Machine (SVM).
In the case of liquefaction susceptibility prediction, the SVM approach is more
capable over the ANN models based on standard penetration data.
This paper presents the prediction models for the assessment of liquefaction poten-
tial of soils using nature-inspired Particle swarm Optimization Algorithm (PSO). For
developing the estimation models total stress of soil (⌐o ), effective stress of the soil
(⌐ o ), percentage fines, mean size of soil particles (D50 ), standard penetration value
(SPT), the equivalent dynamic shear stress (Tav /⌐ o ), maximum horizontal accelera-
tion at the ground surface (a/g) and the earthquake magnitude (M) were considered
Assessment of Seismic Liquefaction of Soils … 297
as input variables and the output variable is Liquefaction Potential of Soils (LPS)
in the field. Furthermore, the data sets used in this paper were compiled from the
previous literature. This paper also explores the importance of PSO in the field of
geotechnical engineering.
1.1 Overview of Particle Swarm Optimization
Many optimization algorithms were emerging day by day, but very few optimization
algorithms applications got a great response from all the disciplines such as Genetic
Algorithm (GA), Particle Swarm Optimization (PSO), Flower Pollination Algorithm
(FPA), Bat Algorithm (BA) and Invasive Weed Optimization Algorithm (IWO) [14].
PSO algorithm is nature-inspired algorithm, it was proposed by Kennedy and Eber-
hart in the year 1995 [15]. PSO has been widely used in many disciplines including
civil engineering [8, 9, 14]. The main advantage of PSO is easy for implementation
and no requirement of gradient information. PSO mechanism works in a way that a
group of birds searching randomly for food in two-dimensional space. The position
of each bird is indicated as a solution to a given problem. Initially, these positions
are randomly searched. In case if they are not found at an initial position they shift
to the next random position where it found food, this is the best solution for that
task. In PSO, from the definition, each element has the following properties, let yik
denotes the element position vector for finding the solutions and next position for
each element finds randomly as
yik+1 = yik + vik+1 (1)
where vik+1 is the particle velocity vector which processes the optimization. The
position of every individual bird is yik and the velocity is vik . The best position of
each particle is pbest and the overall optimized position is gbest . Then the velocity of
each organism in the swarm is updated as
Vik+1 = ωV ik + a1 m 1 ( p best i − xik ) + a2 m 2 (gbest i − xik ) (2)
where, a1 and a2 are acceleration constants mainly helps in the movement of a particle
in a given iteration; m 1 and m 2 are random numbers; ω is the inertia weight which
helps in controlling the behaviour of the object in direction of convergence.
In this study, for the classification of liquefaction potential, single variable and
multivariable approaches were adopted. Moreover, the multilinear prediction model
equation was developed for predicting the liquefaction potential of soil in the field.
For this liquefaction estimation, 8 input variables were considered such as earthquake
magnitude (X1), total stress of soil (X2), effective stress of the soil (X3), standard
penetration value (X4), maximum horizontal acceleration at the ground surface (X5),
the equivalent dynamic shear stress (X6), percentage fines (X7) and mean size of soil
particles (X8), these variables were multiplied by respective coefficients such as c1 ,
c2 , c3 , c4 , c5 , c6 , c7 and c8 , and additional coefficient co was considered. Equation (3)
is expressed as the developed prediction multivariable equation for classification of
field liquefaction potential.
Liq(est) = c1 · X 1 + c2 · X 2 + c3 · X 3 + c4 · X 4 + c5 · X 5
+ c6 · X 6 + c7 · X 7 + c8 · X 8 + co (3)
where c1 , c2 , c3, c4 , c5 , c6 , c7 and c8 are coefficients used to evaluate liquefaction

potential classification. For classification of liquefaction potential, if the value is
found to be (+1) then say yes (Y) for the possibility of liquefaction in field or in case
value is found to be (−1) then say no (N) for the possibility of liquefaction in field.
2 Database Compilation
The data sets were taken from the Tokimatsu and Yoshimi [16] case study report on
soil liquefaction. The data sets include the field behaviour of the soil liquefaction
during the various earthquakes experienced in the past such as Mino-Owari (1891),
Kanto (1923), Tohnankai (1944), Fukui (1948), Niigata (1964), Tokachi-Oki (1968),
Miyagiken-Oki (1978), San Francisco (1957), Alaska (1964), San Fernando (1971),
Guatemala (1976), and Imperial Valley (1979). Moreover, this case study data have
been widely used for developing prediction models based on different approaches
[17, 18]. For the classification of these data sets, a MATLAB code was developed to
evaluate the data sets. These evaluation processes consist of a series of steps such as
objective function, boundary of iterations and acceleration constants. Few data sets
were tabulated in Table 1 [17].
The objective of this section is to evaluate and discuss the prediction models of
liquefaction potential developed by using PSO. The results of the PSO models were
developed by using seven different input variables as mentioned in the previous
section. The data set includes 59 samples. Initially, an attempt has been made to
study the influence of individual variables for predicting liquefaction potential in the
field. Later, by considering the input variable as mentioned in the previous section
were used to develop a linear prediction model for predicting liquefaction potential
in the field (X9) using PSO.
Table 1 Data set for liquefaction potential classification

Earthquake Total Effective SPT a/g Dynamic Fines D50 (mm) Liquefaction
magnitude stress stress value, shear (%) behaviour in
(kPa) (kPa) N stress field
7.9 186.4 96.1 20.0 0.32 0.36 0.0 0.46 Y
7.9 130.5 81.4 10.0 0.32 0.32 5.0 0.28 Y
7.9 93.2 73.6 5.7 0.2 0.16 20.0 0.20 N
7.9 149.1 100.1 2.0 0.2 0.18 0.33 0.15 N
8.0 89.3 59.8 8.0 0.2 0.19 10.0 0.40 Y
8.0 130.5 81.4 10.0 0.16 0.16 5.0 0.28 Y
7.3 66.7 35.3 7.0 0.35 0.39 35.0 0.13 N
7.3 128.5 63.8 20.0 0.40 0.45 0.0 0.45 N
7.5 130.5 71.6 8.0 0.16 0.17 2.0 0.30 Y
7.5 84.2 46.1 6.0 0.16 0.18 0.0 0.40 Y
6.7 61.8 34.4 5.0 0.12 0.12 5.0 0.70 Y
6.7 74.6 49.1 4.0 0.12 0.10 10.0 0.15 N
5.5 111.8 48.1 6.0 0.19 0.10 30 0.20 Y
8.3 56.9 53.0 10.0 0.16 0.22 10.0 0.20 Y
6.6 72.6 86.3 9.0 0.45 0.29 20.0 0.10 Y
7.5 72.6 28.4 8.0 0.14 0.17 3.0 1.00 Y
6.6 107.9 51.0 31.0 0.6 0.45 11.0 0.12 N
6.6 72.6 46.1 7.0 0.2 0.21 34.0 0.09 Y
6.7 118.7 66.7 10.0 0.10 0.09 0.0 0.60 N
6.7 93.2 68.7 9.0 0.14 0.10 20.0 0.15 N
3.1 Effect of Individual Variables on Susceptibility

of Liquefaction Potential in Field
All the input variables X1, X2, X3, X4, X5, X6, X7 and X8 were individually
considered for the prediction of field susceptibility of liquefaction using particle
swarm optimization. Figure 1a–h shows the classification results of liquefaction
susceptibility.
Figure 1a–h shows the relative individual parameters’ influence on the liquefaction
potential assessment by considering each individual parameter as input variables and
liquefaction potential as output variable. All 8 parameters have significant importance
in the liquefaction potential. However, for developing predicting model considering
individual input parameter exhibits complexity in classification of problem.
2 2
1 1
0 0
X9
X9
-1 -1
-2 -2
4 5 6 7 8 9 0 200 400 600 800
X1 X2
a) Earthquake magnitude Vs LPS b) Total stress of soil Vs LPS
2 2
1 1
0 0
X9
X9
-1 -1
-2 -2
0 50 100 150 0 10 20 30 40
X3 X4
c) Effective stress of the soil Vs LPS d) Standard penetration value Vs LPS
Fig. 1 Single variable classification of liquefaction potential based on PSO

2 2
1 1
0 0
X9
X9
-1 -1
-2 -2
0 0.2 0.4 0.6 0.8 0 10 20 30 40
X5 X7
e) a/g Vs LPS f) Equivalent dynamic shear stress Vs LPS
2 2
1 1
0 0
X9
X9
-1 -1
-2 -2
0 0.2 0.4 0.6 0.8 0 0.5 1 1.5 2
X6 X8
g) Percentage fines Vs LPS h) Mean size of soil particles Vs LPS
Fig. 1 (continued)
3.2 Effect of Multivariable Approach on Susceptibility

of Liquefaction Potential in Field
Multivariable linear prediction was developed using PSO. All the input variables as
previously mentioned were considered for estimating liquefaction potential in the
field. Figure 2 shows the classification of liquefaction potential in the field.
1.5
1
Predicted case value
0.5
-0.5
-1
-1.5
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Actual case value
Fig. 2 Multivariable classification based on PSO
The PSO-based linear regression approach results indicate that the PSO model has
been successful in predicting the susceptibility of liquefaction in the field. The devel-
oped relationship between the input data variables and the output of the liquefaction
possibility estimation has 90% convergent predictions.
For predicting the field behaviour of liquefaction of soils using PSO, the Eq. (4)
is developed for classification.
Liq(est) = 0.5023 · X 1 − 0.001 · X 2 + 0.00896 · X 3

− 0.08156 · X 4 + 0.5365 · X 5 + 3.0703 · X 6
− 0.01316 · X 7 − 0.12729 · X 8 − 3.9882 (4)
The developed equation, as aforementioned, classification of liquefaction is

expressed in terms of possibility of filed liquefaction (+1) and not possibility of
field liquefaction (−1). The PSO equation provides the best convergent results.
4 Conclusions
In this study, based on the field liquefaction data of Tokimatsu and Yoshimi’s [16]
case study report on soil liquefaction, an optimization algorithm PSO is used to
estimate the behaviour of soil liquefaction in the field. The conclusions are drawn
from this computational study as follows:
(1) The obtained classification results indicated that multilinear variable model is
best suitable compared to the single variable classification models for estimating
liquefaction potential in the field.
(2) Particle swarm optimization can be an alternative method for the classification
of liquefaction potential in the field with 90% convergent results.
(3) Although many approaches are in practice for predicting liquefaction potential,
the PSO prediction model would be easy to implement and no requirement for
gradient information.
References
1. Seed HB (1987) Design problems in soil liquefaction. J Geotech Eng 113(8):827–845

2. Ishibashi I, Sherif MA, Cheng WL (1982) The effects of soil parameters on pore-pressure-rise
and liquefaction prediction. Soils Found 22(1):39–48
3. Juang CH, Zhang J, Khoshnevisan S, Gong W (2017) Probabilistic methods for assessing soil
liquefaction potential and effect. Geo-Risk 122
4. Ling YH (1992) Liquefaction of granular soils with non-cohesive and cohesive fines. In:
Proceedings of the tenth world conference on earthquake engineering, 19–24 July 1992, vol 3.
Madrid, Spain. CRC Press, p 1491
5. Schneider JA, Moss RES (2011) Linking cyclic stress and cyclic strain based methods for
assessment of cyclic liquefaction triggering in sands. Géotech Lett 1(2):31–36
6. Zhou J, Li E, Wang M, Chen X, Shi X, Jiang L (2019) Feasibility of stochastic gradient boosting
approach for evaluating seismic liquefaction potential based on SPT and CPT case histories. J
Perfor Construct Facilit 33(3):04019024
7. Seo MW, Olson SM, Sun CG, Oh MH (2012) Evaluation of liquefaction potential index along
western coast of South Korea using SPT and CPT. Mar Georesour Geotechnol 30(3):234–260
8. Nagaraju TV, Prasad CD, Raju MJ (2020) Prediction of California bearing ratio using particle
swarm optimization. In: Soft computing for problem solving. Springer, Singapore, pp 795–803
9. Nagaraju TV, Prasad CD, Murthy NGK (2020) Invasive weed optimization algorithm for predic-
tion of compression index of lime-treated expansive clays. In: Soft computing for problem
solving. Springer, Singapore, pp 317–324
10. Young-Su K, Byung-Tak K (2006) Use of artificial neural networks in the prediction of
liquefaction resistance of sands. J Geotech Geoenviron Eng 132(11):1502–1504
11. Hanna AM, Ural D, Saygili G (2007) Neural network model for liquefaction potential in soil
deposits using Turkey and Taiwan earthquake data. Soil Dyn Earthquake Eng 27(6):521–540
12. Bagheripour MH, Shooshpasha I, Afzalirad M (2012) A genetic algorithm approach for
assessing soil liquefaction potential based on reliability method. J Earth Syst Sci 121(1):45–62
13. Samui P, Sitharam TG (2011) Machine learning modelling for predicting soil liquefaction
susceptibility. Nat Hazards Earth Sys Sci 11(1):1–9
14. Yang XS, Gandomi AH, Talatahari S, Alavi AH (eds) (2012) Metaheuristics in water,
geotechnical and transport engineering. Newnes
15. Kennedy J, Eberhart R (1995) Particle swarm optimization. IEEE, 0-7803-2768-3/95, pp 1942–
1948
16. Tokimatsu, K., & Yoshimi, Y. (1983). Empirical correlation of soil liquefaction based on SPT
N-value and fines content. Soils and Foundations, 23(4), 56–74.
17. Goh AT (1994) Seismic liquefaction potential assessed by neural networks. J Geotech Eng
120(9):1467–1480
18. Sen G, Akyol E (2010) A genetic-algorithm approach for assessing the liquefaction potential
of sandy soils. Nat Hazards Earth Syst Sci 10(4)
Liquefaction Potential Evaluation:
Necessity of Developing Energy
Correction Factor for SPT N-value
S. M. Ali Jawaid
Abstract The liquefaction analyses proposed by various researchers such as Seed

et al. (1971) and Seed and Idriss (1985) are based on Cyclic Stress Ratio (CSR) and
Cyclic Resistance Ratio (CRR). The Cyclic Resistance Ratio (CRR) is dependent on
corrected SPT blow count N60 . Since hammer energy efficiency in India is different
in comparison to the USA, it is necessary to develop an energy correction factor for
India as the same was carried out by China and Japan. In this paper, the I.S. code
procedure of liquefaction evaluation as well as energy correction factor developed
for various countries were reviewed, and it is found that there is a need to develop
one such energy correction factor for India.
Keywords Liquefaction · Earthquake · SPT energy correction
1 Introduction
About 54% of land in India is vulnerable to earthquakes. The U.N. has predicted that
around 200 million city dwellers in India will be exposed to storms and earthquakes
by the year 2050. Most of the cities/towns in India do not have planned development
and thus, it is the need of the hour to carryout the micro-zonation of area liable to
liquefy during any future earthquake in order to get prepared well ahead of an actual
disaster. This will help us in the adaptation of suitable mitigation techniques such as
deep dynamic compaction and vibro-compaction/replacement.
The purpose of this paper is to review the I.S. code procedure documented in
IS:1893–2016 for the evaluation of liquefaction potential using SPT blow counts
(N-value). The I.S. code procedure is based on a “Simplified Method” developed by
Seed and Idriss [8].
S. M. A. Jawaid (B)
M. M. M. University of Technology, Gorakhpur, U.P., India
e-mail: smaj@rediffmail.com
306 S. M. A. Jawaid
2 I.S. Code Procedure [I.S. 1983(Part I)-2016]
To evaluate the liquefaction potential of a site, the subsurface evaluation data should
include SPT blow count (N-value) and the location of the water table. The cyclic
stress ratio (CSR) induced by the earthquake may be evaluated using the following
equation:

amax σv0
CSR = 0.65 · rd (1)
g σv0
where
amax = Peak Ground Acceleration (PGA) which may be taken equal to seismic
zone factor,
g = Acceleration due to gravity and
rd = Stress reduction factor
The stress reduction factor rd may be calculated by the following relationship
proposed by Lio and Whiteman [5]:

1 − 0.00765 Z 0 < Z ≤ 9.15 m
rd =
1.174 − 0.0267 Z 9.15 < Z ≤ 23.0 m
where Z is the depth below the ground surface in meters.

In order to obtain the cyclic resistance ratio (CRR), the SPT blow count must be
converted to N60 using the following formula:
N60 = N C60 (2)
where
C60 = CHT . CHW .C55 . CRL . CBD (3)
If the SPT is conducted as per IS 2131 [2], it is suggested to take C60 = 1.0.
The computed N60 is normalized using the overburden correction factor CN
(N1 )60 = CN N60 (4)
where

Pa
CN = ≤ 1.7 (5)
σv0
Liquefaction Potential Evaluation: Necessity of Developing Energy … 307
where
σ’v0 = effective overburden pressure = 100 kPa
Pa = atmospheric pressure measured in the same unit.
(N1 )60 must also be corrected for fines as per the following equation:
(N1 )60 corrected = A + B (N1 )60 (6)
where A and B are coefficients determined from the following relationships:
A = 0 for FC ≤ 5% (7)

A = exp 1.76− 190/FC2 for 5% < FC < 35% (8)
A = 5.0 for FC ≥ 35% (9)
B = 1.0 for FC ≤ 5% (10)

B = 0.99− FC1.5 /1000 for 5% < FC < 35% (11)
B = 1.2 for FC ≥ 35% (12)
The cyclic resistance ratio (CRR) may be either estimated from Fig. 1 for an
earthquake of magnitude 7.5 or using Rauch [6]’s relationship:
1 (N1 )60 50 1
CRR = + + − (13)
34 − (N1 )60 135 10(N1 )60 + 45
2 200
The factor of safety against liquefaction may be defined as
Factor of Safety against Liquefaction = CRR/CSA ≥ 1.00 (14)
3 Discussion
If the Standard Penetration Test is conducted as per IS 2131 [2], the energy transfer
to the drill rod is about 60% and then C60 may be taken as 1.0 [3]. The recommended
standard SPT equipment/procedure as per IS 2131 [2] is given in Table 1.
As per the requirement given in Table 1, two wraps of rope around the pulley and
blow count rates of 30–40 blows per minute are recommended which is not adopted
308 S. M. A. Jawaid
Fig. 1 Graph between CRR

and (N1 )60 for Sand for a
magnitude 7.5 earthquake
(courtesy [3, 9])
Table 1 Recommended standardized SPT equipment as per IS 2131 [2]

S. Element Detailed explanation
No.
1 Sampler “Standard split-spoon sampler with OD = 51 mm and ID =
35 mm (constant, that is, no room for liners in the barrel)”
2 Drill rods “A or AW type for depth less than 15.2 m; N or NW type
for greater depths. Standard (Safety Hammer) with
(a) weight = 63.5 kg
(b) drop Ht = 762 mm (delivered 60% of theoretical
free-fall energy)”
3 Rope “Two wraps of rope around the pulley”
4 Borehole “100–130 mm diameter rotary borehole with bentonite mud
for borehole stability”
5 Drill bit Upward deflection of drilling mud
6 Blow count rate “30 to 40 blows per minute”
7 Penetration resistant count “Measured over a range of 150 mm–450 mm of penetration
into the ground”
(Courtesy [3])
in most parts of India, thanks to the lack of automatic hammers in India. Normally,
SPT is carried out manually using unskilled labors. They use more wraps of rope as
well as the blow count rate is less than 15 blows per minute, thus taking C60 as unity
is not correct.
In U.S. testing practice, Energy Ratio (ER) of 60% is accepted as a reference value
for energy correction. ER depends on the type and weight of SPT hammer, anvil,
lifting mechanism and method of hammer release. If N-value is measured in any other
country using different weight, fall and lifting mechanism, hammer efficiency, etc.,
the correction should be made to SPT N-value for the difference in energy efficiency
using the following equation [4]:

ERm
(N1 )60 = Nm (15)
ERc
where Nm and ERm are the measured N-value and corresponding energy efficiency.
(N1 )60 and ERc are the N-value at an energy ratio of 60% and energy efficiency equal
to 60%. Since hammer efficiency and hammer used in India are different from the
one used in the USA, it is necessary to normalize the measured N-value using energy
ratio to get (N1 )60 .
Yoshimi et al. [11] had studied the effect of different methods of hammer release
on SPT blow counts. They found that blow counts measured with the mechanical trip
release system gave 13% lower blow counts than blow count measured with the rope
and pulley method. The relationship between N-value measured with a mechanical
trip system and manual system in China may be given by the following relationship
[10]:
Nmechanical = 0.83 Nmanual (16)
It shows that the rod energy ratio for manual type hammers is 20% less than that
of a mechanical hammer and thus energy ratio for hammers operated with the rope
and pulley system in China is about 50%. Seed et al. [9] had compiled the energy
ratio (ERm /ERc ) for different countries and the same is presented in Table 2.
Unfortunately, the standardized energy ratio (ERm /ERc ) is not available for Indian
SPT hammers. Gupta et al. [1] had proposed an energy ratio (the correction factor
for 60% rod energy) of 0.75. However, some researchers proposed an energy ratio
of 1.0 [3, 7]. Based on the above discussion, it is the need of the hour to correct the
measured N-value based on actual energy transfer. This can be achieved by measuring
the energy by the SPT analyzer at various sites and developing a correction factor
for 60% energy transfer for India.
Some SPT tests were carried at different locations in and around Gorakhpur
city using an automatic SPT hammer (make: AIMIL) by the author. The energy
calibration rod is attached in between the top SPT A-rod and hammer guide rod and
is then attached with a handheld unit of the SPT analyzer (make: Pile Dynamics,
USA). Energy measurement was carried out at every 1.0 m of penetration. The SPT
310 S. M. A. Jawaid
Table 2 Energy correction factors for different countries

Country Hammer type Hammer Release Estimated Correction Factor for
Rod Energy, % 60% Rod Energy
Japan Donut Free-fall 78 1.30
Donut Rope and Pulley with 67 1.12
special throw release
United States Safety Rope and Pulley 60 1.00
Donut Rope and Pulley 45 0.75
Argentina Donut Rope and Pulley 45 0.75
China Donut Free-fall 60 1.00
Donut Rope and Pulley 50 0.83
(After [9])
analyzer had recorded energy transmitted per blow of the hammer. Only the blows
for the last 30 cm increments were used in the analysis since these corresponding
blows count as the N-value. The SPT data and energy measurement per site were
recorded. It was found that the automatic hammer has energy efficiencies around
86.54%. The standard deviation was 0.869. Based on this study, it is suggested to
apply the correction factor (C60 ) of 0.86 for converting the measured SPT N-value
in terms of N60 in the case of an automatic hammer in India. These values are within
the range of correction factors reported in the literature for other countries. However,
a broad study is needed in order to develop a universally accepted correlation for
India.
4 Concluding Remarks
The Standard Penetration Test (SPT) is currently the most popular and economical
means of obtaining subsurface information. In order to reduce the significant vari-
ability associated with the SPT N-value, it is recommended that N-values be stan-
dardized to N60 . This standardization is to be achieved by correcting the measured
field N-values by the ratio of energy transfer to the standard 60% energy of an SPT
hammer. This requires knowing the performance characteristics of the SPT system.
Various researchers had proposed an energy ratio (the correction factor for 60%
rod energy) of 0.75–1.0 for India. It is the need of the hour to develop a standard
correction factor based on the measurement of actual energy transfer. This can be
achieved by measuring the energy by an SPT analyzer at various sites and developing
a correction factor for 60% energy transfer for India. However, based on a pilot study,
the author had found correction factors (CN ) of 0.86 in the case of automatic hammers
available in India. These values are within the range of correction factors reported in
the literature for other countries. However, a broad study is needed for India in order
to develop a universally accepted correlation for India.
References
1. Gupta S, Sundaram R, Gupta S (2008) Liquefaction susceptibility study at commonwealth

games village, Delhi. In: Proceedings of the Indian geotechnical conference 2008, vol 1.
Bangalore, pp 436–440
2. IS 2131 (2002) Methods for standard penetration test for soils. Bureau of Indian Standards,
New Delhi
3. IS 1893 (2016) Criteria for earthquake design of structures, vol 1. Bureau of Indian Standards,
New Delhi
4. Japanese Geotechnical Society (1998) Remedial measures against soil liquefaction. Balkema,
Rotterdam
5. Lio SSC, Whiteman RV (1986) Catalogue of liquefaction and non-liquefaction occurrences
during earthquake. Research Report, Department of Civil Engineering, MIT, Cambridge, Mass
6. Rauch AF (1998) Evaluating cyclic liquefaction potential using cone penetration test. Can
Geotech J 35(3):442–459
7. Saran S (2006) Soil dynamics and machine foundation. Galgotia Publications, New Delhi
Geotech Eng Div, ASCE, 97(9):1249–1273
9. Seed HB, Tokimatsu K, Harder LF, Chung RM (1985) The influence of SPT procedures in soil
liquefaction resistance evaluations. J Geotech Eng, ASCE, 111(2):1425–1445
10. ShiMing H (1982) Experience on standard penetration test. In: Proceedings of the second
European symposium on penetration testing, Amsterdam
11. Yoshimi Y, Tokimatsu K, Oh-oka H (1983) Comparision of SPT N-values by cat-and-rope
method and trip monkey method. In: Proceedings of the 18th annual meeting, JSSMFE
Assessment of Soil Liquefaction Safety
Factors Under Different Earthquake
Magnitudes
Shiva Shankar Choudhary, Avijit Burman, and Sanjay Kumar
Abstract In the present study, seismic soil liquefaction in terms of factor of safety
against liquefaction (FS) is evaluated by the IS Code: (Criteria for earthquake resistant
design of structures. Bureau of Indian Standards, New Delhi [5]) (Part-I, 2016) for
a site of IIT Patna campus. The FS values against liquefaction are evaluated under
three different earthquake magnitudes, namely, M w = 6.0, 6.5 and 7.0. A design peak
ground acceleration of 0.24 g used as the Patna city lies in the Zone IV of the seismic
zoning map of India as per IS code. In this paper, an evaluation of the severity of
liquefaction in the form of liquefaction potential index (LPI) is also determined. The
LPI is determined at a single borehole location from the obtained factors of safety
(FS) to predict the potential of liquefaction to cause damage at the site of interest. The
evaluated FS and LPI values from IS Code are compared with Idriss and Boulanger
(Soil Dyn Earthq Eng 26:115–130 [4]) procedures to investigate the liquefaction
behavior for cohesionless soils. The FS and LPI results of both SPT-based semi-
empirical procedures indicate that the soil liquefaction probability increases during
earthquakes with magnitude ≥7.0 at this site. It is also observed that the values of
FS from IS code match with Idriss and Boulanger (Soil Dyn Earthq Eng 26:115–130
[4]) results under high earthquake magnitude.
Keywords Earthquake magnitudes · IS code · Liquefaction potential index · Soil

liquefaction
S. S. Choudhary (B) · A. Burman · S. Kumar

National Institute of Technology Patna, Patna 800005, Bihar, India
e-mail: shiva@nitp.ac.in
A. Burman
e-mail: avijit@nitp.ac.in
S. Kumar
e-mail: sanjay@nitp.ac.in
314 S. S. Choudhary et al.
1 Introduction
Soil liquefaction can be defined as a temporary loss in shear strength of satu-

rated sandy and silty soil due to the significant swelling in the pore-water pressure
loudening under the influence of cyclic/dynamic loading generated due to incoming
stress waves reaching the site in the form of seismic loading. Liquefaction is gener-
ally observed in the case of saturated cohesionless soil; however, the clayey soil may
also undergo huge deformation/spreading. Liquefaction could cause serious damage
to the buildings, earth structures, and sloping grounds due to the loss of strength of
the soil, which eventually leads to the failure of the structure. The failure of the soil
may be in the form of vertical ground settlements and slope instability, sand boil,
and lateral spreading. To quantify the severity of soil liquefaction an index known
as liquefaction potential index (LPI) could be computed for the determination of the
degree of severity of liquefaction.
The liquefaction potential index (LPI) and factor of safety against liquefaction
(FS) provide a technique to assess the level of risk of liquefaction-induced damage
to the foundation of a structure. It is the basic and initial investigation from a geotech-
nical point of view and it is more important when the structural project site is under a
higher seismic zone or near river/sea. The potential for soil liquefaction to occur at a
certain depth at a site is quantified in terms of the factors of safety against liquefaction
(FS). Seed and Idriss [9] suggested a simplified approach to evaluate the liquefaction
resistance of soils in terms of factor of safety against liquefaction by taking the ratio of
the capacity of a soil element to resist soil liquefaction to the seismic demand imposed
on it. The liquefaction potential index (LPI) as proposed by Iwasaki et al. [6] was also
used to illustrate the liquefaction vulnerability, which takes into account the effect of
thickness and depth along with the FS of the liquefiable layer. Capacity to resist soil
liquefaction is computed as the cyclic resistance ratio (CRR), and seismic demand
is computed as the cyclic stress ratio (CSR). Factor of safety of a soil layer can be
calculated with the help of several field tests such as standard penetration test (SPT),
conic penetration test (CPT), Becker penetration test (BPT), and shear wave velocity
(V s ) test. Factor of safety along the depth of soil profile is generally calculated using
the surface-level peak ground acceleration (PGA), earthquake magnitude (Mw), and
SPT data, namely, SPT blow counts (N), overburden pressure, fines content (FC),
liquid limits and grain size distribution which is discussed by many researchers [7,
10, 12, 11]. Bao et al. [1] and Bwambale and Andrus [2] analyzed and discussed the
characteristics of the newly developed methods including the mitigating mechanism
and effectiveness, and possible executive problems to understand the progress of
liquefaction resistance. Olson et al. [8] performed a broad parametric study to assess
the viability of using nonlinear site response analysis with validated constitutive
and pore-water pressure generation models to evaluate level-ground liquefaction. In
this study, the nonlinear site response analysis with pore-water pressure generation
correctly predicted liquefaction for dynamic centrifuge tests and field cases, demon-
strating that this approach can assess level ground liquefaction while avoiding highly
uncertain correction factors required in the cyclic stress method. An approach for
Assessment of Soil Liquefaction Safety Factors … 315
soil liquefaction evaluation using a probabilistic method based on the SPT databases
has been presented by Goharzay et al. [3]. In this paper, the uncertainties of the
parameters for liquefaction probability have been taken into account. A calibrated
mapping function is developed using Baye’s theorem in order to capture the failure
probabilities in the absence of the knowledge of parameter uncertainty.
To study the liquefaction vulnerability with different earthquake magnitudes for
possible building construction and design at the Indian Institute of Technology Patna
(IITP) site, initial liquefaction check is done in terms of FS and LPI. Patna (Bihta)
city falls under seismic zone IV and established near the Ganga River. So, it is abso-
lutely necessary to make foundation soil improve before construction to eliminate
the liquefaction risk. There are many empirical relations are developed based on
in situ and laboratory test results. In the current study, two different SPT-based semi-
empirical procedures [4, 5] are used for the liquefaction vulnerability check under
earthquake magnitude (M w ) of 6.0, 6.5, and 7.0.
2 Site Location and Soil Condition
Soil sample of Indian Institute of Technology Patna campus site (located at Bihta
and 35 km from Patna) is investigated. The borehole site lies in the latitude of 25°
32 03.1 North and longitude of 84° 50 54.5 East. The borehole location and soil
property with varying depth is presented in Fig. 1 and Table 1, respectively.
At this site, many borehole tests are performed. However, in this study, one bore-
hole test data is used for the investigation of soil liquefaction potential. The standard
penetration test (SPT) is performed up to the depth of 15 m in order to determine the
SPT-N values in the field. The laboratory tests are also performed on collected soil
Fig. 1 Borehole location at IIT Patna campus site

Table 1 Soil properties at IIT Patna campus

Depth (m) Fine Content, FC (75 η) Ud (g/cc) W (%) SPT-N value
0–1.5 6.50 1.570 12.00 8
1.5–3.0 3.75 1.581 13.50 12
3.0–4.5 5.75 1.602 14.10 17
4.5–6.0 0.75 1.612 8.20 24
6.0–7.5 6.50 1.588 9.30 26
7.5–9.0 4.50 1.595 10.30 29
9.0–10.5 1.75 1.615 7.50 30
10.5–12.0 1.00 1.620 8.50 32
12.0–13.5 88.60 1.534 23.25 32
13.5–15.0 87.30 1.512 24.53 30
samples at the National Institute of Technology Patna (NIT Patna) to determine the
soil properties at different depths. From the SPT and laboratory tests, it is found that
the subsurface soil layers mainly consist of sandy soil.
3 Evaluation of Soil Liquefaction Potential
The empirical relations based on SPT and CPT are widely used in practice for the
determination of factors of safety against liquefaction (FS) and liquefaction potential
index (LPI). In this paper, empirical relations based on the SPT test [4, 5] are used.
The values of FS are determined by comparing the seismic demand expressed in
terms of cyclic stress ratio (CSR) to the capacity of liquefaction resistance of the soil
expressed in terms of cyclic resistance ratio (CRR). CSR distinguishes the seismic
demand produced by a given earthquake and it is evaluated from peak ground surface
acceleration that is based upon the ground motions. CRR requires fines content (FC)
of the soil to correct updated SPT blow count (N 1 )60 to an equivalent clean sand
standard penetration resistance value (N 1 )60cs .
From IS Code: [5] (Part-I), 2016,

amax σv
CSR = 0.65 rd (1)
g σv
1 (N1 )60cs 50 1
CRR = + + 2 − (2)
34 − (N1 )60cs 135 10 × (N1 )60cs + 45 200
From Idriss and Boulanger [4],


amax σv 1 1
CSR = 0.65 rd (3)
g σv MSF Kσ
2 3 4
(N1 )60cs (N1 )60cs (N1 )60cs (N1 )60cs
CRR = exp + + + − 2.8 (4)
14.1 126 23.6 25.4
where Amax = peak horizontal ground acceleration; = acceleration due to gravity;

r d = depth-dependent stress reduction factor; σ v and σ v’ = total vertical overburden
stress and effective vertical overburden stress, respectively, at a given depth below the
ground surface; MSF = magnitude scaling factor and K σ = overburden correction
factor. For the soil profiles with the depth less than 20 m, LPI is calculated using the
following expression:

n
Fi = 1 − FSi for FSi < 1.0
LPI = ωi Fi Hi with (5)
i=1
Fi = 0 for FSi ≥ 1.0
where H i = thickness of the discretized soil layers; n = number of layers;

F i = liquefaction severity for i-th layer; FS i = factor of safety for i-th layer; wi
= weighting factor (=10–0.5 zi ); and zi = depth of i-th layer.
In this study, FS and LPI for earthquake magnitudes (Mw) of 6.0, 6.5, and 7.0 are
determined with the use of SPT-based IS code: [5], 2016 and Idriss and Boulanger [4]
methods. For calculation of soil liquefaction potential, the earthquake magnitudes
with a peak ground acceleration level of 0.24 g are considered. Two basic definitions
of factors of safety against liquefaction (FS) for soil liquefaction potential are used
in the present study: liquefiable (FS ≤ 1.0) and non-liquefiable (FS > 1.0). However,
it cannot be established that FS just below 1.0 will certainly liquefy or FS above
1.0 will not, as it may vary based on the method adopted and the level of safety
desired. From in situ and laboratory tests, the typical subsoil profiles of the existing
site indicate sandy soil and the groundwater table is found at a depth of 3.0 m.
The type of soil most susceptible to liquefaction is one in which the resistance to
deformation is mobilized by friction between particles. The LPI and FS with three
different earthquake magnitudes are summarized in Tables 2 and 3 for IS code: [5]
and Idriss and Boulanger [4], respectively. It is found from Tables 2 and 3 that the
values of factor of safety against soil liquefaction are decreases with the increase of
seismic magnitude. This type of increasing pattern indicates that the soil liquefaction
vulnerability increases with an increase in the intensity of seismic events.
The FS values with M w = 6.0 and 6.5 indicate no zone of liquefaction and
M w = 7.0 indicate two zones of potential liquefaction in the poorly graded sand
layer at depths of 3.0 m from ground surface as per IS code. However, in case of
Table 2 Factors of safety (FS) against soil liquefaction potential [5]

Depth CSR CRR6 CRR6.5 CRR7 FS LPI
(m) Mw = Mw = Mw = Mw = Mw = Mw =
6 6.5 7 6 6.5 7
1.5 0.15 0.18 0.15 0.15 1.19 0.97 0.80 – – 2.71
3.0 0.15 0.22 0.18 0.15 1.45 1.18 0.98 – – 0.27
4.5 0.18 0.29 0.24 0.19 1.57 1.28 1.06 – – –
6.0 0.21 0.45 0.37 0.30 2.17 1.76 1.46 – – –
7.5 0.22 0.46 0.38 0.31 2.08 1.69 1.40 – – –
9.0 0.23 0.48 0.39 0.32 2.08 1.70 1.40 – – –
10.5 0.23 0.51 0.42 0.35 2.20 1.79 1.48 – – –
12.0 0.23 0.52 0.42 0.35 2.26 1.84 1.52 – – –
13.5 0.21 0.62 0.51 0.42 2.93 2.38 1.97 – – –
15.0 0.21 0.51 0.42 0.34 2.46 2.01 1.66 – – –
LPI = wi F i H i = 0.00 0.00 2.98
Table 3 Factors of safety (FS) against soil liquefaction potential [4]

Depth (m) CRR CSR6 CSR6.5 CSR7 FS LPI
Mw Mw Mw Mw Mw Mw
=6 = 6.5 =7 =6 = 6.5 =7
1.5 0.10 0.13 0.10 0.08 1.34 1.03 0.79 – – 2.91
3.0 0.11 0.18 0.14 0.10 1.65 1.25 0.93 – – 0.89
4.5 0.16 0.29 0.22 0.15 1.80 1.40 0.93 – – 0.81
6.0 0.16 0.38 0.29 0.21 2.40 1.84 1.33 – – –
7.5 0.17 0.40 0.30 0.23 2.38 1.78 1.33 – – –
9.0 0.17 0.44 0.32 0.24 2.60 1.89 1.41 – – –
10.5 0.15 0.41 0.31 0.23 2.70 2.06 1.52 – – –
12.0 0.19 0.52 0.38 0.27 2.73 2.00 1.41 – – –
13.5 0.20 0.73 0.53 0.38 3.67 2.65 1.88 – – –
15.0 0.20 0.64 0.48 0.32 3.20 2.40 1.59 – – –
LPI = wi F i H i = 0.00 0.00 4.61
Mw = 7.0, top three layers (up to 4.5 m depth) are found liquefiable as per Idriss
and Boulanger [4] calculation. The results of soil liquefaction potential indicated
that soil liquefaction probability increases when the M w ≥ 7.0, whereas the higher
the seismic magnitude, the greater the liquefaction potential and the more damage it
causes. Comparisons of FS values evaluated from IS code and Idriss and Boulanger
[4] method is presented in Fig. 2 for different earthquake magnitude (M w ). It is found
from the comparison curves that the evaluated values of FS from IS code are similar
to Idriss and Boulanger [4] results for M w = 7.0. Liquefaction susceptibility for
Fig. 2 Comparison of FS values between IS code and Idriss and Boulanger [4]
sites with LPI > 15 is very high, and the liquefaction is very unlikely at sites with
LPI < 5. From this study, it is found that the site is in a liquefaction-susceptible area,
indicating the possibility of damage during high earthquake.
5 Conclusions
In the present study, SPT-based semi-empirical procedures [4, 5] are used to eval-
uate the liquefaction potential of cohesionless soils with three different earthquakes
magnitude (M w = 6.0, 6.5, and 7.0). The SPT-based soil liquefaction in terms of FS
and LPI, revealed that the soil liquefaction probability increases during earthquakes
with a magnitude greater and equal to 7.0 at the IIT Patna campus site. So, miti-
gation techniques such as densification methods, modification of site geometry, or
drainage to lower the groundwater table are needed for this site. It is also found that
the values of FS from IS code matches with Idriss and Boulanger [4] results under
higher earthquake magnitude.
References
1. Bao X, Jin Z, Cui H, Chen X, Xie X (2019) Soil liquefaction mitigation in geotechnical
engineering: an overview of recently developed methods. Soil Dyn Earthq Eng 120:273–291
2. Bwambale B, Andrus RD (2019) State of the art in the assessment of aging effects on soil
liquefaction. Soil Dyn Earthq Eng 125:1–13
3. Goharzay M, Noorzad A, Ardakani AM, Jalal M (2020) Computer-aided SPT-based reliability

model for probability of liquefaction using hybrid PSO and GA. J Comput Des Eng 7:1–21
4. Idriss IM, Boulanger RW (2006) Semi-empirical procedures for evaluating liquefaction
potential during earthquakes. Soil Dyn Earthq Eng 26:115–130
5. IS: 1893 (Part 1): Criteria for earthquake resistant design of structures. Bureau of Indian
Standards, New Delhi
6. Iwasaki T, Tokida K, Tatsuoka F, Watanabe S, Yasuda S, Sato H (1982) Microzonation for soil
liquefaction potential using simplified methods. In: Proceedings of 2nd international conference
on microzonation, pp 1319–1330. Seattle
7. Karakan E, Eskişar T, Altun S (2018) The liquefaction behavior of poorly graded sands rein-
forced with fibers. Adv Civ Eng 1–14, Hindawi. Article ID 4738628 (2018)
8. Olson SM, Mei X, Hashash YMA (2020) Nonlinear site response analysis with pore-water
pressure generation for liquefaction triggering evaluation. J Geotech Geoenviron Eng 146(2):1–
17. ASCE (2020)
Soil Mech Found 97(9):1249–1273. ASCE (1971)
10. Seed HB, Tokimatsu K, Harder LF, Chung RM (1985) The influence of SPT procedures in soil
liquefaction resistance evaluations. J Geotech Eng. ASCE 111(12):1425–1445
11. Shelley EO, Mussio V, Rodríguez M, Chang JGA (2015) Evaluation of soil liquefaction from
surface analysis. Geofísica Int 54(1):95–109
12. Youd TL, Idriss IM (2001) Liquefaction resistance of soils. Summary report from the (1996)
NCEER and 1998 NCEER/NCF workshops on evaluation on liquefaction resistance of soil. J
Geotech Geoenviron Eng 297–313
Reliability-Based Assessment
of Liquefaction Potential Using SPT
Approach
G. S. Bhatia, K. Bhargava, and A. Mondal
Abstract Earthquake-induced liquefaction is one of the most complex and inter-

esting phenomenon in Geotechnical Earthquake Engineering. Liquefaction is a very
significant phenomenon in alluvial soil deposits consisting of silty sand or sandy silt
type of soils. Many different methods are available at present to assess the liquefac-
tion potential of soil using in situ field test data. Generally, field tests like SPT are
carried out and liquefaction assessment is done for each borehole. Different bore-
holes provide different depths up to which soil has potential for liquefaction for
particular peak ground acceleration (PGA) and magnitude of Earthquake. Maximum
liquefaction depth obtained from different boreholes is used to convey the liquefac-
tion depth of the area in consideration. The greatest drawback of this approach is
that results of only one borehole are used to conservatively predict the liquefaction
potential of the entire area and results of remaining boreholes are neglected as they
have predicted lesser depth of liquefaction. In case some more boreholes are drilled
and assessment is done then there is quite a good probability that liquefaction depth
of that area will change depending upon the SPT results from new boreholes. In this
paper liquefaction potential assessment of alluvial soil site is carried out using SPT-
based approach as proposed by NCEER [5]. SPT data of 25 boreholes are analyzed
and converted into the equivalent single borehole using probabilistic approach to
assess the liquefaction potential of alluvial soil site. Probabilistic approach is used
to assess 95 percentile values of all variables required to assess Liquefaction poten-
tial like SPT blow counts, percentage fines, and soil density. As there is uncertainty
G. S. Bhatia (B)
Nuclear Power Corporation of India Limited, Hisar, Haryana, India
e-mail: bhatiasgurpreet09@gmail.com
K. Bhargava
Nuclear Recycle Board, Bhabha Atomic Research Centre, Mumbai, India
Homi Bhabha National Institute, Mumbai, India
K. Bhargava
e-mail: kapilesh_66@yahoo.co.uk
A. Mondal
Nuclear Power Corporation of India Limited, Mumbai, India
e-mail: apurba@npcil.co.in
322 G. S. Bhatia et al.
present in the evaluation of these parameters at the site and in the laboratory, it is
appropriate to evaluate these parameters based on a probabilistic approach using
the best fit probabilistic distribution curve. Parameters required for assessing CSR
(Cyclic Stress Ratio) like overburden pressure and effective overburden pressure are
also analyzed using probabilistic distribution curves considering data from 25 bore-
holes and converting them into an equivalent single borehole. Factor of Safety (FOS)
obtained using 50, 95, and 98 percentile values of different parameters is compared
for converted equivalent single borehole.
Keywords Liquefaction · Cyclic stress ratio · Cyclic resistance ratio · 95

percentile values
1 Introduction
Liquefaction is one of the most spectacular phenomenon associated with an earth-

quake that has caused trouble to engineers and mankind. Some of the most devastating
damages to structures in history have occurred when soil deposits have lost their
strength during seismic loading. In liquefaction, the soil strength is reduced drasti-
cally to the point where the soil is unable to support structures over it. It is essential
to clearly predict the liquefaction potential of soil deposits before the construction of
any particular structure in order to mitigate the effects of liquefaction by the improve-
ment of strata. There are various in situ methods available to assess the liquefaction
potential of soil like SPT, CPT, BPT, and Shear Wave Velocity. These methods have
been developed and refined over the years by different researchers to predict the
liquefaction potential with quite a good probability. SPT-based liquefaction assess-
ment is quiet popular worldwide as it is one of the most convenient and effective
ways of determining liquefaction potential.
Seed and Idriss [6] developed a simplified procedure for evaluating soil liquefac-
tion potential. This procedure was further refined by various researchers including
some landmark studies by Seed and Idriss [7], Cetin et al. [2], Idriss and Boulanger
[4]. Youd and Idriss [9] published a paper that summarizes the recommendation
from NCEER [9] workshop that has incorporated research findings on liquefaction
assessment since many decades. The proceedings of this workshop on liquefaction
potential procedures [5] have shaped the state of practice to evaluate liquefaction
potential.
There are no clear guidelines regarding the extent of geotechnical investigation
required to assess the liquefaction potential of area with a high degree of certainty.
For the construction of important structures like Nuclear Power Plants, Dams, Chem-
ical storage facilities, etc., the need for stringent geotechnical investigation studies is
essential to the safety of the complete structure. Currently, liquefaction assessment
of a particular area based on SPT field data is carried out using the most conservative
approach, wherein liquefaction potential depth of soil deposit is calculated using the
data of worst borehole. Generally, liquefaction assessment of individual boreholes
Reliability-Based Assessment of Liquefaction Potential … 323
is carried out and the results of the worst borehole are used to define the liquefaction
potential depth of soil deposit. In case some new boreholes are drilled in the same
area at later stage, there are possibilities that the liquefaction depth of the area might
change depending upon data from new boreholes. This approach gives all due atten-
tion to the results of single boreholes and neglects the data from other boreholes to
conservatively predict the liquefaction depth. Here the word ‘conservative approach’
is misleading, as there are chances that some other borehole may give much more
worst results and the liquefaction depth may change considerably thus eliminating
the importance of earlier boreholes.
In order to eliminate this problem of pseudo-conservative results and to give due
respect to the data of every borehole, reliability-based assessment has been performed
on SPT data. Reliability-based assessment is helpful in dealing with uncertainties
associated with the field as well as laboratory data. The data were collected by a
single operator using the same equipment in order to reduce the data uncertainty at
the site. Data from 25 boreholes are analyzed at each depth and different probabilistic
distribution curves (like normal, lognormal, and gamma) are used to find the best
suitable curve for each parameter at each depth under consideration. This approach
is adopted for all variables associated with the determination of cyclic resistance
ratio (CRR) like SPT blow counts, soil density and percentage fines. 95 percentile
values of these parameters are determined at each depth thereby converting the data
from 25 boreholes to an equivalent single borehole for all analysis purposes. Factor
of Safety has been determined using this approach and also compared with the FOS
values determined using 50 percentile and 98 percentile values of these parameters.
2 Probabilistic Determination of Field Parameters

Required for Liquefaction Assessment
Liquefaction is a state, in which effective shear strength of saturated cohesionless soil

is reduced to negligible value for the all engineering purpose due to an increase in pore
water pressure. In this condition, the soil tends to behave like a fluid mass. In order to
assess liquefaction resistance of soils, estimation of two variables is required: (1) the
seismic demand on a soil layer, expressed in terms of CSR (Cyclic Stress Ratio); and
(2) the capacity of the soil to resist liquefaction, expressed in terms of CRR (Cyclic
Resistance Ratio). CSR is evaluated using the formulation given by Seed and Idriss
[6], which involves parameters like PGA and density of soil layers determined from
laboratory tests. On the other hand, CRR is evaluated using the formulations given by
NCEER [5] modified from Seed et al. [8]. The parameters involved in the evaluation
of CRR are determined from in situ field tests like corrected SPT blow counts and
from laboratory tests like percentage fines and soil density.
Data obtained from the field tests have randomness due to various reasons like
local conditions, faulty equipment, manual errors, various environmental factors, etc.
Randomness means that more than one outcome is possible, or it can be said that
usually a range of measured or observed values is possible. Also, it is possible that

within this range some values may occur more frequently than others. Represen-
tation of a random variable or mathematical modeling is thus a major step in any
probabilistic formulation, which needs to be conducted systematically.
The first step in converting all boreholes to an equivalent single borehole is to
group the data of all boreholes at each depth. Data from each borehole like corrected
SPT blow counts; percentage fines and soil densities are grouped together for each
depth and analyzed separately. Figure 1 shows the distributed data of SPT values
at the same depth from 25 different boreholes in a particular area after making the
required corrections as suggested by NCEER [5]. While designing any structure
values of different soil parameters are required at a depth where the structure will
rest and also below it. To quantify such data and extract the meaningful value from
this is not an easy task because the mean value of data will not give the overall
estimate of the data obtained. As can be seen from Fig. 1, the mean of corrected
N-values at depth 16.5 m is 32.48. However, at this depth, the corrected SPT blow
counts have maximum and minimum values as 79 and 12, respectively. This shows
a considerable difference between the mean values with extreme values of SPT at a
particular depth. It is therefore very much essential to estimate a value that can be used
conservatively for the estimation of required parameters for further evaluations. Thus,
probabilistic determination is used to determine the non-exceedance probability value
at a particular depth from available borehole data.
Corrected N-values as shown in Fig. 1 are arranged in Table 1 for better under-
standing and evaluation of essential parameters. In Table 1, the maximum value for
SPT is 79 and the minimum value is 12. The information on minimum and maximum
value is very useful, but if one of these values is used for design purposes or for the
Fig. 1 Plot of corrected N-values obtained from 25 different bore holes at a depth of 16.5 m
Table 1 Corrected SPT blow counts at 16.5 m depth
Bore hole No. (N1 )60 Bore hole No. (N1 )60 Bore hole No. (N1 )60 Bore hole No. (N1 )60 Bore hole No. (N1 )60
BH-1 29 BH-6 34 BH-11 79 BH-16 40 BH-21 60
BH-2 39 BH-7 21 BH-12 24 BH-17 32 BH-22 42
BH-3 28 BH-8 27 BH-13 12 BH-18 26 BH-23 33
BH-4 33 BH-9 39 BH-14 21 BH-19 22 BH-24 30
Reliability-Based Assessment of Liquefaction Potential …
BH-5 24 BH-10 31 BH-15 25 BH-20 31 BH-25 30

325
determination of other ground parameters, it may not predict the actual behavior of
soil, since the actual N-value will probably be something different. The behavior of
soil, in that case, will be either underestimated or overestimated. Furthermore, it is
possible that these values may not be the absolute minimum or maximum values. If
more samples are collected, the minimum and maximum values can change. Thus,
working with minimum and maximum values may not be desirable and are usually
not accepted as a standard design practice.
To overcome the deficiency of minimum–maximum value approach, one common
practice is to go for average or mean or expected mean values of SPT data obtained for
design purpose. For 25 samples of SPT values in Table 1, the mean N-value is 32.48.
However, the mean value alone cannot provide the entire information. Obviously,
information on the dispersion of the values with respect to the mean is needed. The
measure of dispersion can be expressed in terms of variance, standard deviation, or
coefficient of variation. It is also helpful to know if the dispersion is symmetrical
or unsymmetrical. The degree of symmetry can be measured using the concept of
skewness.
Therefore, for corrected SPT blow counts at 16.5 m depth as given in Table 1, the
following information can be calculated. Mean is 32.48; Variance is170.81; Standard
Deviation is 13.07; COV is 0.402; Skewness is 0.433 × 104 ; Skewness Coefficient
is 1.941. Thus for the given corrected N-value data, the uncertainty in SPT values is
relatively high, the randomness is unsymmetrical, and the dispersion is more above
the mean than below the mean.
It is always better to depict data in pictorial form. Plotting of information graph-
ically in form of histograms describes it completely. Figure 2 and Fig. 3 shows the
Fig. 2 Histogram and frequency distribution curve of SPT values at 16.5 m depth from 25 boreholes
Fig. 3 CDF plot of SPT values at 16.5 m depth from 25 boreholes showing different distributions
PDF and CDF plots respectively for the data given in Table 1 using different distri-
butions. It can be seen from Fig. 2, that it is difficult to deduce the best distribution
(among normal, lognormal, and gamma) for the corrected SPT data at 16.5 m depth.
In order to find out the 95 percentile value of data from these distributions, it is
first required to plot the PDF and CDF of the random data using all the probability
distribution functions. In case random data satisfies all the probability functions, it
is then required to choose the best function amongst the available functions that best
suits the randomness of data. Comparing probability functions with each other can be
done by performing statistical tests like the Chi-square test, Kolmogorov–Smirnov
test, or Anderson-Darling test.
The K-S and A-D tests are based on the error between the observed and assumed
CDF of the distribution, while Chi-square test is based on the error between the
observed and assumed PDF of the distribution. Chi-square goodness of fit test requires
a large number of data points for the test statistic to converge to its underlying Chi-
square distribution, something that K-S and A-D tests do not require. Therefore, K-S
test is used to estimate the best distribution available for parameters like corrected
N-values, density, and % fines as data from only 25 boreholes are used in this study.
K-S test has the advantage over Chi-square test that it is not necessary to divide the
data into intervals; thus the error or judgment associated with the number or size of
the interval is avoided.
Different probability distribution functions are used to determine the best fit curve
for each parameter at each depth. 95 percentile values of corrected SPT blow counts,
soil density, and percentage fines at each depth are determined from the data of 25 bore
holes at each depth. This 95 percentile value of each parameter at each depth becomes
an input for an equivalent single borehole. The equivalent single borehole concept is
used to assess the liquefaction potential of area with fairly good probability. In case
some more boreholes are drilled at later stage, results are not affected drastically.
'N' values
0 10 20 30 40 50
0
10
Depth (mtr)
15
20
25
30
Average 'N' values 95 percenle 'N' values from Probabilisc Distribuons

35
Fig. 4 Comparative plot between corrected SPT blow counts obtained using average values and
95 percentile values using the best probabilistic distribution curve at each depth
Comparison has been made between the 95 percentile values obtained from the
best probabilistic function as determined using K-S test and mean values of these
parameters at every depth. It can be observed from Fig. 4 that values obtained from
probabilistic distribution are more conservative in comparison to mean values of these
parameters. While assessing liquefaction potential for critical facilities like nuclear
power plants, hydropower plants, military bases, etc., it is always recommended
to go for reliability-based approach to assess critical parameters so as to deal with
the uncertainties associated with the estimation of such parameters. Therefore, it is
beneficial to use reliability method to assess liquefaction potential rather than going
for deterministic or mean value approach.
Figure 4 shows the graphical comparison between corrected average N-values and
corrected 95 percentile N-values at each depth. It can be observed from Fig. 4, that
N-values obtained from probabilistic distributions are almost half of the N-values
obtained from mean of N-values at each depth. This is due to the large variability
present in N-values collected from different boreholes at each depth. Minimum and
maximum N-values obtained at each depth from different boreholes have large vari-
ations which affect the mean value as already explained above. On the other hand,
values obtained from probabilistic distribution are 95 percentile values which give
more reliable values as compared to average values. A similar comparison is shown
for calculated values of density and percentage fines content in Fig. 5 and Fig. 6,
respectively.
Density (kN/m3)
17.2 17.4 17.6 17.8 18 18.2 18.4 18.6 18.8 19
0
10
Depth (mtr)
15
20
25
30
Average Density Values

35 95 percenle Density Values from probablisc distribuons
Fig. 5 Comparative plot between soil densities obtained from average values and 95 percentile
values using the best probabilistic distribution curve at each depth
3 Assessment of CRR and CSR Using Inputs

from Probabilistic Distributions
SPT was conducted in 25 boreholes up to the depth of 110 m from ground level.
However, Liquefaction assessment is carried out up to 31.5 m depth from ground
level. The probability of liquefaction decreases drastically below 20 m depth due to
the overburden pressure of soil layers. The soil at shallower depths is more prone
to liquefaction as compared to the soil layers at deeper depths. Laboratory tests
were conducted on the soil samples obtained from the locations of SPT to determine
percentage fines, dry density, and natural moisture content of the soil samples. Water
table in the area varies from 2 to 3 m depth below ground level. However, for analysis
purpose, water table is considered at ground level.
SPT values (N) as obtained from the in situ testing are corrected to obtain the (N1 )60
by using correlation given in NCEER [5]. These corrected N-values from 25 bore-
holes are then used to determine the 95 percentile values at each depth as explained
earlier. A similar assessment has been made for soil densities and percentage fines.
Corrected SPT values (N1 )60 are further corrected to account for fines content present
in the soil using the approach given in NCEER [5].
% Fines
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70
0
10
Depth (mtr)
15
20
25
30
Average % Fine Values

35 95 percenle % Fine Values from probablisc distribuons
Fig. 6 Comparative plot between % Fines obtained using average values and 95 percentile values
using the best probabilistic distribution curve at each depth
3.1 Evaluation of CRR
SPTs are generally preferred to evaluate liquefaction potential of soil due to the vast
database available and past experiences. The added advantage of SPT is that soil
samples can be retrieved during the process and can be tested at the laboratory to
assess important soil properties. CRR is evaluated using the curve recommended by
NCEER [5]. This curve gives CRR for a magnitude 7.5 earthquake. The clean-sand
base curve as suggested by NCEER [5] can be approximated with Eq. (1). Scaling
factors are used to adjust the CRR for different earthquake magnitudes. Different
formulations for scaling factors are given by various researchers.
1 (N1 )60 50 1
CRR7.5 = + + − (1)
34 − (N1 )60 135 [10.(N1 )60 +45] 2 200
The above equation is valid for (N1 )60 < 30. NCEER [5] suggest that for (N1 )60 ≥
30, clean sands are too dense to liquefy and hence termed as non-liquefiable. Table 2
shows the data of an equivalent single borehole that will be used to assess CRR
at different depths. This data is derived from 25 boreholes using the probabilistic
distribution curves as explained above. This is the 95 percentile data for corrected
SPT blow counts, soil density, and percentage fines. In order to assess the liquefaction
potential of the same area more conservatively even 98 percentile values of these
Table 2 Data for equivalent borehole to assess liquefaction potential

Depth (m) 95 percentile values of 95 percentile values of % 95 percentile values of field
(N1 )60 for equivalent fines for equivalent density for equivalent
borehole borehole borehole (kN/m3 )
1.5 3.8 28.07 17.8
4.5 7.2 27.49 17.7
7.5 11.9 37.65 17.9
10.5 15.5 26.34 17.6
13.5 17.4 16.15 17.8
16.5 23.3 9.39 17.7
19.8 19.1 12.56 17.6
22.5 23.0 5.84 17.5
25.5 22.2 8.98 17.6
28.5 25.8 8.62 17.7
31.5 21.7 11.40 17.6
parameters can be adopted. 95 percentile values of any parameter suggest that the
possibility of exceeding this value is only 5%. Therefore, evaluation of Factor Safety
using this approach suggests that there is only a 5% probability that liquefaction in
this area can occur beyond the calculated depth. Thus all uncertainties associated
with the parameters required for estimation of CRR can be dealt with using this
approach with fairly good probability.
3.2 Evaluation of CSR
Seismic demand on the soil layer, expressed in terms of CSR is calculated using the
formulation given by Seed and Idriss [6] as given in Eq. (2).

amax σvo
CSR = 0.65 rd (2)
g σ vo
where amax = peak horizontal acceleration at the ground surface generated by

the earthquake; g = acceleration of gravity; σvo and σ’vo are total and effective
vertical overburden stresses, respectively; and rd = stress reduction coefficient. This
coefficient accounts for the flexibility of the soil profile.
Overburden and effective overburden stresses are calculated using soil density of
equivalent single borehole as determined using the probabilistic approach explained
earlier. CSR has been calculated for peak ground acceleration values of 0.2g.
4 Evaluation of Factor of Safety (FOS)
Factor of safety is calculated as the ratio of CRR and CSR. Soil liquefaction and
subsequently depth of treatment will be assessed on the basis of this factor of safety.
As per NCEER [5] soil is non-liquefiable for FOS > 1. The CRR curves (or clean-sand
base curves) for SPT apply only to magnitude 7.5 earthquakes. To adjust the clean-
sand curves to magnitudes smaller or larger than 7.5, Seed and Idriss [7] introduced
correction factors termed “Magnitude Scaling Factors (MSFs)”. These factors are
used to scale the CRR base curves upward or downward on CRR versus (N1 )60 plots
as given in Eq. (3).
CRR = (CRR7.5 ).MSF (3)
where CRR is CRR for earthquake magnitudes other than 7.5.

To illustrate the influence of magnitude scaling factors on calculated hazard, the
equation for factor of safety (FOS) against liquefaction is written in terms of CRR,
CSR, and MSF as follows:

CRR7.5
FOS = .MSF (4)
CSR
where CSR is calculated cyclic stress ratio generated by the earthquake shaking;
CRR7.5 is cyclic resistance ratio for magnitude 7.5 earthquakes as calculated earlier.
NCEER [5] proposed MSF to adjust liquefaction resistance for magnitudes (M)
other than 7.5 as given in Eq. (5).
2.24
MSF = 10 Mw2.56 (5)
where Mw is the Magnitude of Earthquake used in the analysis.

Figure 7 shows the Factor of Safety for equivalent borehole using 50 percentile,
95 percentile and 98 percentile values at amax equal to 0.2g and 7.5 Magnitude of
Earthquake.
5 Conclusions
Standard penetration test (SPT) was conducted in 25 boreholes up to the depth of 110
m from the existing ground level. However, analysis for liquefaction potential was
done up to 31.5 m depth as the probability of liquefaction potential decreases beyond
20 m depth due to overburden pressure of soil layers. Undisturbed and disturbed soil
samples were collected from the same boreholes to determine the soil density and
percentage fines of soil. There are many uncertainties involved in the assessment
Factor of Safety
0 0.5 1 1.5 2 2.5 3 3.5

0
10
Depth (mtr)
15
20
25
30
FOS using 50 percenle values FOS using 95 percenle values

35 FOS using 98 percenle values
Fig. 7 Comparison between FOS using parameters obtained from different percentile values
of input parameters involved in the determination of liquefaction potential (like N-

values, density of soil, %fines of soil). Therefore, to assess these parameters the
probabilistic approach has been followed in this study. 95 percentile values of all
parameters are calculated based on the best distribution followed by the input param-
eters obtained from the field as well as laboratory results. These values were then
used to convert data from 25 boreholes to an equivalent single borehole, which was
used to determine liquefaction potential of the entire area. The analysis was done at
amax equal to 0.2g and at 7.5 magnitude of earthquake.
Cyclic resistance ratio (CRR) for this equivalent borehole was determined using
the relationship given in NCEER [5] using the 50, 95, and 98 percentile values
of different parameters. A comparison was made between the FOS obtained using
the different percentile values of these parameters. 95 percentile value means that
there are only 5% chances that the value of a particular parameter will go beyond
the calculated value. Therefore, higher percentile values predict more conservative
values as compared to lesser percentile values. This can also be seen from Fig. 7 that
a more conservative assessment of Factor of Safety was done using higher percentile
values. The major advantage of this approach is that results from all boreholes have
been used to assess the values of parameters with a certain probability. In case some
more bore holes will be drilled at a later stage, values from these boreholes will not
affect the overall liquefaction potential depth of that area significantly.
References
1. BIS (1981) IS 2131:1981, Method for standard penetration test for soils. Bureau of Indian
Standards, New Delhi
2. Cetin KO, Seed RB, Kiureghian AD, Tokimatsu K, Harder LF Jr, Kayen RE, Moss RE
(2004) Standard penetration test-based probabilistic and deterministic assessment of seismic
soil liquefaction potential. J Geotech Geoenviron Eng 130(12):1314–1340
3. Haldar A, Mahadevan S (2000) Probability, reliability and statistical methods in engineering
design. Wiley, New York
4. Idriss IM, Boulanger RW (2008) Soil liquefaction during earthquakes. Earthquake Engineering
Research Institute, USA
5. NCEER (1997) Proceedings of the NCEER workshop on evaluation of liquefaction resistance
of soils. National Centre for Earthquake Engineering Research, Report No. NCEER-97-0022,
State University of New York at Buffalo, Buffalo, NY
6. Seed HB, Idriss IM (1971) Simplified procedure for evaluating soil liquefaction potential. Am
Soc Civ Eng J Soil Mech Found Div 97(SM9):1249–1273
7. Seed HB, Idriss IM (1982) Ground motions and soil liquefaction during earthquakes. Monograph
series. Earthquake Engineering Research Institute, University of California, Berkeley, CA
8. Seed HB, Tokimatsu K, Harder LF, Chung R (1985) Influence of SPT procedures in soil
liquefaction resistance evaluations. Am Soc Civ Eng J Geotech Eng Div 111(12):861–878
9. Youd TL, Idriss IM (2001) Summary report from the 1996 NCEER and 1998 NCEER/NSF
workshops on evaluation of liquefaction resistance of soils. J Geotech Geoenviron Eng
127(4):297–313
Building Performance and Geotechnical
Failures in 7.5M Palu Earthquake
and Tsunami 28 September 2018
Sumedha Koul and Harish Mulchandani
Abstract On 28 September 2018, a 7.5 magnitude earthquake and subsequent

tsunami hit Palu and Donggala in Central Sulawesi, Indonesia killing at least 2,245
people. Thousands of people were missing and over 10,000 were injured, of whom
over 4,000 were with severe injuries. Nearly 75,000 were displaced in the three most
affected areas: Donggala, Palu City, and Sigi. This paper discusses the performance
of reinforced concrete buildings, bridges, port facilities, and lifelines in the Palu
Earthquake. It was observed that a number of residential and commercial buildings
were partially or fully collapsed and many sustained significant structural damages.
The iconic twin steel arch cable-suspended Palu Bridge IV over the Palu River
was collapsed. Lateral spreading induced by liquefaction resulted in huge housing
damages and farm fields. It was seen from field investigations that possible reasons
for building failure include the lack of confinement bars, improper confinement in
beam and column joints, Ssrong beams weak columns and presence of soft story in
multi-story buildings.
Keywords Palu earthquake · Building performance · Geotechnical failure
1 Introduction
The Palu earthquake with a magnitude of 7.5 Mw occurred on 28 September, 2018, at

18:02 WITA time. Palu-Koro Fault experienced a strike with a mechanism of “strike-
slip”. The earthquake epicenter is located 72 km north of the city of Palu with an
earthquake depth of 10 km. This earthquake triggered a tsunami, a shift in the fault
(rupture) along The Palu-Koro Fault and its secondary fault, and the occurrence of
extensive and massive landslides in the Petobo, Jono Oge, Sibalaya, Balaroa and Lolu
villages [1–7]. There is evidence of earthquake rupture occurring due to supershear
velocities [8]. The earthquake triggered an instantaneous tsunami that reached up to
the height of 5 m [9, 10].
S. Koul · H. Mulchandani (B)

Department of Civil Engineering, Birla Institute of Technology & Science Pilani, Pilani, India
e-mail: mulchandani1010@gamil.com
336 S. Koul and H. Mulchandani
Major building failures occurred in the region were due to soft storey shear
failure, beam-column joint failure, buildings with strong beam and weak column.
It leads to complete collapse, discontinuity in lateral load resisting systems, and due
to landslides resulting from wet rice fields [11].
2 Building Performance
Many of the buildings suffered huge losses because of the earthquake. A few notable
multi-storey buildings included the Tatura shopping mall, Roa Roa Hotel and several
mosques. The major reasons for structural failure have been listed below.
2.1 Lack of Confinement in the Joints
The majority of buildings suffered beam-column joint failures. It occurred due to

a lack of confinement reinforcement bars at the joints. The design check for joints
is not very crucial for buildings designed to sustain normal gravity loads. However,
this becomes an essential criterion for buildings situated in earthquake-prone areas.
Most of the reinforced concrete structures in Palu also collapsed due to the same
reason.
Shown below are the images of various failed joints of the Tatura shopping
complex. The shopping complex was a four-storeyed building that suffered a partial
collapse of the roof and third floor due to the failure of multiple columns and
beam-column joints.
The absence of stirrups in beam column joints led to buckling of the vertical
column reinforcement due to which unconfined concrete in these joints failed. The
lack of shear stirrups also caused many columns to fail due to shear. Some interior
columns also collapsed due to shear or flexure failure in the hinging region. However,
there was no damage in the frame or beams due to these columns as can be seen in
Fig. 1.
The Dunia Baru restaurant, a three-storeyed building, close to the shopping
complex also suffered complete collapse since the bars used for reinforcement were
smooth. Despite the complete collapse of the structure, the beams were not damages
hinting toward a strong beam weak column condition like in the Tatura shopping
complex. Similarly, the Mecure hotel also faced structural damages due to the lack
of shear stirrups in the beam-column joints and the strong beam weak columns
condition.
Building Performance and Geotechnical Failures … 337
(a) (b)
(c) (d)
Fig. 1 Beam-column joint failure due to lack of confinement bars
2.2 Soft Story Failure
A soft storey refers to a multi-storey building in which at least one floor has open-
ings in places like windows, wide doors, etc., where shear walls would have been
necessary for stability in earthquake engineering design. Technically, a soft storey
can be defined as if a floor has a stiffness less than 70% stiffness of subsequent floor
or less than 80% stiffness in comparison to the average stiffness of the upper three
floors. Such buildings are more vulnerable to damage due to a moderate or severe
earthquake. Due to the inadequate bracing, the floors are less resistant to the lateral
forces due to the earthquake.
Many buildings have collapsed in Palu due to this phenomenon. The Tatura shop-
ping complex suffered from many columns and joint failures since these were not
appropriately braced. The failure of the columns led to the partial collapse of the
third floor and the roof. The figure below shows the damaged building.
(a) (b)
(c)
Fig. 2 Soft storey failure at a Ramayana Shopping centre, b Dunia Baru restaurant, c Mecure Hotel
at Palu Indonesia [1]
The Dunia Baru restaurant, a three-storeyed building was also completely

collapsed due to the failure of the columns which led to the failure of the floors
as can be seen in the image. The Mecure Hotel, a five-storeyed building, also had
weak columns that failed leading to the complete collapse of the ground floor and
partial collapse of all the other floors as shown in Fig. 2.
3 Damage Due to Lateral Spreading
The earthquake ground shaking induced liquefaction over three large areas in and
around Palu City, namely the Balaroa neighbourhood, the Petobo sub-district and
Jono Oge Village. Despite the slope being as low as 1%, these areas experienced
lateral spreading that extended up to 3.5 km.
Fig. 3 Google Earth images

of Balaroa region of the Palu
a before the lateral spreading
b after the lateral spreading
[1]
(a)
(b)
3.1 Balaroa
Balaroa is a district in Palu which experienced a lateral spread of up to 1000 m on

an average slope of 3.4%. Figure 3 shows an aerial view of the site before and after
the lateral spreading. The upper layer of the soil along with the structures moved
downslope due to the liquefaction of the soil caused due to the earthquake.
3.2 Petobo Sub-district
The Petobo sub-district lies towards the south of the Palu airport and 7.2 km east of
Palu-Koro fault line. This region is a residential area that experienced destruction
due to liquefaction which had a lateral spreading of up to 2200 m on an average slope
of only 2%. Lateral spreading lead to massive damage to the building in the region,
which were either buried or moved along with the liquefied soil. Figure 4 shows the
aerial view of the site before and after the earthquake transpired.
(a)
(b)
Fig. 4 Google Earth images of Petobo sub-district region of the Palu a before the lateral spreading
b after the lateral spreading [1]
3.3 Jono Oge Village
Due to the liquefaction, the lateral spread in this area spread up to 3500 m on a
terrain that had an average slope of only 1.2%. Despite it being a rural setup, many
structures were destroyed because of the severity of the slide. Figure 5 shows the
village before and after the slide occurred.
Fig. 5 Images of Jono Oge

village of the Palu a before
the lateral spreading b after
the lateral spreading [1]
(a)
(b)
4 Damage to Palu Bridge IV
Palu bridge IV was a prominent steel double arch suspension bridge that connected
the east and west sides of the bay of Palu. This bridge, as per eye witness, got damaged
before the tsunami arrived. Some of the possible reasons for its failure may be due to
the lateral collapse because of the impact caused by heavy concrete deck suspended
from the arches by steel cables on the base of the arches or due to the lateral deck
motion which could have been sufficient to fail the structure. Figure 6 shows the
images of Palu bridge IV before and after damage.
5 Conclusion
This paper discusses the damages caused to infrastructure due to the Palu earthquake
which hit Indonesia on September 28th, 2018. In this study, the major focus has been
laid on the Tatura shopping complex, the Dunia Baru restaurant and, the Mecure
hotel amongst other RCC buildings that were damaged. Particular emphasis was put
on these buildings since all three of them failed more or less due to the same reasons
(a) (b)
Fig. 6 Palu bridge IV a before the earthquake b after the earthquake [1]
i.e. weak-column-strong-beams and the lack of shear reinforcements in the beam-

column joints. Along with these, the paper also lays emphasis on damages caused
due to the liquefaction of the soil because of the earthquake. Balaroa, Petobo and
Jono Oge village were completely washed out by great distances despite the gentle
slope highlighting the magnitude of the earthquake.
By studying the causes of damage of structures, one can understand the need for
earthquake-resistant designs. We can reduce the damage caused to life to infrastruc-
ture by an earthquake, by carrying out proper geotechnical analysis of the area and
studying its liquefaction tendency and designing the foundations accordingly.
References
1. Robertson I et al (2019) StEER—Palu earthquake and tsunami, Sulawesi, Indonesia: Field

assessment team 1 (FAT-1) early access reconnaissance report (EARR). DesignSafe-CI. https://
doi.org/10.17603/DS2JD7T
2. Harnantyari AS, Takabatake T, Esteban M, Valenzuela P, Nishida Y, Shibayama T, Kyaw TO
(2020) Tsunami awareness and evacuation behaviour during the 2018 Sulawesi Earthquake
tsunami. Int J Disaster Risk Reduct. https://doi.org/10.1016/j.ijdrr.2019.101389
3. Putra PS, Aswan A, Maryunani KA, Yulianto E, Kongko W (2019) Field survey of the 2018
Sulawesi Tsunami deposits. Pure Appl Geophys. https://doi.org/10.1007/s00024-019-02181-9
4. Paulik R, Gusman A, Williams JH, Pratama GM, Lin S, Prawirabhakti A, Suwarni NWI (2019)
Tsunami hazard and built environment damage observations from Palu City after the September
28th 2018 Sulawesi Earthquake and Tsunami. Pure Appl Geophys. https://doi.org/10.1007/s00
024-019-02254-9
5. Widiyanto W, Santoso PB, Hsiao S-C, Imananta RT (2019) Post-event field survey of September
28th 2018 Sulawesi Earthquake and Tsunami. Nat Hazards Earth Syst Sci Discuss. https://doi.
org/10.5194/nhess-2019-91
6. Syamsidik B, Umar M, Margaglio G, Fitrayansyah A (2019) Post-tsunami survey of the
September 28th 2018 tsunami near Palu Bay in Central Sulawesi, Indonesia: impacts and
challenges to coastal communities. Int J Disaster Risk Reduct. https://doi.org/10.1016/j.ijdrr.
2019.101229
7. Rajindra R, Wekke IS, Sabara Z, Pushpalal D, Samad MA, Yani A, Umam R (2019) Diversity,
resilience, and tragedy: three disasters in Palu of Indonesia. Int J Innov Creat Change
8. Socquet A, Hollingsworth J, Pathier E, Bouchon M (2019) Evidence of supershear during the

2018 magnitude 7.5 Palu earthquake from space geodesy. Nat Geosci. https://doi.org/10.1038/
s41561-018-0296-0
9. Sassa S, Takagawa T (2019) Liquefied gravity flow-induced tsunami: first evidence and compar-
ison from the 2018 Indonesia Sulawesi earthquake and tsunami disasters. Landslides. https://
doi.org/10.1007/s10346-018-1114-x
10. Carvajal M, Araya-Cornejo C, Sepúlveda I, Melnick D, Haase JS (2019) Nearly instantaneous
tsunamis following the Mw 7.5 2018 Palu Earthquake. Geophys Res Lett. https://doi.org/10.
1029/2019GL082578
11. Bradley K, Mallick R, Andikagumi H, Hubbard J, Meilianda E, Switzer A, Hill EM (2019)
Earthquake-triggered 2018 Palu Valley landslides enabled by wet rice cultivation. Nat Geosci.
https://doi.org/10.1038/s41561-019-0444-1
Numerical Modeling of Three-Tiered
Reinforced Soil Wall with Different
Offset Distances Subjected to Dynamic
Excitation
Sudipta Sikha Saikia and Arup Bhattacharjee
Abstract Reinforced retaining wall using geogrids is an effective method to deal

with high and steep soil slope under complicated geological engineering conditions.
There have been comparatively lesser studies on multi-tiered walls due to its limited
application. This paper aims at making a comparative study of the behavior of three-
tiered reinforced soil wall with small and large offset distances subjected to seismic
excitations. A 2.8 m high numerical model of reinforced soil wall is developed using
finite element software PLAXIS 2D. The numerical model is subjected to dynamic
excitations of 0.4 g Kobe earthquake and results of the response of the numerical
model are validated with shake table tests results of Ling et al. [7]. A 9 m high
three-tiered wall with small offset distances (i.e., 0.5 m and 0.75 m) and large offset
distances (1.5 m and 2.5 m) is simulated with validated model parameters. The tiered
walls are subjected to seismic excitations of the Kobe earthquake at a peak accelera-
tion of 0.4 g and the variation of lateral pressure, maximum reinforcement loads, and
acceleration amplification factors of three-tiered walls with various offset distances
are compared. The results showed that the lateral stress, maximum reinforcement
load, and acceleration amplification factor decreases with the increasing tier offset.
Keywords Reinforced retaining wall · Finite element model · Dynamic excitation
1 Introduction
Soil reinforcement is a modern and efficient technique for improving the mechanical
properties of soil, which takes advantage of the frictional interaction between the
soil and reinforcement. The most common application of reinforced earth is in the
construction of geosynthetic reinforced soil (GRS) retaining walls. Retaining walls
are the structures that are built to retain vertical or nearly vertical earth banks or any
S. S. Saikia (B) · A. Bhattacharjee

Jorhat Engineering College, Jorhat 785007, India
e-mail: sudiptasikhasaikia@gmail.com
A. Bhattacharjee
e-mail: bhatta_arup@yahoo.com
346 S. S. Saikia and A. Bhattacharjee
other materials (Murthy [1]). Geo-synthetically reinforced soil wall is cost-effective,

durable, and perform better during earthquakes as compared to conventional gravity
retaining walls. They are also suitable for uneven settlements and can adapt to new
conditions. The tensile stress in the reinforcement increases rapidly with height
while designing single-tiered walls. To take advantage of both the aesthetics and the
economics of GRS walls while considering high heights, multi-tiered walls are often
used.
The study of multi-tiered GRS wall has not achieved much progress as compared
to the single-tiered GRS wall. Though few guidelines and numerical studies are
available [1–8] etc., for the design of multi-tiered walls but still these are not adequate
for the execution of multi-tiered reinforced soil wall in different conditions (like fill
height, quality of retaining soil, etc.).
The objective of the paper is to understand the behavior of three-tiered rein-
forced soil walls with small as well as large offset distances under dynamic loading
conditions and making a comparative study on the variation of lateral stress of back-
fill, maximum reinforcement load, acceleration amplification factor with tier height
through numerical simulations.
2 Development of Numerical Model of Reinforced Soil Wall

in PLAXIS 2D
2.1 Full-Scale Model: Background
Ling et al. (2005) [9] conducted large-scale shaking table tests on modular block
geosynthetic reinforced soil retaining wall subjected to Kobe earthquake motions.
The wall was 2.8 m high, 4 m long, and 2 m wide constructed on a 20 cm thick
soil foundation. The facing blocks were 20 cm high, 30 cm deep, and 45 cm wide.
The wall was backfilled with medium dense Tokachi port sand (Dr = 55%) and
reinforced with PET geogrid. Geogrids length of h = 205 cm were placed at vertical
intervals of 60 cm and its ultimate strength was 35 KN/m.The foundation soil has the
same properties as the backfill soil but at a relative density of 90%. To prevent waves
reflecting from the steel walls during shaking, 10 cm thick expanded polystyrene
(EPS) boards had been placed at the front and back ends of the steel container. The
geometry of the wall is shown in Fig. 1. The wall was subjected to horizontal shakings
of Kobe earthquake (1995) scaled to a peak acceleration of 0.4 g.
Numerical Modeling of Three-Tiered Reinforced Soil … 347
Fig. 1 Geometry of the shake table test used for validation (after Ling et al. [9])
2.2 Numerical Modeling in PLAXIS 2D
The components of shake table test as reported by Ling et al. (2005) [10] are simulated
using finite element program PLAXIS 2D. The analysis is conducted under two-
dimensional plane strain conditions and 15-noded triangular elements are selected
[11]. The input model parameters used in PLAXIS 2D are given in Table 1. The
model was fixed at the base and assumed to have roller boundaries at the sides. In
the finite element model, the bottom boundary is fixed in both horizontal and vertical
directions while the side boundaries are fixed in only horizontal directions. To allow
for absorption of stress waves, the absorbent boundary condition in PLAXIS 2D
is created by placing the boundaries far apart. For dynamic excitation prescribed
displacement option is used to define the seismic load. The wall is excited with a
maximum horizontal acceleration of 0.4 g. The acceleration time history for 0.4 g is
shown in Fig. 2.
The geometry of the model in PLAXIS 2D is shown in Fig. 3.
Table 1 Input model parameters used in finite element program PLAXIS 2D

Properties Backfill Foundation Facing wall Geogrid
Material model Mohr–Coulomb Mohr–Coulomb Linear elastic Linear elastic
Elastic modulus (kPa) 156E3 156E3 2E6
Cohesion (kPa) 1 1
Angle of friction (z) 38 40
Dilatancy angle (ψ) 8 8
Mass density (kN/m3 ) 14.30 14.30
Poisson’s ratio 0.33 0.33 0.2
Fig. 2 Input dynamic excitation at peak acceleration of 0.4 g
Fig. 3 Geometry of numerical model used for validation
2.3 Validation of the Numerical Model with the Full Scale

Model
To verify the accuracy of the finite element model developed in the study, the results
of the shake table test as reported by Ling et al. [9] are compared with the results
obtained from the finite element model.
Horizontal displacement of facing: The comparison between measured and
predicted results of horizontal displacement of the facing wall is shown by in Fig. 4.
Fig. 4 Horizontal 2.8

2.4
Height(m)
displacement measured by
2
Ling et al. [9] and FE 1.6 Ling et al.
analysis 1.2 Plaxis 2D
0.8
0.4
0
0 20 40 60 80 100
Horizontal displacement(mm)
4
plaxis 2D 4 plaxis 2D
3 measured 3 measured
Force
Force
2 2
1 1
0 0
2 1.5 1 0.5 0 2 1.5 1 0.5 0
Distance (m)
Distance (m)
(a) Geogrid layer at h=0.2m (b) Geogrid layer at h=1.4m
4
plaxis 2D
3 measured
Force
2
1
0
0 1 2
Distance (m)
(c) Geogrid layer at h=2.6m
Fig. 5 Strain in various primary geogrid layers
The maximum horizontal displacement at the top of the wall is found to be in

reasonably good agreement with the measured value as discussed by Ling et al.
[9].
Strain in primary geogrid layers: The strain distributions in the geogrid layers
at wall height h = 0.2 m, 1.4 m, 2.6 m are shown in Fig. 5. The tensile strain measured
in the various geogrid layers were converted to tensile force. Although the increase
in tensile stress under dynamic load is not proportional to the increase in wall height;
but the distribution within the wall height shows similar trends with the measured
results.
3 Numerical Modeling of Three-Tiered Reinforced Soil

Walls
A 9 m high wall is selected for numerical analysis of three-tiered geosynthetic rein-

forced soil retaining walls. Three-tiered walls with small offset (0.5 m, 0.75 m) as
well as large offset distances (1.5 m, 2.5 m) were considered for the analysis. The
Table 2 Reinforcement
No. of tiers Position of tier Reinforcement length
lengths for different tiers of a
three-tiered wall as per Three tier (H = 9 m, Upper tier 0.7 H2 = 2.1 m
FHWA [3] H 1 = H2 = 3 m) Middle tier 0.7 H1 = 2.1 m
Lower tier 0.6 H = 5.4 m
(a) (b)
Fig. 6 Geometry of three-tiered walls with a 0.5 m offset, b 1.5 m offset distances
reinforcement lengths of the three-tiered walls are calculated as per FHWA [3] and
are shown in Table 2.
The different models of three-tiered soil walls are reinforced using 12 number of
geogrid layers at a spacing of 0.6 m. The model parameters and boundary conditions
used for modeling the three-tiered walls are the same as that used in the validation
analyses. The numerical models are subjected to a maximum horizontal acceleration
of 0.4 g at the base of the foundation (Fig. 6).
3.1 Comparison of Results
Lateral stress of backfill: Fig. 7 shows the distribution of lateral soil pressure at the
face of the wall for small as well as large offsets of three-tiered walls. From the plot,
it can be inferred that the lateral stress decreases almost linearly with an increase in
wall height, however at the junction of the tiers the magnitude of stresses increases.
The maximum lateral stress on the facing wall is not found exactly at the base of the
wall, but somewhere near the position of the first geogrid layer. The reason behind
such behavior may be due to the rigid interaction between the foundation and the
lowermost facing panel as considered in the analysis.
Maximum reinforcement load: Fig. 8 shows the comparison of maximum rein-
forcement load of three-tiered walls for different offsets. With an increase in the tier
offset, the maximum reinforcement load decreases but due to the small reinforcement
1
0
9
8
0.5 m offset
Wall height
7
6 0.75 m offset
5
4 1.5 m offset
3 2.5 m offset
2
1
0 0 20 40 60 80 100
Lateral stress (kPa)
Fig. 7 Variation of lateral stress versus wall height for different offsets of three-tiered walls
subjected to Kobe excitations
1
0
9
Wall height
8
0.5 m offset
7
6 0.75 m offset
5
1.5 m offset
4
3 2.5 m offset
2
1
0 0 2 4 6 8
Maximum reinforcement load (kN/m)
Fig. 8 Maximum reinforcement load versus tier height for different offsets of three-tiered walls
subjected to Kobe excitations
length in upper tiers, there is not much decrease in reinforcement load for different
offsets in three-tiered walls. The maximum reinforcement load at the mid-height is
found to be 7.2 kN/m, 6.9 kN/m, 6 kN/m, and 3.9 kN/m for offset distances of 0.5 m,
0.75 m, 1.5 m, and 2.5 m, respectively.
Acceleration amplification: The amplification factor is given for the ratio of
maximum acceleration in the backfill, typically at the top of backfill, to the acceler-
ation at the foundation level. The excitation was given at the base of the wall. After
the excitation, the acceleration amplification in the backfill was determined behind
the facing blocks. The acceleration amplifications at different elevations of the wall
are quantified as root mean square acceleration (RMSA) amplification factor. The
Fig. 9 Variation of 1
acceleration amplification 0
Wall height (m)

factor with wall height for 8 0.5 m offset
three-tiered walls subjected 0.75 m offset
to Kobe excitations 6
1.5 m offset
4
2.5 m offset
2
0
0 1 2 3
Acceleration amplification factor
RMS acceleration can be calculated as

1/2
RMS = 1/td td a(t)2 dt
0
where a(t) is acceleration time history, t d is the duration of the acceleration record.
dt is the time interval of the acceleration record.
Figure 9 shows the acceleration amplification of three-tiered walls at 10 m away
from the toe. Acceleration amplification factors decreased with increasing tier offset
in the model walls analyzed in the study. As can be seen in the figure, the amplifi-
cation factor is high for smaller offset and for higher offsets amplification factor is
decreasing.
4 Important Outcomes
The following important outcomes are drawn from the study:

• A numerical model of shake table test was developed in finite element program
PLAXIS 2D and validated by comparing the numerical results with physical
measurements taken from shake table test on reinforced soil wall as reported
by Ling et al. [9]. The closed agreement of numerical results with the physical
data indicated that finite element model is capable of simulating the construction
sequence of geosynthetic reinforced soil structures.
• For three-tiered walls, the values of lateral stresses are smaller for larger offsets
as compared to that of small offset distances. The lateral stress decreases
almost linearly with height except at the junction of the tiers where it shows
a contradictory behavior.
• With the increase in tier offset, the maximum reinforcement load in the three-
tiered walls with different tier-offset decreases. For all the walls considered in the
analysis, the maximum reinforcement load is found almost at the midheight of
the walls.
• The acceleration amplification factor was found to increase along the height of
the wall. For smaller offsets, the values are found slightly higher for three-tiered
walls as compared to larger offset distances.
References
1. AASHTO (1998) Standard specifications for highway bridges, American Association of State
Highway and Transportation Officials, Washington, DC.
2. FHWA Publication (1995) Geosynthetic design and construction guideline. FHWA Publication
No: FHWA-HI-95–038
3. FHWA Publication (2010) Mechanically stabilized earth walls and reinforced soil slopes design
and construction guidelines, vol I & II. Publication No. FHWA-NHI-10–024, US Department
of Federal Highway Administration (FHWA)
4. Leshchinsky D, Han J (2004) Geosynthetic reinforced multitiered walls. J Geotech Geoenviron
Eng 130(12)
5. Liu H, Yang G, Ling HI (2014) Seismic response of multi-tiered reinforced soil retaining
walls. Soil Dyn Earthquake Eng 61–62, 1–12
6. Mohamed SBA, Yang K-H, Hung W-Y (2014) Finite element analyses of two tier geosynthetic-
reinforced soil walls: Comparison involving centrifuge tests and limit equilibrium results.
Comput Geotech 61:67–84
7. NCMA (1997) Design manual for segmental retaining walls. In: Collin J (ed) National concrete
masonry association, 2nd edn. Herndon, Virginia, USA
8. Yoo C, Kim S (2008) Performance of two-tier geosynthetic reinforced segmental retaining
wall under a surcharge load: full scale load test and 3D finite element analysis. Geo-Textiles
Geomemb 26:460–472
9. Ling et al (2005) Large scale shaking table tests on modular block reinforced soil retaining
walls. J Geotech Geoenviron Eng ASCE
10. Kramer SL (1996) Geotechnical earthquake engineering. Prentice Hall, Upper Saddle River,
NJ
11. Kumari S, Bhattacharjee A (2018) Numerical stimulation of tiered reinforced soil retaining
wall subjected to dynamic excitation. In: Proceedings of the Indian geotechnical conference
2018, Bengaluru, India
Study on Earth Pressure Distribution
and Displacements of Narrow Reinforced
Earth (RE) Wall Under Static and Cyclic
Loading
Shivani R. Patel, P. J. Mehta, and M. V. Shah
Abstract To study the reduction in lateral earth pressure due to narrow reinforced
earth (RE) wall, which is used for various earth retaining structures such as bridge
abutments, retaining walls, and also where the available space for the reinforced earth
(RE) walls is less than required. In a narrow RE wall, interface connection will be
provided to prevent extensive pressure and cracks which are developed in-between
existing wall and narrow RE wall. The main objective of this paper is to evaluate
the earth pressure distribution for a narrow RE wall under static and cyclic loading
considering the elastic (Flexible) and non-elastic (Rigid) behavior of the wall keeping
the same relative density. Two major conditions, i.e., rigid boundary condition and
flexible condition, which is used to perform a series of load–displacement and load-
settlement test on the RE wall model using uniaxial geogrid reinforcement. The
load–displacement-settlement is measured by using a conventional high capacity
compressive mechanical jack and dial gauges. To validate the experimental results,
the earth pressure distribution using Arching theory, Rankine theory, Coulomb theory.
From results and analysis, there was a considerable variation is determined in load–
displacement characteristics for both flexible and rigid boundary conditions. The
percentage reduction in earth pressure was also observed in the case of a flexible RE
wall as compared to a rigid narrow RE wall.
Keywords Narrow RE wall · Displacement · Static and cyclic loading
S. R. Patel · P. J. Mehta (B) · M. V. Shah (B)

Applied Mechanics Department, L.D. College of Engineering, Ahmedabad 380015, Gujarat, India
e-mail: mehta.priti67@gmail.com
M. V. Shah
e-mail: drmvs2212@gmail.com
S. R. Patel
e-mail: shivani.patel2996@gmail.com
356 S. R. Patel et al.
1 Introduction
The concept of associating two materials of different strength characteristics to form

a composite material of greater strength is quite familiar with civil engineering prac-
tices. Reinforced earth technique is commonly used in earth retaining structures such
as bridge abutments, retaining walls, steep terrain, etc. It comprises of incorporating
reinforcement such as steel strips, fibers, geogrids, etc., into soil mass in such a way
that it enhances overall structures stability. Mechanically stabilized earth (MSE)
walls, are commonly used structures for retaining the earth, especially in urban areas
[1]. These walls are constructed from the bottom-up by placing alternating layers of
soil and reinforcement. This creates friction between soil and adjacent reinforcement
and thus increasing the strength of the composite material as a whole [2].
As the population has increased, transportation demand has increased so that the
development of urban areas has become a priority, which has led to the widening
of existing highways to improve traffic flow. Widening has been placed in front of
existing stable walls.
The existing state of practice suggests a minimum wall width and MSE rein-
forcement length equal to 70 percent of the wall height [2]. Due to the high cost of
additional rights-of-way and limited space available at job sites, construction of those
walls is often done under a constrained space. This leads to MSE walls narrower than
those in current design guidelines. Narrow MSE walls are referred to as MSE walls
having an aspect ratio L/H (ratio of wall width L to wall height H) of less than 0.7, the
minimum value suggested in Federal Highway Administration (FHWA) MSE-wall
design guidelines [3]. This design guideline for shored mechanically stabilized earth
(SMSE) wall systems [4] provides design methods specifically for wall aspect ratios
ranging from 0.3 to 0.7. A minimum value of L/H (L being the length of reinforce-
ment and H is the height of the wall) equal to 0.3 was recommended for constructing
narrow walls.
The behavior of narrow walls is different from those of conventional walls because
of constrained space and interaction with stable walls. Such differences include the
magnitude of earth pressures and the location of failure planes [5].
Woodruff [6] performed a series of centrifugal tests on RE wall with confined
space [7]. A total of 24 tests were undertaken of wall aspect ratio ranging from 0.17
to 0.9. He observed that when L/H > 0.6, the wall fails in an internal mode with
weaker reinforcement or no failure with stronger reinforcement. If a wall fails, the
critical failure plane is linear and passes through the entire reinforced area. When
L/H is between 0.25 and 0.6, the wall fails internally in a compound mode with
bilinear failure surfaces that pass through the reinforced soil and the edge between
the reinforced backfill and stable wall. The inclination angle of the failure plane in
a compound mode is less than that predicted by the Rankine theory. The compound
failure has a failure surface formed partially through the reinforced soil and partially
along with the interface between the MSE and stabilized wall faces. When L/H <
0.25, failure mode changes from internal to external resulting in overturning failure
Study on Earth Pressure Distribution and Displacements … 357
Table 1 Summary of wall failure modes and corresponding design guidelines [7]
L/H L/H < 0.25 0.25 < L/H < 0.3 0.3 < L/H < 0.6 0.6 < L/H < 0.7 L/H > 0.7
Failure External Compound failure Internal failure
mode
Design Cement FHWA SMSE-wall design FHWA MSE-wall design
method stabilized wall guidelines [4] guidelines [3]
Table 2 Geotechnical
Soil properties Test Determination
properties of sand
Index properties Specific gravity 2.65
Grain size analysis Cc = 1.07
Cu = 2.77
SP
Relative density Umax = 19.10
kN/m3
Umin = 16.95
kN/m3
Engineering Direct box shear c = 0.05 kg/cm2
properties F = 36.9°
of RE wall. The dominant failure modes and current design methods with various
aspect ratios are summarized in Table 1.
2 Laboratory Investigation
Laboratory investigation includes material properties and experimental setup.

Material properties
See Table 2.
Geogrid
Uniaxial geogrid having ultimate tensile strength 40 kN/m in machine direction and
20 kN/m in the cross-machine direction. Mechanical properties of geogrid are shown
in Table 3.
3 Experimental Setup
The model tank was fabricated of size 900 mm * 400 mm * 930 mm using angle
sections and metal sheets. A metal sheet of size 395 mm * 925 mm was fixed at the
base. Two sides of the model tank were fixed with acrylic sheets so that the behavior
Table 3 Mechanical
Mechanical properties Value
properties of geogrid
Ultimate tensile strength-MD (kN/m) 40
Ultimate tensile strength-CMD (kN/m) 20
Creep reduction factor 1.49
Partial factor-installation damage 1.11
Partial factor-environmental effects 1.15
Pull-out interaction coefficient 0.8
of the narrow RE wall can be observed during the test. The front side of the wall is
flexible and the backside of the wall is rigid. The front side of the wall is kept open
for measuring the displacement of the narrow RE wall. The plywood sheet was used
to construct a hollow shoring wall model. Then the hollow wall model is filled with
some filler material to achieve the desired weight. The front side of the shoring wall
has facing better of 1H: 9 V. The backside of the shoring wall is kept vertical. For
proper positioning of the shoring wall inside the tank, capsule welding was done
on both sides of the tank and two angle sections were bolted with a model tank
having the same slope as the shoring wall. Aspect ratio (L/H) for reinforced earth fill
is decided based on the SMSE wall design guideline which states that aspect ratio
for narrow RE wall should range from 0.25 to 0.6. Aspect ratio of the minimum
geogrid reinforcement length (measured at the bottom of the wall) was 400 mm,
corresponding to an aspect ratio (L/H) of approximately 0.44 [8]. The reinforcement
length increased from bottom to top with the largest reinforcement provided at the
top of length 500 mm, corresponding to an aspect ratio (L/H) of 0.55 (Figs. 1, 2, 3
and 4).
Fig. 1 Schematic diagram of narrow RE wall model

Fig. 2 Side view of model tank
Fig. 3 Dial gauge

arrangement
Fig. 4 Fitting of geogrid

with facing panel
4 Test Procedure
Load–displacement and load-settlement tests will be carried out in a narrow RE

wall model to know displacement and settlement characteristics. First, the tank will
be filled in layer by layer by compacting the backfill material to achieve the desired
density. In each layer, the volume of the sand was calculated ahead, and by controlling
the weight of corresponding soil mass to achieve relative density. Geogrid is used as
the reinforcement in the narrow RE wall and placing at the 0.4H (225 mm) vertical
distance of the facing panel. The tank will be filled in four layers and geogrid is
placed in between sand layers. For the unconnected system, geogrid is mechanically
connected with a facing panel and curtail down near the shoring wall. This process
was continued for all three layers of geogrid. Footing plate of size 360 mm × 150 mm
and 25 mm thick was placed at the top of the material having firm contact with the
backfill material to have the uniform transfer of load. Displacements and settlement
measurements can be made using dial gauges of suitable capacity. Totally 6 dial
gauges will be set up in a narrow RE wall model out of which 2 dial gauges will be
set up to measure vertical settlement, 4 dial gauges will be set up at the front side of
the model tank to measure lateral displacement. In flexible condition 1, dial gauge
will be set up at the backside of the model tank. To carry out tests on the narrow RE
wall model, the vertical downward load was applied using a proving ring of 5 tones
capacity. A seating load of 7 kN/m2 was applied on the plate using a mechanical
jack. Proving ring of capacity 50 kN was used to maintain the loading at the constant
interval of the 1.66 kN. When there is no perceptible increase in settlement or when
the rate of settlement reduces to 0.02 mm per minute as given in IS 1888-1982, the
dial gauge reading was noted (Table 4).
Table 4 Dial gauge location

Dial gauge H1 (front H2 (front H3 (front H4 (front H5 (backside)
side) side) side) side)
Horizontal 245 155 245 155 200
distance from
left to right (mm)
Vertical distance 105 305 505 705 300
from top to
bottom (mm)
Fig. 5 Variation of lateral

earth pressure with depth
Table 5 Value of lateral

Z (in m) σh (Rankine’s σh (Coulomb’s σh (Arching
earth pressure in kN/m2
theory) theory) theory)
0.1 0.47 0.463 0.493
0.2 0.941 0.926 0.941
0.3 1.411 1.389 1.345
0.4 1.882 1.852 1.712
0.5 2.353 2.315 2.042
0.6 2.823 2.778 2.342
0.7 3.294 3.241 2.612
0.8 3.764 3.7 2.864
0.9 4.235 4.167 3.081
5 Result and Discussion
5.1 Lateral Earth Pressure (σ h) Distribution
Figure 5 illustrated the distribution of lateral earth pressure by Rankine’s theory,

Coulomb’s theory, and Arching theory for F = 36.9°. From the figure, it is observed
that as depth increases, the value of lateral earth pressure increases linearly in
Rankine’s and coulomb’s theory thus shows a straight-line diagram. While in arching
theory, the lateral earth pressure distribution is nonlinear, thus the shape of lateral
earth pressure distribution is curve type (Table 5).
5.2 Load Versus Displacement Curve for Static Loading
The test was performed on a narrow RE wall model with sand as a backfill with
reinforcement. The result obtained from displacement and settlement at a particular
location of dial gauge is shown in various graphs (Figs. 6, 7, 8 and 9).
At location of dial gauge H5 (Fig. 10).
Fig. 6 Load versus Load (kN)

displacement characteristics 0 5 10 15 20
for static loading in rigid 0
condition and flexible
Displacement (mm)
condition with reinforcement 0.2
at dial gauge H1 0.4
0.6
0.8
1
1.2
Sta c Rigid Sta c flexible

0.2
Displacement (mm)
condition with reinforcement

0.6
0.8
1
1.2
1.4
Sta c rigid Sta c Flexible
Fig. 8 Load versus Load in KN

condition and flexible 0.2
Displacement (mm)

0.8
1
1.2
1.4
1.6
Sta c rigid Sta c flexible
The displacement behavior narrow of RE wall is shown from the above graph at the
location of dial gauge H1, H2, H3, and H4 for backfill material sand with geogrid as
reinforcement. The total displacement for an applied load of 12.78 kN is observed as
for rigid condition and flexible condition is 0.55 mm, 1.15 mm, 1.36 mm, 0.68 mm
and 0.85 mm, 0.96 mm, 0.63 mm, 0.23 mm, respectively. In flexible conditions
the total displacement at the backside of the wall for an applied load of 12.78 KN is
observed as 3.78 mm. In flexible condition, dial gauge shows maximum displacement

0.1
Displacement (mm)
0.2
0.3
0.5
0.6
0.7
0.8
Sta c rigid Sta c flexible

displacement characteristics 0 2 4 6 8 10 12 14 16
for static loading in flexible 0
Displacement (mm)
at dial gauge H5 1
1.5
2
2.5
3
3.5
4
at the backside of the wall and the front side of wall dial gauge shows minimum
displacement compared to a rigid condition.
5.3 Load Versus Displacement Curve for Cyclic Loading
See Figs. 11, 12, 13 and 14.

At the location of dial gauge H5 (Fig. 15).
The displacement behavior of the narrow RE wall is shown from the above graph
at the location of dial gauge H1, H2, H3, and H4 for backfill material sand with
geogrid as reinforcement. The total displacement for an applied load of 11.629 kN
is observing as, for rigid condition and flexible condition is 0.34 mm, 1.09 mm,
1.09 mm, 0.52 mm and 0.5 mm, 0.89 mm, 0.91 mm, 0.37 mm, respectively. In
flexible condition, the total displacement at the backside of the wall for an applied
load of 12.78 KN is observed as 3.57 mm. In flexible condition, dial gauge shows
maximum displacement at the backside of the wall and the front side of the wall dial
gauge shows minimum displacement compared to the rigid condition.

displacement characteristics 0 5 10 15
for cyclic loading in rigid 0
Displacement (mm)
at dial gauge H1
0.2
0.3
0.4
0.5
0.6
Cyclic rigid Cyclic flexible

0.2
Displacement (mm)
0.6
0.8
1.2
1.4
Cyclic rigid Cyclic Flexible

Displacement(mm)
0.6
0.8
1.2
1.4
Cyclic rigd Cyclic flexible


at dial gauge H4
Displacement (mm)
0.2
0.3
0.4
0.5
0.6
0.7
Fig. 15. Load versus Load in KN

displacement characteristics 0 2 4 6 8 10 12 14
for cyclic loading in flexible 0
at dial gauge H5
Displacement (mm)
1
1.5
2
2.5
3
3.5
4
5.4 Load Versus Settlement Curve for Static Loading

and Cyclic Loading
The settlement behavior of the narrow RE wall is shown from the above graph. The
total displacement in static loading for an applied load of 12.78 KN is observing as, for
rigid condition and flexible condition is 6.05 and 13.71 mm. In flexible condition, dial
gauge shows maximum settlement as compared to rigid condition (Figs. 16 and 17).
The total displacement in cyclic loading for an applied load of 11.629 KN is
observing as, for rigid condition and flexible condition is 5.66 and 11.24 mm. In
flexible condition, dial gauge shows maximum settlement as compared to rigid
condition.

settlement characteristics for 0 5 10 15 20
static loading in rigid 0
condition and flexible 2
Se lement (mm)
condition with reinforcement 4
at dial gauge V1 6
8
10
12
14
16
Sta c rigid Sta c Flexible

settlement characteristics for 0 5 10 15
static loading in rigid 0
condition and flexible Se lement (mm) 2
4
at dial gauge V1
6
8
10
12
6 Conclusion
In this paper, a comparison between cyclic loading and static loading is carried out on
a narrow RE wall model with two conditions, i.e., rigid condition, flexible condition.
In flexible condition settlement and displacement of soil is more compared to rigid
condition [9]. Also, displacement of backside of the tank plays a major role in flexible
condition. The dial gauge at the backside of the tank shows the highest displacement
among all dial gauges.
In this study, Janssen’s arching theory was used to provide a reasonable estimate
of the lateral earth pressures on narrow RE wall and compare the results of lateral
earth pressure with conventional earth pressure theories such as Rankine’s theory
and Coulomb’s theory. It is observed that as depth increases, the value of lateral
earth pressure increases linearly in Rankine’s and coulomb’s theory thus shows a
straight-line diagram. While in arching theory, the lateral earth pressure distribution
is nonlinear, thus shape of lateral earth pressure distribution is curve type.
Acknowledgements The authors are very thankful to Prof. (Dr.) R. K. Gajjar, Principal, L. D.
College of Engineering, Ahmedabad, for providing all the necessary research facilities.
References
1. Chao X, Luo Y-S (2016) Effects of interface connections on narrow mechanically stabilized
earth walls. Environ Earth Sci 75:1411
2. Tavakolian R, Grien M (2011) Narrow shored reinforced earth wall with friction-based
reinforcing strip connection, as an innovative solution to expand urban highways. Geo-Frontiers
3. Elias V, Christopher BR, Berg RR (2001) Mechanically stabilized earth walls and reinforced soil
slopes design and construction guidelines. Report No. FHWA-NHI-00-043, National Highway
Institute, Federal Highway Administration, Washington, D.C. March
4. Morrison KF, Harrison FE, Collin JG, Dodds A, Arndt B (2006) Shored Mechanically Stabilized
Earth (SMSE) wall systems design guidelines. Report No. FHWA-CFL/TD-06-001, Federal
Highway Administration, Central Federal Lands Highway Division.
5. Yang KH, Zornberg JG, Hung WY, Lawson CR (2011) Location of failure plane and design
considerations for narrow geosynthetic reinforced soil wall systems. J Geoeng
6. Woodruff R (2003) Centrifuge modeling of MSE-shoring composite walls. Master Thesis,
University of Colorado, Boulder, Colorado
7. Yang KH, Kniss KT, Zornberg JG, Wright SG (2007) Finite element analyses for centrifuge
modelling of narrow MSE walls
8. Xiao C, Han J, Zhang Z (2015) Experimental study on the performance of geosynthetic reinforced
soil model walls on rigid foundations subjected to static footing loading. Geotext Geomem 1–14
9. Paik KH, Salgadoy R(2003), Estimation of active earth pressure against rigid retaining walls
considering arching effects. Geotechnique 53(7):643–653
Seismic Mitigation
Liquefaction––An Extensive Study
on New Concepts
Jiji Krishnan and Shruti Shukla
Abstract Liquefaction of soil significantly affects the life of buildings as well as the
life of humans. Liquefaction develops when the shear strength of the soil is lesser to
resist the shear stresses induced when subjected to dynamic loadings during vibra-
tion or an earthquake. In order to improve the load-bearing capacity so as to mitigate
the liquefaction characteristics, the addition of chemicals in soil can also be used.
Addition of chemicals can be done in two ways. The first method in which void
spaces can be filled by grouting/stabilising material, whereas the second method
is the mechanical stabilisation of external materials. The modifications of soil to
upgrade its properties through grouting technologies are extensively popular these
days. The present study is an attempt to study the non-conventional seismic liquefac-
tion mitigation methods. Also, this study reviews the most significant laboratory tests
with respect to liquefaction mitigation and compares colloidal silica with many other
recent liquefaction mitigation techniques such as bentonite suspension grouting, bio-
cementation, colloidal silica grout and sand–rubber tire shred mixtures. The current
study revealed that the two main grouts, which can be used as a prospective lique-
faction reduction materials in the upcoming era are colloidal silica and bentonite
suspensions.
Keywords Liquefaction mitigation · Grouting · Ground improvement · Colloidal

silica
1 Introduction
Soil liquefaction is a foremost threat for structures built on saturated sandy soils.
More than 300 lives and loss of 200 billion dollars were estimated in the 1995
Kobe earthquake due to liquefaction [1]. The significantly functional technologies
for liquefaction mitigation are vibro compaction, vibro flotation and dynamic
compaction, mixing with wet soil, vertical earthquake drains, and compacted stone
J. Krishnan (B) · S. Shukla

Sardar Vallabhbhai National Institute of Technology, Surat, India
e-mail: jijiktu@gmail.com
370 J. Krishnan and S. Shukla
columns [2–6]. In order to progress the load-bearing capacity so as to mitigate

the liquefaction characteristics, the addition of chemicals in soil also can be used.
Addition of chemicals can be done in two ways. The first method in which void
spaces can be filled by grouting/stabilising material, whereas the second method
is the mechanical stabilisation of external materials. The modifications of soil to
upgrade its properties through grouting technologies are extensively popular these
days. Improving the seismic responses of the ground by various new concepts such as
grouting with colloidal silica and bentonite suspension, bio-cementation, using tire
chips have been popular for the past two decades with the advances of science and
technology. Site remediation without disturbing the existing structures should be the
main concern while selecting a suitable stabiliser for chemical grouting as well as bio
grouting. The research works on liquefaction mitigation by new concepts in ground
improvement is still in its flourishing phase. Regardless of such problems, many
researchers have made an effort to discover the prospective research zones associated
with the new concepts in seismic liquefaction mitigation. In order to improve the
research in this field, every work nevertheless of the amount of work is significant.
The current study makes an effort to summarise all the new phases of soil grouting
and its economic viability. Also, this study reviews the most significant laboratory
tests with respect to liquefaction mitigation and compares colloidal silica with
many other recent liquefaction mitigation techniques such as bentonite suspension
grouting, bio-cementation, colloidal silica grout and sand–rubber tire shred mixtures.
2 Motivation and Objective(s)
Traditional liquefaction mitigation methods such as Vibro flotation, dynamic

compaction, Vibro compaction cause disturbance to the existing structures with
limited areas of applications. Recently developed new methods in mitigating the
hazards of seismic liquefaction such as grouting with colloidal silica grouting
and bentonite, bio-cementation, microfine cement grout, extended set control
admixtures grout, sodium silicate grout, acrylate grout, iron precipitation grout,
epoxy grout and polysiloxane grout are non-disruptive and environmental friendly.
The new methods mentioned above suggest new perceptions and methodologies
for improving the research and explaining present-day engineering problems for
mitigating liquefaction.
The primary purpose of the current study is to review the new soil enhancement
approaches for mitigating liquefaction. This study also aims to review the economic
viability of various grouting methods in mitigating liquefaction for passive site stabil-
isation. Passive site stabilisation is a technique in ground stabilisation of liquefiable
soil under already developed structures. Moreover, the significance of these novel
methods in resolving the reduction of seismic hazard in ground improvement and
future research orientations are also discussed.
Seismic Mitigation Liquefaction … 371
3 Practical Records of New Soil Improvement Methods
3.1 Dynamic Properties of Sand Mixed with Shredded

Rubber
Rubber fibres in soil foundations can absorb low amplitude mechanical waves gener-
ated by vibration caused by machines, locomotives and micro-tremors. Shredded
rubber tires are used in soil mixtures due to its high damping behaviour [7, 8].
Madhusudhan et al. [9] investigated the shear modulus and damping ratio of the
sand mixed with shredded tires at around seven number of loading cycles. They
conducted the strain-controlled two-way cyclic triaxial tests in the sand mixed with
shredded tires under undrained conditions. In order to avoid the segregation on
rubber-sand mixtures and to ensure homogeneity and uniformity, researchers selected
the sizes smaller than 2 mm for both sand as well as rubber particles. The mixtures
of the tires used for the experiment were 0, 10, 30, 50 and 100%. They observed
a reduction in shear modulus until the 5th cycle for all the mixtures, after which it
remains constant (Fig. 1). Also, Fig. 1 shows that during critical earthquakes, sand
mixed with rubber tires becomes softer significantly at large strains which in turn
results in period lengthening. They reported that the 10 wt.% rubber tire mixture in
sand mixture improved static and dynamic properties. The enhancement in properties
of treated sand could be made use for seismic base isolation in low-rise buildings.
Bahadori and Manafi [10] conducted experiments to calculate the properties of
rubber tire chips on the dynamic properties of saturated sands. In their study, four
Fig. 1 Shear strain versus shear modulus of rubber-sand mixtures at different rubber volume
fractions [9]
different mixture ratios of 10, 20, 30 and 40 were used. In order to study the improve-
ment in mitigation as well of sand-tire mixtures, 1 g shaking table prototypical tests
were conducted. They noticed a decrease in settlement caused by liquefaction with
the addition of tire chips in sands. Also, they noticed an increment of mean damping
ratio and decrement of shear modulus with the increment of tire particles in the
reinforced soil.
Senetakis et al. [11] conducted experiments to measure the dynamic properties of
rubber mixed in dry sand as well as gravel in an extensive range of shearing strain
amplitudes. They proposed the generic normalised shear modulus and damping ratio
versus shearing strain amplitude curves for dry mixtures of sand-rubber (0–35%) and
gravel-rubber (0–40%) appropriate for the engineering practice. They also mentioned
the different factors which affect the strain shear modulus and damping ratio of
soil-rubber and gravel-rubber mix. The factors are rubber content, grain size, the
comparative size of soil and rubber and also the dynamic properties of the intact soil.
Ali et al. [12] reported the dynamic properties of dense sand-rubber mixtures with
small particle size ratio. In order to determine the damping ratio and shear modulus,
bender elements were inserted into a triaxial cell. The rubber chips used for the study
were having sizes changing from 0.06 to 0.60 mm and also had the shape of angular to
sub-angular. They investigated the results which had the confining pressures ranging
from 50 to 500 kPa with the addition of less than 50% rubber volume fractions in the
sand. They notice a decrement in shear modulus and increment in damping ratio with
the addition of rubber chips in the sand for all the confining pressures selected. They
also proposed that a 10% rubber volume fraction in sand increases the damping ratio
without substantial loss of rigidity and negligible sensitivity to confining pressure.
3.2 Dynamic Properties of Colloidal Silica Grouted Sand
Colloidal silica is considered to be a potential stabiliser in ground engineering due to

its comparable costs with other grouting methods, fewer disturbances to structures
while stabilising, controllable gel time as well as low viscosity [13–15].
Gallagher and Mitchell [16] investigated the effect of colloidal silica suspension
in loose sand on the improvement of liquefaction potential and cyclic undrained
behaviour. The experimental tests were conducted on sand samples treated with
colloidal silica at four different concentrations (5, 10, 15 and 20%). They investigated
an enhancement in deformation behaviour by adding colloidal silica grout in sand.
In order to study the effects of the curing period, they have conducted tests on
10% colloidal silica-treated sand samples for different curing periods of 11, 21, 32
and 56 days. They noticed that strain level reached during cyclic loading decreases
with an increase in the curing period (Fig. 2). They concluded that cyclic resistance
of colloidal silica-treated sand will progress with time. They also recommended a
concentration of up to 10 wt.% colloidal silica solution in sand to offer satisfactory
stabilisation with tolerable strain levels.
Fig. 2 Cyclic stress ratio versus strain during cyclic loading at different curing times [16]
Kodaka et al. [17] conducted laboratory tests such as monotonic and cyclic
torsional shear test in colloidal silica-treated sand. They noticed that the treated
colloidal silica sand exhibits greater liquefaction strength than the untreated sample.
They have also proposed an elastoplastic model to define the performance of treated
sand. This proposed model as then employed in a finite element analysis program,
which is based on fully coupled effective stress. From the FEM analysis, they
proposed that treating sand with colloidal silica grout is an active method in increasing
the resistance against liquefaction.
Díaz-Rodríguez et al. [18] studied the cyclic strength characteristics of sand
stabilised with colloidal silica grout. They have conducted a series of experimental
tests on cyclic simple shear apparatus on liquefiable silty sand with and without
sample enhancement with colloidal silica grout. The tests were conducted on poorly
graded sand having non-plastic silt. They have noticed that the sand sample treated
with colloidal silica-treated grout substantially improved the liquefaction resistance
compared with untreated soil specimens. Also, upon cyclic loading, a considerable
reduction in shear strain and pore pressure generation was also observed.
Conlee et al. [19] conducted centrifuge tests to study the liquefaction mitigation
using colloidal silica stabiliser in sands. The study was conducted on model geometry
in order to determine the properties of lateral spreading in a 4.8 m thick liquefiable
layer, which was overlain by silty clay sloping towards a central channel. They have
noticed an increment in cyclic resistance ratios and decrement in cyclic shear strains
with the increase in colloidal silica concentrations.
Spencer et al. [20] performed resonant column experiments in order to notice the
dynamic properties such as shear modulus and damping ratios in colloidal silica-
treated sand samples. They have selected a colloidal silica grout percentage of 5, 7, 9
wt.% in sand to study the dynamic properties. They noticed only a lesser improvement
in the shear modulus with the cumulative concentration of colloidal silica in sand.
And, they observed a negligible variation in damping ratio with the addition of
colloidal silica in sand over the strain range examined.
3.3 Dynamic Properties of Bentonite Sand Suspensions
Mohtar et al. [21] investigated the cyclic resistance of sand bentonite specimens. The
sand specimens were mixed with different proportions of the dry mass of bentonite
(0, 3 and 5%) and were tested to find out the cyclic shear strength of bentonite-
treated sand specimens. Also, undrained cyclic triaxial tests were conducted on sand
specimens as well as sand treated with 10% bentonite grout with the addition of
0.5% sodium pyrophosphate. They noticed an increment in resistance of cyclic shear
strength of sand with the incorporation of bentonite suspensions in sand. The tests
of 10% bentonite suspension treated sand show a tenfold increment in the number
of cycles to liquefaction with reference to untreated sand. Also, it has been noted
that after a delay of 12 h, bentonite permeated sand was capable to reconstruct the
resistance beyond the first loading.
Mohtar et al. [22] investigated the effects of cyclic response on bentonite-treated
sand. They have conducted cyclic triaxial as well as resonant column experiments
on sand treated with bentonite grout. Cyclic tests were conducted at the rate of 1
cycle/s with the reversal in stress. Also, in order to simulate the dynamic earthquake
loading in the field, undrained conditions were chosen. They noticed an increase
in the order of magnitude in the number of cycles that produce liquefaction at any
given cyclic stress ratio. Figure 3 shows the cyclic stress ratio in bentonite treated
as well as untreated sand vs the number of cycles to liquefaction obtained from
experimental results as a function of the cyclic stress ratio applied. Also, it compared
the test results of Salgado et al. [23] in which tests were conducted in the same sand.
Resonant column tests indicated that there is a need for strain increment to initiate
excess pore pressure generation.
Rugg et al. [24] investigate the effect of static liquefaction mitigation with the help
of undrained shear tests in sand permeated with bentonite suspensions. A reduction
in excess pore water generation was observed by conducting consolidated undrained
tests. Also, they noticed hardly any undrained instability state, which indicates that
static loading didn’t cause any strain softening. The results from their findings
concluded that permeation grouting by adding 2% sodium pyrophosphate mixed
in bentonite suspensions could be an operational technique for static liquefaction
mitigation.
Yoon and Mohtar [25] conducted tests on sodium pyrophosphate modified
bentonite suspensions to study the dynamic rheological properties for liquefaction
Fig. 3 Cyclic stress ratio in bentonite treated as well as untreated sand versus number of cycles to
liquefaction [22]
mitigation. Sequences of oscillatory shear experiments were performed at a constant

oscillation rate of 1 Hz. Investigations were conducted on varying weight percentages
of bentonite suspensions (5, 7.5, 10 and 12 wt.%) modified with sodium pyrophos-
phate. The results showed a decrement in critical storage modulus with the increase
in sodium pyrophosphate. Also, they have noticed an increase in critical storage
modulus with the increase in resting time. The normalised modulus of the treated
sand specimens was degraded more gradually than in the untreated suspensions,
which is an appropriate property of the suspensions for mitigation of liquefaction.
3.4 Dynamic Response of Liquefiable Sand Induced

by Bio-cementation
Montoya et al. [26] investigated the improvement in dynamic properties of liquefi-

able sand with the addition of microbial-induced calcite precipitation by conducting
dynamic centrifuge model experiments. Cyclic strength, as well as stiffness of the
sand, showed significant improvement after bio-cementation treatment in sands.
Also, a reduction in excess pore water generation and settlements were observed
in treated sand during dynamic loading. Treatment in sand increases the maximum
accelerations at the ground level as well. So great care to be taken while designing the
treatment level to upgrade liquefaction resistance and reducing unwanted amplified
soil surface motions.
Burbank et al. [27] performed tests on sands treated by calcite precipitation to
study the effects of seismic-induced liquefaction. Cyclic triaxial shear testing was
conducted to determine the increment in soil resistance to liquefaction by bio-induced
calcite precipitation. A substantial increment in CSR was noticed in cyclic triaxial
shear tests in treated sands. Calcite precipitation ranges from 2.1 to 2.6% in sand
increases the cyclic resistance ratio to twofold. Whereas calcite precipitation ranges
from 3.8 to 7.4% in sand increases the cyclic resistance ratio to four- to fivefold.
Xiao et al. [28] investigated the effect of bio-cementation on the relative density
and cyclic response of sand with the help of cyclic triaxial tests. Also, 0.5 mol/L
urea mixed with 0.5 mol/L CaCl2 was used for the bio-cementation purpose. Cyclic
triaxial experiments were executed on treated sand with varying relative densities
(10, 50 and 80%). The experiments were performed at a frequency of 1 Hz cyclic
stress ratios varied from 0.125 to 0.500. They noticed that bio-cementation reduces
the generation of excess pore pressure in sand specimens with the advancement of
axial strain. Also, during liquefaction in treatment of calcareous sand, the number
of cycles in sand increases. The increase in relative density, as well as the increase
in the magnitude of bio-cementation, increases the number of cycles to liquefaction.
Figure 4 shows the variation in the cyclic resistance ratio of bio-cemented sand at
different relative densities by varying unit weight. However, they mentioned that the
Fig. 4 Variation in cyclic resistance ratio of bio-cemented sand at different relative densities by
varying unit weight [28]
Table 1 Colloidal silica grouting method versus non-conventional method [29]

Bio-cementation Bentonite Rubber tires Colloidal silica
suspension grout
grouting
Passive site No Yes No Yes
stabilisation
Non-toxic Yes Yes Yes Yes
Process for Calcification by the Bentonite gel Reduction in soil Colloidal silica
liquefaction microorganisms formation saturation by gel formation
mitigation mixing rubber tires
Cost-effective Yes Yes Yes Yes
degree of increase in cyclic resistance with the technique of bio-cementation could

be a more successful soil improvement method than densifying the sand strata for a
given variation in relative density (not including cost considerations).
4 Feasibility Study Analysis of Colloidal Silica Grouting

Method with Other Non-conventional Methods
of Liquefaction Mitigation
The two main grouts which can be used as prospective liquefaction mitigation mate-
rials in the near future are colloidal silica and bentonite suspensions. Huang and
wen [29] performed large-scale tests, and the cost-effective comparison was made
as per the tests in comparison with conventional techniques of soil improvement and
liquefaction mitigation (Table 1).
5 Conclusions
The current research attempts to sweep the entire new concepts and methods on
liquefaction mitigation.
1. New liquefaction mitigation methods are environmental friendly and non-
disruptive, which in turn rectifies the limitations in traditional methods.
2. Economic viability study of the new methods in liquefaction mitigation recom-
mends the use of bio-cementation-bentonite suspensions, shredded rubber tires
and colloidal silica as preferable grouting material.
3. Colloidal silica and bentonite suspensions could be scrutinised as a prospec-
tive grouting material in liquefaction mitigation for passive site stabilisation
techniques.
4. Colloidal silica can be used as a prospective liquefaction mitigation materials in

the near future since it is environmental friendly, comparable cost, less distur-
bance to existing structures during injection, controllable gel time, low viscosity
as well as a lifetime in excess of 25 yrs.
References
1. Werner SD, Dickenson SE, Taylor CE (1997) Seismic risk reduction at ports: case studies and
acceptable risk evaluation. J Waterway Port Coastal Ocean Eng 123(6):246–337
2. Mitchell JK, Baxter CDP, Munson TC (1995) Performance of improved ground during Earth-
quakes. In: Soil improvement for liquefaction hazard mitigation, Geotech special Pub No 49,
ASCE, pp 1–36
3. Marcuson WF, Hadala PF, Ledbetter RH (1996) Seismic rehabilitation of earth dams. J Geotech
Eng ASCE 122(1):7–20
4. Elias V, Welsh J, Wareen J, Lukas R, Collin J, Berg R (2006) Ground improvement methods:
reference manual, vol I. NHI Course No 13204, Federal Highway Administration, Washington,
DC
5. Mitchell JK (2008) Mitigation of liquefaction potential of silty sands. In: From research to
practice in geotechnical engineering, ASCE, pp 433–451
6. Chu J, Varaksin S, Klotz U, Menge P (2009) Construction Process. In: Proceedings of 17th
international conference on soil mechanics and geotechnical engineering, IOS Press, Fairfax,
VA, pp 3006–3135
7. Hazarika H, Yasuhara K, Kikuchi Y, Karmokar AK, Mitarai Y (2010) Multifaceted potentials of
tire-derived three dimensional geosynthetics in geotechnical applications and their evaluation.
Geotext Geomembranes. https://doi.org/10.1016/j.geotexmem.2009.10.011
8. Tafreshi SNM, Norouzi AH (2013) Bearing capacity of a square model footing on sand rein-
forced with shredded tire—an experimental investigation. Constr Build Mater (2012). https://
doi.org/10.1016/j.conbuildmat.2012.04.092
9. Madhusudhan BR, Boominathan A, Banerjee S (2017) Static and large-strain dynamic proper-
ties of sand-rubber tire shred mixtures. J Mater Civ Eng. https://doi.org/10.1061/(ASCE)MT.
1943-5533.0002016
10. Bahadori H, Manafi S (2015) Effect of tyre chips on dynamic properties of saturated sands. Int
J Phys Model Geotech. https://doi.org/10.1680/ijpmg.13.00014
11. Senetakis K, Anastasiadis A, Pitilakis K (2012) Dynamic properties of dry sand/rubber (SRM)
and gravel/rubber (GRM) mixtures in a wide range of shearing strain amplitudes. Soil Dyn
Earthq Eng. https://doi.org/10.1016/j.soildyn.2011.10.003
12. Brara A, Brara A, Daouadji A, Bali A, Daya EM (2017) Dynamic properties of dense sand-
rubber mixtures with small particles size ratio. Eur J Environ Civ Eng. https://doi.org/10.1080/
19648189.2016.1139509
13. Krishnan J, Shukla S (2019) The behaviour of soil stabilised with nanoparticles: an extensive
review of the present status and its applications. Arab J Geosci 12:436. https://doi.org/10.1007/
s12517-019-4595-6
14. Gallagher PM, Pamuk A, Abdoun T (2007) Stabilisation of liquefiable soils using colloidal
silica grout. J Mater Civ Eng 19:33–40. https://doi.org/10.1061/(ASCE)0899-1561(2007)19:
1(33)
15. Consoli (2004) Effect of material properties on the behavior of sand-cement–fiber composites.
Proc Inst Civ Eng Gr Improv 8:77–90. https://doi.org/10.1680/Grim.2004.8.2.77
16. Gallagher PM, Mitchell JK (2002) Influence of colloidal silica grout on liquefaction potential
and cyclic undrained behavior of loose sand. Soil Dyn Earthq Eng 22:1017–1026. https://doi.
org/10.1016/S0267-7261(02)00126-4
17. Kodaka T, Oka F, Ohno Y, Takyu T, Yamasaki N (2005) Modeling of cyclic deformation and
strength characteristics of silica treated sand. Geotech Spec Pub. https://doi.org/10.1061/407
97(172)11
18. Díaz-Rodríguez JA, Antonio-Izarraras VM, Bandini P, López-Molina JA (2008) Cyclic strength
of a natural liquefiable sand stabilized with colloidal silica grout. Can Geotech J 45:1345–1355.
https://doi.org/10.1139/T08-072
19. Conlee CT, Gallagher PM, Boulanger RW, Kamai R (2012) Centrifuge modeling for lique-
faction mitigation using colloidal silica stabilizer. J Geotech Geoenviron Eng 138:1334–1345.
https://doi.org/10.1061/(ASCE)GT.1943-5606.0000703
20. Spencer L, Rix GJ, Gallagher P (2008) Colloidal silica gel and sand mixture dynamic properties.
Geotech Earthq Eng Soil Dyn IV:1–10. https://doi.org/10.1061/40975(318)101
21. Mohtar CSE, Bobet A, Santagata MC, Drnevich VP, Johnston CT (2013) Liquefaction miti-
gation using bentonite suspensions. J Geotech Geoenviron Eng. https://doi.org/10.1061/(ASC
E)GT.1943-5606.0000865
22. El Mohtar CS, Clarke J, Bobet A, Santagata M, Drnevich V, Johnston C (2008) Cyclic response
of a sand with thixotropic pore fluid. Geotech Spec Pub. https://doi.org/10.1061/40975(318)63
23. Salgado R, Bandini P, Karim A (2000) Shear strength and stiffness of silty sand. J Geotech
Geoenviron Eng. https://doi.org/10.1061/(ASCE)1090-0241(2000)126:5(451)
24. Rugg DA, Yoon J, Hwang H, El Mohtar CS (2011) Undrained shearing properties of sand
permeated with a bentonite suspension for static liquefaction mitigation. Geotech Spec Pub.
https://doi.org/10.1061/41165(397)70
25. Yoon J, Mohtar CE (2013) Dynamic rheological properties of sodium pyrophosphate-modified
bentonite suspensions for liquefaction mitigation. Clays Clay Miner. https://doi.org/10.1346/
CCMN.2013.0610411
26. Montoya BM, DeJong JT, Boulanger RW (2013) Dynamic response of liquefiable sand
improved by microbial-induced calcite precipitation. Geotechnique. https://doi.org/10.1680/
geot.SIP13.P.019
27. Burbank M, Weaver T, Lewis R, Williams T, Williams B, Crawford R (2013) Geotechnical
tests of sands following bioinduced calcite precipitation catalysed by indigenous bacteria. J
Geotech Geoenviron Eng. https://doi.org/10.1061/(ASCE)GT.1943-5606.0000781
28. Xiao P, Liu H, Stuedlein AW, Evans TM, Xiao Y (2019) Effect of relative density and bioce-
mentation on cyclic response of calcareous sand. Can Geotech J. https://doi.org/10.1139/cgj-
2018-0573
29. Huang Y, Wen Z (2015) Recent developments of soil improvement methods for seismic
liquefaction mitigation. https://doi.org/10.1007/s11069-014-1558-9

Book (Assessment of Soil Liquefaction Safety..)

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

Book (Assessment of Soil Liquefaction Safety..)

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

Book (Assessment of Soil Liquefaction Safety..)

Uploaded by

Copyright:

Available Formats

Lecture Notes in Civil Engineering

To submit a proposal or request further information, please contact the appropriate

More information about this series at http://www.springer.com/series/15087

Local Site Effects

ISSN 2366-2557 ISSN 2366-2565 (electronic)

Guwahati, India T. G. Sitharam

Rajib Sarkar, IIT(ISM) Dhanbad

Ground Response Analysis with Deep Bedrock Depth

One-Dimensional Ground Response Analysis to Arrive at Surface

GIS-Based Landslide Hazard Zonation and Risk Studies Using

Prof. T. G. Sitharam is currently the Director of Indian Institute of Technology

Prof. Ravi Jakka is working as Associate Professor in the Department of

Prof. L. Govindaraju is currently Professor in the Department of Civil

Priyanka Sharma, M. L. Sharma, and V. A. Sawant

Keywords MASW · Microtremor · Joint fit inversion · Response spectra

P. Sharma (B) · M. L. Sharma

2 Estimation of Bedrock Depth in Indo-Gangetic Plains

3 Study Region: Roorkee (Uttarakhand)

Fig. 2 Roorkee city division

Shear wave velocity is an important parameter for understanding the dynamic

4.1 Multichannel Analysis of Surface Waves (MASW)

• The distance between each pair of receivers d = d 2 –d 1 is known which can be

• Finally, a dispersion curve can be plotted between V R -λ or VR -depth.

4.2 Micro Tremor Method

4.3 Experimental Setup

5 Results and Discussions

6 Ground Response Analysis

(a) Shear Wave Velocity (Vs) in m/s

(b) Shear Wave Velocity (Vs) in m/s

(c) Shear Wave Velocity (Vs) in m/s

Table 1 Description about Input Motions

Type II(Medium Soil) U arkashi Earthquake Chamoli Earthquake

of Uttarkashi earthquake, the amplification is observed around 0.3 s in reference to

Seismic site characterization of Roorkee (Uttarakhand) was carried out by using

Large amplification is observed for the Chamoli earthquake at a period around

1. Anbazhagan P, Sitharam TG (2008) Spatial variability of the weathered and engineering

Himanshu Shukla, Pradeep Muley, and Sajjan Kumar

Keywords Amplification factors · Peak ground accelerations · Ground motion

H. Shukla · P. Muley (B) · S. Kumar

Table 1 Location of sites in AIIMS Campus in Kalyani region

4 Ground Response Analysis

BH-01 BH-02 BH-03

Fig. 3 Acceleration time 0.2

5 Results and Analysis

6 Summary and Conclusion

Table 3 a PGA and Soil

Anushka Joshi, Sateesh K. Peddoju, and Mohit Pandey

Abstract Subsurface shear wave velocity plays an important role in designing

Keywords HVSR · Simulated annealing · Shear wave velocity

u i (x A , ω)u ∗j (x B , ω) ∝ Im[G i j (x A , x B ; ω)] (1)

where G i j (x A , x B ; ω) is the Green’s function defined as the displacement in the ith

E i (x, ω) = ρω2 u i (x, ω)2 ∝ Im[G ii (x, x; ω)] (2)

In this expression ρ is the mass density. Therefore, the HV can be expressed as

In above expression the subscripts 1 and 2 represent the horizontal directions

ans (t) −→ Ans (f) (4)

aew (t) −→ Aew (f) (5)

av (t) −→ Av (f) (6)

E i (x, ω) = ρω2 u i (x, ω)2 ∝ Im[G ii (x, x; ω)] (2)