See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/222569818

A goal programming model for paper recycling system

Article  in  Omega · June 2008

DOI: 10.1016/j.omega.2006.04.014 · Source: RePEc

CITATIONS READS
275 1,593

3 authors:

Rupesh Kumar Pati Prem Vrat

Indian Institute of Management, Kozhikode The Northcap University, Pro- Chancellor, Ptofessor of Eminence & Chief …
32 PUBLICATIONS   609 CITATIONS    199 PUBLICATIONS   4,450 CITATIONS   

SEE PROFILE SEE PROFILE

Pradeep Kumar
Indian Institute of Technology Roorkee
428 PUBLICATIONS   5,967 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Modelling and Analysis for Manufacturing Growth in Indian economy. View project

sustainability in chemical supply chain, agro supply chain and neutraceutical supply chain.. View project

All content following this page was uploaded by Rupesh Kumar Pati on 17 October 2017.

The user has requested enhancement of the downloaded file.

 Omega 36 (2008) 405 – 417
www.elsevier.com/locate/omega

A goal programming model for paper recycling system夡

Rupesh Kumar Patia , Prem Vratb , Pradeep Kumarc,∗
a PPEG, SHAR Centre, Indian Space Research Organisation, Sriharikota, Nellore, Andhra Pradesh-524124, India
b Indian Institute of Technology Roorkee, Roorkee, Uttaranchal- 247667, India
c Mechanical and Industrial Engineering Department, Indian Institute of Technology Roorkee, Roorkee, Uttaranchal-247667, India

Received 26 January 2005; accepted 15 April 2006

Available online 16 November 2006

Abstract
The conflict between economic optimization and environmental protection has received wide attention in recent research programs
for waste management system planning. This has also resulted in a set of new waste management goals in reverse logistics system
planning. The purpose of this analysis is to formulate a mixed integer goal programming (MIGP) model to assist in proper manage-
ment of the paper recycling logistics system. The model studies the inter-relationship between multiple objectives (with changing
priorities) of a recycled paper distribution network. The objectives considered are reduction in reverse logistics cost; product quality
improvement through increased segregation at the source; and environmental benefits through increased wastepaper recovery. The
proposed model also assists in determining the facility location, route and flow of different varieties of recyclable wastepaper in
the multi-item, multi-echelon and multi-facility decision making framework. The use of the model has been illustrated through a
problem of paper recycling in India.
䉷 2006 Elsevier Ltd. All rights reserved.

Keywords: Environmental studies; Mixed integer goal programming; Multicriteria; Reverse logistics; Paper recycling

1. Introduction

Sustainable development is emerging as a dominant paradigm that is likely to play an important role in the design of
any societal and economical policies. According to Petek et al. [1], there are three main requirements for sustainable
development: resource conservation, environmental protection, and social as well as economic development. Reverse
logistics concept of a supply chain provides the best strategy to reduce and reutilize waste.
In the past, various deterministic mathematical programming models have been used for planning waste management
systems. The spectrum of these techniques includes linear programming (LP) [2,3], Mixed integer programming (MIP)
[4,5]. Chang et al. [6,7] and Chang and Wang [8,9] combined the effects of the environmental impacts, such as air
pollution, noise control and traffic congestion, as a set of risk constraint in an economic-oriented locational model for
the solid waste management systems.
Goal programming (GP), proposed by Charnes et al. [10], is most widely used approach within the multi-criteria
decision-making (MCDM). It attempts to combine the logic of optimization in mathematical programming with the

夡 This paper was processed by Guest Editors Margaret M. Wiecek, Matthias Ehrgott, Georges Fadel and José Rui Figueira.
∗ Corresponding author. Tel.: +91 1332 285602; fax: +91 1332 285665, +91 1332 273560.
E-mail addresses: rupesh_pati@rediffmail.com (R.K. Pati), director@iitr.ernet.in (P. Vrat), kumarfme@iitr.ernet.in (P. Kumar).
406 R.K. Pati et al. / Omega 36 (2008) 405 – 417

decision maker’s desire to satisfy several goals. Carlson [11] and Chang [12] used weighted non-linear goal program-
ming to discuss the economic impacts of material recycling on energy recovery facilities. Perlack and Willis [13] further
presented the analysis of multi-objective decision making in waste disposal planning. Chang and Wang [14,15] applied
the compromise programming and goal programming techniques to ease the potential conflict among land filling,
incineration, and recycling in a growing metropolitan region. Short-term planning of vehicle routing and scheduling
problems would be a valuable subsequent analysis after the completion of long-term regional planning for solid waste
management [16,17].

2. Background information of Indian paper industry

The paper industry is one of the key industrial sectors in India. However, the domestic per capita consumption of
paper and board in the country is very low at 6 kilograms (kg) compared to South Asian and world average of 11 and
53 kg, respectively [18]. The pulp and paper industry in India has a tremendous growth potential, which is currently
estimated at 8% per annum. But with the existing resources, Nagwekar [18] have projected a shortage of about 0.7
million tonnes per annum by 2010.
At present several problems impede the future growth prospects of Indian paper industry [19]. These are: (i) increase
in literacy rates, rapidly growing urbanization and increasing economic growth put tremendous pressure on the limited
resources available, (ii) lack of raw material viz. wood; as forest cover is not an abundant resource in India. In 1993
only 19.5% of the total land surface was covered with forest. This forest cover is decreasing rapidly at an annual rate
of 0.6%, (iii) low productivity of Indian forests, i.e. Indian forests grow at maximum rate of 0.5 m3 per hectare as
compared to 2.5 m3 per hectare in Europe and USA, (iv) energy consumption is relatively higher than the international
standard, due to interruptions in production, poor quality of fuel and equipment and relatively low rate of utilization of
wastepaper in the production.
The study on physical composition municipal solid waste collected in some of the Indian cities by Kumar et al.
[20] reveals that the proportion of plastic and paper in waste generated are very significant and needs immediate
attention in order to reduce environmental pollutants. Recycling provides a better option to reduce paper and plastics
wastes. However, the inherent complexity of various priorities in wastepaper recycling system like waste composition
(or segregation at source), waste recovery quantity and the total relevant reverse logistics cost may cause additional
difficulties in decision making. While the task of manufacturing recycled paper at a lower reverse logistics costs may be
an important objective, trade-offs among quality of recycled paper, wastepaper recovery as a social responsibility and
other factors may need to be incorporated in an optimization process. GP was found to be an efficient tool for such an
analysis compared to other mathematical programming techniques. A mixed integer goal programming (MIGP) model
has been proposed for a reverse distribution network facility location problem. The model can be used in the analysis
of tradeoff between the achievements of goals in different combinatorial strategies.

3. Formulation of MCDM model for paper recycling distribution network

The reverse distribution network for the wastepaper recycling is shown in Fig. 1. There are five entities in the
network i.e. vendor–customer (initial source of wastepaper), dealer, godown owner, supplier and the manufacturer.
Vendor–customer represents a bin or collection area where the end user of the finished paper/paper products can
assemble the paper for recycling after its use. Dealer collects the wastepaper (which contains the mixture of relevant
and non-relevant waste) from vendor–customers and supplies it to the godown owner stage, where the segregation
for relevant wastepaper occurs. The supplier collects the relevant wastepaper for recycling from the godown owner
and supplies it to the final customer of the reverse distribution network, i.e. the recycling plant/manufacturer. At each
echelon of the network, numbers of sub-stages/sub-entities exist. The direction of the material flow in the entire network
can be seen in Fig. 1.
Based on the requirements of the paper recycling distribution network, the goal programming model consists of three
objectives/goals. These are explained below:

1. Reverse logistics cost: From the manufacturer’s point of view, it is essential that there is no overrun of the logistics
cost associated with recovery of recyclable wastepaper from various possible sources. Hence the objective is to
 R.K. Pati et al. / Omega 36 (2008) 405 – 417 407

Vendor Godown
dis Customer (V) Dealer (D) Owner (G) Supplier (S) Manufacturer (M)
39
73 Q111I = 24748.57
72 V1 Q 411I = 3212.43 dis 69 65 51 17 63
75 dis 582 478 230 698
37 Q111I = 24748.57
D1 G1

Q341I = 11633.43 dis 950

dis QM311 = 17324
2
87 V2 S1
98 Q411I = 3212.43
10 dis 85 37 89 76 71
47 dis 217 419 143 595
D2 G2
QM152 = 28125
Q342I = 14938
dis 286
dis
93 QM231 = 19006
21 V3 S2
95 QM322 = 1758 QM241 = 4455
97 dis 34 11 27 10 59
69 QM232 = 15953
QM232 = 15953 M
D3 Q336I = 64.00 QM242 = 4164
G3
Q
QM311 = 17324
dis dis 689 137 096 204 QM321 = 2012
41 Q436I = 9100.43 QM331 = 4203
91
QM331 = 4203
80 V4 S3 QM322 = 1758
76 Q344I = 4976.28 Q245I = 5385.71 QM241 = 4455 QM242 = 4164
59
dis 577
Q341I = 11633.43
D4 Q342I = 14938.00 G4
Q344I = 4976.28
Q245I = 5385.71 Q345I = 2433 4.0
dis dis 774 291 726 893
63 Q345I = 24334.0 dis 87 08 08 89 42
78
87 V5 S4
47
QM152 = 28125
56
dis 912
Q336I = 64.00 Q556I = 40178.57
dis D5 G5
Q436I = 9100.43
22 Flow of P1 relevant wastepaper
19
16 dis 202 844 599 726
V6 dis 79 79 97 26 06 Flow of P2 relevant wastepaper
78
03 Q556I = 40178.57
Flow of unsegregated wastepaper

Fig. 1. Reverse logistics distribution network for paper recycling problem with ‘CNW’ priority structure * the distance (in kilometers) between the
sub-entities of successive stages are given along with notation ‘dis’ and the quantities of material flow in the represented over the lines in tonnes.

minimize the reverse logistics cost i.e. (RL)c . The corresponding goal is stated as: minimize the positive deviation
from the planned budget allocated for reverse logistics activities (dc+ ) (Eq. (2)).
2. Non-relevant wastepaper target: The quality of the recycled paper can be improved to certain extent by the separation
of the lower grade (non-relevant) of wastepaper at the source (i.e. vendor–customer stage). The objective may
therefore be stated as, minimizing the quantity of non-relevant wastepaper in the reverse distribution network. The
corresponding goal is stated as: minimize the positive deviation from the maximum limit of non-relevant wastepaper
target (dq+ ) (Eq. (3)).
3. Wastepaper recovery target: The recycling of the paper consumes less energy; conserves the natural resources
viz. wood pulp and decreases the environmental pollution. Hence, the objective of a recycled paper manufacturer
should be to maximize the wastepaper collection at the source. In order to encourage the wastepaper recovery at the
source, the goal may be stated as: minimize the negative deviation from the minimum desired waste collection (de− )
(Eq. (4)).
Therefore, the proposed MIGP model can be formulated as given in Eqs. (1)–(27). The notations used in the model
are listed in Appendix A.

Lexicographically minimize: {dc+ , dq+ , de− } (1)

408 R.K. Pati et al. / Omega 36 (2008) 405 – 417

Subject to:

reverse logistics
(RL)c + dc− − dc+ = TRLC (2)
cost constraint,


  non relevant
QgdvI − QMsgp + dq− − dq+ = WA (3)
wastepaper target,
g d v s g p


 wastepaper
QgdvI + de− − de+ = CT (4)
recovery target,
g d v


 manufacturer’ s demand
QMsgp Dp ∀s, g, p (5)
satisfaction for product‘p’,
s g


  godown owner’ s demand
QgdvI ∗ (1 − w) = QMsgp ∀g (6)
satisfaction for product‘p’,
d v s p

 ⎫
QgdvI TdI ∗ d ∀d ⎪⎪

g v ⎪
⎬
QMsgp Tgp ∗ g ∀g, p annual throughput (7,8,9)

s ⎪
⎪ constraint,
QMsgp Tsp ∗ s ∀s, p ⎪
⎭
g


 supply capacity constraint
QgdvI (SUP)v ∗ v ∀v (10)
of vendor customer,
g d


QMsgp BMsgp ∗ psg ∀s, g, p
route capacity constraint
(11,12)
after segregation,
QMsgp BMsgp ∗ pMs ∀s, g, p


QgdvI BgdvI ∗ I dv ∀g, d, v
route capacity constraint
(13,14)
before segregation,
QgdvI BgdvI ∗ Igd ∀g, d, v
 ⎫
d Od ⎬ limits on the number of

d open dealers and godown (15,16)
g Og ⎭
g owners,
 
s Os limits on the number of

s open suppliers and vendor (17,18)
v Ov
v customers,
 ⎫
I dv Odv ⎬ limits on number

d 
v
of open routes of (19,20)
Igd Ogd ⎭
g d the system,
 R.K. Pati et al. / Omega 36 (2008) 405 – 417 409

 ⎫
psg Osgp ∀p ⎬ limits on number
s g of open routes of
 ⎭ the system,
(21,22)
pMs OMsp ∀p
s


I dv ∈ {0, 1}; Igd ∈ {0, 1}; psg ∈ {0, 1}; pMs ∈ {0, 1};
integrality constraint, (23)
d ∈ {0, 1}; g ∈ {0, 1}; s ∈ {0, 1}; v ∈ {0, 1}

⎫
QgdvI 0 ∀ g, d, v ⎪
⎬
non-negativityconstraint, (24,25)
QMsgp 0 ∀ g, s, p ⎪
⎭
dc+ , dc− , dq+ , dq− , de+ , de− 0

complementary
dc+∗ dc− = 0; dq+∗ dq− = 0; de+∗ de− = 0 (27)
constraints.

In the model, the structural constraints (5), (6) ensure that the demand for ‘pth’ variety of recyclable wastepaper
and unsegregated wastepaper are satisfied at the manufacturer ‘M’ and godown owner ‘g’, respectively. At the godown
owner stage the effect of proportion of non-relevant wastepaper (w) accompanying the unsegregated wastepaper has
also been considered. The presence of non-relevant/undesired wastepaper adversely affects the quality of recycled
paper. The limit on the annual throughput for dealer ‘d’, godown owner ‘g’ and supplier ‘s’ are imposed by constraints
(7)–(9). Restriction on the supply capacity for the source/vendor–customer ‘v’ is imposed by constraint (10). Supply
capacity of source, largely depends on the population and recycling habit of members of the society residing in the area
under consideration. Constraints (11)–(14) describe the route capacity limit on movement of wastepaper in a particular
route from source (vendor–customer ‘v’) to godown owner ‘g’ as well as from godown owner ‘g’ to manufacturer
‘M’. Route capacity depends on the mode of transportation, frequency of transportation and capacity at source/godown
owner. Constraints (15)–(18), limit the number of dealers, godown owners, suppliers and vendor–customers that can
remain open. Restrictions on the number of open routes between source-dealer, dealer-godown owner, godown owner-
supplier and finally supplier-manufacturer are laid down by constraints (19)–(22), respectively. Set of integrality
restrictions for decision variables I dv , Igd , psg , pMs , d , g , s , v is imposed by constraint (23) while constraints
(24)–(26) impose non-negativity restriction on the decision variables QgdvI , QMsgp and the deviational variables
dc+ , dc− , dq+ , dq− , de+ , de− . Finally, the complementary constraints imposed by the deviational variables are shown in
Eq. (27).
A sub-model to represent the total reverse logistics cost for paper recycling is required to support the effective
operation of the above optimization model. The total reverse logistics cost comprises of all relevant logistics cost
incurred for the collection and distribution of the wastepaper from source (vendor–customer) to the ultimate stage in
reverse network i.e. recycling unit/manufacturer stage. It has been assumed in this formulation that the collection as
well as ordering at various facilities follows the fixed period ordering/collection principle Appendix B describes the
components of the reverse logistics cost. The expression for the total reverse logistics cost of paper recycling distribution
network is given as

Total reverse logistics costs = (RL)c

  
  CCdv   1
= ∗ I dv + Fvc ∗ v + ∗ (%Sd ) ∗ CI ∗ QgdvI
v
t dv v g v
2
d d

   
    COgd
+ TgdvI ∗ lgdv ∗ QgdvI + Fdc ∗ d + ∗ Igd
g v g
tgd
d d d
410 R.K. Pati et al. / Omega 36 (2008) 405 – 417

 
 
+ Sgp ∗ QMsgp + Dg ∗ w ∗ QgdvI
s g p g d v
  
 1 
+ ∗ (%Spg ) ∗ Cp ∗ QMsgp + Fgc ∗ g
s g p
2 g
  
   COsgp  
+ ∗ psg + Fsc ∗ s x + TMsgp ∗ lMsg ∗ QMsgp
s g p
tsgp s s g p
  
 1   COMsp
+ ∗ (%Sps ) ∗ Cp ∗ QMsgp + ∗ pMs . (28)
s g p
2 s p
tMsp

GP has the ability to produce Pareto inefficient or dominated solution. In the case of integer GP an objective is Pareto
inefficient if a different integer solution can be found that improves the objective without degrading the value of any
other objective. If no such point exists, then that objective is termed as Pareto efficient. Tamiz et al. [21] proposed
Pareto efficiency detection and restoration techniques for integer goal programming. Hence, after solving the above
MIGP problem, the Pareto optimality/efficiency status of the solution is detected by constructing a new achievement
function from the non-weighted deviational variables. The objective of the new achievement function is to maximize
the sum of deviational variables which are not present in the achievement function of the MIGP problem i.e. maximize
{dc− , dq− , de+ }, and subject to the same set of constraints i.e. constraints (2)–(27). The new solution is then compared
with the initial optimal solution and the objective are then given the relevant Pareto status.

4. Implementation of model and solution procedure

Using the three objectives of the MIGP model, a set of six priority structures can be constructed. Hence, following
priority structures are considered for the purpose of this study: CNW, CWN, WCN, WNC, NWC, and NCW. Where,
the position of the characters (C, W, N) in the sequence indicates the priority assigned to reverse logistics cost, waste
recovery and non-relevant waste goals, respectively. The analysis of the priority structures will assist policy makers
to understand the effect of target values of individual goals on the system behavior and also guide the managers in
deciding the best priority structure for their reverse distribution network under the given condition. The priority structure
mentioned appears to capture the interrelationship among the goals of various actors involved in the paper recycling
system.
The model developed in this paper has been illustrated through a ‘real world’ problem. The cost data were approx-
imated from a survey conducted by authors on paper recycling units in India, which has a considerable number of
small and large-scale paper recycling plants. It is assumed that the manufacturer has demands for two varieties of
wastepaper (relevant) from the reverse logistics network for recycling. Annual demands for the two varieties P 1, P 2
are 50 000 tonnes (T) and 47 000 tonnes (T), respectively. It is estimated that a sum of Indian rupees (INR) one thousand
(103 ) millions is the total reverse logistics cost (TRLC ) available for varieties P 1 and P 2. The minimum desired annual
target waste recovery/collection (CT ) at the source is 150 000 tonnes (T), whereas, the maximum limit of non-relevant
wastepaper in reverse network is 15 000 tonnes (T). The model assumes that the degree of segregation at the source
is 70% i.e. percentage of non-relevant waste reaching godown owner stage is 30% (w = 30%). The number of avail-
able sub-entities for the source/vendor–customer, dealer, godown owner and supplier are six (V 1, V 2 . . . V 6), five
(D1, D2 . . . D5), five (G1, G2 . . . G5) and four (S1, S2 . . . S4), respectively. Other parameters required for solving
the problem were generated from the random number table to capture the wide range of problem structure. To solve the
optimization problem LINDO-32 (version 6.1) software has been used. The solution procedure entails the partitioning
of the objective function according to the priority levels and the sequential solution of the resultant mixed integer
linear programming models. The solution obtained at each priority levels is used as a constraint at the lower level. The
example discussed here is intended to serve as an illustration for applicability of the model to a practical sized problem.
The results of the MIGP reverse logistics problem formulated with priority structure ‘CNW’ and solved by LINDO-
32 (version 6.1) are depicted in Fig. 1. The sub-entities that should be opened or closed for ‘CNW’ priority structure are
represented in Fig. 1 along with the associated path and quantity of flow for different varieties of papers. For the given
problem with ‘CNW’ priority, vendor–customer V 3, dealer D2 and supplier S4 are not essential for wastepaper distri-
 R.K. Pati et al. / Omega 36 (2008) 405 – 417 411

Reverse logistics cost-Non relevant waste-Waste recovery (CNW)

30 G2 non relevant waste

25 G3 wastepaper recovery
20

Deviation from target (%)

15
10
5
G2
0
-5
-10
G3
-15
-20
20 40 60 80 100
% increase in RL cost target

Fig. 2. Effect of target reverse logistics cost (TRLC ) with priority structure ‘CNW’.

bution; hence these sub-entities should be closed. The solid lines represent the flow of unsegregated wastepaper between
the vendor–customer (V ) and godown–owner (G) stage. The dotted lines and bold solid lines respectively denotes the
flow of segregated P 1 and P 2 relevant wastepaper from the godown–owner (G) stage to the manufacturer stage for re-
cycling. The quantities of the flow of wastepaper between the entities are represented over the above mentioned lines in
Fig. 1. In order to check the Pareto efficiency of the solutions generated by the MIGP model, Pareto detection techniques
was applied on the model with ‘CNW’ priority structure by using LINDO-32 (version 6.1) As no other integer points
could be found in the feasible dominating area i.e. the pareto state of the three objectives was found to be efficient, and
hence the optimum MIGP point/solution is classified as Pareto efficient. Similarly, the reverse logistics network design
with other priority structures has been carried out using the proposed MIGP model and the objectives are found to be
Pareto efficient.

5. Results and discussions

Application of the proposed MIGP model to analyze the inter-relationships among the various goals of paper recycling
system resulted in the following observations.

5.1. Resource requirements

It is observed that for the priority structure ‘CNW’, where segregation at source (non-relevant wastepaper target) is
given a higher priority compared to the wastepaper recovery, achievement of latter goal is improved for reverse logistics
cost levels greater than the target value. With a minimum of 60% increase in reverse logistics cost target (TRLC ) all the
desired goals can be achieved (Fig. 2). Increase in TRLC improves the recovery of wastepaper from the source due to
availability of additional resources for collection/recovery.
On the other hand, if the reverse logistics cost is increased by 50% for the priority structure ‘CWN’ all the goals
can be achieved. Prior to 50% increase in TRLC , substantial deviation form non-relevant wastepaper target is observed
(Fig. 3). This decreases the quality of recycled paper. Increase in TRLC facilitates increased segregation at the source and
hence the amount of non-relevant wastepaper decreases to an acceptable quantity. The comparison of the two priority
structures (CNW and CWN) confirms that an increase in target reverse logistics cost (minimum 60%) is necessary for
fulfilling the decision maker’s desire to satisfy the three stated goals irrespective of the priority of non-relevant and
wastepaper recovery goals. This will indirectly benefit the environment as well as improve the quality of wastepaper
reaching the recycling unit.

5.2. Impact of wastepaper recovery at source

The sensitivity analysis with the priority structure ‘WCN’ reveals the effect of change in wastepaper recovery target
on the reverse logistics cost as well as non-relevant wastepaper goals. With the increase in wastepaper recovery target
412 R.K. Pati et al. / Omega 36 (2008) 405 – 417

Reverse logistics cost- Waste recovery- Non relevant waste (CWN)

30 G2 wastepaper recovery
G3 non relevant waste
25
G3

Deviation from target (%)

20

15

10

5
G2
0

-5
10 20 30 40 50 60 70 80
% increase in RL cost target

Fig. 3. Effect of target reverse logistics cost (TRLC ) with priority structure ‘CWN’.

Waste recovery-Reverse logistic cost-Non relevant waste (WCN)

60

Deviation from non relevant waste target

1200
G3 non relevant waste
Deviation from RL cost target (%)

1000 G2 reverse logistics cost 50

40
800
30

(%)
600
20
400 G2
G3 10
200 0

0 -10
0 20 40 60 80
% increase in wastepaper recovery target

Fig. 4. Effect of wastepaper recovery target (CT ) with priority structure ‘WCN’.

Waste recovery- Non relevant waste- Reverse logistics cost (WNC)

190.0 G2 non relevant waste

170.0 G3 reverse logistics cost
150.0
Deviation in target (%)

130.0
110.0
90.0
70.0 G3
50.0 G2
30.0
10.0
-10.0
0 10 20 30 40 50 60 70 80
% increase in wastepaper recovery target

Fig. 5. Effect of wastepaper recovery target (CT ) with priority structure ‘WNC’.

(CT ), steep increase in TRLC and WA was observed. Both the goals were overachieved (Fig. 4). The increase in the
reverse logistics cost could be attributed to the increase in collection, transportation, inventory, segregation and the
disposal costs.
In contrast, ‘WNC’ priority structure (Fig. 5) with non-relevant wastepaper goal given higher priority compared to the
reverse logistics cost goal, overachievement of the prior goal is observed after 20% increase in the wastepaper recovery
target. The reverse logistics cost goal is always overachieved with the increase in waste recovery due to increase in
collection, transportation, inventory, segregation and the disposal costs. In both the cases the overachievement of the
non-relevant wastepaper goal is attributed to the increased quantity of non-relevant wastepaper accompanying the
wastepaper recovered at the source. The implication of the analysis for wastepaper recovery at the source signifies that
a recycling unit with prime concern for social and environmental benefit approach has to invest an increased amount
in reverse logistics activities—may be as a premium for their social concern.
 R.K. Pati et al. / Omega 36 (2008) 405 – 417 413

Non relevant waste-Waste recovery- Reverse logistics cost (NWC)

60
G2 wastepaper recovery
G3 G3 reverse logistics cost
50

Deviation in target (%)

40

30

20

10
G2
0

-10
0 10 20 30 40 50 60 70 80
% increase in non relevant waste target

Fig. 6. Effect of decreased segregation (increased non-relevant waste target) with priority structure ‘NWC’.

Non relevant waste-Reverse logistics cost-Waste recovery (NCW)

G2 reverse logistics cost
10
G2 G3 wastepaper recovery
5
Deviation fromtarget (%)

0
-5
-10
-15
-20
G3
-25
-30
0 10 20 30 40 50 60 70 80
% increase in non relevant waste target

Fig. 7. Effect of decreased segregation (increased non-relevant waste target) with priority structure ‘NCW’.

5.3. Effect of degree of segregation at the source

Decrease in the degree of segregation at the source increases the proportion of non-relevant wastepaper reaching
the godown owner stage. It is observed that for priority structure ‘NWC’ (giving wastepaper recovery a higher priority
compared to reverse logistics cost) overachievement of the reverse logistics cost decreases with increase in the non-
relevant wastepaper target value (Fig. 6). This decrease is due to the reduced cost of segregation at the source. On the
other hand, improvement in wastepaper recovery goal with decrease in degree of segregation at the source is observed for
the priority structure ‘NCW’ (Fig. 7). Thus, it is observed that source separation could provide good quality recyclables
to the recycling industry but could also strongly effect the reverse logistics cost (for ‘NWC’) and waste recovery (for
‘NCW’) goals of the system.

6. Conclusion

Substantial quantities of wastes (recyclable and non-recyclable) are generated every day in the modern era leading to
increase in environmental pollution level. Considerable effort should be directed towards decreasing the environmental
load by recycling. But, multiple goals with appropriate priority structure must be taken into consideration when planning
the recycling network system.
A mixed integer goal programming (MIGP) model proposed in this paper appears to capture the inter-relationships
among the three most important goals identified in the context of paper recycling network system. The goals considered
in the model have economical, social as well as quality implications on the paper recycling industry. The model can be
used to address many of the problems and issues associated with the management of recycling distribution system such
as the need to increase reverse logistics cost for achievement of good quality recyclables by better segregation at source
and benefiting environment through increased wastepaper recovery. The proposed model also assists in determining
414 R.K. Pati et al. / Omega 36 (2008) 405 – 417

the facility location, route and flow of different varieties of recyclable wastepaper in the multi-item, multi-echelon and
multi-facility environment. The use of the model has been demonstrated, through a problem of paper recycling in India.
The results obtained show that the model is a viable tool and can be used to assist in making appropriate decisions
regarding the management of reverse distribution network for paper recycling system.
It is envisaged that the quantitative analysis presented in the paper can play a vital role in the decision making process
of a manufacturing company for recycling paper. The selection of the priority structure of goals could now be justified
quantitatively. Future studies on this topic can be undertaken by inclusion of non-linearities, stochasticity of parameters
in the proposed linear model. The model can also be extended to other similar reverse logistics problem areas involving
the environmental issues and conservation of natural resources such as recycling of plastic wastes.

Appendix A: Notations used in model formulation

Definition
I set of varieties of wastepaper collected by dealer stage from the vendor–customer’s
P set of varieties of wastepaper desired by manufacturer for paper production (p ¥ I)
G set of godown owners
V set of vendor–customers or source of wastepaper
D set of dealers
S set of suppliers
CCdv collection cost per trip for collecting unsegregated wastepaper from the source/vendor–customer ‘v’ by
the dealer ‘d’
tdv cycle time for collection of unsegregated wastepaper by dealer ‘d’ from vendor–customer ‘v’
Fvc fixed operating cost of vendor–customer ‘v’
CI unit value of unsegregated wastepaper
%Sd annual inventory holding cost as percentage of unsegregated wastepaper cost at the dealer ‘d’
TgdvI unit transportation cost of unsegregated wastepaper from vendor–customer ‘v’ via dealer ‘d’ to godown
owner ‘g’
lgdv total distance traveled by unsegregated wastepaper from vendor–customer ‘v’ via dealer ‘d’ to godown
owner ‘g’
Fdc fixed operating cost of dealer ‘d’
COgd unit ordering cost at godown owner ‘g’ for ordering unsegregated wastepaper from dealer ‘d’
tgd cycle time for ordering the unsegregated wastepaper by godown owner ‘g’ from dealer ‘d’
Sgp unit segregation cost of ‘pth’ variety of relevant wastepaper at godown owner ‘g’
Dg unit disposal cost of non-relevant wastepaper at godown owner ‘g’
w proportion of non-relevant wastepaper accompanying initial collection from source
%Spg annual inventory holding cost as percentage of ‘pth’ wastepaper cost at the godown owner ‘g’
Cp unit value of ‘pth’ variety of wastepaper after segregation
Fgc fixed operating cost of godown owner ‘g’
COsgp unit ordering cost at supplier ‘s’ for ordering ‘pth’ variety of wastepaper from godown owner ‘g’
tsgp cycle time for ordering the ‘pth’ variety of wastepaper by supplier ‘s’ from godown owner ‘g’
Fsc fixed operating cost of supplier ‘s’
TMsgp unit transportation cost of ‘pth’ variety of wastepaper from a godown owner ‘g’ via a supplier ‘s’ to the
manufacturer ‘M’
lMsg Total distance traveled by ‘pth’ variety of wastepaper from a godown owner ‘g’ via a supplier ‘s’ to the
manufacturer ‘M’
%Sps Annual inventory holding cost as percentage of ‘pth’ wastepaper cost at supplier ‘s’
COMsp unit ordering cost at manufacturer ‘M’ for ordering ‘pth’ variety of wastepaper from supplier ‘s’
tMsp cycle time for ordering the ‘pth’ variety of wastepaper by manufacturer ‘M’ from supplier ‘s’
Dp demand for the ‘pth’ variety of paper by the manufacturer
TdI annual throughput capacity at dealer ‘d’ for unsegregated wastepaper
Tgp annual throughput capacity at godown owner ‘g’ for ‘pth’ variety of wastepaper
 R.K. Pati et al. / Omega 36 (2008) 405 – 417 415

Tsp annual throughput capacity at supplier ‘s’ for ‘pth’ variety of wastepaper
BMsgp route capacity limit of ‘pth’ wastepaper on route from godown owner ‘g’, supplier ‘s’ and manufacturer
‘M’
BgdvI route capacity limit of unsegregated wastepaper on route from vendor–customer ‘v’, dealer ‘d’ and godown
owner ‘g’
(SUP)v unsegregated wastepaper supply capacity at the source i.e. vendor–customer ‘v’
Od maximum number of permissible dealers that can be operated/opened
Og maximum number of permissible godown owners that can be operated/opened
Os maximum number of permissible suppliers that can be operated/opened
Ov maximum number of permissible vendor–customer/source or collection point that can be operated
Odv maximum number of allowed routes between the dealer stage and vendor–customer stage
Ogd maximum number of allowed routes between the godown owner stage and dealer stage
Osgp maximum number of allowed routes between the supplier stage and godown owner stage for ‘pth’ variety
of wastepaper
OMsp maximum number of allowed routes between the manufacturer stage and the supplier stage for ‘pth’ variety
of wastepaper
TRLC total available budget for reverse logistics activities
WA maximum limit of non-relevant wastepaper permitted in reverse network
CT minimum desired limit of wastepaper collection at the source

Decision variables

QgdvI quantity of unsegregated wastepaper (I) shipped from a source (vendor–customer ‘v’) via a dealer‘d’ to a
godown owner ‘g’.
QMsgp quantity of ‘p’ varieties of relevant segregated wastepaper shipped from a godown owner ‘g’ via a supplier
‘s’ to the manufacturer ‘M’.
I dv binary variable; equal to 1 if unsegregated wastepaper is collected by a dealer ‘d’ from a vendor–customer
‘v’, otherwise 0.
Igd binary variable; equal to 1 if order for unsegregated wastepaper is placed by a godown owner ‘g’ to a dealer
‘d’, otherwise 0.
psg binary variable; equal to 1 if order for segregated/relevant ‘pth’ of wastepaper is placed by a supplier ‘s’
to a godown owner ‘g’, otherwise 0.
pMs binary variable; equal to 1 if order for segregated/relevant ‘pth’ of wastepaper is placed by the manufacturer
‘M’ to a supplier‘s’, otherwise 0.
d binary variable; equal to 1 when a dealer‘d’ is open, otherwise 0.
g binary variable; equal to 1 when a godown owner ‘g’ is open, otherwise 0.
s binary variable; equal to 1 when a supplier‘s’ is open, otherwise 0.
v binary variable; equal to 1 when a source/vendor–customer ‘v’ is open, else 0.

Appendix B: Components of total reverse logistic cost (RL)c

Component description Mathematical formulation

1. Costs from vendor–customer to godown owner stage

  CCdv
Wastepaper collection cost by dealer stage from source(V) tdv ∗ I dv

d v
Fixed cost of operating the vendor–customer stage Fvc ∗ v

v 
Inventory holding cost of unsegregated wastepaper at dealer stage 1
2 ∗ (%Sd ) ∗ CI ∗
g d v
QgdvI
416 R.K. Pati et al. / Omega 36 (2008) 405 – 417

Component description Mathematical formulation


Transportation cost of unsegregated wastepaper from vendor–customer TgdvI ∗ lgdv ∗
(source) to godown owner stage through dealer stage g d v
QgdvI

Fixed cost of operating the dealer stage Fdc ∗ d
d

2. Costs at godown owner stage

  COgd
Ordering cost at godown owner stage, for ordering unsegregated tgd ∗ Igd
wastepaper from dealer stage g d

Segregation cost for relevant wastepaper at godown owner stage Sgp ∗ QMsgp
s 
 g 
p
Disposal cost of non-relevant wastepaper at godown owner stage Dg ∗ w ∗ QgdvI
  1
g d v
Inventory holding cost of segregated wastepaper at godown owner 2 ∗ (%Spg ) ∗ Cp ∗
stage s g p
QMsgp
 c
Fixed cost of operating the godown owner stage Fg ∗  g
g

3. Cost after godown owner stage to manufacturer stage

   COsgp
Ordering cost at supplier stage, for ordering segregated wastepaper from tsgp ∗ psg
godown owner stage s g p

Fixed cost of operating the supplier stage Fsc ∗ s

s
Transportation cost of segregated wastepaper from godown owner to TMsgp ∗ lMsg ∗
manufacturer stage through supplier stage s g p
Q
Msgp
 1
Inventory carrying cost of segregated wastepaper at supplier stage 2 ∗ (%Sps ) ∗ Cp ∗
s g p
QMsgp
  COMsp
Ordering cost at manufacturer stage, for ordering segregated wastepaper tMsp ∗ pMs
from supplier stage s p

References

[1] Petek J, Glavic P. An integral approach to waste minimization in process industries. Resource, Conservation and Recycling 1996;17:169–88.
[2] Hsieh HN, Ho KH. Optimization of solid waste disposal system by linear programming technique. Journal of Resource Management Technology
1993;21(4):194–201.
[3] Lund JR, Tchobanoglous G, Anex RP, Lawver RA. Linear programming for analysis of material recovery facilities. Journal of Environmental
Engineering 1994;120(5):1082–94.
[4] Jenkins L. Parametric mixed integer programming: an application to solid waste management. Management Science 1982;28(11):1271–84.
[5] Kirca O, Erkip N. Selecting transfer station locations for large solid waste systems. European Journal of Operational Research 1988;38:
339–49.
[6] Chang NB, Shoemaker CA, Schuler RE. Solid waste management system analysis with air pollution and leachate impact limitations. Waste
Management Research 1996;14:463–81.
[7] Chang NB, Yang YC, Wang SF. Solid waste management system analysis with noise control and traffic congestion limitations. Journal of
Environmental Engineering 1996;122(2):122–31.
[8] Chang NB, Wang SF. A locational model for the site selection of solid waste management facilities with traffic congestion constraint. Civil
Engineering Systems 1994;11:287–306.
[9] Chang NB, Wang SF. Comparative risk analysis of solid waste management alternatives in a metropolitan region. Environmental Management
1996;20(1):65–80.
[10] Charnes A, Cooper WW, Ferguson R. Optimal estimation of executive compensation by linear programming. Management Science 1955;1:
138–51.
[11] Carlson RL. The impact of materials recycling programs on energy recovery economics. Journal of Resource Management and Technology
1985;15(1):28–36.
 R.K. Pati et al. / Omega 36 (2008) 405 – 417 417

[12] Chang NB. Economic impacts of recycling requirements on waste-to-energy incinerators. In: Proceedings of the 1993 ASCE–CSCE joint
conference on environmental engineering, Montreal, Canada; 1993. p. 1007–14.
[13] Perlack RD, Willis CE. Multiobjective decision-making in waste disposal planning. Journal of Environmental Engineering 1985;111(3):
373–85.
[14] Chang NB, Wang SF. Solid waste management system analysis by multi-objective mixed integer programming. Journal of Environmental
Management 1996;48:17–43.
[15] Chang NB, Wang SF. A goal programming approach for multi-objective optimal planning of an integrated solid waste management system.
Waste Management Research 1997;15:1–14.
[16] Chang NB, Lu HY, Wei YL. GIS technology for vehicle routing and scheduling in the solid waste collection systems. Journal of Environmental
Engineering 1997;123(9):901–10.
[17] Chiplunkar AV, Mehndiratta SL, Khanna P. Optimization of refuse collection systems. Journal of Environmental Engineering 1981;107(6):
1203–10.
[18] Nagwekar V. Paper industry update; www.indianinfoline.com on 11th March, 2005.
[19] Beukering P.van, Duraiappah A. The economic and environmental impacts of the waste paper trade and recycling in india: a material balance
approach. CREED working paper series no 10. Institute for Environmental Studies, Amsterdam, 1996.
[20] Kumar S, Gaikwad SA, Sehkdar AV, Kshirsagar PS, Singh RN. Estimation method of national methane emission from solid waste landfills.
Atmospheric Environment 2004;38:3481–7.
[21] Tamiz M, Mirrazavi SK, Jones DF. Extensions of pareto efficiency analysis to integer goal programming. Omega: International Journal of
Management Science 1999;27:179–88.

View publication stats