USING OF THE FUZZY TOPSIS AND FUZZY AHP METHODS

FOR WASTEWATER TREATMENT PROCESS SELECTION

Key words: Fuzzy TOPSIS; Fuzzy AHP; MCDM; Anaerobic wastewater treatment process; Industrial
estates.

1. INTRODUCTION

Appropriate treatment process selection is an important issue before designing and implementing each
wastewater treatment plant (WWTP). According to quantity diversity of industrial wastewater and local condition
of effluent sources, it is impossible to use general criteria in treatment process selection. However, some points are
available in process selection which is applied for almost all kinds of industrial wastewater to achieve the prior
treatment process. The general procedure for making process selection usually consists of the following steps:
 Decide on the criteria that will be used to evaluate alternatives.
 Identify criteria that are important.
 Develop treatment alternatives.
 Evaluate alternatives and select the best one.
Designers should consider both quantitative and qualitative criteria very well [1-3]. There are many
criteria that influence the treatment process selection. In this study, following four criteria and their sub-criteria
take into consideration.
1) Technical criteria, such as performance, reliability, process applicability, resistance to hydraulic
shocks, consistency to organic loading shocks, adaptability, compatibility, less electromechanical facilities.
2) Economic criteria, such as capital cost, operation & maintenance cost, sludge disposal cost, land
requirement and energy requirement.
3) Environmental criteria, such as treatment degree requirement, odor generation, visual and safety.
4) Administrative criteria, such as technical skills requirements, simple operation, simple maintenance,
local availability to facilities and stability of wastewater treatment plant operation and use of online monitoring.
In real-world situation, the evaluation data of the treatment process suitability for various subjective
criteriaand the weights of the criteria are usually expressed in linguistic terms. So, to efficiently resolve the
ambiguityfrequently arising in available information and do more justice to the essential fuzziness in human
judgment andpreference, the fuzzy set theory has been used to establish multiple criteria decision-making
(MCDM) problems [4].In this paper, Technique for Order Preference by Similarity to Ideal Solution
(TOPSIS) and Analytical Hierarchy Process (AHP) methods in a fuzzy environment are proposed for treatment
process selection, in whichthe ratings of various alternatives under various subjective criteria and the weights of
all criteria are represented by fuzzy numbers. Although, there are many studies in the literature that use fuzzy
TOPSIS and fuzzy AHP methodsfor different MCDM problems; however, we propose these methods for the
anaerobic treatment process selection.
Then, the related results are compared with each other.
There are many applications of fuzzy TOPSIS in the literature. Chu (2002) presented a fuzzy TOPSIS
method under group decisions for solving the facility location selection problem [5]. Chen et al. (2006) presented
a fuzzy TOPSIS approach to deal with the supplier selection problem in a supply chain system [6]. Yang and
Hung (2007) used TOPSIS and fuzzy TOPSIS methods for a plant layout design problem [7].
 The AHP method, which was first introduced by Saaty (1980), is an effective method for solving MCDM
problems [8]. It has been widely used for multi-criteria decision making and applied to many practical problems
successfully. Traditional AHP requires exact or crisp judgments. However, due to the complexity and uncertainty
involved in real-world decision problems, decision makers may be more reluctant to provide crisp judgments than
fuzzy ones. Furthermore, even when they use the same words, individual judgments of events are invariably
subjective, and the interpretations attached to the same words may differ. This is why fuzzy numbers and fuzzy
sets have been introduced to characterize linguistic variables used to represent the imprecise nature of human
cognition when we try to translate people’s opinions into spatial data. The preferences in AHP are essentially
human judgments based on human perceptions, so fuzzy approaches allow for a more accurate description of the
decision-making process [9]. A number of methods have been developed to handle fuzzy AHP. The first study of
fuzzy AHP was proposed by Van Laarhoven and Pedrycz (1983) who compared fuzzy ratios described by
triangular fuzzy numbers [10]. Anagnostopoulos et al., (2007) performed the fuzzy extension of AHP in order to
evaluate alternative wastewater treatment process with the use of economic, environmental and social criteria [11].

2. MATERIALS AND METHODS

To consider anaerobic treatment processes and related efficiency in industrial estates, a field study is
carried out. The data analysis and related questionnaires are used for determining the processes efficiency. Process
selection criteria have been issued on the basis of objectivity in industrial estates and the fuzzy TOPSIS and fuzzy
AHP methods specified for the processes assessment and selection.

2.1. Anaerobic Treatment Alternatives

This paper considers five anaerobic treatment processes, which are operating in Iran industrial estates.
These are as follows: 1) Up-flow Anaerobic Sludge Blanket (UASB) operating in 9 industrial estates in Iran; 2)
Up- flow Anaerobic Fixed Bed (UAFB),operating with 7 reactors currently; 3) Anaerobic Baffled Reactors (ABR)
used as an anaerobic system of many treatment plants in Iran's industrial estates; 4) Contact anaerobic process
operating as an anaerobic system of Abbarik Industrial estate's treatment plant successfully, and designing in some
other estates; and 5) Anaerobic lagoons operating in some wastewater treatment plants of industrial estates in Iran.

2.2. Fuzzy TOPSIS Method

The TOPSIS method was firstly proposed by Hwang and Yoon (1981). The basic concept of this
method is that the chosen alternative should have the shortest distance from the positive ideal solution and the
farthest distance from a negative ideal solution [12]. A positive ideal solution is a solution that maximizes the
benefit criteria and minimizes cost criteria; whereas, a negative ideal solution maximizes the cost criteria and
minimizes the benefit criteria [13]. In the classical TOPSIS method, the weights of the criteria and the ratings of
alternatives are known precisely and crisp values are used in the evaluation process. However, under many
conditions crisp data are inadequate to model real-life decision problems. Therefore, the fuzzy TOPSIS method is
proposed, in which the weights of criteria and ratings of alternatives are evaluated by linguistic variables
represented by fuzzy numbers to deal with the deficiency in the traditional TOPSIS [14].
This paper presents an extension of the TOPSIS method proposed by Chen (2000) and Chen et al.
(2006). The related algorithm can be described as follows [6, 15]:
Step 1: A committee of the decision-makers is formed. Fuzzy rating of each decision maker, Dk= (k=1,
~
2,...,K), can be represented as triangular fuzzy number Rk  (k  1, 2,..., K ) with membership function μR~k (x)
Step 2: Criteria evaluation is determined.
Step 3: After that, appropriate linguistic variables are chosen for evaluating criteria and alternatives.
Step 4: Then the weight of criteria are aggregated. The aggregated fuzzy rating can be determined by:
~
R  (a,b,c), k  1,2,..., K .
K
where, a  min ak , b   b , c  max c 
1
k K

k 1
k
k
k (1)

k
  1 K
aij  min a ijk , bij  K 
k
b , cij  max cijk
1 ijk
  (2)

 
k
~
Then, the aggregated fuzzy weights w of each criterion are calculated by:
ij

w~ j  w j1, wj 2 , w j3 

(3)

wj1  minw jk 1, wj 2   w jk 3 

1 K
where,
k
 wjk 2 , wj3  max
k k 1 k
(4)

Step 5: Then the fuzzy decision matrix is constructed.

Step 6: The above matrix is normalized.
Step 7: Considering the different weight of each criterion, the weighted normalized decision matrix is
computed by multiplying the importance weights of evaluation criteria and the values in the normalized fuzzy
decision matrix.
 Step 8: Then, the fuzzy positive ideal solution (FPIS, A ) and fuzzy negative ideal solution (FNIS, A ) are
*
determined by:
A*  (v~* , ~
v * ,..., ~
v *) , (5)
1 2 n
A  (v~1 , v~2 ,..., ~v n ) (6)
Where, v  maxv 
~* and v   min v
~ 
j ij 3 j ij1
i i
i=1,2,…,m ; j=1,2,…,n
Step 9: Then, the distance of each alternative from FPIS and FNIS are calculated by:
d *   d (~
n
v ,~
v *) i=1, 2, …, m (7)
i v ij j
j 1

d   d (~
n

v ,~
v ) i=1, 2, …, m (8)
i v ij j
j1
where, dv(.,.) is the distance measurement between two fuzzy numbers.
Step 10: A closeness coefficient (CCi) is defined to rank all possible alternatives. The closeness
coefficient represents the distances to the fuzzy positive ideal solution (A* and fuzzy negative ideal solution
)
( A ) simultaneously. The closeness coefficient of each alternative is calculated by:
d i ,
CCi  i=1, 2, …, m (9)
d *  d 
i i
Step 11: According to the closeness coefficient, the ranking of the alternatives can be determined.
Obviously, according to Eq. (9), alternative Ai will be closer to FPIS and farther from FNIS as CCi approaches
to1.

2.3. Fuzzy analytic hierarchy process

In the fuzzy extension of AHP, the weights of the nine level fundamental scales of judgments are
expressed via the triangular fuzzy numbers (TFN) in order to represent the relative importance among the
hierarchy's criteria [16]. A TFN is fully characterized by a triple of real numbers (l, m, u), where parameter m
gives the maximal grade of the membership function µ(x), and parameters l and u are the lower and upper bounds
that limit the field of the possible evaluation [16-17].
 (x  l) /(m  l ) x [l, m]

μ (x)  (u
  x) /(u  m) x [m, u] (10)

 0 otherwise
The Fuzzy AHP method is a popular approach for MCDM that has been widely used in the
literature. In this paper, the extent fuzzy AHP is utilized, which was originally introduced by Chang
(1996) [18]. The steps of the Chang’s (1996) extent analysis can be given as follows:
Step 1: The value of the fuzzy synthetic extent with respect to the i-th object is defined by:
m n m 
1

Si   M j    M j  (11)
gi gi
j 1  i1 j1 
where, all the M gj (j=1,2,…,m) are triangular fuzzy numbers.
i

Step 2: As M1= (l1, m1, u1) and M2= (l2, m2, u2) are two triangular fuzzy numbers, the degree of possibility
of M1≥ M2 is defined by:

1, if m2  m1
V (M  M )   u (12)
2 1 0, if l1 2
 l1  u2
 , otherwise
 (m2  u2 )  (m1  l1 )
Step 3: To compare M1and M2, we need both the values of V(M1≥ M2) and V(M2≥ M1).
The degree possibility for a convex fuzzy number to be greater than k convex fuzzy numbers
Mi (i  1,2,...,k) can be defined by:
V (M  M1, M2,...,Mk ) (13)
 V [(M  M 1 ) and (M  M 2 ) and...and (M  M k )
 min V (M  M i ) , i  1,2,..., k
 Assume that: d ( Ai )  minV (Si  Sk ) For k =1,2,...,n ; k≠i.
Then the weight vector is given by:
W   (d ( A ), d ( A ),...,d ( A ))T (14)
1 2 n

where, Ai (i  1,2,..., n) are n elements.

Stop 4: Via normalization, the normalized weight vectors are:
W  (d ( A , d ( A ),...,d ( A ))T (15)
1 2 n
where, W is a non-fuzzy number.

3. RESULTS AND DISCUSSION

Combined anaerobic–aerobic processes have more attended due to industrial unit and related wastewater
diversity, and quantitative and qualitative variation of effluents. According to object of this paper, a different
method of operating anaerobic processes in wastewater treatment plants of industrial estates in Iran have been
surveyed. This process including UASB, UAFB, ABR, contact process and anaerobic lagoons has investigated.
Fuzzy TOPSIS and fuzzy AHP methods as assessment tools have been used to select an effective treatment
process.
Tables 1 and 2 illustrate linguistic variables and related triangular fuzzy numbers to assess the criteria
weight importance and alternatives grading, respectively.

Table 1. Linguistic variables for the importance weight of each criterion

Triangular fuzzy
Linguistic variables numbers
Very low (VL) (0, 0, 0.2)
Low (L) (0.1, 0.2, 0.3)
Medium low (ML) (0.2, 0.35, 0.5)
Medium (M) (0.4, 0.5,0.6)
Medium high (MH) (0.5, 0.65, 0.8)
High (H) (0.7, 0.8, 0.9)
Very high (VH) (0.8, 1,1)

Table 2. Linguistic variables for ratings

Triangular fuzzy
Linguistic variables numbers
Very poor (VP) (0, 0, 2)
Poor (P) (1, 2, 3)
Medium poor (MP) (2, 3.5, 5)
Fair (F) (4, 5, 6)
Medium good (MG) (5, 6.5, 8)
Good (G) (7, 8, 9)
Very good (VG) (8, 10, 10)

3.1. Application with a Fuzzy TOPSIS method

In this section, fuzzy TOPSIS method is used for the anaerobic treatment process selection. Decision- makers
evaluated the importance of criteria by using the linguistic variables which are illustrated in Table 1. The importance weights
of the criteria are shown in Table 3.
Linguistic variables presented in Table 2 are used for rating of alternatives assessment regarding to each criterion.
Regarding to the mentioned baselines, grades of 5 alternatives have been issued according to four criteria as shown in Table
4. Then, the fuzzy decision matrix is formed on the basis of triangular fuzzy numbers related to criteria and alternatives.
Finally, the fuzzy weights of alternatives are determined. Table 5 shows the result of the mentioned functions. The normalized
fuzzy decision matrix is formed as shown in Table 6. Finally, the weighted normalized fuzzy decision matrix is formed on the
basis of Table 6 and the related results are presented in Table 7.

Table 3. Importance weight of the criteria from three decision-makers

Linguistic variables Triangular fuzzy

Criteria DM1 DM2 DM3 numbers
Technical VH H VH (0.7, 0.93,1)
Economical H MH MH (0.5, 0.7, 0.9)
Environmental H VH VH (0.7, 0.93,1)
Administrative MH H H (0.5, 0.75, 0.9)
 Table 4. Ratings of alternatives by the decision-makers under four criteria

Linguistic variables Triangular fuzzy

Criteria Alternatives
DM1 DM2 DM3 numbers
UASB MG MG MG (5, 6.5, 8)
UAFB G G G (7, 8, 9)
Technical ABR MG G G (5, 7.5, 9)
Contact Process G MG MG (5, 7, 9)
Anaerobic Lagoon G G G (7, 8, 9)
UASB G G G (7, 8, 9)
UAFB G G G (7, 8, 9)
Economical ABR VG VG VG (8, 10, 10)
Contact Process MG MG MG (5, 6.5, 8)
Anaerobic Lagoon F F F (4, 5, 6)
UASB G MG MG (5, 7, 9)
UAFB G G G (7, 8, 9)
Environmental ABR MG MG G (5, 7, 9)
Contact Process MG MG MG (5, 6.5, 8)
Anaerobic Lagoon F F F (4, 5, 6)
UASB F F F (4, 5, 6)
UAFB MG MG MG (5, 6.5, 8)
Administrative ABR MG G G (5, 7.5, 9)
Contact Process G G G (7, 8, 9)
Anaerobic Lagoon G MG G (5, 7.5, 9)

Table 5. Fuzzy decision matrix and fuzzy weights of alternatives

Alternatives
Criteria Contact Anaerobic Weight
UASB UAFB ABR
Process Lagoon
Technical (5, 6.5, 8) (7, 8, 9) (5, 7.5, 9) (5, 7, 9) (7, 8, 9) (0.7, 0.93,1)
Economical (7, 8, 9) (7, 8, 9) (8, 10, 10) (5, 6.5, 8) (4, 5, 6) (0.5, 0.7, 0.9)
Environmental (5, 7, 9) (7, 8, 9) (5, 7, 9) (5, 6.5, 8) (4, 5, 6) (0.7, 0.93,1)
Administrative (4, 5, 6) (5, 6.5, 8) (5, 7.5, 9) (7, 8, 9) (5, 7.5, 9) (0.5, 0.75, 0.9)

Table 6. Normalized fuzzy decision matrix

Contact
Criteria UASB UAFB ABR Anaerobic Lagoon
Process
Technical (0.5, 0.65, 0.8) (0.7, 0.8, 0.9) (0.5, 0.75, 0.9) (0.5, 0.7, 0.9) (0.7, 0.8, 0.9)
Economical (0.7, 0.8, 0.9) (0.7, 0.8, 0.9) (0.8, 1, 1) (0.5, 0.65, 0.8) (0.4, 0.5, 0.6)
Environmental (0.5, 0.7, 0.9) (0.7, 0.8, 0.9) (0.5, 0.7, 0.9) (0.5, 0.65, 0.8) (0.4, 0.5, 0.6)
Administrative (0.4, 0.5, 0.6) (0.5, 0.65, 0.8) (0.5, 0.75, 0.9) (0.7, 0.8, 0.9) (0.5, 0.75, 0.9)

Table 7. Weighted normalized fuzzy decision matrix

Contact Anaerobic
Criteria UASB UAFB ABR Process Lagoon
Technical (0.35,0.6,0.8) (0.49,0.74,0.9) (0.35,0.7,0.9) (0.35,0.65,0.9) (0.49,0.74,0.9)
Economical (0.35,0.56,0.81) (0.35,0.56,0.81) (0.4,0.7,0.9) (0.25,0.45,0.72) (0.2,0.35,0.54)
Environmental (0.35,0.65,0.9) (0.49,0.74,0.9) (0.35,0.65,0.9) (0.35,0.6,0.8) (0.28,0.46,0.6)
Administrative (0.2,0.38,0.54) (0.25,0.49,0.72) (0.25,0.56,0.81) (0.35,0.6,0.81) (0.25,0.56,0.81)

After the weighted normalized fuzzy decision matrix is formed, the fuzzy positive ideal solution (FPIS) andfuzzy
negative ideal solution (FNIS) are determined by:
A* = [(0.9,0.9,0.9),(0.9,0.9,0.9),(0.9,0.9,0.9),(0.81,0.81,0.81)]
A- = [(0.35,0.35,0.35),(0.2,0.2,0.2),(0.28,0.28,0.28),(0.2,0.2,0.2)]
Then, the distance of each alternative from the FPIS and FNIS with respect to each criterion is calculatedby using
the vertex method by:

d (A1 , A*) 
1
(0.9  0.35)2  (0.9  0.6)2  (0.9  0.8)2   0.37
3

d ( A1 , A ) 
1
3

(0.35  0.35)2  (0.35  0.6)2  (0.35  0.8)2  0.30 
Here only the calculation of the distance of the first alternative to the FPIS and FNIS for the first criterion is shown,
as the calculations are similar in all steps. The results of all alternatives’ distances from the FPIS and FNIS are shown in Tables
8 and 9.
 Table 8. Distances between alternatives and A* with respect to each criterion

Technical Economical Environmental Administrative

*
d(A1,A ) 0.37 0.38 0.35 0.46
*
d(A2,A ) 0.25 0.38 0.25 0.38
*
d(A3,A ) 0.34 0.31 0.35 0.35
*
d(A4,A ) 0.35 0.47 0.37 0.29
*
d(A5,A ) 0.25 0.55 0.47 0.35

Table 9. Distances between alternatives and A− with respect to each criterion

Technical Economical Environmental Administrative

-
d(A1,A ) 0.30 0.42 0.42 0.22
-
d(A2,A ) 0.40 0.42 0.46 0.34
-
d(A3,A ) 0.38 0.51 0.42 0.41
-
d(A4,A ) 0.36 0.34 0.36 0.43
-
d(A5,A ) 0.40 0.21 0.21 0.41

Then closeness coefficients of alternatives are calculated by Eq. (6). According to the closeness coefficient of
alternatives, the ranking order of alternatives is determined. The first alternative is determined as the most appropriate
anaerobic process for industrial estates. Value of this parameters and final ranking order of alternatives are presented in Table
10.

Table 10. Computation of di*, di- and CCi and the rating order of alternatives

Contact Anaerobic
UASB UAFB ABR process lagoon Ranking order
*
di 1.55 1.26 1.36 1.47 1.63
di
- UAFB > ABR > Contact process
1.36 1.62 1.72 1.48 1.24
CCi > UASB > Anaerobic lagoon
0.467 0.563 0.559 0.502 0.431

3.2. Application with fuzzy AHP method

In this section, the fuzzy AHP method is proposed for the same problem of the anaerobic treatment process. A group
decision is used based on fuzzy AHP. Firstly, each decision-maker (Dp), individually carry out pair-wise comparison as shown
in Eq. (16):
 b11 p b12 p b1mp  
b 
b221 p b2mp 
Dp   p=1, 2, …, t (16)
21 p
 ⁝ ⁝ ⋱ ⁝ 
 
bm1 p bm2 p bmmp 
The pair-wise comparison of three decision-makers for the four criteria is as follows:
C1 C2 C3 C4
C1 1 3 3 5
1/ 3 
D1  C2 
1 1 3
C3 1/ 3 1 1 3
 
C4 1/ 5 1/ 3 1/ 3 1
C1 C2 C3 C4
C1  1 3 1/ 3 1 
 
D2  C2 1/ 3 1 1/ 5 1/ 3
C3  3 5 1 3 
 
C 4  1 3 1/ 3 1 
 C1 C2 C3 C4
 C1  1 5 1 3 
 
D3  C2 1/ 5 1 1/ 5 1/ 3
C3  1 1/ 5 1 3 
 
C4 1/ 3 3 1/ 3 1 



 Then, a comprehensive pair-wise comparison matrix is built as shown in Table 11 by integrating the
grades of three decision-makers via Eq. (17). By this way, the pair-wise comparison values of the decision-makers
are transformed into triangular fuzzy numbers.
t

b
p 1
jep

l je  min(bjep ) , mje  , uje  max(b jep ) (17)

p
~ p=1,2,…,t j=1,2,…,m e=1,2,…,m
bje  (l je , mje ,uje ) , j=1,2,…,m e=1,2,…,m

Table 11. Fuzzy evaluation matrix with respect to goal

Tech. Criteria (C1) Eco. Criteria (C2) Env. Criteria (C3) Admin. Criteria (C4)
Tech. Criteria (C1) (1, 1, 1) (3, 3.67, 5) (0.33, 1.44, 3) (1, 3, 5)
Eco. Criteria (C2) (0.2,0.29, 0.33) (1, 1, 1) (0.2, 0.47, 1) (0.33, 1.22, 3)
Env. Criteria (C3) (0.33, 1.44, 3) (0.2, 2.07, 5) (1, 1, 1) (3, 3, 3)
Admin. Criteria (C4) (0.2, 0.51, 1) (0.33, 2.11, 3) (0.33, 0.33, 0.33) (1, 1, 1)

From Table 11, according to the extent analysis synthesis, the value respect to the main goal is calculated
as the same as Eq. (11):
Sc1  (5.33, 9.11, 14)  (1/ 36.66, 1/ 23.55, 1/13.45) = (0.145, 0.387, 1.041)
Sc 2  (1.73, 2.98, 5.33)  (1/ 36.66, 1/ 23.55, 1/13.45) = (0.047, 0.127, 0.396)
Sc3  (4.53, 7.51, 12)  (1/ 36.66, 1/ 23.55, 1/13.45) = (0.124, 0.387, 0.892)
Sc 4  (1.86, 3.95, 5.33)  (1/ 36.66, 1/ 23.55, 1/13.45) = (0.051, 0.168, 0.396)
These fuzzy values are compared by using Eq. (12), and these values are obtained by:
V (Sc1  Sc 2 )  1 , V (Sc1  Sc3 )  1, V (Sc1  Sc4 )  1
V (Sc2  Sc1)  0.491, V (Sc 2  Sc3 )  0.511, V (Sc 2  Sc 4 )  0.894
V (Sc3  Sc1 )  1 , V (Sc 3  Sc 2 )  1 , V (Sc3  Sc 4 )  1
V (Sc 4  Sc1)  0.534, V (Sc 4  Sc 2 )  1, V (Sc 4  Sc 3 )  0.554
Then, priority weights are calculated by using Eq. (13):
d (C1 )  min(1, 1, 1)  1
d (C2 )  min(0.491, 0.511, 0.894)  0.491
d (C3 )  min(1, 1, 1)  1
d (C4 )  min(0.534,1, 0.554)  0.534
Priority weights form W   (1, 0.491, 1, 0.534 ) vector. After the normalization of these values priority
weight respect to main goal is calculated as (0.331, 0.162, 0.331, 0.176). After the priority weights of the criteria are
determined, the priority of the alternatives is determined for each criterion. From the pair-wise comparison of the
decision makers for five alternatives, evaluation matrixes are formed as illustrated in Tables 12 to 15. Then, priority
weights of alternatives for each criterion are determined by making the same calculation as shown Table 16.

Table 12. Fuzzy evaluation matrix with respect to technical criteria (C1)

Anaerobic
Tech. Criteria (C1) UASB UAFB ABR Contact process Lagoon
UASB (1, 1, 1) (0.56, 0.81, 1.14) (0.56, 0.87, 1.6) (0.56, 0.93, 1.6) (0.56, 0.81, 1.14)
UAFB (0.88, 1.23, 1.8) (1, 1, 1) (0.78, 1.07, 1.8) (0.78, 1.14, 1.8) (0.78, 1, 1.29)
ABR (0.63, 1.15, 1.8) (0.56, 0.94, 1.29) (1, 1, 1) (0.56, 1.07, 1.8) (0.56, 0.94, 1.29)
Contact process (0.63, 1.08, 1.8) (0.56, 0.88, 1.29) (0.56, 0.93, 1.8) (1, 1, 1) (0.56, 0.88, 1.29)
Anaerobic Lagoon (0.88, 1.23, 1.8) (0.78, 1, 1.29) (0.78, 1.07, 1.8) (0.78, 1.14, 1.8) (1, 1, 1)

Table 13. Fuzzy evaluation matrix with respect to economic criteria (C2)

Contact Anaerobic
Eco. Criteria (C2) UASB UAFB ABR
process Lagoon
UASB (1, 1, 1) (0.78, 1, 1.29) (0.7, 0.8, 1.13) (0.88, 1.28, 1.8) (1.17, 1.6, 2.25)
UAFB (0.78, 1, 1.29) (1, 1, 1) (0.7, 0.8, 1.13) (0.88, 1.23, 1.8) (1.17, 1.6, 2.25)
ABR (0.89, 1.25, 1.43) (0.89, 1.25, 1.43) (1, 1, 1) (1, 1.54, 2) (1.33, 2, 2.5)
Contact process (0.56, 0.83, 1.14) (0.56, 0.83, 1.14) (0.5, 0.66, 1) (1, 1, 1) (0.83, 1.32, 2)
Anaerobic Lagoon (0.44, 0.63, 0.86) (0.44, 0.63, 0.86) (0.4, 0.5, 0.75) (0.5, 0.77, 1.2) (1, 1, 1)
 Table 14. Fuzzy evaluation matrix with respect to environmental criteria (C3)

Contact Anaerobic
Env. Criteria (C3) UASB UAFB ABR
process Lagoon
UASB (1, 1, 1) (0.56, 0.88, 1.29) (0.56, 1, 1.8) (0.63, 1.08, 1.8) (0.83, 1.4, 2.25)
UAFB (0.78, 1.14, 1.8) (1, 1, 1) (0.78, 1.14, 1.8) (0.88, 1.23, 1.8) (1.17, 1.6, 2.25)
ABR (0.56, 1, 1.8) (0.56, 0.88, 1.29) (1, 1, 1) (0.63, 1.08, 1.8) (0.83, 1.4, 2.25)
Contact process (0.56, 0.93, 1.6) (0.56, 0.81, 1.14) (0.56, 0.93, 1.6) (1, 1, 1) (0.83, 1.3, 2)
Anaerobic Lagoon (0.44, 0.71, 1.2) (0.44, 0.63, 0.86) (0.44, 0.71, 1.2) ((0.5, 0.77, 1.2) (1, 1, 1)

Table 15. Fuzzy evaluation matrix with respect to administrative criteria (C4)

Anaerobic
Admin. Criteria (C4) UASB UAFB ABR Contact process
Lagoon
UASB (1, 1, 1) (0.5, 0.77, 1.2) (0.44, 0.67, 1.2) (0.44, 0.63, 0.86) (0.44, 0.67, 1.2)
UAFB (0.83, 1.3, 2) (1, 1, 1) (0.56, 0.87, 1.6) (0.56, 0.81, 1.14) (0.56, 0.87, 1.6)
ABR (0.83, 1.5, 2.25) (0.63, 1.15, 1.8) (1, 1, 1) (0.56, 0.94, 1.29) (0.56, 1, 1.8)
Contact process (1.17, 1.6, 2.25) (0.88, 1.23, 1.8) (0.78, 1.07, 1.8) (1, 1, 1) (0.78, 1.07, 1.8)
Anaerobic Lagoon (0.83, 1.5, 2.25) (0.63, 1.15, 1.8) (0.56, 1, 1.8) (0.56, 0.94, 1.29) (1, 1, 1)

Table 16. Summary of priority weights of the main criteria

Tech. Criteria Eco. Criteria Env. Criteria Admin. Criteria Alternative

(C1) (C2) (C3) (C4) priority weight
Weight 0.331 0.162 0.331 0.176
Alternatives
UASB 0.183 0.221 0.211 0.153 0.202
UAFB 0.215 0.221 0.229 0.195 0.215
ABR 0.206 0.279 0.211 0.215 0.219
Contact process 0.2 0.179 0.195 0.222 0.197
Anaerobic Lagoon 0.196 0.101 0.154 0.215 0.168

According to Table 16, the weight vectors of the technical, economic, environmental, administrative,
criteria are calculated as follows, respectively.
(0.183, 0.215, 0.206, 0.2, 0.196),
(0.221, 0.221, 0.279, 0.179, 0.101),
(0.211, 0.229, 0.211, 0.195, 0.154), and
(0.153, 0.195, 0.215, 0.222, 0.215).
The ranking order of the alternatives with the fuzzy AHP method is ABR > UAFB > UASB > Contact
process > Anaerobic Lagoon. These results are near to the results obtained by the fuzzy TOPSIS method.

4. CONCLUSIONS

By using fuzzy TOPSIS and fuzzy AHP methods, uncertainty and vagueness from subjective perception
and the experiences of the decision-maker can be effectively represented and reached to a more effective
decision. In this paper, the anaerobic process selection with fuzzy TOPSIS and fuzzy AHP methods has been
proposed. The decision criteria were technical, economical, environmental, and administrative criteria and their
sub-criteria. These criteria were evaluated to determine the order of anaerobic alternatives for selecting the most
appropriate one.
In fuzzy TOPSIS, the decision makers have used the linguistic variables to assess the importance of the
criteria and evaluate the each alternative with respect to each criterion. These linguistic variables were converted
into triangular fuzzy numbers, and the fuzzy decision matrix was formed. Then, the normalized fuzzy decision
matrix and weighted normalized fuzzy decision matrix were formed. After FPIS and FNIS were defined, the
distance of each alternative to FPIS and FNIS was calculated. Then, the closeness coefficient of each alternative
was calculated separately. According to the closeness coefficient of alternatives, the ranking order of alternatives
has been determined as UAFB> ABR> Contact process> UASB>Anaerobic lagoon. In fuzzy AHP, the decision-
makers made pair-wise comparisons for the criteria and alternatives under each criterion. Then, these comparisons
have been integrated, and the pair-wise comparison values of the decision-makers were transformed into triangular
fuzzy numbers. The priority weights of criteria and alternatives were determined by the Chang’s extent analysis.
According to the combination of the priority weights of criteria and alternatives, the best alternative was determined.
According to fuzzy AHP, the best alternative was ABR and the ranking order of the alternatives is ABR > UAFB >
UASB > Contact process > Anaerobic Lagoon. The results of these two methods were near to each other, and
UAFB and ABR were the appropriate anaerobic treatment process for industrial estates in Iran.

5. ACKNOWLEDGEMENTS

Herein after, there are special thanks for Iran Small Industries and Industrial Estates Organization along
with financial supports as well as the following grant: "Comprehensive investigation of wastewater treatment plant
which are operating in Iran industrial estates and appropriate treatment process selection"
REFERENCES

1. A.J. Balkema, H.A. Perisig, R. Otterpohl and F.J.D. Lambert. Indicators for the sustainability
assessment of wastewater treatment systems. Urban water, 4:153-161 (2002).
2. M. Gratziou. Alternative choices of wastewater management in Thrace. Proc. of the International
conference on ecological protection of the planet earth bio-environment and bio-culture, Sofia,
Bulgaria. pp. 303-308 (2003).
3. Metcalf and Eddy. Wastewater engineering: treatment, disposal, reuse. 4th Ed., Tata McGraw-Hill
Co., New Delhi (2003).
4. G.S. Liang. Fuzzy MCDM based on ideal and anti-ideal concepts. Eur J Oper Res, 112:682-
691(1999).
5. T.C Chu. Selecting plant location via a fuzzy TOPSIS approach. Int J Adv Manuf Technol,
20:859–864 (2002).
6. C.T. Chen, C.T. Lin and S.F. Huang. A fuzzy approach for supplier evaluation and selection in
supply chain management. Int J Prod Econ, 102:289–301 (2006).
7. T. Yang and C.C. Hung. Multiple-attribute decision making methods for plant layout design
problem. Robot Comput-Integr Manuf, 23:126–137 (2007).
8. T.L. Saaty. The Analytic Hierarchy Process: Planning, Priority Setting, Resource Allocation.
McGraw Hill, NY (1980).
9. M.F. Chen, G.H. Tzeng and C.G. Ding. Combining fuzzy AHP with MDS in identifying the
preference similarity of alternatives. Applied Soft Computing, 8:110–117 (2008).
10. P.J.M. Van Laarhoven and W. Pedrcyz,. A fuzzy extension of Saaty’s priority theory. Fuzzy Sets
Syst, 11:229–241 (1983).
11. K.P. Anagnostopoulos, M. Gratziou and Vavatsikos. Using the fuzzy hierarchy process for
selecting wastewater facilities at prefecture level. European Water, 19/20:15-24 (2007).
12. C.L. Hwang and K. Yoon. Multiple attributes decision making methods and applications.
Springer, Berlin (1981).
13. Y.M. Wang, T.M.S. Elhag. Fuzzy TOPSIS method based on alpha level sets with an application
to bridge risk assessment. Expert Syst Appl, 31:309–319 (2006).
14. İrfan Ertuğrul and Nilsen Karakaşoğlu. Comparison of fuzzy AHP and fuzzy TOPSIS methods for
facility location selection. Int J Adv Manuf Technol, 39:783–795 (2008).
15. C.T. Chen. Extensions of the TOPSIS for group decision making under fuzzy environment. Fuzzy
Sets Syst, 114:1–9 (2000).
16. K.J. Zhu, Y. Jing and D.Y. Chang. A discussion on extent analyses method and applications of
fuzzy AHP. European Journal of Operational Research, 116: 450-456 (1999).
17. M.T. Lamata. Ranking of alternatives with ordered weighted averaging operators. International
Journal of Intelligent Systems. 19:473-482 (2004).
18. D.Y. Chang. Applications of the extent analysis method on fuzzy AHP. Eur J Oper Res, 95:649-
655 (1996).
Supplier selection using fuzzy AHP and fuzzy multi-objective linear
programming for developing low carbon supply chain
⇑
Krishnendu Shaw a, , Ravi Shankar a , Surendra S. Yadav a , Lakshman S. Thakur b
1. Introduction
Supplier selection plays an important role to make a
supply chain green (Rao, 2002). A positive relation between
green supplier selection and green supply chain
implementation has been ob- served in a study of Seuring
and Müller (2008). Many researchers have addressed
supplier selection issue in the green supply chain from the
perspectives of environmental sustainability (Bai & Sarkis,
2010; Enarsson, 1998; Handfield, Walton, Sroufe, &
Melnyk, 2002; Humphreys, Wong, & Chen, 2003a;
Humphreys, McIvor, & Chan, 2003b; Hsu & Hu, 2009; Lee,
Kang, Hsu, & Hung, 2009a; Noci, 1997; Rao, 2005;
Walton, Handfield, & Melnyk, 1998). However, very few
studies have addressed the carbon emission and the related
issues for supplier evaluation. Recently, Lash and Welling-
ton (2007) have discussed the impacts of climate change
over the business operations. They suggested that
companies have to handle climate change risk properly for
gaining the competitive advantage. Some leading
companies have already started working to develop next
generation carbon emissions management for their supply
chain to survive in the business.
An interesting survey conducted by a consulting
company (Tru- cost, 2009) showed that only 19 percent
of the total green house
gas (GHG) emission in the supply chain is generated
from direct operational activities of the company and rest
of the 81 percent emission is generated from other
indirect activities such as, emis- sion from first tier
supplier, electricity supplier and emission from other
supply chain members. In this scenario, supplier selection
plays an important role to minimize carbon emission in
supply chain. According to a survey report CDP (2010),
more than half of the participants said that in the future
they would cease business with the suppliers, if they do
not manage their carbon emissions. Due to increase
consciousness about climate change, companies are
imposing pressure on their suppliers to manage their
GHG emissions as one of the conditions for doing
business with them. Supplier propensity to minimize
green house gas emission is becoming one of the criteria
for supplier selection (CDP, 2010). Therefore, suppliers
need to make a thorough assessment of their current
capabilities in terms of carbon emission management and
set appropriate targets for further reduction of their
emissions. Wal–Mart in US can be taken as an example of
a global supply chain which has been trying to achieve
environmental sustainabil- ity. Its aim is to become a
world leader in environmental sustain- ability. To achieve
this, it has suggested the suppliers to reduce their energy
consumption for processing of products (Wal–Mart,
2010). Suppliers who measure and publish their own
emission are strategically more preferable than others
because they help the buyers to manage their carbon
emission. However, only a little number of supply chain
members has extensive knowledge abo
the low-carbon material procurement for their supply chain. This
paper deals with the low-carbon material procurement and carbon
management for supplier selection. The relevant literature on green
supplier selection is discussed below.
Noci (1997) proposed a green vendor rating framework for the
assessment of suppliers’ environmental performance. Green com-
petence, green image, life cycle cost, and environmental efficiency
were the important considerations for supplier evaluation.
Humphreys et al. (2003b) developed a knowledge-based system
to evaluate suppliers’ environmental performance. Cost, manage-
ment competencies, green image, green design, environmental
management system, and environmental competencies were con -
sidered as the evaluation factors in the model. Lu, Wu, and Kuo
(2007) proposed an analytic hierarchy process (AHP) and fuzzy
logic based model for Green supplier evaluation. Further, analytic
network process (ANP) based framework was suggested by Hsu
and Hu (2009) to construct an assessment framework of the
supplier for Taiwanese Electronics Company. Five criteria such as
procurement management, R& D management, process manage -
ment, incoming quality control, and management system were
considered in the model. Lee et al. (2009a) suggested an integrated
model to select green suppliers for high-tech industry considering
six factors. The considered factors were quality, technology capa-
bility, pollution control, environmental management, green prod -
uct, and green competencies.
Bai and Sarkis (2010) developed a green supplier evaluation
model considering economic, environmental, and social issues.
Rough set theory was used to deal with the information vagueness.
Kuo, Wang, and Tien (2010) developed a green supplier selection
model applying artificial neural network (ANN) and two multi-
attribute decision analysis (MADA) methods that consists of data
envelopment analysis (DEA) and analytic network process (ANP).
Awasthi, Chauhan, and Goyal (2010) developed a fuzzy multi-crite-
ria model for evaluating environmental performance of suppliers.
Fuzzy TOPSIS was applied in this model. Buyukozkan and Cifci
(2011) proposed fuzzy multi-criteria decision framework for sus-
tainable supplier selection considering incomplete information.
Fuzzy analytic network process within the multi-person deci-
sion-making scheme under incomplete preference relations was
used in their model.
Earlier studies have limited focus on the carbon management is -
sue for supplier evaluation. Earlier studies have mostl y focused on
multi-criteria decision making approaches such as AHP, fuzzy AHP,
fuzzy ANP, TOPSIS, Rough set theory etc. for supplier evaluation.
These types of the models are less robust because quantification
of order quantity to a particular supplier is not possible. To solve
this drawback a hybrid model using fuzzy AHP, fuzzy linear pro-
gramming is proposed for selection of supplier. In few of these stud-
ies, product carbon footprint is taken as one of the criteria of
supplier selection. Product carbon footprint can be measured by
using Publicly Available Specification (PAS) 2050 (2008) standard
developed by British Standard Institution. The buyer can fix certain
amount of carbon emission cap, which acts as a constraint in the
decision model. The present article is organized as follows. Section 2
explores the literature related to supplier selection methodologies.
Section 3 discusses the fuzzy set theory. In Section 4, multi-objec-
tive mathematical model is shown. Section 5 represents case study,
related results and discussions. Section 6 presents the conclusions.
2. Supplier selection problem
Business environment is continuously changing due to diversifi -
cation of customer demands. This diversification of demand leads
to increase in operating cost and followed by the decrease in profit.
Therefore, purchasing decision from a particular supplier is a crucial
strategic decision to ensure profitability and long term survival of
the company. Most of the companies are trying to reduce their
operating costs while satisfying customer needs by increasing their
core competencies and outsourcing other functions (Lee, 2009). A
careful assessment is needed to select right supplier who can main-
tain a continuous replacement of product in proper time. Most of
the times supplier strength and weakness are varied, which leads
to complex decision making of supplier selection. Many researches
in supplier selection area used mathematical programming.
Ghodsypour and O’Brien (1998) solved a supplier selection prob-
lem using a hybrid approach involving AHP and linear programming.
A mixed integer non-linear programming model considering multi-
ple sourcing opportunities was solved by Ghodsypour and O’Brien
(2001). Total cost of logistics with budget constraint, quality, service,
etc. were considered in their model. Karpak, Kumcu, and Kasuganti
(1999) proposed a goal programming model that minimized costs
and maximized delivery reliability and quality for supplier selection
and quota allocation.
Gao and Tang (2003) formulated a multi-objective linear pro-
gramming model to purchase raw materials for a large-scale steel
plant in China. Kumar, Vrat, and Shankar (2004) developed a fuzzy
goal programming approach for vendor selection considering the
effect of information uncertainty in the decision making. Similar
type of problem was solved by Amid, Ghodsypour, and O’Brien
(2006). They used fuzzy multi objective linear programming to
determine the order quantity from many suppliers by considering
the criteria of lowest cost and highest quality. Hong, Park, Jang, and
Rho (2005) proposed a mathematical model for supplier selection
considering the change in suppliers’ supply capabilities and
customers’ needs over a period of time. This model optimized the
revenue and customer needs simultaneously.
There are numerous studies, which applied the dual methodol -
ogies for supplier selection.
Weber, Current, and Desai (2000) formulated a combined multi
objective programming (MOP) and the DEA based framework for
supplier selection. They applied MOP to calculate the order quan-
tity and used DEA for suppliers’ efficiency evaluation. Further, Cebi
and Bayraktar (2003) solved a supplier order allocation problem
considering the quantitative as well as qualitative criteria. Wang,
Huang, and Dismukes (2004) applied AHP method to choose a
strategy from agile/lean supply chain. Further, they used pre-emp-
tive goal programming (PGP) to obtain the optimal order quantity
from the suppliers.
Chan and Kumar (2007) developed a fuzzy extended analytic
hierarchy process (FEAHP) model for global supplier selection. Fur-
ther, Kumar, Shankar, and Yadav (2008) solved a supplier selection
problem using AHP and fuzzy linear programming. Ku, Chang, and
Ho (2010) and Lee, Kang, Hsu, and Hung (2009b) used fuzzy AHP
and fuzzy goal for supplier selection. In their model, fuzzy AHP
was applied first to calculate the weights of the criteria. The crite-
ria’s weights were subsequently used in fuzzy goal programming
to select the supplier. Amin, Razmi, and Zhang (2011) developed
a supplier selection model using fuzzy SWOT analysis and fuzzy
linear programming. Yücel and Güneri (2010) proposed a weighted
additive fuzzy programming approach for supplier selection. They
used TOPSIS and fuzzy linear programming in their framework.
3. Fuzzy set theory
Decision making is very difficult for vague and uncertain envi-
ronment. This vagueness and uncertainty can be handled by using
fuzzy set theory, which was proposed by Zadeh (1965). Fuzziness
and vagueness are normal characteristics of a decision making
problem. This fuzziness and vagueness can be managed by increas -
ing robustness of the model (Yu, 2002). If we do not consider the
fuzziness during the decision making process then the results
evolved from the model may mislead the decision maker. Fuzzy
theory is very useful to solve such practical problems.
Many times decision makers provide an uncertain answer
rather than a precise value. Therefore, it is very difficult to quantify
this qualitative value (Lee, Kang, & Wang, 2005). In AHP, the crisp
value is taken for the pair-wise comparison but this method is not
appropriate for real life decision making problem where fuzziness
is present. To solve this problem, a degree of uncertainty is to be
considered in the decision model ( Lee, 2009; Yu, 2002). Incorpora-
tion of the fuzzy theory in AHP is more appropriate and more effec -
tive than conventional AHP. In fuzzy AHP, the concept of fuzzy set
theory is used, and calculation is done as per normal AHP method
for selecting the alternatives (Bozbura, Beskese, & Kahraman, Fig. 2. Two triangular fuzzy numbers M1 and M2 (Lee, 2009).
2007). There are many areas where fuzzy AHP has been applied
for decision making; and many researchers have developed differ-
ent methodologies for calculating the fuzziness (Boender, DeGraan, The strongest grade of membership is the parameter b that is,
& Lootsma, 1989; Buckley, 1985; Chen, 1996; Chang, 1996; Csutora fM(b) = 1 while a and c are the lower and upper bounds. Two trian-
& Buckley, 2001; Laarhoeven & Pedrycz, 1983; Lee et al., 2005; Lee, gular fuzzy number M1 m—1; m1; mþ1 and M 2 m—2; m2; mþ2 shown in
2009). There are many different methods available for ranking of Fig. 2 is compared by Lee et al. (2009a).
the fuzzy number but every method has its own advantage and when m— P m —; m 1 P m2; mþ P m þ ð2Þ
1 2 1 2
disadvantage (Klir & Yan, 1995).
Lee and Li (1988) proposed intuition ranking method that calcu- The degree of the possibility is defined as (3):
lates the ranks of triangular fuzzy numbers by drawing their mem -
bership function curve. Adamo (1980), Yagar (1978) proposed VðM1 P M2 Þ ¼ 1 ð3Þ
/-cut method and centroid method respectively to rank the fuzzy Otherwise, the ordinate of the highest intersection point is calcu-
numbers. The extent analysis method proposed by Chang (1996) is lated as (Chang, 1996; Lee, 2009; Zhu, Jing, & Chang, 1999)
applied here because the computation is much easier than other m1— — m2þ
fuzzy AHP processes and it takes little time to calculate. Another P
M1 Þ¼ hgtðM1 \ M1 Þ¼
l ðdÞ¼
VðM2 — mþ — m — m—
advantage of this method is that it can overcome the deficiencies m2 2 1 1
of conventional AHP process. This fuzzy AHP not only handles ð4Þ
the uncertainty imposed by the decision maker during decision
making process, but it also provides the robustness and flexibility The value of the fuzzy synthetic extent can be calculated as follows
during the decision making (Chan & Kumar, 2007). Triangular fuz- (5)–(11) (Chang, 1996; Lee, 2009; Zhu et al., 1999).
zy number is used to calculate the priority of different decision var - " # —1
X
m X
n
X
m
iable by pair-wise comparison, and the extent analysis is used to Fi ¼ M jgi M jg i ; i ¼ 1; 2; .. . ; n ð5Þ
calculate the synthetic value from pair-wise comparison. A brief j¼1 i¼1 j¼1
introduction of the fuzzy set theory is given below. Triangular
X
m X
m
X
m
X
m !
fuzzy number is used extensively for most of the fuzzy applica-
^ M jgi ¼ m —ij; mij; mþij
tions. A triangular fuzzy number M is shown in Fig. 1. A fuzzy num- ; j ¼ 1; 2; . . . ; m ð6Þ
j¼1
"
j¼1 j¼1 j¼1
!
# —1 Xm
ber can be represented by (a, b, c) and the membership function can X
n X
m
1 1
1 n
—
M jg ¼
i¼1 j¼1 i¼1 j¼1 ij i¼1 j¼1 ij i¼1 M
be defined as Pn P ; Pn Pm ;P
j¼1 ij
ð7Þ
x—afollows (1) (Cheng, 1999; Lee et al., 2005; Lee, 2009).
a6x6b
i m þ
< b—a A convex fuzzy number can M
be defined as, M
lM ðxÞ ¼
^ c—x
c—b b6x6c ð1Þ
0 Otherwise VðF P F1; F2; .. . ; F k Þ¼ min V ðF P FiÞ; i ¼ 1; 2; . . . ; k ð8Þ
dðF i Þ ¼ min V ðF i P F k Þ ¼ W 0i k ¼ 1; 2; . . . ; n and k – i ð9Þ
with —1 < a 6 b 6 c 6 1. Based on the above procedure, the weights, W0 i of factors are
T
W 0 ¼ W 01 ; W 02; . . . ; W 0n ð10Þ

After normalization, the priority weights are as follows

W ¼ ðW1; W2; .. . ; WnÞT ð11Þ

4. Supplier selection model

The following sets of assumptions, index set, decision variable

and parameters are considered for formulating the multi-objective
supplier selection model.

(i) Only one type of product is purchased from one supplier.

(ii) This model does not consider quantity discounts.
(iii) No shortage of the item is allowed for any of the suppliers.
Fig. 1. Triangular fuzzy number. (iv) It is assumed that lead time is constant.
 Index applied here to deal with the problem. In this method, it is desired
i index for suppliers, for all i = 1, 2,.. . , n. to maximize the overall aspiration level rather than strictly satisfy-
j index for objectives, for all j = 1, 2,.. . , J. ing the constraints (Kumar et al., 2006).
k index for constraints, for all k = 1, 2, ... , K. Zimmermann (1978) developed a multi objective fuzzy linear
Decision variable programming, which can handle linguistics issues properly in
xi order quantity given to the supplier i. decision making. Fuzzy decision is classified into two categories,
Parameters symmetric and asymmetric fuzzy decision-making. In symmetrical
D aggregate demand of the item over a fixed planning period. fuzzy decision same weights are considered for objectives and con-
n number of suppliers competing for selection. straints, but in case of asymmetric fuzzy decision-making, the
Pi price of unit item of the ordered quantity xi to the supplier i. weights are different for objectives and constraints ( Sakawa,
Qi percentage of the rejected units delivered by the supplier i. 1993; Zimmermann, 1978). Multi-objective programming consid-
Li percentage of the late delivered units by the supplier i. ering the fuzzy goals and fuzzy constraints can be transformed into
Gi green house gas emission (GHGE) for product supplied by crisp linear programming formulation (Zimmermann, 1978). We
supplier i. have adopted the weighted additive model proposed by Tiwari,
Ui upper limit of the quantity available for the supplier i. Dharmahr, and Rao (1987) for computing the supplier selection
Bi budget constraint allocated to supplier i. problem because in the real situation all the objective functions
Ccap total carbon emission cap for sourcing of material. and constraints have different weights. The weights have been cal-
culated by using fuzzy AHP extent method proposed by Chang
Model (1996).
A typical linear model for supplier selection problem can be for -
mulated as follows (Amid et al., 2006; Kumar, Vrat, & Shankar, 4.1. Fuzzy linear programming
2006):
Fuzzy linear programming was proposed by Zimmermann
X
n
Minimise Z1 ¼ Pixi ð12Þ (1978). Fuzzy linear programming consists of fuzzy goals, and fuz-
i¼1 zy constraints can be reformulated in such a way that it can be
X
n solved like a normal linear programming problem.
Minimise Z1 ¼ Qixi ð13Þ Conventional LP problem proposed by Zimmermann (1978) is
i¼1
given below (21)–(23).
Xn
Minimise Z3 ¼ Li x i ð14Þ Minimise Z ¼ Cx ð21Þ
i¼1
Subject to
Xn
Minimise Z4 ¼ Gixi ð15Þ Ax 6 b ð22Þ
xP 0
i¼1
ð23Þ
Subject to;
After fuzzification the equation can be represented like this (24)–
X
n
xi ¼ D ð16Þ (26),
i¼1
Čx - Z ð24Þ
xi 6 Ui ð17Þ
X n
e
Gixi 6 Ccap ð18Þ Ax - b ð25Þ
i¼1

Pi xi 6 Bi ð19Þ xP0 ð26Þ

xi P 0 and integer ð20Þ
Objective function (12) minimizes the total cost of ordering.
Objective function (13) minimizes the rejection due to the qual-
ity problem.
Objective function (14) minimizes the late delivered items of
the suppliers.
Objective function (15) minimizes the total green house gas
emissions for procurement.
Constraint (16) shows the total aggregate demand of the item.
Constraint (17) ensures the maximum available capacities of
the suppliers.
Constraint (18) puts restrictions on carbon footprint for
sourcing.
Constraint (19) puts restrictions on the budget amount allo -
cated to the suppliers for supplying the items.
Constraint (20) ensures all the variables greater than zero and
integer.
In real life problem of supplier selection, there are many factors,
which are not known properly, create vagueness in the decision
environment. This vagueness cannot be interpreted by the deter-
ministic problem. Therefore, the deterministic models are not suit-
able for real life problems (Kumar et al., 2006). Fuzzy technique is
The symbol - in the constraint set denotes ‘essentially smaller than
or equal to’ and allows one reach some aspiration level where Č and
Ǎ represent the fuzzy values.
4.2. Membership function
Fuzzy set was proposed by Bellman and Zadeh (1970). The fuzzy
set A in X is defined as (27):
A ¼ fx; lA ðxÞ=x 2 Xg ð27Þ
where lA(x): x ? [0, 1] is called the membership function of A and
lA(x) is the degree of membership to which x belongs to A. The fuz-
zy set A is thus uniquely determined by its membership function
lA(x) and the range of membership function is a subset of the
non-negative real numbers whose value is finite and usually finds
a place in the interval [0, 1].
A linear membership function has been considered in this mod-
el for all fuzzy parameters. A linear membership function has char -
acteristics of continuously increasing or decreasing value over the
range of the parameter. It is defined by the lower and upper values
of the acceptability for that parameter.
A fuzzy objective Ž 2 X is a fuzzy subset of X characterized by its
membership function l A(x): x ? [0, 1]. The linear membership
function for the fuzzy objectives is given as:
 1 if ZjðxÞ 6 Zmin Minimise ZjðxÞ for all j; j ¼ 1; 2; . . . ; J ð39Þ
j
>
< ½Z max —Z ðxÞ] Subjected
gkðxÞ 6 bk to
þ dk for all k; k ¼ 1; 2; .. . ; K ð40Þ
lZj ðxÞ ¼ >
>
j j
if Z min 6 ZjðxÞ 6 Z max where j ¼ 1; 2; .. . ; J:
: ½Z max —Z min ]
j j j j
j Ax 6 b for all the deterministic constant ð41Þ
0 if Zj ðxÞ P Zmax ð28Þ x P 0 and integer ð42Þ
min * max ⁄ ⁄
In (28) Zj is minj Zj(x ) and Zj is maxjZj (x ) and x is the optimum
solution. The upper bound of the optimal values Zmaxj is obtained by solving
A fuzzy constraint Č 2 X is a fuzzy subset of X characterized by a similar supplier selection problem as a linear programming prob-
lem (43)–(46).
its membership function lC (x): x ? [0, 1]. The linear membership
function for the fuzzy constraints is given by (29): Maximise ZjðxÞ for all j; j ¼ 1; 2; .. . ; J ð43Þ
1 if gkðxÞ 6 bk; Subjected to
lC ðx Þ ¼ ½1 — fg k ðxÞ — bk g=d k ] if bk 6 g k ðxÞ 6 bk þ dk ; ð29Þ gkðxÞ 6 bk þ dk k ¼ 1; 2; .. . ; K ð44Þ
8 for all k;
>
<

k >
:
0 if bk þ dk 6 gkðxÞ Ax 6 b for all the deterministic constant ð45Þ

for all fuzzy parameters k = 1, 2, ..., K. The interpretation of d k is the

xP0 and integer: ð46Þ
tolerance interval. According to Zimmermann (1978), the weight of the objective func-
tions and constraints are same in the crisp formulation of the sup-
4.3. Solution of the formulation plier selection problem. However, for a real life supplier selection
problem, all objective functions cannot be given same weights. By
A fuzzy solution is the intersection of all the fuzzy sets repre - using same weights, the value of important objective function is
senting either fuzzy objectives or fuzzy constraints (Bellman & Za- decreased. As a result, an optimal solution for supplier selection
deh, 1970). The membership function of the fuzzy solution is may not be obtained for that case. To avoid this problem, we are
represented by (30). adopting the weighted additive model. The weighted additive mod-
lS ðxÞ ¼ lZ ðxÞ \ lC ðxÞ ¼ min½lZ ðxÞ; lC ðxÞ] ð30Þ el is widely used in multi objective optimization problems. Linear
weighted utility function is obtained by multiplying each member-
In the Eq. (30) lZ (x), lC (x) and lS(x) represent the membership ship function of fuzzy goals by their corresponding weights and
functions of objectives, constraints and solutions, respectively. then adding the results together.
The fuzzy solution of the supplier selection model for the J fuzzy The weighted additive model proposed by Tiwari et al. (1987) is
multiple objectives and K constraints may be represented as (31),
\ ! \ ! X
J
X
K
wjl z ðxÞþ bklg ðxÞ
J K
lS ðxÞ ¼ lD ðxÞ¼ ð47Þ
lZ ðxÞ \ lC ðxÞ j¼1
j
k¼1
k

j¼1 k¼1
X
J
X
K
¼ min min lZ ðxÞ; k¼1;2;...;K
min lC ðxÞ ð31Þ wj þ bk ¼ 1; wj; bk P 0 ð48Þ
j¼1;2;...;J j k
j¼1 k¼1

Highest degree of the membership value is the optimum solution of

the supplier selection problem (32). In (47) and (48), wj and bk are the weights coefficients that present
the relative importance among the fuzzy goals and fuzzy
l ðx ω Þ¼ max l ðxÞ¼ max min min l ðxÞ; min l ðxÞ ð32Þ constraints.
s S Zj Ck
x2S x2x j¼1;2;...;J k¼1;2;...;K The following crisp single objective programming (49)–(55) is
equivalent to the above fuzzy model.
4.4. Crisp formulation of the supplier selection model J K
X X
Maximise w j kj þ bk c k ð49Þ
j¼1 k¼1
A fuzzy programming model consists of J objectives and K con-
straints are transformed into the following crisp formulation. Subject to;
Crisp formulation can be represented by (33) and (38) (Kumar kj 6 l zðxÞ; j ¼ 1; 2; . . . ; J ð50Þ
j
et al., 2006),
ck 6 lg ðxÞ;
k
k ¼ 1; 2; . .. ; K ð51Þ
Maximise k ð33Þ
gpðxÞ 6 bp; p ¼ 1; . . . ; M ð52Þ
Subjected to; k ; c 2 ½0; 1]; j ¼ 1; 2; . . . ; J and k ¼ 1; 2; . . . ; K ð53Þ
j k
k Zj
max
— Zj
min
þ ZjðxÞ 6 Zjmax for all j; j ¼ 1; 2; .. . ; J ð34Þ J K
X X
kðdx Þ þ gk ðxÞ 6 bk þ d k for all k; k ¼ 1; 2; .. . ; K ð35Þ j¼1 wj þ k¼1b k ¼ 1; wj ; b k P 0 ð54Þ
Ax 6 b for all the deterministic constant; ð36Þ xi P 0; i ¼ 1; 2; .. . ; n ð55Þ
x P 0 and integer ð37Þ
06k61 ð38Þ
4.5. Application of fuzzy linear programming for supplier selection
According to the Zimmermann (1978) the optimum lower bound
Zmin
j
and upper bound Zmaxj
can be calculated by solving the same The fuzzy linear programming for supplier selection is proposed
objective function two times like minimization and maximization below. In this mathematical model, we are considering cost,
respectively. rejection percentage, late delivery percentages, green house gas
min
The lower bound of the optimal values j Z emission per product and demand are fuzzy information. After
solving is
theobtained
supplierby
selection problem as a linear programming fuzzification, we can represent the equations as follows (56)–(64)
problem (39)–(42). (Kumar et al., 2006):
 X
n
Step 6: The first objective is selected and solved. After solving the
Pi xi - Zf1 ð56Þ first objective, we obtained the lower bound optimal value
i¼1
of first objective.
X
n
f
Q i xi - Z 2 ð57Þ Step 7: The process is repeated for the remaining objectives one
i¼1 by one. The lower bound and upper bound for each of
the objectives are calculated using the same set of
X
Li xi - Zf3
n
ð58Þ constraints.
i¼1
Step 8: The crisp formulation is done using the weighted additive
X n
Gi xi - Zf4
model proposed by Tiwari et al. (1987). The weights of the
ð59Þ
i¼1 factors which are calculated earlier by EAM are used to
X n formulate the crisp formulation.
xi ffi D ð60Þ Step 9: The crisp formulation of the fuzzy optimization problem is
i¼1 solved and result is obtained.
xi 6 U i ð61Þ
X n 5. A case illustration
Gixi 6 Ccap ð62Þ
i¼1
The effectiveness of the model is discussed through a case con-
Pi xi 6 Bi ð63Þ ducted for an Indian based garment manufacturing company
xi P 0 and integer ð64Þ (ABC). The company is fully export oriented and fulfills demand
of American and European customers. It produces a variety of gar-
ments such as Jeans, T-Shirt, formal shirting, suiting and ladies
4.5.1. Computational procedure
garments, etc. The company procures finished fabric from different
In this study, a combined approach of fuzzy-AHP and fuzzy
suppliers and transformed it into garments in Delhi based plant in
multi-objective linear programming is used to solve the problem.
India. Subsequently finished and packed garments are exported to
Fuzzy-AHP is used to determine the relative weights of supplier
American and European markets.
selection criteria. These weights are multiplied with each member -
The company operates in pull based system and procurement
ship function of fuzzy linear programming to formulate the crisp
of raw material starts after placement of order by buyers. Most
equation. By using fuzzy-AHP, we can calculate the relative
of the foreign buyers prefer to buy green and environmental sus -
weights of each membership function of fuzzy goals in the
tainable products. To fulfill the demand of the customers, the
different strategic environment (Ku et al., 2010).
management of ABC Company has decided to incorporate envi-
The solution steps to solve this model are given below.
ronmental criteria into their suppliers’ evaluation process. Man -
agement has wanted to improve the environmental efficiency
Step 1: Identification of supplier selection criteria is done first.
and cost effectiveness of the purchasing process. The manage-
Step 2: Questionnaire is developed for pair wise comparison of
ment has realized that a loyal supplier manufacturer relationship
factors. Experts in the fields of supply chain and operations
is needed to minimize the carbon footprint of sourcing. The rela -
management were asked to fill the nine-point-scale ques-
tionship should be such that suppliers would share the informa -
tionnaire. The consistency property of each expert’s com-
tion with manufacturer regarding the carbon footprint of their
parison results has to be checked first. If there is any
manufactured product. Management has formed a special com-
inconsistency, the questionnaire is to be filled again and
mittee that consists of managers from different departments
the whole process is to be repeated until the consistency
such as purchasing, production, marketing, quality control, re -
requirement is met.
search and development. The aim of the committee is to find
Step 3: Fuzzy importance weight of the criteria is calculated using
out the best supplier.
the response of the experts. A triangular fuzzy number Ď is The committee has decided to take four criteria such as cost,
obtained by combining the experts’ opinions (Lee, 2009). quality rejection, percentage of late delivered item and green
Ď ¼ ðh— ; h; hþ Þ house gas emission per product for supplier selection. After decid-
ing the factor the committee has chosen four potential suppliers
where for sourcing of the material. After deciding the selection criteria,
a brain-storming session was conducted in the presence of pur-
Y
s
!1=s
chasing and operations managers to prioritize these criteria by
h— ¼ lt ; 8t ¼ 1; 2; .. . ; s: using the FAHP method. The membership functions of triangular
t¼1 fuzzy numbers are given in Table 1 (Lee, 2009). The fuzzy pair wise
Y
s
!1=s
comparisons among the criteria are shown in Table 2.
h¼ mt ; 8t ¼ 1; 2; . . . ; s:
t¼1

Y
s
! 1=s
hþ ¼ ut ; 8t ¼ 1; 2; . . . ; s:
t¼1
Table 1
and (lt , mt , ut ) is the importance weights from expert t. Characteristic function of the fuzzy numbers.
Step 4: Crisp relative importance weight (priority vectors) for fac-
tors is calculated using the extent analysis method (EAM) Fuzzy number Characteristic (membership) function
proposed by Chang (1996). By using, Eqs. (2)–(11), the ~
1 (1, 1, 2)
weights of the factors are calculated. ~
x (x — 1, x, x + 1) for x = 2, 3, 4, 5, 6, 7, 8
Step 5: Supplier selection objective functions are formulated. ~
9 (8, 9, 9)
These objective functions are cost minimization, rejection ~
1=1 (2—1,1—1,1—1)
minimization, late delivery minimizations and green- 1=~x ((x + 1) —1, x—1, (x — 1)1 ) for x = 2, 3, 4, 5, 6, 7, 8
1=9~ (9—1,9—1,8—1)
house gas emission minimization.
X
n
X
m Table 2
M jg i ¼ ð1; 1; 1Þþ ð1; 1:15; 2:16Þ þ · · · þ ð1; 1; 1Þ Fuzzy pair wise comparisons among the criteria.
i¼1 j¼1
Cost Quality Lead time GHGE Demand
¼ ð19:36; 25:17; 37:33Þ
" # —1
Cost (1, 1, 1) (1, 1.15, 2.16) (1, 1.32, 2.35) (1, 1.32, 2.35) (1, 1.74, 2.76)
XX j
n m
1 1 1 Quality (0.46, 0.87, 1) (1, 1, 1) (1, 12) (1, 1, 2) (1, 1.52,
M gi ¼ ; ; 2.55)
i¼1 j¼1
37:33 25:17 19:36 Lead (0.43, 0.75, 1) (0.5, 1, 1) (1, 1, 1) (0.5, 1, 1) (1, 1, 2)
time
¼ ð0:0267; 0:0397; 0:0516Þ GHGE (0.43, 0.75, 1) (0.5, 1, 1) (1, 1, 2) (1, 1, 1) (1, 1.15, 2.16)

X
m Demand (0.36, 0.57, 1) (0.39, 0.66, 1) (0.33, 0.5, 1) (0.46, 0.87, 1) (1, 1, 1)
M jg 1 ¼ ð1; 1; 1Þþ ð1; 1:15; 2:16Þ þ · · · þ ð1; 1:74; 2:76Þ
j¼1

¼ ð5; 6:53; 10:62Þ

From the above fuzzy-AHP analysis, it is observed that cost has the
Xm
highest weight for supplier selection. The weights of quality, GHG
M jg 2 ¼ ð4:46; 5:39; 8:55Þ
emission, lead time and demand come after that. The management
j¼1
of ABC Company expressed that the quality and lead time are
X
m X
m important factors for supplier selection. They also commented that
M gj M gj
j¼1 3 ¼ ð3:43; 4:75; 6Þ; j¼1 4 ¼ ð3:93; 4:9; 7:16Þ; green house gas emission per product is given importance for sup-
plier selection because these would help the company to minimize
X
m

M g 5 ¼ ð2:54; 3:6; 5Þ
j
their carbon footprint.
j¼1
" #—1
Xm Xn X m

F1 ¼ M ¼ ð5; 6:53; 10:62Þ

M jg1 i¼1
jg
i
5.1. Fuzzy linear programming
j¼1 j¼1 In this supplier selection model, we considered four suppliers.
ð0:0267; 0:0397; 0:0516Þ¼ ð0:14; 0:31; 0:56Þ The purchasing criteria such as cost, quality rejection, late delivery
F2 ¼ ð4:46; 5:39; 8:55Þ ð0:0267; 0:0397; 0:0516Þ and green house gas emission per product are considered in this
¼ ð0:10; 0:26; 0:48Þ model. Capacity constraint, budget constraint and total purchasing
F ¼ ð3 43 4 75 6 0 0267 0 0397 0 0516 carbon cap (Ccap) are considered as constraints in this model. These
3 : ; : ; Þ ð : ; : ; : Þ
constraints are deterministic in nature. We have considered de-
¼ ð0:05; 0:10; 0:23Þ mand as a fuzzy variable. The demand is predicted to be about
F4 ¼ ð3:93; 4:9; 7:16Þ ð0:0267; 0:0397; 0:0516Þ 20,000, and it is assumed that it can vary from 19,950 to 20100.
¼ ð0:07; 0:18; 0:34Þ The Ccap value is taken 30,000 in this model. Supplier quantitative
information is given in Table 3.
F5 ¼ ð2:54; 3:6; 5 Þ ð0:0267; 0:0397; 0:0516Þ¼ ð0:05; 0:09; 0:16Þ
Numerical example of multi objective linear programming is gi-
V ðF1 P F 2 Þ ¼ 1; V ðF1 P F 3 Þ¼ 1; ven below. Objective Z1 minimizes the total purchasing cost of the
V ðF1 P F 4 Þ ¼ 1; V ðF1 P F 5 Þ ¼ 1 material. Objective Z2 minimizes the rejection due to quality prob-
V ðF2 P F 1 Þ ¼ 0:8716; VðF2 P F 3 Þ ¼ 1; lem of the product. Objective Z3 minimizes the number of late
delivered item. Objective Z4 minimizes the total carbon footprint
V ðF2 P F 4 Þ ¼ 1; V ðF2 P F 5 Þ ¼ 1 of the purchased item.
V ðF3 P F 1 Þ ¼ 0:7135; VðF3 P F 2 Þ ¼ 0:8824;
V ðF3 P F 4 Þ ¼ 0:9715; Z1 ¼ 6x 1 þ 7x 2 þ 4x 3 þ 3x 4
V ðF3 P F 5 Þ ¼ 1; V ðF4 P F 1 Þ ¼ 0:7847; Z2 ¼ 0:05x1 þ 0:03x2 þ 0:02x3 þ 0:04x4
V ðF4 P F 2 Þ ¼ 0:9280; V ðF4 P F 3 Þ ¼ 1 Z3 ¼ 0:03x1 þ 0:02x2 þ 0:08x3 þ 0:04x4
V ðF4 P F 5 Þ ¼ 1; V ðF5 P F 1 Þ ¼ 0:5170; V ðF 5 P F 2 Þ ¼ 0:6634; Z4 ¼ 1:3x1 þ 1:5x2 þ 1:2x3 þ 1:6x4
Subject to;
V ðF5 P F 3 Þ ¼ 0:7850; V ðF5 P F 4 Þ ¼ 0:7479
x1 þ x 2 þ x3 þ x4 ¼ 20; 000
The weight vectors are calculated as follows. x1 6 6000
dðF 1 Þ¼ Min V ðF1 P F2; F3; F4; F5Þ x2 6 14; 500
¼ Minð1; 1; 1; 1Þ ¼ 1 x3 6 7000
dðF 2 Þ¼ Min V ðF2 P F1; F3; F4; F 5 Þ¼ Minð0:8716; 1; 1; 1Þ x4 6 4000
¼ 0:8716 1:3x1 þ 1:5x2 þ 1:2x3 þ 1:6x4 6 30; 000
6x1 6 24; 000
dðF 3 Þ¼ Min V ðF3 P F1; F2; F4; F 5 Þ¼ Minð0:7135; 0:8824; 0:9715; 1Þ
¼ 0:7135 7x2 6 70; 000

dðF 4 Þ¼ Min V ðF4 P F1; F2; F3; F 5 Þ¼ Minð0:7847; 0:9280; 1; 1Þ 4x3 6 60; 000
3x4 6 10; 000
¼ 0:7847
x1 P 0; x2 P 0; x3 P 0; x4 P 0
dðF 5 Þ¼ Min V ðF5 P F1; F2; F3; F4Þ x1; x2; x3; x4 are integer
¼ Minð0:5170; 0:6634; 0:7850; 0:7479Þ¼ 0:5170
According to computational procedure discussed earlier, the objec -
W 0 ¼ ðdðF 1 Þ; dðF 2 Þ; dðF 3 Þ; dðF 4 Þ; dðF 5 ÞÞT
tive function Z1 is minimized using the set of constraints for getting
¼ ð1; 0:8716; 0:7135; 0:7847; 0:5170ÞT the lower-bound of the objective function. The same objective func -
¼ ð0:257; 0:224; 0:184; 0:202; 0:133Þ tion (Z1) is again maximized using the same set of constraints for
getting the upper-bound of the objective function. This procedure
 Table 5
Table 3
Comparison between Hybrid and Zimmermann approach.
Supplier’s quantitative information.
SN Objective function Hybrid approach Zimmermann approach
Supplier Pi($) Qi Li GHGE Ui Bi($)
(Percentage) (Percentage) (kg) 1 Z1 105,668 110,506
2 Z2 563 628
1 6 0.05 0.03 1.3 6000 24,000
3 Z3 887 806
2 7 0.03 0.02 1.5 14,500 70,000
3 4 0.02 0.08 1.2 7000 60,000 4 Z4 28,233 27,983
4 3 0.04 0.04 1.6 4000 10,000

tion is obtained as follows.

Table 4 Objective value is k = 0.6336, and the value of k1 = 0.755,
The data set for membership function calculation.
k2 = 0.9684, k 3 = 0.1513, k 4 = 0.3059, k 5 = 1, and the value of x 1 = 0,
Serial number Objective function l=1 l=0 x 2 = 9667, x 3 = 7000, x 4 = 3333.
1 Z1 101666.7 118,000
Z1 ¼ 105; 668; Z2 ¼ 563; Z3 ¼ 887; Z4 ¼ 28; 233
2 Z2 560 686.6667
3 Z3 666.6667 926.6667 Supplier selection problem is again solved by Zimmermann (1978)
4 Z4 27,100 28733.33
approach. In Zimmermann approach the weights of the all member -
ship functions is considered same. In this approach k is considered
is repeated for rest three objective functions (Z2, Z3 and Z4) for get- the overall membership function for all the objective functions
ting the lower and upper bound of these objective functions. The (Z1, Z2, Z3 and Z4) and the constraints. The overall membership func -
minimum and maximum value of cost, quality rejection, late deliv- tion (k) is maximized in this case.
ery and GHGE are presented in Table 4. Approach 2: Zimmermann approach
The crisp formulation of the supplier selection problem is for-
mulated using the weighted additive model proposed by Tiwari Maximise k
et al. (1987) (49)–(55). The weights calculated by fuzzy-AHP are Subject to;
used for crisp formulation of the supplier selection problem. In 118; 000 — ð6x1 þ 7x2 þ 4x3 þ 3x4Þ
the crisp formulation, the additive value of membership functions k6
16333:3
of the objectives and the constraints is maximized. In crisp formu- 686 — ð0:05x1 þ 0:03x2 þ 0:02x3 þ 0:04x4Þ
lation, first four terms are the membership functions of the objec- k6
126:6667
tive functions (Z1, Z2, Z3 and Z4) and the fifth term (c1) is the 926 — ð0:03x1 þ 0:02x2 þ 0:08x3 þ 0:04x4Þ
membership function of the demand constraint. k6
260
28; 733 — ð1:3x1 þ 1:5x2 þ 1:2x3 þ 1:6x4Þ
Approach 1: hybrid approach k6
1633:33
Crisp formulation for supplier selection problem 20; 100 — ðx1 þ x2 þ x3 þ x4Þ
k6
100
Maximise 0:257 ω k 1 þ 0:224 ω k 2 þ 0:184 ω k 3
ðx1 þ x2 þ x3 þ x 4 Þ— 19; 950
þ 0:202 ω k4 þ 0:133 ω c1 k6
50
Subject to; x1 6 6000
k1 6
118; 000 — ð6x1 þ 7x2 þ 4x3 þ 3x 4Þ x2 6 14; 500
16333:3 x3 6 7000
686 — ð0:05x1 þ 0:03x2 þ 0:02x3 þ 0:04x4Þ
k2 6 126:6667 x4 6 4000
926 — ð0:03x1 þ 0:02x2 þ 0:08x3 þ 0:04x4Þ 1:3x1 þ 1:5x 2 þ 1:2x3 þ 1:6x4 6 30; 000
k3 6 260 6x1 6 24; 000
28; 733 — ð1:3x1 þ 1:5x2 þ 1:2x3 þ 1:6x4Þ 7x2 6 70; 000
k4 6
1633:33 4x3 6 60; 000
20; 100 — ðx þ x þ x þ x Þ
1 2 3 4
c1 6 100 3x 4 6 10; 000
ðx1 þ x2 þ x3 þ x 4 Þ — 19; 950 x1 P 0; x2 P 0; x3 P 0; x4 P 0;
c1 6
50 and x1; x2; x3; x4 integer
x1 6 6000
We get k = 0.4586 and x 1 = 2922, x 2 = 9088, x 3 = 5469, x 4 = 2494.
x2 6 14; 500
x3 6 7000
Z1 ¼ 110; 506; Z2 ¼ 628; Z3 ¼ 806; Z4 ¼ 27; 981
x4 6 4000
1:3x1 þ 1:5x2 þ 1:2x3 þ 1:6x4 6 30; 000 The solutions are summarized in Table 5. It is observed that for a
range of demand between 19,950 and 20,100, the optimized cost,
6x1 6 24; 000
net rejection, late delivered item and GHGE is $105,668, 563 unit,
7x2 6 70; 000 887 unit, and 28,233 kg respectively. When this problem is solved
4x3 6 60; 000 using Zimmermann approach, the optimized cost, net rejection, lat e
3x4 6 10; 000 delivered item and GHGE is $110,506, 628 unit, 806 unit and
x 1 P 0; x 2 P 0; x 3 P 0; x 4 P 0 and 27,983 kg respectively. In our proposed approach, the quota alloca -
x1; x2; x3 and x4 integer: tion to Supplier 1 is zero. However, in case of Zimmermann ap-
proach 14.6 percent quota is allocated to Supplier 1. Table 6
Linear programming based software LINGO (Ver. 11) has been used shows that the quota allocation to Suppliers calculated by two dif -
to solve this problem. The optimal solution for the above formula- ferent methods. It is observed that the quota allocation to Supplier 1
Table 6
Suppliers’ quota allocation.

Supplier Ui Solution using Quota allocation Solution using Quota allocation percentage
hybrid approach percentage for Hybrid Zimmermann approach for Zimmermann

1 6000 0 0 2922 14.6

2 14,500 9667 48.3 9088 45.5
3 7000 7000 35 5469 27.4
4 4000 3333 16.7 2494 12.5

Fig. 3. Variation of achievement of cost goal (k1) and quality goal (k2 ) with respect to overall goal (k).

Fig. 4. Variation of achievement of lead time goal (k3 ) and GHGE goal (k4 ) with respect to overall goal (k).

varies significantly for two methods. Supplier 1 has lost its entire
quota because the product cost of Supplier 1 is higher than the
other suppliers. The quality supplied by Supplier 1 is much more
inferior to other suppliers. In our proposed approach, more stress
is given on cost, quality and carbon footprint for selecting the sup-
plier but in Zimmermann approach all variables are given same
weights. That’s why we observed the different quota allocation to
the suppliers for these two approaches. 35 percent quota is allo-
cated to Supplier 3 when the calculation is done using hybrid ap -
proach, and 27.4 percent quota is allocated to supplier 3 when it
is solved by Zimmermann approach. Supplier 3 has received the
maximum order as compared to the supplying capacity of the sup-
pliers. The maximum order is due to its lower cost product, lowest Fig. 5. Variation of achievement of demand fulfillment ( c1) goal with respect to
quality rejection and lower GHGE. Late delivery is given lower overall goal (k).
weight in this model that enhances the order quantity to Supplier
3. Supplier 4 is allocated 16.4 percent quota in spite of the lowest tion than other suppliers. The highest quota allocation is due to the
cost of product among the suppliers. The lower quota allocation is lower quality rejection percentage and highest supplying capacity.
due to highest carbon footprint of supplied products. The remaining Supplier 3 is ranked the best on the basis of low cost, lowest quality
quota is fulfilled by Supplier 2. It is observed that 48.3 percent qu o- rejection and lower GHGE. However, the supplier does not have the
ta has been allocated to supplier 2. It has the highest quota alloca- sufficient capacity to supply.
 Fig. 6. Variation of supplier quota allocation with respect to overall goal (k).
Variability of the individual goal achievement of hybrid model
is checked by fixing the value of objective function. Fig. 3 shows
the individual variability of achievement of cost goal (k 1) and qual-
ity goal (k 2) with respect to total achieved goal (k). Fig. 4 shows the
fluctuation of lead time goal (k 3) and GHGE goal (k 4) with respect
to the overall achieved goal (k). Fig. 5 shows the variability of
demand fulfillment goal ( c1) with respect to the overall achieved
goal (k). Fig. 6 shows the quota allocations to the suppliers. It is
observed that quota allocation to different suppliers is different
for different (k) values.
As the degree of the achievement of these fuzzy goals changes,
the quota allocation to the suppliers also changes. It is observed
that quota allocation to Supplier 1 is not changing, when k value
is changing from 0 to 0.52, after that it drops to 0 and then again,
comes back to the value 4000. For k value ranging from 0.54 to
0.62, there is no change in supplier quota allocation to Supplier
1. After k value of 0.62 the quota allocation to Supplier 1 is de -
creased to 0. Quota allocation to Supplier 4 is zero up to k value
0.46, and then the allocated quota is increased and stabilized up
to k value 0.6336. For k value 0.56 the quota allocation to Supplier
4 is zero. Supplier 2 and Supplier 3 follow the mixed trend, and
their quota allocations are changing according to the value of k.
For k value 0.6336 the quota allocation to Supplier 2 is 9667 and
for the same k value quota allocation to Supplier 3 is 7000.
In our proposed approach, the degree of achievement of late
delivered item goal (k 3) is obtained as 0.1513. This achievement le-
vel may not be sufficient to satisfy decision makers’ objective of
lesser late delivered items. In the realistic situation, one cannot ar-
gue that a poor performance of one criterion can be balanced with
a good performance of other criteria. To solve this problem, the
presented model is again reformulated by including an additional
condition such that the achievement level of membership func-
tions should not be less than an allowed value ( Amid et al.,
2006). The a-cut approach is used for reformulation of the prob-
lem. The a-cut approach ensures that the degree of achievement
of any goal is more than the specified value of a. The weighted
additive model is to be reformulated by incorporating new
constraints, kj P a and ck P a and a 2 [ a— , a+] to other system con-
straints (Amid et al., 2006). In this approach, an appropriate value
of a is to be chosen to avoid the infeasible solutions (Chen, 1985).
After discussing this issue with decision makers, we have reformu-
lated the model considering the following constraints k 1 > 0.5; k 2
> 0.5; k 3 > 0.4; k 4 > 0.4 and c1 = 1. After solving the equation, we
obtained the maximum achievable goal which is k = 0.5432. The
individual achievable goals are k 1 = 0.5002; k 2 = 0. 5684; k 3 = 0.4;
k4 = 0.4 and c1 = 1. The final solution is obtained as: x1 = 2264;
x 2 = 9316; x 3 = 5774 and x 4 = 2646. The optimized cost is
$109,830, rejection due to quality problem is 614, late delivered
item is 822 and carbon emission is 28,079 kg.
6. Conclusions
By nature, supplier selection process is a complicated task. The
profitability and customer satisfaction are directly proportional to
the effectiveness of supplier selection. Therefore, supplier selection
is a crucial strategic decision for long term survival of the firm. In
the present supplier selection model, a combined approach of fuz-
zy-AHP and fuzzy multi-objective linear programming are used.
This formulation integrates carbon emission in the objective func-
tion, and carbon emission cap (Ccap ) of sourcing as a constraint
while selecting a supplier. In this model fuzzy AHP is used first
to calculate the weights of the criteria and then fuzzy linear pro -
gramming is used to find out the optimum solution of the problem.
Vagueness and imprecision can be effectively handled in this mod-
el. The proposed model is very useful to solve the practical prob -
lem. In the practical situation, all objective functions do not
possess same weights therefore; the weights of the objective func-
tions can be changed according to the requirement of the manager.
The individual priority can be easily calculated by using the fuzzy-
AHP method. The proposed method is a very useful decision-mak-
ing tool for mitigating environmental challenges. A case study has
been used to demonstrate the implication of fuzzy-AHP and fuzzy
linear programming for supplier selection problem.
References
Adamo, J. M. (1980). Fuzzy decision trees. Fuzzy Sets and Systems, 4(3), 207–219.
Amid, A., Ghodsypour, S. H., & O’Brien, C. (2006). Fuzzy multi-objective linear model
for supplier selection in a supply chain. International Journal of Production
Economics, 104(2), 394–407.
Amin, S. H., Razmi, J., & Zhang, G. (2011). Supplier selection and order allocation
based on fuzzy SWOT analysis and fuzzy linear programming. Expert Systems
with Applications, 38(1), 334–342.
Awasthi, A., Chauhan, S. S., & Goyal, S. K. (2010). A fuzzy multi criteria approach for
evaluating environmental performance of suppliers. International Journal of
Production Economics, 126(2), 370–378.
Bai, C., & Sarkis, J. (2010). Green supplier development: Analytical evaluation using
rough set theory. Journal of Cleaner Production, 18(12), 1200–1210.
Bellman, R. E., & Zadeh, L. A. (1970). Decision making in a fuzzy environment.
Management Science, 17(4), 141–164.
Boender, C. G. E., DeGraan, J. G., & Lootsma, F. A. (1989). Multiple -criteria decision
analysis with fuzzy pairwise comparisons. Fuzzy Sets and Systems, 29(2), 133–
143.
Bozbura, F. T., Beskese, A., & Kahraman, C. (2007). Prioritization of human capital
measurement indicators using fuzzy AHP. Expert Systems with Applications,
32(4), 1110–1112.
Buckley, J. J. (1985). Fuzzy hierarchical analysis. Fuzzy Production, 17(2), 255–264.
Sets and Systems, 17(3), 233–247. Humphreys, P. K., McIvor, R., & Chan, F. T. S. (2003b).
Buyukozkan, G., & Cifci, G. (2011). A novel fuzzy Using case based reasoning to evaluate supplier
environmental management performance. Expert
multi-criteria decision framework for sustainable
Systems with Applications, 25(2), 141–153.
supplier selection with incomplete information.
Humphreys, P. K., Wong, Y. K., & Chen, F. T. S. (2003a).
Computers in Industry, 62(2), 164–174.
Integrating environmental criteria into the supplier selection
Carbon Disclosure Project. 2010. Supply Chain
Report 2010. Available from process. Journal of Materials Processing Technology,
<www.cdproject.net/CDPResults/CDP-Supply-Chain- 138(1–3), 349–356.
Report_2010.pdf>. Karpak, B., Kumcu, E., & Kasuganti, R. (1999). An
Cebi, F., & Bayraktar, D. (2003). An integrated application of visual interactive goal programming: a case
approach for supplier selection. in vendor selection decisions. Journal of Multi-Criteria
Logistics Information Management, 16(6), 395–400. Decision Analysis, 8(2), 93–105.
Chang, D. Y. (1996). Applications of the extent Klir, G. I., & Yan, B. (1995). Fuzzy sets and fuzzy logic theory
analysis method on fuzzy AHP. and applications. London: Prentice – Hall International.
European Journal of Operational Research, 95(3), 649– Ku, C. Y., Chang, C. T., & Ho, H. P. (2010). Global supplier
655. selection using fuzzy analytic hierarchy process and fuzzy
Chan, F. T. S., & Kumar, N. (2007). Global supplier goal programming. Quality and Quantity, 44(4), 623–640.
development considering risk factors using fuzzy Kumar, P., Shankar, R., & Yadav, S. S. (2008). An integrated
extended AHP based approach. Omega, 35(4), 417– approach of analytic hierarchy process and Fuzzy linear
431. programming for supplier selection. International Journal
Chen, S. H. (1985). Ranking fuzzy numbers with
maximizing set and minimizing set. of Operational Research, 3(6), 614–631.
Fuzzy Sets and Systems, 17, 113–129. Kumar, M., Vrat, P., & Shankar, R. (2004). A fuzzy goal
Chen, S. M. (1996). Evaluating weapon systems using programming approach for vendor selection problem in a
fuzzy arithmetic operations. supply chain. Computers and Industrial Engineering,
Fuzzy Sets and Systems, 77(3), 265–276. 46(1), 69–85.
Cheng, C. H. (1999). Evaluating weapon systems using Kumar, M., Vrat, P., & Shankar, R. (2006). A fuzzy
ranking fuzzy numbers. Fuzzy Sets and Systems, programming approach for vendor selection problem in a
107(1), 25–35. supply chain. International Journal of Production
Csutora, R., & Buckley, J. J. (2001). Fuzzy hierarchical Economics, 101(2), 273–285.
analysis: The lambda–max method. Fuzzy Sets and Kuo, R. J., Wang, Y. C., & Tien, F. C. (2010). Integration of
Systems, 120(2), 181–195. artificial neural network and MADA methods for green
Enarsson, L. (1998). Evaluation of suppliers: How to supplier selection. Journal of Cleaner Production, 18(12),
consider the environment. International Journal of 1161–1170.
Physical Distribution and Logistics Management,
28(1), 5–17.
Gao, Z., & Tang, L. (2003). A multi-objective model for
purchasing of bulk raw materials of a large-scale
integrated steel plant. International Journal of
Production Economics, 83(3), 325–334.
Ghodsypour, S. H., & O’Brien, C. (1998). A decision
support system for supplier selection using an
integrated analytic hierarchy process and linear
programming. International Journal of Production
Economics, 56–57(1), 199–212. Ghodsypour, S. H., &
O’Brien, C. (2001). The total cost of logistics in
supplier selection, under conditions of multiple sourcing,
multiple criteria and capacity
constraint. International Journal of Production
Economics, 73(1), 15–27.
Handfield, R., Walton, S., Sroufe, R., & Melnyk, S.
(2002). Applying environmental criteria to supplier
assessment: A study in the application of the analytical
hierarchy process. European Journal of Operational
Research, 141(1), 70–87.
Hong, G. H., Park, S. C., Jang, D. S., & Rho, H. M. (2005).
An effective supplier selection method for constructing a
competitive supply-relationship. Expert Systems with
Applications, 28(4), 629–639.
Hsu, C. W., & Hu, A. H. (2009). Applying hazardous
substance management to supplier selection using
analytic network process. Journal of Cleaner
 Trucost. (2009). Carbon emissions – measuring the
Laarhoeven, P. J. M., & Pedrycz, W. (1983). A fuzzy risks, Available from:
extension of Saaty’s priority theory. Fuzzy Sets <www.nsf.org/business/sustainability/SUS_NSF_Trucost_
and Systems, 11(1–3), 229–241. Report.pdf>.
Lash, J., & Wellington, F. (2007). Competitive Wal-Mart. (2010). Global Sustainability Report,
advantage on a warming planet. Available from: <http://
Harvard Business Review (3), 1–11.
cdn.walmartstores.com/sites/sustainabilityreport/201
Lee, A. H. I. (2009). A fuzzy supplier selection model
with the consideration of benefits opportunities, 0/ WMT2010GlobalSustainabilityReport.pdf>,
costs and risks. Expert Systems with Applications, downloaded on 23 October 2010.
36(2), 2879–2893. Walton, S. V., Handfield, R. B., & Melnyk, S. A. (1998).
Lee, H. I., Kang, H. Y., Hsu, C. F., & Hung, H. C. The green supply chain: Integrating suppliers into
(2009a). A green supplier selection model for high- environmental management processes. International
tech industry. Expert Systems with Applications, Journal of Purchasing and Materials Management,
36(4), 7917–7927. 34(2), 2–11.
Lee, H. I., Kang, H. Y., Hsu, C. F., & Hung, H. C. Wang, G., Huang, S. H., & Dismukes, J. P. (2004).
(2009b). Fuzzy multiple goal programming applied to Product-driven supply chain selection using integrated
TFT-LCD supplier selection by downstream multi-criteria decision-making methodology.
manufacturers. Expert Systems with Applications, International Journal of Production Economics, 91(1),
36(3), 6318–6325. 1–15.
Lee, A. H. I., Kang, H. Y., & Wang, W. P. (2005). Weber, C. A., Current, J. R., & Desai, A. (2000). An
Analysis of priority mix planning for semiconductor optimization approach to determining the number of
fabrication under uncertainty. International Journal vendors to employ. Supply Chain Management: An
of Advanced Manufacturing Technology, 28(3–4), International Journal, 5(2), 90–98.
351–361. Yagar, R. R. (1978). On a general class of fuzzy
Lee, E. S., & Li, R. L. (1988). Comparison of fuzzy connective. Fuzzy Sets and Systems, 4(3), 235–242.
Yu, C. S. (2002). A GP-AHP method for solving group
numbers based on the probability measure of fuzzy
decision-making fuzzy AHP problems. Computers and
events. Computers and Mathematics with
Operations Research, 29(14), 1969–2001.
Applications, 15(10), 887–896.
Lu, Y. Y., Wu, C. H., & Kuo, T. C. (2007). Yücel, A., & Güneri, A. F. (2010). A weighted additive
Environmental principles applicable to green fuzzy programming approach for multi-criteria
supplier evaluation by using multi-objective supplier selection. Expert Systems with Applications,
decision analysis. International Journal of 38(5), 6281–6286.
Production Research, 45(18–19), 4317–4331. Zadeh, L. A. (1965). Fuzzy sets. Information and Control, 8,
338–353.
Noci, G. (1997). Designing ‘Green’ vendor rating
Zhu, K. J., Jing, Y., & Chang, D. Y. (1999). A discussion on
systems for the assessment of a supplier’s
extent analysis method and applications of fuzzy AHP.
environmental performance. European Journal of
European Journal of Operational Research, 116(2), 450–
Purchasing and Supply Management, 3(2), 103–
456.
114.
Zimmermann, H. J. (1978). Fuzzy programming and
Publicly Available Specification, PAS 2050.
(2008). Available from: linear programming with several objective functions.
<www.bsigroup.com/Standards-and- Fuzzy Sets and Systems, 1(1), 45–55.
Publications/How-we-can-help-you/
Professional-Standards-Service/PAS-2050>.
Rao, P. (2002). Greening the supply chain a new
initiative in south East Asia.
International Journal of Operations & Production
Management, 22(6), 632–655. Rao, P. (2005). The
greening of suppliers in the South East Asian context.
Journal of
Cleaner Production, 13(9), 935–945.
Sakawa, M. (1993). Fuzzy Sets and Interactive Multi
Objective Optimization. New York: Plenum Press.
Seuring, S., & Müller, M. (2008). Core issues in
sustainable supply chain management a delphi
study. Business Strategy and the Environment,
17(8), 455–466.
Tiwari, R. N., Dharmahr, S., & Rao, J. R. (1987). Fuzzy
goal programming-an additive model. Fuzzy Sets and
Systems, 24(1), 27–34.
Refrensi :

https://www.google.com/search?client=ms-android-xiaomi-
rvo2&sca_esv=2fcdd62dc0011717&sca_upv=1&sxsrf=ACQVn09L6CqiIAZ2ZyigCnvG3n4mLh-
oOQ:1712149636765&q=jurnal+internasional+dengan+METODE+GOAL+PROGRAMMInG+ANALY
TICAL+HIERARCHY+PROCESS+(AHP)+FUZZY+AHP+Analytical+network+process+(ANP)+Metod
e+Simple+Additive+Weighting+(SAW)+bahasa+Inggris&uds=AMwkrPs2XaEwwsdETHaBXiaUtx0sH
x88UkUR-wSr784HWmjkkrJFhVdmf2BOSPqTI7E9dU_425nhRbotpNm7Av5WGOc-
qt25aT4XT18dumdLM8qVs5zlPE0bp1cZlEY_8vKz4lYJZH_SX1nO-q81buk-JDum-
uJiIW55wSLo2D0xTNJ3X0NdiWl60sgISslBMzZ6_7BCuJFv_H1BAmF5JQmpwIFYVUNkyNzDqLg2
BpwEaXJ4penTdEXbO3jALyOG6HTwuCTJms7dCpDydwz8qEiuI2nuB1xGyYwJyFdHyiIGDHizjMd2
LHkL527fy-neR0aBAS3UivrJSltD-
a4k8WPk2EVxLzgDrm7pOZknrnhp7Q0yAyH3GMYrzTOZw4LJwgfP_TezWGIu2Cv6fiJlIz-
725MGk3XPTEqZerYyoRRfaxipIYuPqB-43VTNftURrzwMFeBf6pTsKESEFyPUoZL5-
tyQzyfNKQ645CF1BVuJmDl9s-7lugX1igo6SJEZIcWnVX-sizz6TjvagB3DbL7-Z_VIVS5-
ew&udm=2&prmd=isvnbmtz&sa=X&ved=2ahUKEwi17Zq3jqaFAxUX7zgGHUgOCXgQtKgLegQIDR
AB&biw=393&bih=732&dpr=2.75#imgrc=mXoshitnXoHBGM&imgdii=nA6906YM0

https://www.google.com/search?client=ms-android-xiaomi-
rvo2&sca_esv=2fcdd62dc0011717&sca_upv=1&sxsrf=ACQVn09L6CqiIAZ2ZyigCnvG3n4mLh-
oOQ:1712149636765&q=jurnal+internasional+dengan+METODE+GOAL+PROGRAMMInG+ANALY
TICAL+HIERARCHY+PROCESS+(AHP)+FUZZY+AHP+Analytical+network+process+(ANP)+Metod
e+Simple+Additive+Weighting+(SAW)+bahasa+Inggris&uds=AMwkrPs2XaEwwsdETHaBXiaUtx0sH
x88UkUR-wSr784HWmjkkrJFhVdmf2BOSPqTI7E9dU_425nhRbotpNm7Av5WGOc-
qt25aT4XT18dumdLM8qVs5zlPE0bp1cZlEY_8vKz4lYJZH_SX1nO-q81buk-JDum-
uJiIW55wSLo2D0xTNJ3X0NdiWl60sgISslBMzZ6_7BCuJFv_H1BAmF5JQmpwIFYVUNkyNzDqLg2
BpwEaXJ4penTdEXbO3jALyOG6HTwuCTJms7dCpDydwz8qEiuI2nuB1xGyYwJyFdHyiIGDHizjMd2
LHkL527fy-neR0aBAS3UivrJSltD-
a4k8WPk2EVxLzgDrm7pOZknrnhp7Q0yAyH3GMYrzTOZw4LJwgfP_TezWGIu2Cv6fiJlIz-
725MGk3XPTEqZerYyoRRfaxipIYuPqB-43VTNftURrzwMFeBf6pTsKESEFyPUoZL5-
tyQzyfNKQ645CF1BVuJmDl9s-7lugX1igo6SJEZIcWnVX-sizz6TjvagB3DbL7-Z_VIVS5-
ew&udm=2&prmd=isvnbmtz&sa=X&ved=2ahUKEwi17Zq3jqaFAxUX7zgGHUgOCXgQtKgLegQIDR
AB&biw=393&bih=732&dpr=2.75#imgrc=nA6906YM0lJUgM&imgdii=UW9vgByXc