Int. J. Environ. Sci. Technol.

(2016) 13:1377–1392
DOI 10.1007/s13762-016-0977-4

ORIGINAL PAPER

An extended fuzzy TOPSIS–GRA method based on different

separation measures for green logistics service provider selection
E. Celik1 • M. Erdogan2 • A. T. Gumus2

Received: 4 October 2012 / Revised: 11 November 2015 / Accepted: 8 March 2016 / Published online: 4 April 2016
Ó Islamic Azad University (IAU) 2016

Abstract In this paper, first, the criteria that make Introduction

logistics service providers more ‘‘green’’ are determined
as: cooperation with customer’s company and its customer, Increased environmental pollution is forcing companies to
green government regulations, environmental management become more environmentally friendly. Companies that
system, green process design, reduction in energy con- produce logistics services should obtain sensitivity to
sumption and green network design. The criteria weights environmental awareness and environmentally friendly
are determined by fuzzy AHP, based on expert opinions. policies. Third-party logistics service providers (3PLs) play
Then, a new method is proposed, which is the combination a major role between outsourcing companies, market and
of fuzzy TOPSIS and GRA, and used to evaluate green customers (Liu and Wang 2009). The 3PLs include using
3PLs based on different separation measures, as an exten- external companies to perform logistics functions, which
sion, using trapezoidal fuzzy numbers. Five Turkish 3PLs have been traditionally operational within an organization
serve in Istanbul are selected in order to apply a case study (Liu and Wang 2009; Işıklar et al. 2007). The main
to show the applicability of the proposed method. Finally, advantages of 3PLs can be ranged as let to customer firms
the proposed method is verified with respect to different to concentrate on the core competencies improve the ser-
resolving coefficient values and separation measures and vice, reduce the transportation cost, restructure the supply
also compared with fuzzy TOPSIS and fuzzy VIKOR chains and establish the marketplace legitimacy (Bhatnagar
method results. Different multi-criteria decision-making et al. 1999). In the competitive business world, usage of
methods can be applied and compared to check validity of 3PLs by the companies that put to use recycling, reuse and
our results for future studies. The proposed method can remanufacturing functions, has valuable effects on their
also be implemented to 3PLs in other countries. performances (Govindan et al. 2012). In the use of logistics
outsourcing, 3PL selection and evaluation is a critical
Keywords Fuzzy GRA  Fuzzy TOPSIS  Green logistics  process; by selecting the right 3PL, customer companies
Logistics service provider selection  Trapezoidal fuzzy can reduce capital investments in facilities, equipment,
numbers information technology and manpower, increase the flexi-
bility in adapting to changes in the market, reduce inven-
tory and improve inventory turnover rate, improve on-time
delivery, reduce transportation cost (Liu and Wang 2009;
Ho et al. 2012; Razzaque and Sheng 1998). Selection of the
best 3PL requires more than scanning price lists; it also
& E. Celik depends on many factors both quantitative and qualitative.
[email protected]
Many multi-criteria decision-making methods, such as
1
Department of Industrial Engineering, Tunceli University, analytic hierarchy process (AHP), analytic network process
62000 Tunceli, Turkey (ANP), artificial neural networks (ANN), case-based rea-
2
Department of Industrial Engineering, Yildiz Technical soning (CBR), data envelopment analysis (DEA), rule-
University, 34349 Besiktas, Istanbul, Turkey based reasoning (RBR) and technique for order preference

123
1378 Int. J. Environ. Sci. Technol. (2016) 13:1377–1392

by similarity to ideal solution (TOPSIS), are used for Lai and Christina (2012) describe green logistics as a
selection of 3PLs (Ho et al. 2012). management approach that considers product return and
In this paper, first, the criteria that make 3PLs more recycling, environmental management systems and eco-
‘‘green’’ are determined as: cooperation with customer efficiency as viable ways to comply with environment-
company and its customer, green government regulations, based regulations in international trade. Green logistics
environmental management system, green process design, management provides resource conservation, waste
reduction in energy consumption and green network reduction and organizational skills to meet social expec-
design. The criteria weights are determined by fuzzy AHP, tations for environmental protection (Lai and Christina
based on expert opinions. Then, a combined fuzzy TOP- 2012; Ramanathan and Yunfeng 2009). Evangelista et al.
SIS–GRA (grey relational analysis) method is proposed to (2011) conduct a detailed literature survey suggesting
evaluate green 3PLs based on different separation mea- methods to be more ‘‘green’’ for companies on their
sures for trapezoidal fuzzy numbers. Five Turkish 3PLs transportation and logistics activities. They list the studies
serve in Istanbul are selected in order to apply a case study as: modal changes and intermodal solutions (McKinnon
to show the applicability of the proposed method. The 2010a), advances in technology solutions (McKinnon
proposed method is verified with respect to different 2010b), tools for assessing logistics carbon footprint
resolving coefficient values and separation measures and (Eglese and Black 2010; Lieb and Lieb 2010; McKinnon
also compared with fuzzy TOPSIS and fuzzy VIKOR 2010c), green transport management (Lieb and Lieb 2010)
method results. and green logistics system design (Erdogan and Gumus
The rest of paper is organized as follows: In ‘‘Literature 2012; Harris et al. 2010).
review’’ section, a detailed literature review is presented
about green logistics, green applications of 3PLs and Green applications of 3PLs
selection of green 3PLs. ‘‘Materials and methods’’ section
gives brief literatures and computations of fuzzy TOPSIS For the physically distribution companies, outsourcing
and GRA methods. ‘‘The proposed combined fuzzy TOP- logistics activities has become an integral part of supply
SIS and GRA method’’ section describes the proposed chain management (Ageron et al. 2011). 3PLs are put into
method combining fuzzy TOPSIS and GRA, to evaluate strategic role because of growing interest in logistics ser-
green 3PLs. In ‘‘Results and discussion’’ section, a case vices. Many researches claim that if logistics firms do not
study considering five 3PL companies and consequently measure company performance and monitor it in a flow of
verification of the proposed method are presented. Finally, functions rather than individual activities, supply chain will
‘‘Conclusion’’ section concludes our study. not be operative (Robertson et al. 2002). 3PLs provide
integrated logistics services and are closely related to
supply chain management (Tezuka 2011). The value-added
Literature review services provided by 3PLs can be ranged as repair, re-
manufacturing, re-packaging and re-labelling (Min and Ko
In this section, the literature background of green logistics, 2008).
green applications of 3PLs and selection of green 3PLs is The 3PLs have a critical position in improving envi-
presented. ronmental sustainability of supply chain operations (Azadi
and Saen 2011). In Hamdan and Roger’s (2008) study,
Green logistics DEA is used to evaluate warehouse logistics operations of
a group of 3PLs. They determine the impact of each input
Greening has become a key word for environmental con- and output on the efficiency of each warehouse, studied on
cerns. When ‘‘logistics’’ term is put next to ‘‘green’’, it specific warehouse characteristics and made suggestions
becomes an environmentally friendly and efficient trans- for improvement and design of more efficient operations.
port and distribution system (Rodrigue et al. 2001). Green Evangelista et al. (2011) present a guideline for buyers to
logistics is an entire part of production logistics applica- consider their awareness, initiatives as well as drivers and
tions to increase green degree and also includes green barriers affecting 3PLs’ sustainability initiatives.
packaging and reverse logistics (Ying and Li-Junb 2008). Azadi and Saen (2011) use DEA to select third-party
Whilst the traditional logistics provides the flow of forward reverse logistics providers (3PRLPs). The relationship
activities from supplier to consumer, environmental con- between 3PLs and environmental consciousness is inves-
cerns cause to arise a new concept ‘‘reverse logistics’’ by tigated in Tezuka’s (2011) study. He analyses the envi-
taking into consideration waste management, recycling, ronmental issues from the point of economy and names this
etc. (Erdogan and Gumus 2012). as ‘‘externality’’ problem. Krumwiede and Sheu (2002)

123
Int. J. Environ. Sci. Technol. (2016) 13:1377–1392 1379

develop a reverse logistics decision-making model for method for green supplier selection. Humphreys et al.
strategic reverse logistics decision-making. Kuo et al. (2003) present a framework for integrating environmental
(2015) present a framework of the supplier evaluating factors into the supplier selection process. Kuo et al. (2010)
process for carbon management by integrating fuzzy ANP develop a green supplier selection model that integrates
and fuzzy TOPSIS approaches. Govindan et al. (2012) ANN and two multi-attribute decision analysis methods:
describe a case study considering a company in the tire DEA and ANP. Amindoust et al. (2012) present a study for
industry aiming to show how a 3PRLP may be chosen as a a ranking model based on fuzzy inference system for sus-
partner from n possible provider alternatives. Min and Ko tainable supplier selection. Walton et al. (1998) propose a
(2008) propose a dynamic reverse logistics model for 3PLs. number of supply chain environmentally friendly practices
Here, a mixed-integer programming model and a genetic (EFP) and define their top ten environmental supplier
algorithm are used for the location and allocation of repair evaluation criteria. Buyukozkan (2012) presents a decision
facilities for 3PLs. model for supplier performance evaluation by considering
various environmental performance criteria. One of the rare
Selection of green logistics service providers publications using GRA for green supplier evaluation is
Tseng’s (2010) paper. His study attempts to develop the
There are lots of studies in the literature about selecting the fuzzy GRA to rank the best supplier prior to environmental
right 3PLs, but nowadays companies are looking for green knowledge management capacities. Fu et al. (2012) intro-
3PLs for improving the environmental sustainability. duce a formal structured managerial approach for organi-
Rational selection of green 3PLs gives companies the zations to help evaluating the influence relationships
advantage of professionalism and takes the advantage of amongst green supplier development programs using a
cost to achieve the goals as logistics routing optimization formalized grey-based DEMATEL methodology.
(Shan 2012). There are limited numbers of studies in the
literature related to green 3PLs. The studies are usually
concerned with the selection of 3PLs. Jharkharia and Materials and methods
Shankar (2007) focus on the issues concerning evaluation
and selections of providers. Kannan et al. (2009) develop a The fuzzy TOPSIS and GRA methods
multi-criteria group decision-making model based on fuzzy
TOPSIS in fuzzy environment to guide the selection pro- Here, brief literatures and computations of fuzzy TOPSIS
cess of best 3PRLP. Aghazadeh (2003) determine the most and GRA methods are presented.
effective ways of choosing 3PLs. Azadi and Saen (2011)
propose the approach of a new chance-constrained DEA Fuzzy TOPSIS
(CCDEA) to assist the decision-makers to determine the
most appropriate 3PRLPs in the presence of both dual-role TOPSIS, one of the classical MCDM methods, was pro-
factors and stochastic data. Ho et al. (2012) develop an posed by Hwang and Yoon (1992). TOPSIS is based on the
integrated approach, combining quality function deploy- concept that the chosen alternative should have the shortest
ment (QFD), fuzzy set theory and AHP approach, to distance from the positive ideal solution (PIS) and the far-
evaluate and select the optimal 3PLs in their study. thest from the negative ideal solution (NIS) for solving a
Lee et al. (2009) determine criteria and sub-criteria for multiple criteria decision-making problem. Chen and Hwang
evaluating traditional and green suppliers. They propose a (1992) applied fuzzy numbers to establish fuzzy TOPSIS.
model to evaluate green suppliers or to select the best green The algorithm and the steps of the fuzzy TOPSIS
supplier for cooperation. Buyukozkan and Ciftci (2012) use method are given below (Chen 2000):
a novel hybrid MCDM approach based on fuzzy DEMA-
Step 1 In the first step, assume that a decision group has
TEL, fuzzy ANP and fuzzy TOPSIS to evaluate green
K people, and then, the importance of the criteria and the
suppliers for the need of improving green supply chain
rating of alternatives with respect to each criterion can be
management initiatives. According to their references,
calculated as:
there are the five major evaluation criteria for green sup-
pliers, and these are listed as organization, financial per- 1h 1 i
formance, service quality, technology and green x~ij ¼ x~ij þ x~2ij þ    þ x~Kij
K
competencies (Awasthi et al. 2010; Humphreys et al. 2003;
1h 1 i
Kuo et al. 2010; Lee et al. 2009; Walker et al. 2008). wj ¼ w~j þ w ~2j þ    þ w ~Kj
K
Tuzkaya et al. (2009) present a methodology for the
evaluation of suppliers’ environmental performances. Yang where x~Kij and w
~Kj are the rating and the importance weight
and Wu (2007) present a research using grey entropy of the Kth decision-maker.

123
1380 Int. J. Environ. Sci. Technol. (2016) 13:1377–1392

Step 2 A decision matrix is formed. Step 6 The distance of each alternative from the positive
2 3 ideal solution diþ and the negative ideal solution di is
x~11 x~12 . . . x~1n
6 7 calculated.
~ ¼ 6 x~21 x~22 . . . x~2n 7;
D 4 ... ... ... ... 5 Xn qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

x~m1 x~m2 . . . x~mn diþ ¼ vij  v~j Þ2 i ¼ 1; 2; . . .; m

ð~
j¼1
~
_
n qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
W ¼ ½w
~1 ; w ~n 
~2 ; . . .; w X 2
di ¼ ð~vij  v~ j Þ i ¼ 1; 2; . . .; m
where x~ij ; 8i; j and w ~j ; j ¼ 1; 2; . . .; n are linguistic vari- j¼1
ables. These linguistic variables can be described by
  where dð; Þ is the distance measurement between two
trapezoidal fuzzy numbers, x~ij ¼ aij ; bij ; cij ; dij and
  fuzzy numbers.
w~j ¼ wj1 ; wj2 ; wj3 ; wj4 .
Step 7 Then, the fuzzy closeness coefficient CCi is
Step 3 To avoid complexity of mathematical operations determined.
in a decision process, the linear scale transformation is
used here to transform the various criteria scales into CCi ¼ di =ðdi þ diþ Þ
comparable scales. The set of criteria can be divided into Step 8 Rank the preference order. Using this index, the
benefit criteria (B) and cost criteria (C). Therefore, the alternatives can be ranked in decreasing order.
normalized fuzzy decision matrix can be represented as:
 
R~ij ¼ r~ij mn Fuzzy GRA

where B and C are the sets of benefit criteria and cost GRA is a MCDM method that was originally proposed by
criteria, respectively, and, Deng (1982, 1989). It has been applied in solving a variety
! of MCDM problems by many authors. For example, Wang
aij bij cij dij
r~ij ¼ ; ; j 2 B; (2009) apply fuzzy GRA to evaluate financial performance
dj dj dj dj of Taiwan container lines. Wei (2010) proposes a GRA
     with intuitionistic fuzzy information in which the infor-
aj aj aj aj
r~ij ¼ ; ; j 2 C; mation about attribute weights is incompletely known.
dij cij bij aij
Zhang and Liu (2011) propose a GRA-based intuitionistic
dj ¼ max dij ; j 2 B; fuzzy MCDM method and apply to personnel selection
i
problem. The steps of the fuzzy GRA algorithm can be
a
j ¼ min aij ; j 2 C:
i outlined as follows (Chen and Tzeng 2004; Wei 2010):
Step 4 Considering the different weights of each Step 1 In the first step, a panel of decision-makers (DMs)
criterion, the weighted normalized decision matrix is who are knowledgeable about 3PLs selection process is
computed by multiplying the importance weight of established.
evaluation criteria and the values in the normalized
1h 1 i 1X K
decision matrix. The weighted normalized decision x~ij ¼ x~ij þ    þ x~Kij ¼ x~e
K K e¼1 ij
matrix V~ for each criterion is defined as:
 
V~ ¼ v~ij mn for i ¼ 1; 2; . . .; m and j ¼ 1; 2; . . .; n Step 2 Calculate the normalized decision matrix.
x~ij  minð~ xij Þ
where v~ij ¼ r~ij  w
~j here v~ij denotes normalized positive Larger the better r~ij ¼
maxð~ xij Þ  minð~ xij Þ
trapezoidal fuzzy numbers.
minð~ xij Þ  x~ij
Step 5 Then, fuzzy positive A~þ and fuzzy negative A~ Smaller the better r~ij ¼
xij Þ  minð~
maxð~ xij Þ
ideal solutions are determined. The fuzzy positive ideal
x~ij  x~oj
solutions (FPIS, A~þ ) and the fuzzy negative ideal solution Nominal - the - best r~ij ¼
   
(FNIS, A~ ) can be defined for beneficial criteria. max max x~ij  x~oj ;xoj  min x~ij

A~ ¼ ð~
v1 ; v~2 ; . . .; v~n Þ where x~oj is optimal value of jth criterion.
A~ ¼ ð~
v ; v~ ; . . .; v~ Þ
1 2 n Step 3 Determine the reference series. The reference
where v~j ¼ ð1; 1; 1; 1Þ and v~
j ¼ ð0; 0; 0; 0Þ for j ¼ 1; . . .; n: series can be defined as:

123
Int. J. Environ. Sci. Technol. (2016) 13:1377–1392 1381

R~0 ¼ ½r~01 ; r~02 ; . . .; r~0n ; where r~0j ¼ maxð~

rij Þ By the extension principle, the fuzzy sum and
j ¼ 1; . . .; n subtraction of any two trapezoidal fuzzy numbers are
also trapezoidal fuzzy numbers. But the multiplication of
Step 4 Establish the distance matrix. The distance d~ij any two trapezoidal fuzzy numbers is only an approximate
between the reference value and each comparison value is trapezoidal fuzzy number. Given any two positive
given as: trapezoidal fuzzy numbers, A~ ¼ ða1 ; a2 ; a3 ; a4 Þ, B~ ¼
ðb1 ; b2 ; b3 ; b4 Þ and k is real number, some operations of
d~ij ¼ r~0j  r~ij
fuzzy numbers A~ and B~ can be calculated as follows:
Then, the distance matrix D can be obtained as:
A~ þ B~ ¼ ða1 þ b1 ; a2 þ b2 ; a3 þ b3 ; a4 þ b4 Þ
2 3
d~11 d~12 . . . d~1n A~  B~ ¼ ða1  b1 ; a2  b2 ; a3  b3 ; a4  b4 Þ
6 d~21 d~22 . . . d~2n 7
D¼6
4 ...
7
A~  B~ ffi ða1  b1 ; a2  b2 ; a3  b3 ; a4  b4 Þ
... ... ... 5
d~m1 d~m2 . . . d~mn A~  k ffi ða1  k; a2  k; a3  k; a4  kÞ
Step 5 Calculate the fuzzy grey relational coefficient. In the proposed method, the criteria weights are
The fuzzy grey relational coefficient n~ij is defined as: n~ij ¼ generated by a fuzzy AHP procedure. It is easy to extend
d~min þfd~max to the fuzzy case and guarantees a unique solution to the
d~ij þfd~max
d~max ¼ maxðd~ij Þ; d~min ¼ minðd~ij Þ and f resolving
reciprocal comparison matrix, and the steps of this
coefficient f 2 ½0; 1: approach are relatively easier than the other fuzzy AHP
Step 6 Estimate the fuzzy grey relational grade c~i by the approaches. The ranking of 3PLs is calculated by the
relation combined fuzzy TOPSIS and GRA method. The steps
(1–3) used for the Buckley’s fuzzy AHP algorithm
Xn
c~i ¼ ~j n~ij ; i ¼ 1; . . .; m
w (Buckley 1985; Gumus 2009; Chen 2009), and the steps
j¼1 (4–13) of the proposed combined fuzzy TOPSIS and GRA
Pn algorithm can be summarized as follows:
~j is the weight of the jth criterion, and
where w j¼1 ~j ¼ 1~
w
Step 1 Construct pairwise comparison matrices amongst
The proposed combined fuzzy TOPSIS and GRA all the criteria in the hierarchical structure. Assign linguistic
method terms as given in Table 1, to the pairwise comparisons by
asking which is the more important of each two criteria.
In this section, a fuzzy MCDM method is presented to Step 2 Use geometric mean technique to define the fuzzy
evaluate greenness of 3PLs. The proposed method can be geometric mean as follows:
applied to a complex decision-making problem containing " #1=n
imprecise, indefinite and subjective data or vague infor- Y
n

mation. The combined fuzzy TOPSIS and GRA method is r~j ¼ r~jl for j; l ¼ 1; 2; . . .; n ð1Þ
l¼1
proposed to solve fuzzy MCDM problems. A triangular or
trapezoidal fuzzy number is usually adopted to express the where r~jl is fuzzy comparison value for criterion j to cri-
decision-maker’s evaluation on alternatives with respect to terion l; thus, it is geometric mean of fuzzy comparison
each criterion. Indeed, a triangular fuzzy number is a value of criterion j to each criterion.
special version of a trapezoidal fuzzy number. When the
Step 3 Calculate the fuzzy weights of each criterion using
two most promising values are equal, the trapezoidal fuzzy " #1
number becomes a triangular fuzzy number. Therefore, a Xn

trapezoidal fuzzy number can represent more general sit- ~j ¼ r~j

w r~j ð2Þ
j¼1
uations. Membership function lA~ð xÞ of a trapezoidal fuzzy
number A~ ¼ ða1 ; a2 ; a3 ; a4 Þ is defined as: where w ~j is the fuzzy weight of the jth criterion, can be
8 indicated by w ~j ¼ ðwj1 ; wj2 ; wj3 ; wj4 Þ. Here ðwj1 ; wj2 ; wj3 ;
>
> 0; x\a1
>
> x  a1 wj4 Þ stand for the lower, middle and upper values of the
> a1  x  a2
< a2  a1 ;
>
fuzzy weight of the jth criterion, respectively.
lA~ð xÞ ¼ 1; a2  x  a3
>
> x  a1 Step 4 In this step, a panel of decision-makers (DMs)
>
> ; a 3  x  a4
>
>
: a1  a1 who are knowledgeable about 3PLs and evaluation process
0; x [ a4 is established.

123
1382 Int. J. Environ. Sci. Technol. (2016) 13:1377–1392

Table 1 Linguistic scale for Linguistic scale for importance Trapezoidal fuzzy scale Trapezoidal fuzzy reciprocal scale
importance
Equal (E) (1,1,1,1) (1,1,1,1)
Weakly important (WI) (1,3,3,5) (1/5,1/3,1/3,1)
Fairly important (FI) (3,5,6,7) (1/7,1/6,1/5,1/3)
Very strongly important (VSI) (5,7,8,9) (1/9,1/8,1/7,1/5)
Absolutely important (AI) (7,9,9,10) (1/10,1/9,1/9,1/7)

1h 1 i 1X K Step 8 Then, fuzzy positive A~þ and fuzzy negative A~

x~ij ¼ x~ij þ x~2ij þ    þ x~Kij ¼ x~e ð3Þ ideal solutions are determined as the referential sequences.
K K e¼1 ij
It can be defined for beneficial criteria as follows.
where x~tij is the fuzzy value assigned by tth judgment.  
 
A~þ ¼ max vij ¼ v~þ ~þ
1 ;v vþ
2 ; . . .;~m ; i ¼ 1; 2; . . .; m:
Step 5 A decision matrix is formed. j
2 3 ð9Þ
x~11 x~12 . . . x~1n
6 x~21 x~22 . . . x~2n 7    
A~ ¼ 6 7 ¼ x~ij ð4Þ  
4 ... ... ... ... 5 mn A~ ¼ min vij ¼ v~ ~
1 ;v v
2 ; . . .~m ; i ¼ 1; 2; . . .; m:
j
x~m1 x~m2 . . . x~mn
ð10Þ
Step 6 Normalization based on the characteristics of  
where vþ þ þ þ þ     
i ¼ ðai ; bi ; ci ; di Þ and vi ¼ ai ; bi ; ci ; di ;
three types of criteria, namely larger-the-better (benefit),
i ¼ 1; 2; . . .; m:
smaller-the-better (cost) or nominal-the-best (optimal), is
used here to transform the various criteria scales into Step 9 The different separation measures of each alter-
comparable scales. native are calculated in order to determine the grey rela-
x~ij  minð~ xij Þ tional coefficient of each alternative from PIS and NIS. The
Larger the better r~ij ¼ ð5Þ separation between alternatives can be measured by
maxð~ xij Þ  minð~ xij Þ
Hamming distance (d), normalized Hamming distance (l),
minð~ xij Þ  x~ij Euclidean distance (q) and normalized Euclidean distance
Smaller the better r~ij ¼ ð6Þ
maxð~xij Þ  minð~ xij Þ (h). Several definitions are considered that proposed by
Park et al. (2008) and consist of Hamming, Euclidean and
x~ij  x~oj
Nominal the best r~ij ¼
    their normalized version, which are proposed by Burillo
max max x~ij  x~oj ; xoj  min x~ij and Bustince (1996) and Grzegorzewski (2004), respec-
ð7Þ tively. Park et al. (2011a; 2011b) applied these separation
measures to TOPSIS and VIKOR under interval-valued
Step 7 Considering the different weights of each
intuitionistic fuzzy information, respectively. These sepa-
criterion, the weighted normalized decision matrix is
ration measures are applied for trapezoidal fuzzy numbers
computed by multiplying the importance weights of
in the proposed combined fuzzy TOPSIS–GRA method.
evaluation criteria and the values in the normalized
decision matrix. The weighted normalized decision
Separation measures based on the Hamming distance (d)
matrix V~ for each criterion is defined as:
 
V~ ¼ v~ij mn for i ¼ 1; 2; . . .; m and j ¼ 1; 2; . . .; n 1. The extension of Burillo and Bustince’s method,
n h i
ð8Þ 1X   þ  þ  þ
Adj1 ¼ aij  aþ
i þ bij  bi þ cij  ci þ dij  di
where v~ij ¼ r~ij  w
~j here v~ij denotes normalized positive 4 i¼1
 n h i
trapezoidal fuzzy numbers and v~ij ¼ aij ; bij ; cij ; dij ; i ¼ 1X       
Adj1 ¼ aij  a
i þ bij  bi þ cij  ci þ dij  di
4 i¼1
1; 2; . . .; m; j ¼ 1; 2; . . .; n: r~ij is obtained by using
Eqs. (5)–(7). ð11Þ

123
Int. J. Environ. Sci. Technol. (2016) 13:1377–1392 1383

2. The extension of modified Burillo and Bustince’s 3. The extension of Grzegorzewski’s method,
method,

n h i
1X   þ  þ  þ   aþ  bþ þ c  d þ cþ  dþ
Adj2 ¼ aij  aþ
i þ b ij  b i þ c ij  c i þ d ij  d i þ a ij  b ij  i i ij ij i i
4 i¼1
n h i
1X          a  b þ c  d þ c  d
Adj¼2 ¼ aij  a
i þ b ij  b i þ c ij  c i þ d ij  d i þ a ij  b ij  i i ij ij i i
4 i¼1
ð12Þ

3. The extension of Grzegorzewski’s method, 1 Xn h 

 þ
n h  AljH ¼ max aij  aþ i ; bij  bi
1X  þ 2m i¼1
AdjH ¼ max aij  aþ i ; bij  bi  i
2 i¼1  þ
 i þ max cij  cþ
i ; d ij  d i
 þ
þ max cij  cþ
i ; dij  di n h 
1 X  
n h  AljH ¼ max aij  a i ; bij  bi
1X   2m i¼1
AdjH ¼ max aij  a i ; b ij  b i  i
2 i¼1  
 i þ max cij  c
i ; dij  di ð16Þ
 
þ max cij  c
i ; d ij  d i ð13Þ
Separation measures based on the Euclidean distance (q)
Separation measures based on the normalized Hamming
1. The extension of Burillo and Bustince’s method,
distance (l) (
n  2  2
1X
1. The extension of Burillo and Bustince’s method, e1
Aj  ¼ aij  aþ
i þ bij  bþi
4 i¼1
n h i
1 X   þ  þ  þ
Alj1 ¼ aij  aþ
i þ bij  bi þ cij  ci þ dij  di
 2  2 12
4m i¼1  þ
þ cij  ci þ dij  di  þ
n h i
1 X        (
n 
Alj1 ¼ aij  a
i þ b ij  b i þ c ij  c i þ d ij  d i 1X 2  2
4m i¼1 e1
Aj  ¼ aij  a þ bij  b
i i
4 i¼1
ð14Þ
 2  2 12
2. The extension of modified Burillo and Bustince’s þ cij  c
i þ d 
ij  d 
i ð17Þ
method,

n h  i
1 X   þ  þ  þ   þ þ  þ þ
Alj2 ¼ aij  aþ
i þ bij  bi þ cij  ci þ dij  di þ aij  bij  ai  bi þ cij  dij þ ci  di
4m i¼1
n h  i
1 X              
Alj2¼ ¼ aij  a
i þ bij  bi þ cij  ci þ dij  di þ aij  bij  ai  bi þ cij  dij þ ci  di
4m i¼1
ð15Þ

123
1384 Int. J. Environ. Sci. Technol. (2016) 13:1377–1392

2. The extension of modified Burillo and Bustince’s 2. The extension of modified Burillo and Bustince’s
method method,

( ) 1
n  2  2  2  2  2   2 2
1X  þ  þ  þ  þ   þ þ  þ þ
Aej2
¼ aij  ai þ bij  bi þ cij  ci þ dij  di þ aij  bij  ai  bi þ cij  dij þ ci  di
4 i¼1
( ) 1
n  2  2  2  2  2   2 2
e2 1X               
Aj¼ ¼ aij  ai þ bij  bi þ cij  ci þ dij  di þ aij  bij  ai  bi þ cij  dij þ ci  di
4 i¼1
ð18Þ

( ) 1
n  2  2  2  2  2   2 2
1 X   þ  þ
Aqj2
¼  þ  þ  þ  þ þ
aij  ai þ bij  bi þ cij  ci þ dij  di þ aij  bij  ai  bi þ
þ cij  dij þ ci  di
4m i¼1
( ) 1
n  2  2  2  2  2   2 2
1 X  
q2
Aj¼ ¼ aij  ai þ bij  bi þ cij  ci þ dij  di þ aij  bij  ai  bi þ cij  dij þ ci  di
            
4m i¼1
ð21Þ

3. The extension of Grzegorzewski’s method 3. The extension of Grzegorzewski’s method,

( (
n   2 n   2
1X  þ 1 X  þ
AejH ¼ max aij  aþ i ; b ij  b i AqjH ¼ max aij  aþ i ; b ij  b i
2 i¼1 2m i¼1
  2 12   2 12
 þ  þ
þ max cij  cþi ; d ij  d i þ max cij  cþi ; d ij  d i
( (
n   2 n   2
1X   1 X  
AejH ¼ max aij  a i ; b ij  b i AqjH ¼ max aij  a i ; b ij  b i
2 i¼1 2m i¼1
  2 12   2 12
   
þ max cij  ci ; ij
d  d i ð19Þ þ max cij  ci ; ij
d  d i ð22Þ

Separation measures based on the normalized Euclidean Step 10 Calculate the grey relational coefficient of each
distance (h) alternative from PIS and NIS using the following equation,
respectively.
1. The extension of Burillo and Bustince’s method,
( mini minj Aj þ f maxi maxj Aj
n  2  2 cþ ðAj Þ ¼ ð23Þ
1 X Aj þ f maxi maxj Aj
Aqj1 ¼ aij  aþ
i þ b 
ij  b þ
i
4m i¼1
mini minj Aj þ f maxi maxj Aj
 2  2 12 c ðAj Þ ¼ ð24Þ
 þ  þ Aj þ f maxi maxj Aj
þ cij  ci þ dij  di
(
n  2  2
1 X
Aqj1 ¼ aij  a
i þ bij  b i Step 11 The grey relational grade of each alternative
4m i¼1
from PIS and NIS is determined as follows.
 2  2 12 Xn
   
þ cij  ci þ dij  di ð20Þ diþ ¼ cþ ðAj Þ ð25Þ
j¼1

123
Int. J. Environ. Sci. Technol. (2016) 13:1377–1392 1385

X
n
opinions, working in the logistics sector as manager and
di ¼ c ðAj Þ ð26Þ
j¼1
related studies such as Murphy et al. (1994), Noci (1997),
Walton et al. (1998), Handfield et al. (2002), Humphreys
Step 12 Then, the closeness coefficient CCi is et al. (2003), Yang and Wu (2007), Lee et al. (2009),
determined. Tuzkaya et al. (2009), Viswanathan, (2009), Kuo et al.
CCi ¼ diþ =di ð27Þ (2010), Amindoust et al. (2012) and Buyukozkan (2012).
Three decision-makers are consulted to obtain the weights
Step 13 Rank the preference order. Using this index, the of the criteria and the importance of green 3PLs with
alternatives can be ranked in decreasing order. respect to criteria. Our experts are working as manager in
the logistics sector. They work in the different companies
within the same job position. Because the companies which
Results and discussion worked by experts in are pioneers in their field and deci-
sion-makers that served as the manager, it is almost
A case study impossible to find any senior managers to score them. So,
the weights of experts are considered as equal.
In this section, first, a case study is structured considering Five Turkish 3PLs that serve in Turkey are selected.
five 3PL companies under several decision criteria which These companies are listed alphabetically as Borusan
make 3PLs more ‘‘green’’. Then, the criteria weights and Logistics, Ekol Logistics, Horoz Logistics, Omsan Logis-
final ranking of alternatives are determined based on the tics and Reysas Logistics. Only the names of 3PLs are
proposed combined fuzzy TOPSIS–GRA method. Finally, mentioned because of privacy. In the final ranking, A1, A2,
the verification of the proposed methodology is conducted. A3, A4 and A5 are used for each 3PLs without giving their
names. The criteria explanations are as below:
The hierarchical structure Cooperation with customer company and its customers
(C1): That means not only 3PLs should take cognizance of
Here, an empirical case is considered for evaluating green greening issues, also customers and their customers should
3PLs serve in Istanbul, to illustrate the proposed method’s pay attention to protect environment (Walton et al. 1998).
applicability. The aim of the case is to show the advantages Green government regulations (C2): Government agen-
of the proposed method and select the greenest 3PL with cies at state and local levels should work to build sus-
respect to green criteria. The decision criteria and alter- tainable communities. Technologies and policies that are
natives hierarchy are shown in Fig. 1. greening business need to be constituted. (Handfield et al.
In this part of our study, the most important criteria are 2002; Kuo et al. 2010).
determined for selecting and evaluating green 3PLs. The Environmental management system (C3): These activi-
evaluation criteria are determined based on expert ties may include environmentally relevant certificates (such

GREEN LOGISTICS SERVICE

PROVIDER EVALUATION

Cooperation with
Customer Green government Environmental Green Process Reduction of Energy Green
Company and its Regulations Management System Design Consumption Network Design
Customer

A1 A2 A3 A4 A5

Fig. 1 The decision criteria and alternatives

123
1386 Int. J. Environ. Sci. Technol. (2016) 13:1377–1392

as ISO 14000, WEEE) and reverse logistics management manager, is presented in Table 3 as linguistic evaluations.
programs. This should be containing to check supplier’s Then, it is converted into trapezoidal fuzzy numbers to
environmental policies, implementation and certification determine the average fuzzy performance values by using
(Buyukozkan 2012; Walton et al. 1998; Noci 1997; Mur- Eq. (3), with the guidance of experts, as given in Table 4.
phy et al. 1994; Tuzkaya et al. 2009; Yang and Wu 2007; Then, the normalized fuzzy decision matrix is calculated
Humphreys et al. 2003; Amindoust et al. 2012). by using Eqs. (5)–(7). The weighted fuzzy normalized
Green process design (C4): Arise in pollution reveals decision matrix is calculated by using Eq. (8) with respect
inefficient use of inputs and eco-unfriendliness in one or to fuzzy weights of the criteria which are obtained by fuzzy
more logistics activities, spanning from product design, AHP (Eqs. 1–2); it can be seen from Table 4. Also, the
production and distribution to disposal. Such problems can fuzzy positive and fuzzy negative ideal solutions are
be prevented through redesigning the product and pro- determined by using Eqs. (9)–(10), and they are given in
duction processes (Noci 1997; Handfield et al. 2002; Lee Table 4.
et al. 2009; Humphreys et al. 2003; Amindoust et al. 2012). Then, the grey relational coefficients of each alternative
Reduction in energy consumption (C5): With green from PIS and NIS are calculated by using Eq. (11). In this
logistics practices, 3PLs can reduce the amount of diesel step, as an example, the extension of Burillo and Bustince’s
consumptions by procuring state-of-the-art engines for method based on separation measures based on the Ham-
trucks, using the scarce resources efficiently, converting ming distance (d) is calculated in Table 5.
roadway to railway if possible and using advanced fuel In this study, as an extension, different separation
management system (Walton et al. 1998; Lee et al. 2009; measures are proposed for calculating the grey relational
Tuzkaya et al. 2009; Amindoust et al. 2012). coefficient using Eqs. (11)–(22). The grey relational coef-
Green network design (C6): Green network enables ficient of each alternative from PIS and NIS is calculated
companies to achieve the purpose of green supply chain using Eqs. (23)–(24). Here, the grey relational grade is
movement by optimizing resources and reducing waste calculated (Eqs. 25–26). Finally, the closeness coefficient
(Viswanathan 2009).
Table 3 Importance of green 3PLs with respect to criteria
Determination of criteria weights
C1 C2 C3 C4 C5 C6

The importance values of criteria are determined via fuzzy D1 A1 A FG VG G G NB

AHP. The pairwise comparison scores are carried out by A2 P A P G P G
three experts using Table 2. In order to obtain the impor- A3 P P A FG FG G
tance of each criterion, the experts applied a nine-point A4 VG VG FG G G NB
scale given in Table 1. A5 G FG G G G NB
Consistency ratio (CR) for the defuzzified version is D2 A1 VG PF PF G G FG
calculated as 0.080, and it is \0.10. The results show that A2 VG A A A VG FG
the decision matrix for the proposed hierarchical structure A3 VG FG VG VG FG FG
is consistent. The fuzzy weights of the criteria are given in A4 G FG FG FG PF FG
Table 2. These weight values are used as inputs of the A5 G FG FG FG PF FG
proposed method. D3 A1 VG G FG G G PF
A2 A A A VG VG FG
The final ranking A3 A VG A FG FG G
A4 G FG G G PF PF
The importance of 3PLs with respect to criteria, assessed
A5 G G G PF PF NB
by decision-makers working in the logistic sector as

Table 2 Pairwise comparisons Criteria C1 C2 C3 C4 C5 C6 Fuzzy weights

of evaluation criteria
linguistically and trapezoidal C1 E WI FI FI FI FI (0.17;0.43;0.43;0.98)
fuzzy values
C2 WLI E WI FI WI FI (0.09;0.25;0.25;0.67)
C3 FLI FLI E WI WLI E (0.03;0.08;0.08;0.22)
C4 FLI FLI WLI E FLI WLI (0.02;0.03;0.03;0.12)
C5 FLI WLI WI FI E WI (0.05;0.14;0.14;0.4)
C6 FLI FLI E WI WLI E (0.03;0.07;0.07;0.19)

123
Int. J. Environ. Sci. Technol. (2016) 13:1377–1392 1387

(0.241;0.241;0.669;0.031)
(0.154;0.205;0.594;0.026)
(0.092;0.172;0.532;0.005)
(Eq. 27) of each alternative is shown in the first column of

(0.803;0.907;0.943;0.86)
(0.68;0.84;0.897;0.493)
(0.56;0.727;0.79;0.407)

(0.01;0.028;0.044;0.14)

(0.015;0.04;0.06;0.185)
(0.4;0.453;0.593;0.65)
Table 6. The resolving coefficient is assumed as f = 0.5.

(0.033;0.116;0.392;0)

(0.033;0.116;0.392;0)
(0.98;0.98;1;0.953)
The CCi values of each 3PLs A1, A2, A3, A4 and A5 are
0.6078, 2.9659, 1.9397, 0.6481 and 0.4073, respectively.
Therefore, the ranking order of the five 3PLs is
A2 [ A3 [ A4 [ A1 [ A5 . Thus, it is clear that the most
appropriate candidate is A2. In this paper, the all separation
C6

measures are also used to calculate CCi values of green

3PLs, and it is presented in Table 6. For example, the
(0.767;0.803;0.907;0.943)

(0.075;0.241;0.241;0.669)
(0.047;0.154;0.205;0.594)
(0.023;0.092;0.172;0.532)

(0.038;0.11;0.134;0.403)
second column shows result of CCi values with respect to
(0.627;0.68;0.84;0.897)
(0.493;0.56;0.727;0.79)

(0;0.023;0.059;0.213)

(0;0.033;0.116;0.392)

(0;0.033;0.116;0.392)
(0.32;0.42;0.58;0.65)

the extension of Burillo and Bustince’s method (Eq. 11),

(0.93;0.98;0.98;1)

and the third column shows the result of CCi values with
respect to the extension of modified Burillo and Bustince’s
Table 4 Average fuzzy performance values and the weighted fuzzy normalized decision matrix with fuzzy positive and negative ideal solutions

method (Eq. 12).

C5

Verification and discussion

(0.978;0.075;0.241;0.241)
(0.978;0.047;0.154;0.205)
(0.638;0.023;0.092;0.172)

The resolving coefficient value is used for sensitivity

(0.99;0.767;0.803;0.907)

(0.01;0.024;0.031;0.118)
(0.757;0.627;0.68;0.84)
(0.65;0.493;0.56;0.727)

analysis in order to verify the proposed method, if it is

(0;0.006;0.016;0.068)

(0.482;0;0.033;0.116)

(0.482;0;0.033;0.116)
(0.32;0.42;0.58;0.65)
(0.99;0.93;0.98;0.98)

rational and stable, or not. During verification process, it is

observed that various resolving coefficient values, as the
extension of Burillo and Bustince’s method based on
Hamming distance, do not affect the ranking order of the
3PLs (Fig. 2). Also, it can be said that the application of
C4

other separation measures does not affect the ranking

order. But as the f value grows, the CCi values of alter-
(0.005;0.017;0.036;0.129)
(0.414;0.978;0.075;0.241)
(0.414;0.978;0.047;0.154)
(0.239;0.638;0.023;0.092)

(0.031;0.074;0.074;0.223)
(0.967;0.99;0.767;0.803)
(0.693;0.757;0.627;0.68)

natives converge.
(0.49;0.543;0.693;0.75)
(0.967;0.99;0.93;0.98)

(0.58;0.65;0.493;0.56)

(0.167;0.482;0;0.033)

(0.167;0.482;0;0.033)

Table 7 shows the CCi values based on different sepa-

ration measures with respect to different f values. As seen
from the table, A2 is the greenest 3PL with respect to
closeness coefficient CCi. The ranking order of the 3PLs
does not change with the separation measure and f value
C3

changes. From the results of CCi, finally, the order of

priority for 3PLs is obtained as A2 [ A3 [ A4 [ A1 [ A5 ,
with respect to different separation measures.
(0.384;0.414;0.978;0.075)
(0.384;0.414;0.978;0.047)
(0.141;0.239;0.638;0.023)

(0.075;0.241;0.241;0.669)
(0.54;0.693;0.757;0.627)
(0.493;0.56;0.727;0.79)

(0.92;0.967;0.99;0.767)

A comparative study is implemented with other

(0.92;0.967;0.99;0.93)

(0.42;0.58;0.65;0.493)

(0;0.033;0.116;0.392)

(0.064;0.167;0.482;0)

(0.064;0.167;0.482;0)

methods to validate the efficiency and applicability of the

proposed methods. The proposed method result, based on
different separation measures, is compared with fuzzy
TOPSIS and fuzz VIKOR results in order to verify the
The weighted fuzzy normalized decision matrix

effectiveness of it, as seen in Table 8. The first compar-

C2

ative analysis is conducted with the results obtained by

Chen (2000) for hiring a system analysis engineer for a
(0.116;0.337;0.397;0.964)
(0.139;0.384;0.414;0.978)
(0.139;0.384;0.414;0.978)
(0.033;0.141;0.239;0.638)

(0.139;0.384;0.414;0.978)
Average fuzzy performance values

(0.453;0.54;0.693;0.757)

software company as a personnel selection problem. The

(0.883;0.92;0.967;0.99)
(0.883;0.92;0.967;0.99)
(0.79;0.847;0.94;0.98)

(0;0.064;0.167;0.482)

(0;0.064;0.167;0.482)
(0.32;0.42;0.58;0.65)

basic concept of the TOPSIS method is that the chosen

alternative should display the shortest distance from the
positive ideal solution and the farthest distance from the
negative ideal solution. The ranking result of the alter-
Fuzzy PIS and NIS

natives yielded by the fuzzy TOPSIS method using a

C1

closeness coefficient approach (Chen 2000) is same to

that obtained by the proposed method. The second com-
parative analysis is conducted with the results obtained by
~
A~
A1
A2
A3
A4
A5

A1
A2
A3
A4
A5

Kaya and Kahraman (2010) for selection of the best

123
1388 Int. J. Environ. Sci. Technol. (2016) 13:1377–1392

Table 5 Grey relational PIS NIS d? d-

coefficient and grade of each
alternative from PIS and NIS C1 C2 C3 C4 C5 C6 C1 C2 C3 C4 C5 C6
based on Hamming distance
A1 0.86 0.33 0.36 0.36 0.33 0.56 0.35 1 0.81 1 1 0.45 2.805 4.614
A2 1 1 1 0.97 1 1 0.33 0.33 0.33 0.34 0.34 0.33 5.965 2.011
A3 1 0.6 0.85 1 0.47 0.7 0.33 0.43 0.36 0.33 0.54 0.39 4.615 2.379
A4 0.41 0.46 0.39 0.44 0.33 0.56 0.64 0.55 0.71 0.64 1 0.45 2.587 3.992
A5 0.33 0.33 0.33 0.56 0.33 0.33 1 1 1 0.47 1 1 2.228 5.471

Table 6 CCi values for each d1 d2 dh l1 l2 lh e1 e2 eh q1 q2 qh

separation measure
A1 0.61 0.61 0.61 0.61 0.61 0.61 0.61 1.14 0.61 0.61 1.14 0.61
A2 2.97 2.98 2.95 2.97 2.98 2.95 2.94 1.37 2.97 2.94 1.37 2.97
A3 1.94 1.93 1.96 1.94 1.93 1.96 1.97 1.32 1.96 1.97 1.32 1.96
A4 0.65 0.65 0.65 0.65 0.65 0.65 0.66 1.03 0.66 0.66 1.03 0.66
A5 0.41 0.41 0.41 0.41 0.41 0.41 0.41 0.90 0.41 0.41 0.90 0.41

12.00 between two method’s results and a perfect Spearman

A1 A2 A3 A4 A5 correlation must be equal to 1 or -1. The correlation
10.00 coefficient between the proposed method, fuzzy TOPSIS
and fuzzy VIKOR is high, positive and negative,
8.00 respectively, as should be.

6.00
Conclusion
4.00
Increased environmental pollution is forcing companies to
2.00 become more environmentally friendly. Companies that
produce logistics services should obtain sensitivity to
0.00
ζ=0,1 ζ=0,2 ζ=0,3 ζ=0,4 ζ=0,5 ζ=0,6 ζ=0,7 ζ=0,8 ζ=0,9 ζ=1,0
environmental awareness and environmentally friendly
policies. The interaction of 3PLs with the environment is
Fig. 2 Variation analysis of CCi values for each alternative based on much more than the other companies, so that these
Hamming distance companies should make further efforts in order to protect
it.
energy policy and wind production site. VIKOR finds out In this study, a fuzzy MCDM method is presented
the compromise ranking list, the compromise solution. It combining fuzzy TOPSIS and GRA methods, based on
ranks and selects best alternative from a set of alternatives different separation measures, in order to evaluate and
in the presence of conflicting attributes. Consider the select the greenest 3PL. First, the criteria that make 3PLs
ranking results yielded by Kaya and Kahraman (2010) more ‘‘green’’ are determined as: cooperation with cus-
and by our proposed method. A2 is the greenest 3PL with tomer company and its customer, green government regu-
respect to six green criteria. Because the two approaches lations, environmental management system, green process
yielded the identical ranking results, it could once again design, reduction in energy consumption and green net-
confirm that the proposed methods are effective to deal work design. The criteria weights based on expert opinions
with the problem of personnel selection problem. On the are determined by fuzzy AHP. Then, the proposed com-
other hand, the Spearman rank correlation coefficient is bined fuzzy TOPSIS–GRA method is used to evaluate
applied to measure the correlation between fuzzy VIKOR green 3PLs based on different separation measures. Five
(Kaya and Kahraman 2010) and fuzzy TOPSIS (Chen Turkish 3PLs serve in Istanbul are selected in order to
2000). Spearman rank correlation coefficient is a non- apply a case study to show applicability of the proposed
parametric measure of statistical dependence between two method. The proposed method is verified with respect to
method’s results. It evaluates how well the relationship different resolving coefficient values and separation

123
Int. J. Environ. Sci. Technol. (2016) 13:1377–1392 1389

Table 7 CCi values based on d1 d2 dh l1 l2 lh e1 e2 eh q1 q2 qh

each resolving coefficient for
different separation measures f = 0.1
A1 0.31 0.32 0.32 0.31 0.32 0.32 0.32 1.16 0.32 0.32 1.16 0.32
A2 10.57 10.68 10.45 10.57 10.68 10.45 10.33 1.61 10.63 10.33 1.61 10.63
A3 4.44 4.39 4.51 4.44 4.39 4.51 4.50 1.52 4.60 4.50 1.52 4.60
A4 0.38 0.39 0.39 0.38 0.39 0.39 0.40 1.09 0.41 0.40 1.09 0.41
A5 0.14 0.14 0.14 0.14 0.14 0.14 0.14 0.86 0.14 0.14 0.86 0.14
f = 0.2
A1 0.43 0.44 0.44 0.43 0.44 0.44 0.44 1.17 0.44 0.44 1.17 0.44
A2 5.86 5.90 5.82 5.86 5.90 5.82 5.76 1.53 5.86 5.76 1.53 5.86
A3 2.99 2.97 3.03 2.99 2.97 3.03 3.04 1.45 3.05 3.04 1.45 3.05
A4 0.49 0.49 0.50 0.49 0.49 0.50 0.51 1.06 0.51 0.51 1.06 0.51
A5 0.23 0.23 0.23 0.23 0.23 0.23 0.23 0.87 0.23 0.23 0.87 0.23
f = 0.3
A1 0.51 0.51 0.51 0.51 0.51 0.51 0.51 1.16 0.51 0.51 1.16 0.51
A2 4.26 4.28 4.23 4.26 4.28 4.23 4.20 1.46 4.26 4.20 1.46 4.26
A3 2.43 2.42 2.46 2.43 2.42 2.46 2.47 1.40 2.47 2.47 1.40 2.47
A4 0.56 0.56 0.56 0.56 0.56 0.56 0.58 1.05 0.57 0.58 1.05 0.57
A5 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.88 0.31 0.30 0.88 0.31
f = 0.4
A1 0.57 0.57 0.57 0.57 0.57 0.57 0.57 1.15 0.57 0.57 1.15 0.57
A2 3.45 3.47 3.44 3.45 3.47 3.44 3.41 1.41 3.45 3.41 1.41 3.45
A3 2.13 2.12 2.15 2.13 2.12 2.15 2.17 1.35 2.16 2.17 1.35 2.16
A4 0.61 0.61 0.61 0.61 0.61 0.61 0.62 1.04 0.62 0.62 1.04 0.62
A5 0.36 0.36 0.36 0.36 0.36 0.36 0.36 0.89 0.36 0.36 0.89 0.36
f = 0.5
A1 0.61 0.61 0.61 0.61 0.61 0.61 0.61 1.14 0.61 0.61 1.14 0.61
A2 2.97 2.98 2.95 2.97 2.98 2.95 2.94 1.37 2.97 2.94 1.37 2.97
A3 1.94 1.93 1.96 1.94 1.93 1.96 1.97 1.32 1.96 1.97 1.32 1.96
A4 0.65 0.65 0.65 0.65 0.65 0.65 0.66 1.03 0.66 0.66 1.03 0.66
A5 0.41 0.41 0.41 0.41 0.41 0.41 0.41 0.90 0.41 0.41 0.90 0.41
f = 0.6
A1 0.64 0.64 0.64 0.64 0.64 0.64 0.64 1.13 0.65 0.64 1.13 0.65
A2 2.64 2.65 2.63 2.64 2.65 2.63 2.62 1.34 2.64 2.62 1.34 2.64
A3 1.80 1.80 1.82 1.80 1.80 1.82 1.83 1.29 1.82 1.83 1.29 1.82
A4 0.68 0.68 0.68 0.68 0.68 0.68 0.69 1.02 0.69 0.69 1.02 0.69
A5 0.45 0.45 0.45 0.45 0.45 0.45 0.45 0.91 0.45 0.45 0.91 0.45
f = 0.7
A1 0.67 0.67 0.67 0.67 0.67 0.67 0.67 1.13 0.67 0.67 1.13 0.67
A2 2.41 2.42 2.40 2.41 2.42 2.40 2.39 1.31 2.41 2.39 1.31 2.41
A3 1.71 1.70 1.72 1.71 1.70 1.72 1.73 1.27 1.72 1.73 1.27 1.72
A4 0.70 0.70 0.71 0.70 0.70 0.71 0.71 1.02 0.71 0.71 1.02 0.71
A5 0.49 0.49 0.49 0.49 0.49 0.49 0.49 0.91 0.49 0.49 0.91 0.49
f = 0.8
A1 0.69 0.70 0.69 0.69 0.70 0.69 0.70 1.12 0.70 0.70 1.12 0.70
A2 2.23 2.24 2.23 2.23 2.24 2.23 2.22 1.29 2.23 2.22 1.29 2.23
A3 1.63 1.62 1.64 1.63 1.62 1.64 1.65 1.25 1.64 1.65 1.25 1.64
A4 0.73 0.72 0.73 0.73 0.72 0.73 0.73 1.02 0.73 0.73 1.02 0.73
A5 0.52 0.52 0.52 0.52 0.52 0.52 0.52 0.92 0.52 0.52 0.92 0.52

123
1390 Int. J. Environ. Sci. Technol. (2016) 13:1377–1392

Table 7 continued d1 d2 dh l1 l2 lh e1 e2 eh q1 q2 qh

f = 0.9
A1 0.71 0.72 0.72 0.71 0.72 0.72 0.72 1.11 0.72 0.72 1.11 0.72
A2 2.10 2.10 2.09 2.10 2.10 2.09 2.08 1.27 2.10 2.08 1.27 2.10
A3 1.57 1.56 1.58 1.57 1.56 1.58 1.58 1.23 1.58 1.58 1.23 1.58
A4 0.74 0.74 0.75 0.74 0.74 0.75 0.75 1.02 0.75 0.75 1.02 0.75
A5 0.54 0.55 0.54 0.54 0.55 0.54 0.55 0.92 0.55 0.55 0.92 0.55
f = 1.0
A1 0.73 0.73 0.73 0.73 0.73 0.73 0.73 1.11 0.74 0.73 1.11 0.74
A2 1.99 1.99 1.98 1.99 1.99 1.98 1.97 1.25 1.99 1.97 1.25 1.99
A3 1.52 1.51 1.52 1.52 1.51 1.52 1.53 1.22 1.53 1.53 1.22 1.53
A4 0.76 0.76 0.76 0.76 0.76 0.76 0.77 1.01 0.77 0.77 1.01 0.77
A5 0.57 0.57 0.57 0.57 0.57 0.57 0.57 0.92 0.57 0.57 0.92 0.57

Table 8 Ranking results of Fuzzy TOPSIS Rank Fuzzy VIKOR (Kaya Rank The proposed method (Burillo and Rank
fuzzy TOPSIS, fuzzy VIKOR (Chen 2000) and Kahraman 2010) Bustince’s Hamming distance)
and the proposed method
A1 0.1642 3 0.5900 3 0.6078 3
A2 0.2263 1 0.0000 1 2.9659 1
A3 0.2069 2 0.2024 2 1.9397 2
A4 0.1477 4 0.7313 4 0.6481 4
A5 0.1147 5 1.0000 5 0.4073 5

measures and also compared with fuzzy TOPSIS and fuzzy Burillo P, Bustince H (1996) Entropy on intuitionistic fuzzy sets and
VIKOR method results. As future directions, different interval-valued fuzzy sets. Fuzzy Set Syst 78:305–316
Buyukozkan G (2012) An integrated fuzzy multi-criteria group
MCDM methods can be applied and compared to check decision-making approach for green supplier evaluation. Int J
validity of our results. The proposed method can be applied Prod Res 50(11):2892–2909
to 3PLs in other countries. Buyukozkan G, Ciftci G (2012) A novel hybrid MCDM approach
based on fuzzy DEMATEL, fuzzy ANP and fuzzy TOPSIS to
evaluate green suppliers. Expert Syst Appl 39:3000–3011
Chen CT (2000) Extensions of the TOPSIS for group decision-
making under fuzzy environment. Fuzzy Set Syst 114:1–9
References Chen CC (2009) Environmental impact assessment framework by
integrating scientific analysis and subjective perception. Int J
Ageron B, Gunasekaran A, Spalanzani A (2011) Sustainable supply Environ Sci Technol 6(4):605–618
management: an empirical study. Int J Prod Econ 150(1):168–182 Chen SJ, Hwang CL (1992) Fuzzy multi attribute decision making.
Aghazadeh SM (2003) How to choose an effective third party Lecture notes in economics and mathematical system series, vol
logistics provider. Manag Res News 2(7):50–58 375. Springer, New York
Amindoust A, Ahmed S, Saghafinia A, Bahreininejad A (2012) Chen MF, Tzeng GH (2004) Combining grey relation and TOPSIS
Sustainable supplier selection: a ranking model based on fuzzy concepts for selecting an expatriate host country. Math Comput
inference system. Appl Soft Comput 12:1668–1677 Model 40(13):1473–1490
Awasthi A, Chauhan SS, Goyal SK (2010) A fuzzy multicriteria Deng JL (1982) Control problems of grey systems. Syst Control Lett
approach for evaluating environmental performance of suppliers. 1(5):288–294
Int J Prod Econ 126(2):370–378 Deng JL (1989) Introduction to grey system theory. J Grey Syst
Azadi M, Saen RF (2011) A new chance-constrained data envelop- 1:1–24
ment analysis for selecting third-party reverse logistics providers Eglese R, Black D (2010) Optimizing the routing of vehicles. In:
in the existence of dual-role factors. Expert Syst Appl McKinnon A, Culliane S, Browne M, Whiteing A (eds) Green
38:12231–12236 logistics: improving the environmental sustainability of logistics.
Bhatnagar R, Sohal AS, Millen R (1999) Third party logistics Kogan Page Publishers, London
services: a Singapore perspective. Int J Phys Distrib Logist Erdogan M, Gumus AT (2012) Environmental consciousness of 3PL
Manag 29:569–587 firms: a brief literature review. In: Proceeding of the interna-
Buckley JJ (1985) Fuzzy hierarchical analysis. Fuzzy Set Syst tional conference on IT applications and management, İstanbul,
17:233–247 June 2012, pp 194–200

123
Int. J. Environ. Sci. Technol. (2016) 13:1377–1392 1391

Evangelista P, Huge-Brodin M, Isaksson K, Sweeney E (2011) The Liu H, Wang W (2009) An integrated fuzzy approach for provider
impact of 3PL’s green initiatives on the purchasing of transport evaluation and selection in third-party logistics. Expert Syst
and logistics services: an exploratory study. In: Proceedings of Appl 36:4387–4398
the 20th International Purchasing and Supply Education and McKinnon A (2010a) Product-level carbon auditing of supply chains,
Research Association (IPSERA) conference environmental imperative or wasteful distraction? Int J Phys
Fu X, Zhu Q, Sarkis J (2012) Evaluating green supplier development Distrib Logist Manag 40:42–60
programs at a telecommunications systems provider. Int J Prod McKinnon A (2010b) Environmental sustainability: a new priority for
Econ 140(1):357–367 logistics managers. In: McKinnon A, Culliane S, Browne M,
Govindan K, Palaniappan M, Zhu Q, Kannan D (2012) Analysis of Whiteing A (eds) Green logistics: improving the environmental
third party reverse logistics provider using interpretive structural sustainability of logistics. Kogan Page Publishers, London
modeling. Int J Prod Econ 140(1):204–211 McKinnon A (2010c) Increasing fuel efficiency in the road freight
Grzegorzewski P (2004) Distance between intuitionistic fuzzy sets sector. In: McKinnon A, Culliane S, Browne M, Whiteing A
and/or interval-valued fuzzy sets on the Hausdorff metric. Fuzzy (eds) Green logistics: improving the environmental sustainability
Set Syst 148:319–328 of logistics. Kogan Page Publishers, London
Gumus AT (2009) Evaluation of hazardous waste transportation firms Min H, Ko H (2008) The dynamic design of a reverse logistics
by using a two step fuzzy-AHP and TOPSIS methodology. network from the perspective of third-party logistics service
Expert Syst Appl 36:4067–4074 providers. Int J Prod Econ 113:176–192
Hamdan A, Rogers KJ (2008) Evaluating the efficiency of 3PL Murphy P, Poist RF, Braunschweig CD (1994) Management of
logistics operations. Int J Prod Econ 113:235–244 environmental issues in logistics: current status and future
Handfield R, Walton R, Sroufe R, Melnky SA (2002) Applying potential. Transp J 34:48–56
environmental criteria to supplier assessment: a study in the Noci G (1997) Designing green vendor rating systems for the
application of the analytical hierarchy process. Eur J Oper Res assessment of a supplier’s environmental performance. Eur J
141:70–87 Purch Supply Manag 3(2):103–114
Harris I, Rodrigues SV, Naim M, Mumford C (2010) Restructuring of Park JH, Lim KM, Park JS, Kwun YC (2008) Distances between
logistics systems and supply chains. In: McKinnon A, Culliane interval-valued intuitionistic fuzzy sets. J Phys: Conf Ser
S, Browne M, Whiteing A (eds) Green Logistics: improving the 96(012089):1–8
environmental sustainability of logistics. Kogan Page Publishers, Park JH, Cho HJ, Kwun YC (2011a) Extension of the VIKOR method
London for group decision making with interval-valued intuitionistic
Ho W, He T, Man Lee CK, Emrouznejad A (2012) Strategic logistics fuzzy information. Fuzzy Optim Decis Mak 10:233–253
outsourcing: an integrated QFD and fuzzy AHP approach. Park JH, Park YII, Kwun YC, Tan X (2011b) Extension of the
Expert Syst Appl 3:10841–10850 TOPSIS method for decision making problems under interval-
Humphreys PK, Wong YK, Chan FTS (2003) Integrating environ- valued intuitionistic fuzzy environment. Appl Math Model
mental criteria into the supplier selection process. J Mater 35:2544–2556
Process Technol 138:349–356 Ramanathan R, Yunfeng J (2009) Incorporating cost and environ-
Hwang CL, Yoon K (1992) Fuzzy multiple attribute decision making: mental factors in quality function deployment using data
theory and application. Springer, Berlin development analysis. Omega 37:711–723
Işıklar G, Alptekin E, Büyüközkan G (2007) Application of a hybrid Razzaque MA, Sheng C (1998) Outsourcing of logistics functions: a
intelligent decision support model in logistics outsourcing. literature survey. Int J Phys Distrib Logist Manag 28:89–107
Comput Oper Res 34:3701–3714 Robertson PW, Gibson PR, Flanagan JT (2002) Strategic supply chain
Jharkharia S, Shankar R (2007) Selection of logistics service development by integration of key global logistical process
provider: an analytic network process (ANP) approach. Omega linkages. Int J Prod Res 40:4021–4040
35:274–289 Rodrigue J, Slack B, Comtois C (2001) Green logistics (The
Kannan G, Pokharel S, Kumar PS (2009) A hybrid approach using paradoxes of). The handbook of logistics and supply-chain
ISM and fuzzy TOPSIS for the selection of reverse logistics management. Handbooks in transport #2. Pergamon/Elsevier,
provider. Resour Conserv Recycl 54:28–36 London
Kaya T, Kahraman C (2010) Multicriteria renewable energy planning Shan L (2012) Research on green logistics service providers selection
using an integrated fuzzy VIKOR & AHP methodology: the case based on intuitionistic language fuzzy entropy. J Comput
of Istanbul. Energy 35(6):2517–2527 7:540–546
Krumwiede DW, Sheu C (2002) A model for reverse logistics entry Tezuka K (2011) Rationale for utilizing 3PL in supply chain
by third-party providers. Omega 30:325–333 management: a shippers’ economic perspective. IATSS Res
Kuo RJ, Wang YC, Tien FC (2010) Integration of artificial neural 35:24–29
network and MADA methods for green supplier selection. Tseng ML (2010) Using linguistic preferences and grey relational
J Clean Prod 18(12):1161–1170 analysis to evaluate the environmental knowledge management
Kuo RJ, Hsu CW, Chen YL (2015) Integration of fuzzy ANP and capacity. Expert Syst Appl 37:70–81
fuzzy TOPSIS for evaluating carbon performance of suppliers. Tuzkaya G, Ozgen A, Ozgen D, Tuzkaya UR (2009) Environmental
Int J Environ Sci Technol. doi:10.1007/s13762-015-0819-9 performance evaluation of suppliers: a hybrid fuzzy multi-
Lai K, Christina WYW (2012) Green logistics management and criteria decision approach. Int J Environ Sci Technol
performance: some empirical evidence from Chinese manufac- 6(3):477–490
turing exporters. Omega 40:267–282 Viswanathan N (2009) Supply chain network design: architecting a
Lee AHI, Kang Y, Hsu HCF, Hung HC (2009) A green supplier green future. www.supplychainbrain.com/content/home/single-
selection model for high-tech industry. Expert Syst Appl article-page/article/supply-chain-network-design-architecting-a-
36:7917–7927 green-future. 15 Jul 2012
Lieb K, Lieb R (2010) Environmental sustainability in the third-party Walker H, Sisto LD, McBain D (2008) Drivers and barriers to
logistics (3PL) industry. Int J Phys Distrib Logist Manag environmental supply chain management practices: lessons from
40:524–533 the public and private sectors. J Purch Supply Manag 14:69–85

123
1392 Int. J. Environ. Sci. Technol. (2016) 13:1377–1392

Walton SV, Handfield RB, Melnyk SA (1998) The green supply Yang Y, Wu L (2007) Grey entropy method for green supplier
chain: integrating. suppliers into environmental management selection. In: International conference on wireless communica-
processes. J Supply Chain Manag 34(2):2–11 tions, networking and mobile computing, pp 4682–4685
Wang YJ (2009) Combining grey relation analysis with FMCGDM to Ying J, Li-junb Z (2008) Study on green supply chain management
evaluate financial performance of Taiwan container lines. Expert based on circular economy. Mech Mater 84–85:761–764
Syst Appl 36:2424–2432 Zhang SF, Liu SY (2011) A GRA-based intuitionistic fuzzy multi-
Wei GW (2010) GRA method for multiple attribute decision making criteria group decision making method for personnel selection.
with incomplete weight information in intuitionistic fuzzy Expert Syst Appl 38:11401–11405
setting. Knowl-Based Syst 23:243–247

123