This article was downloaded by: [University of Missouri Columbia]

On: 22 September 2013, At: 12:50

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

International Journal of Production Research

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/tprs20

Supplier selection based on supply chain ecosystem,

performance and risk criteria
a a
N. Viswanadham & A. Samvedi
a
Department of Computer Science and Automation , Indian Institute of Science Bangalore ,
Bangalore , India
Published online: 06 Aug 2013.

To cite this article: N. Viswanadham & A. Samvedi , International Journal of Production Research (2013): Supplier selection
based on supply chain ecosystem, performance and risk criteria, International Journal of Production Research, DOI:
10.1080/00207543.2013.825056

To link to this article: http://dx.doi.org/10.1080/00207543.2013.825056

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
 International Journal of Production Research, 2013
http://dx.doi.org/10.1080/00207543.2013.825056

Supplier selection based on supply chain ecosystem, performance and risk criteria
N. Viswanadham* and A. Samvedi

Department of Computer Science and Automation, Indian Institute of Science Bangalore, Bangalore, India
(Received 27 March 2013; final version received 5 June 2013)

A supply chain ecosystem consists of the elements of the supply chain and the entities that influence the goods,
information and financial flows through the supply chain. These influences come through government regulations,
human, financial and natural resources, logistics infrastructure and management, etc., and thus affect the supply chain
Downloaded by [University of Missouri Columbia] at 12:50 22 September 2013

performance. Similarly, all the ecosystem elements also contribute to the risk. The aim of this paper is to identify both
performances-based and risk-based decision criteria, which are important and critical to the supply chain. A two step
approach using fuzzy AHP and fuzzy technique for order of preference by similarity to ideal solution has been proposed
for multi-criteria decision-making and illustrated using a numerical example. The first step does the selection without
considering risks and then in the next step suppliers are ranked according to their risk profiles. Later, the two ranks are
consolidated into one. In subsequent section, the method is also extended for multi-tier supplier selection. In short, we
are presenting a method for the design of a resilient supply chain, in this paper.
Keywords: supply chain risk management; supply chain ecosystem; supplier selection; fuzzy AHP; fuzzy TOPSIS

1. Introduction
Over the last two decades, companies had worked hard to reduce costs and improve efficiency of the supply chain
processes by which they delivered products to their customers at the right cost and at the right times. They had done
this by implementing techniques such as the lean production, just-in-time manufacturing, single-source suppliers and
global outsourcing from low-cost countries (Viswanadham and Kameshwaran 2013). The supply chains were highly
connected making the flow of goods, information and funds very smooth and easy. The biggest supply chain challenge
pursued was the supply demand matching avoiding obsolescent inventory or loss of sales and customer confidence. The
supply chains of today face much more challenges because of their increased complexity (Luo, Zhou, and Cauill 2001;
Dotoli et al. 2006).
In integrated supply chain networks, connectedness makes individuals, services and organisations accessible over
distance, sourcing from single supplier helps protect the intellectual property, lean operations lead the way to reduce
costs and inventory. But on the negative side, the leaner, global and more integrated supply chains are less resilient to
uncertainties and accidents in any link. Also, the rising costs of human and other resources and the environmental
concerns of transport of raw materials and other goods around the globe are counteracting the low-cost production
advantages. Efficiency encouraged and created giant firms through mergers and acquisitions and geographical concentra-
tion through cluster concepts (e.g. low-cost manufacturing in China, IT clusters in India, etc. Auto and Electronic
clusters in Japan). Damage due to an accident is higher for a concentration rather than for separate owners in several
locations. Protectionism, the insolvency of suppliers or their banks are other concerns.
Supply chains are complex networks of suppliers, contract manufacturers and third-party service providers with
interdependencies among these firms, hence inter-organisational coordination of risks a critical requirement. Many com-
panies are making considerable investments in monitoring the security, continuity, regulatory and performance risks of
their key suppliers (Asar et al. 2006). However there are no appropriate governing structures in place for monitoring
and control of the globally dispersed manufacturing and service networks during normal as well as abnormal times.
There is a high level of awareness of the potential risk arising from interaction and relationships between supply chain
partners. In recent years, a number of writers have sought to broaden the scope of disruption risk management process
from the level of the single company to the level of the entire supply chain (Gaonkar and Viswanadham 2007).
Managing supply risk, thus has become a critical component of managing the supply chain. Consequently, it is
important to an organisation’s success to understand the sources of supply risk and how to best manage them. The risk

*Corresponding author. Email addresses: nv@csa.iisc.ernet.in, n.viswanadham@gmail.com

Ó 2013 Taylor & Francis
 2 N. Viswanadham and A. Samvedi

sources are many and risk avoidance is not a viable strategy. Hence, one needs to carefully design the processes to be
risk resilient and take appropriate action when an undesirable beyond the control happens. For example, procurement or
selection of suppliers is an important supply chain process. Supplier selection is generally done based on the
performance criteria such as unit cost, quality, delivery times, etc. However, in global sourcing several factors including
political, economical, infrastructural issues; natural and manmade disasters; resource price fluctuations will cause
deviations, disruptions or disasters depending on the magnitude of the event. There is a need to identify all such factors
and also list them and create awareness among all concerned of the events that can happen and how they can be dealt
with. One of the aims of our paper is precisely this. We present the supply chain ecosystem and list all the possible
risks that affect the supply chain. We also develop an understanding of relationships between the countries of the
supplier and the manufacturer such as free trade agreements and also the transport infrastructure such as ports, roads
and also the resource productivity (labour, finance, power, etc.).
Traditionally, supplier selection was done mainly based on performance criteria. Risk is increasingly gaining impor-
tance because of the increased uncertainties in the ecosystem elements. Also, supplier selection process is an inherently
multi-objective problem, and many tangible and intangible performance factors (price, quality, delivery performance, ser-
Downloaded by [University of Missouri Columbia] at 12:50 22 September 2013

vice, etc.) need to be considered and evaluated in selecting suppliers. Wang and Yang (2009) considered supplier selec-
tion in a quantity discount environment using multi-objective linear programming, analytical hierarchy process (AHP)
and fuzzy compromise programming. Chan and Kumar (2007) identified and discussed some of the important and criti-
cal decision criteria including risk factors for the development of an efficient system for global supplier selection using
fuzzy AHP. Lu, Wu, and Kuo (2007) adds environmental principles into supplier selection process by applying fuzzy
AHP. Chan et al. (2008) proposed a fuzzy AHP approach for global supplier selection. Chena, Lin, and Huang (2006)
used fuzzy technique for order of preference by similarity to ideal solution (TOPSIS) for supplier selection. Kaya and
Kahraman (2011) proposed a modified fuzzy TOPSIS for selection of the best energy technology alternative.

1.1 Contribution
In this paper, we concentrate on the procurement process which is global and is managed as an inter-organisation
network. This paper is a significant contribution to the literature on this topic. We present a methodology for choice of
suppliers based on performance criteria and also to minimise the risks. Our methodology is based on the ecosystem
framework and applies fuzzy AHP and fuzzy TOPSIS in a unique way, by separating out the performance criteria from
the risk ones and then solving each part separately before consolidating the scores. The performance criteria such as lead
time, cost and quality are evaluated using all the ecosystem parameters. Generally, costs in supply chain include inven-
tory, transport and unit costs. In our case, they include trade, resource and infrastructure-related costs and coordination
costs as well. Similarly, quality in our case includes quality on delivery rather than at the factory thus including
spoilage, theft and damage during transport, loading, unloading, etc. The risk criteria classification used in this study
also differentiates it from other previous studies. Most of the supply chain risk studies, which have tried to do this, con-
sider only supply failures, partner risks, logistics failures, sharp fall in demand, etc. But, risks for the supply chain can
arise from all the four elements of the ecosystem rather than the supply chain alone. The risks come from governments,
political and social networks, resources and delivery systems such as logistics and IT (Viswanadham and Kameshwaran
2013). Therefore, risk mitigation or avoidance strategies should include all the ecosystem entities and plan the strategies
accordingly. The best way of risk avoidance strategy is to take care of risks when selecting the suppliers.
This paper is organised as follows: In Section 2, we present the ecosystem model. We show how the performance is
affected by the human, financial, infrastructural and natural resources, government actions and also the delivery logistics.
We further study the risk contributions of all the ecosystem elements. We then proceed in Section 3, to select the
suppliers to minimise the risk and enhance the performance. This section presents the proposed integrated methodology
which uses fuzzy AHP and fuzzy TOPSIS. In Section 4, we present a numerical illustration to show the applicability
and usability of the approach. Finally, Section 5, concludes the paper with future research directions.

2. Ecosystem model
A supply chain ecosystem consists of the elements of the supply chain and the entities that influence the goods, infor-
mation and financial flows through regulations, technology, management, etc. Accordingly, the supply chain ecosystem
comprises of networks of companies directly and indirectly part of the supply chain, countries of operations/presence
and their governments, industrial, social and political organisations, logistics and information technology services
infrastructure, the third-party service providers that connect the companies and the countries to the external economic
and social environment, resources including natural, financial and human resources with talent, connections and
 International Journal of Production Research 3

Figure 1. Supply chain ecosystem (Viswanadham and Kameshwaran 2013)

Downloaded by [University of Missouri Columbia] at 12:50 22 September 2013

knowledge of the industrial environment, industry clusters, universities, etc. interacting together with the horizontal and
vertical supply chain landscape and economic and social climate. The ecosystem is shown in Figure 1. The four distinct
risk sources in manufacturing and service chain networks include (Viswanadham and Kameshwaran 2013)

(1) Supply chain network

(2) Institutions: governmental and social
(3) Resources including human, natural, financial and industrial (Clusters)
(4) Delivery service mechanisms

We generally conduct the performance, risk and innovation studies using this framework. For this paper, the perfor-
mance and risk are relevant. Specifically, we deal with the supplier selection problem using Fuzzy AHP framework taking
into consideration the lead time, cost and quality as well as the risk emanating from all the ecosystem parameters.

2.1 Performance analysis using the ecosystem model

Performance analysis of supplier’s supply chains depends on all the ecosystem parameters. The desirable enablers in the
ecosystem of these suppliers supply chains have been listed in Table 1.
Design and technology improvements such as product modularisation, process coordination, supply chain visibility,
etc. resources such as clusters, banks, trained human resources, power, water, etc. government regulations, trade and
tariffs, customs, IP protection and inter-country agreements such as free trade agreements and finally the delivery service
practices such as good ports, good road connectivity, software providers and logistics companies that provide stream-
lined procurement, manufacturing and distribution processes have significant impact on customer satisfaction and in
increasing the performance of the supply chains. Availability of natural, human and financial resources, clusters and high
labour productivity will reduce cost and improve the lead times. A favourable institutional framework like good
judiciary, IP protection laws, trade laws, etc. will improve the trade, and will help instil the confidence in original equip-
ment manufacturers (OEM’s) to outsource more work and to help these suppliers with latest technology. Needless to say
that good delivery infrastructure such as ports, roads, 3Pls, IT, software vendors, soft infrastructure and trade facilitation
will result in predictable lead times and low transportation and inventory costs. Product modularisation, process

Table 1. Ecosystem enablers for supplier’s supply chain (adapted from Viswanadham and Kameshwaran 2013).

Supply chain Institutions Delivery infrastructure Resources

Enablers Modular products, JIT, FTAs, customs, IP protection, Port, Road & IT Finance, power, water etc.
TQM, SRM, SC visibility, Good judiciary, trade laws, social infrastructure, 3PLs, clusters, high labour
collaboration acceptance software vendors productivity
Cost High product design cost, Low tariffs, high profits Low transportation and Low factor costs
low production cost inventory costs
Lead Low Low Low Low
time
Quality High quality products High SC service levels High SC service levels High management quality
& market reach
 4 N. Viswanadham and A. Samvedi

standardisation, collaboration with partners and the supply chain visibility using sensor networks, call centres and Inter-
net, late customisation and use of supply hubs will certainly reduce the lead time and increase the efficiencies and prod-
uct flexibility but may also increase the cost of production. This performance analysis of supply chains is given in
Table 1.
The total landed cost has the following components: product cost, transport (shipping) cost, trade-related costs
(processing, customs clearance, port operations and the like), pipeline (in-transit) inventory and safety stock inventory
costs and finally the coordination cost. If a particular country has highly variable processing times for port operations,
supply chain managers need to hold additional safety stock to maintain desired customer service levels in the face of
increased supply uncertainty.

2.2 Risk analysis using the ecosystem model

Table 2 gives the list of risks that the supplier’s supply chain faces from the four ecosystem elements. We consider
below the three kinds of risks that an integrated supply chain faces apart from the supply chain functions and partners.
Downloaded by [University of Missouri Columbia] at 12:50 22 September 2013

2.2.1 Resources related risks

The resources that we consider are the natural, human, financial and industry resources. In the human resources arena,
skills shortages, employee attrition, communicable diseases and strikes affecting the number of working days,
opportunistic behaviour by the senior managers and other staff such as theft of intellectual property are some of the
risks generally faced by the companies. The input material shortages such as grains, fruits and vegetables, live stock,
quality problems due to diseases such as mad cow disease, chicken flu, price fluctuations in oil and food, currency fluc-
tuations all affect the supply chain effectiveness. Equipment failures, failure of power or water resources can lead to
unavailability of plants, warehouses and office buildings. Availability of quality producer services such as accounting,
management consulting, advertising, venture funding, etc is essential for strategy formulation.

2.2.2 Institutional risks

The economic and political related uncertainties affect businesses across all the industries and they include economic
factors such as economic slowdown, country ratings, foreign exchange, political issues such as war, country to country

Table 2. Risk classification by ecosystem approach (Viswanadham and Kameshwaran 2013).

Sr. No Risk classification Risk subclassification

R1 MR1: Supply chain related

• Location risk
R2
• Outsourcing risk
•
R3
R4 Design, manufacturing defects, Inventory deficit
R5 • Delay or unavailability of materials from suppliers
• Breakdown of machines, power failure

R6 MR2: Resources related

• Raw material, human, financial
R7
• Social unrest, war
•
R8
R9 Infrastructure deficit, talent shortage
• Credit squeeze, energy & water shortage

R10 MR3: Institutional risk

• Regulatory risk
R11
• Political
•
R12
R13 Labour issues
• Trade agreements

R14 MR4: Delivery infrastructure related

• Failure of IT infrastructure
R15
• SC visibility decreases
•
R16
R17 Inbound and outbound logistics failure
• Failure of governance mechanism
 International Journal of Production Research 5

relationships, changes in governments, uncertainties in trade agreements (anti-dumping and voluntary export restrictions)
deregulation, etc. Social unrest and regulatory risks are high in emerging markets. In developed countries, the financial
crisis has created a situation of oversight by the government.

2.2.3 Risks due to failure of delivery infrastructure

Delay or unavailability of either inbound or outbound transportation to move goods due to carrier breakdown or weather
problems will cause the supply demand matching problem. Failure of information and communication infrastructure due
to line, computer hardware or software failures or virus attacks, will lead to the inability to coordinate operations and
execute transactions While the physical supply chain handles the movement of documents data and physical goods, the
financial supply chain handles the movement of documents data and money. Thus, any credit squeeze by the financial
institutions will affect the supply chain. Piracy has increased over the years. Warships to protect ships carrying cars and
oil. It is still cheaper and convenient to pay higher insurance fees and take risk being attacked by Somali pirates than to
incur the extra cost of diverting vessels around the Cape of Good Hope.
Downloaded by [University of Missouri Columbia] at 12:50 22 September 2013

The following hierarchy for supplier selection is being proposed here. This hierarchy simultaneously considers both
performance and risk factors and the ecosystem model ensures the inclusivity of all important factors. The hierarchy is
described in Figure 2.

3. An integrated fuzzy approach

In this section, the proposed methodology has been detailed out. A concise description, of the fuzzy multi-criteria
methods which form an important part of the methodology, is also given.

3.1 Fuzzy AHP

The Fuzzy-AHP methodology extends Saaty’s AHP by combining it with fuzzy set theory. In the Fuzzy-AHP, fuzzy
ratio scales are used to indicate the relative strength of the factors in the corresponding criteria. Therefore, a fuzzy
judgement matrix can be constructed. The final scores of alternatives are also represented by fuzzy numbers. The
optimum alternative is obtained by ranking the fuzzy numbers using special algebraic operators. In this methodology, all
elements in the judgement matrix and weight vectors are represented by triangular fuzzy numbers. Using fuzzy numbers
to indicate the relative importance of one risk type over the other, a fuzzy judgement vector is then obtained for each
risk. These judgement vectors form part of the fuzzy pairwise comparison matrix which is then used to determine the
weight of each risk. Table 3 shows the meaning of linguistic expressions in the form of fuzzy numbers. Experts are
asked to give their assessment in the form of these linguistic expressions which are then converted and analysed to
finally get the weights.
Chang’s extent analysis method has been used for determining weights from pairwise comparisons. The extent
analysis method is used to consider the extent of an object to be satisfied for the goal, that is, satisfied extent. In the
method, the ‘‘extent’’ is quantified by using a fuzzy number. On the basis of the fuzzy values for the extent analysis of
each object, a fuzzy synthetic degree value can be obtained, which is defined as follows (Paksoy, Pehlivan, and
Kahraman 2012).

Lead Supply Chain

Supplier
Resources
Cost Performance Selection Risk
Criteria
Institutional
Quality
Delivery
Infrastructure

Supplier 1 Supplier 2 Supplier Supplier n

Figure 2. Combined performance risk based supplier selection hierarchy.

 6 N. Viswanadham and A. Samvedi

Table 3. Triangular fuzzy number equivalents to the corresponding linguistic expressions.

Linguistic expressions Equivalent fuzzy numbers Triangular fuzzy number (l, m, u)

Equal 1 (1, 1, 3)
Little importance 3 (1, 3, 5)
Strong importance 5 (3, 5, 7)
Very strong importance 7 (5, 7, 9)
Extreme importance 9 (7, 9, 11)

Let X ¼ fx1 ; x2 ; . . . ; xn g be an object set and U ¼ fu1 ; u2 ; . . . ; um g be a goal set. According to the method of
Chang’s extent analysis model, each object is taken and extent analysis for each goal gi is performed. Therefore, m
extent analysis values for each object can be obtained as Mg1i , Mg2i ; . . . ; Mgmi , I ¼ 1; 2; . . . ; n. All the Mgj i , j ¼ 1; 2; . . . ; m
are triangular fuzzy numbers. The algorithm of the Chang’s extent analysis model is as follows,
Downloaded by [University of Missouri Columbia] at 12:50 22 September 2013

Step 1 The value of fuzzy synthetic extent with respect to the ith object is defined as

" #1
X
m X
n X
m
Si ¼ Mgj i Mgj i
j¼1 i¼1 j¼1

Pm
To obtain j¼1 Mgj i perform the fuzzy addition operation of m extent analysis for a particular matrix such that

!
X
m X
m X
m X
m
Mgj i ¼ lj ; mj ; uj
j¼1 j¼1 j¼1 j¼1

hP P i1
n m
and to obtain i¼1
j
j¼1 Mgi , perform the fuzzy addition operation of Mgj i ; j ¼ 1; 2; . . . ; m values such that
!
n X
X m X
n X
n X
n
Mgj i ¼ li ; mi ; ui
i¼1 j¼1 i¼1 i¼1 i¼1

and then compute the inverse of the vector in such that

" #1  
X
n X
m
1 1 1
Mgj i ¼ Pn ; Pn ; Pn
i¼1 j¼1 i¼1 ui i¼1 mi i¼1 li

The principles for the comparison of fuzzy numbers were introduced to derive the weight vectors of all elements for
each level of hierarchy with the use of fuzzy synthetic values. To compare the fuzzy numbers, following principles are
used.

Step 2 The degree of possibility of M2  M1 is defined as

8
> 1; if m2  m1
   <
0; l 1  u2
V (M2  M1 ) ¼ sup min lM1 (x); lM2 (y) ¼ hgt(M1 \ M2 ) ¼ lM2 (d) ¼ (l1  u2 )
yx >
: ; otherwise
(m2  u2 )  (m1  l1 )
 International Journal of Production Research 7

~ 1 and M
Figure 3. The intersection between M ~ 2.

where M1 ¼ (l1 ; m1 ; u1 ) and M2 ¼ (l2 ; m2 ; u2 ) and d is the ordinate of the highest intersection point D between lM1 and
lM2 (see Figure 3). To compare M1 and M2 , both V (M2  M1 ) and V (M1  M2 ) are needed. The comparison is shown
Downloaded by [University of Missouri Columbia] at 12:50 22 September 2013

graphically in Figure 3.

Step 3 The degree of possibility for a fuzzy number to be greater than k fuzzy numbers Mi , (i ¼ 1; 2; : . . . ; k) can be
defined by
V (M  M1 ; M2 ; . . . ; Mk ) ¼ min V (M  Mi ); i ¼ 1; 2; . . . ; k

Assume that,
d 0 (Ai ) ¼ min V (Si  Sk ); k ¼ 1; 2; . . . ; n; k – i

Then the weight vector is given by

W 0 ¼ ðd 0 (A1 ); d 0 (A2 ); . . . ; d 0 (An )Þ
T

where Ai , (i ¼ 1; 2; . . . ; n) are n elements.

Step 4 via normalisation, the normalised weight vectors are

W ¼ ðd(A1 ); d(A2 ); . . . ; d(An )ÞT

where W is not a fuzzy number.

3.2 Fuzzy TOPSIS

TOPSIS is a multiple criteria method to identify solutions from a finite set of alternatives. The underlying logic of
TOPSIS is to define the ideal solution and the negative ideal solution. The alternatives are then compared with these
ideal and negative ideal solutions, to find out the distances. These distances are then used to come up with a score. The
one which is closest to the ideal and farthest from the negative ideal generally qualifies for the optimum. Chen (2000)
extends the TOPSIS method to fuzzy group decision-making situations by considering triangular fuzzy numbers and
defining crisp Euclidean distance between two fuzzy numbers. In Chen’s fuzzy TOPSIS, linguistic preferences can easily
be converted to fuzzy numbers which are allowed to be used in calculations. The details of the method as given by
Kutlu and Ekmekçioğlu (2012) is given below.
It is suggested that the decision-makers use linguistic variables to evaluate the ratings of alternatives with respect to
criteria. Table 3 gives the linguistic scale for evaluation of the alternatives. Assuming that a decision group has K
people, the ratings of alternatives with respect to each criterion can be calculated as

1 h~1 i
X~ ij ¼ X ij ( þ )X~ ij ( þ ) . . . ( þ )X~ ij ;
2 K

K
 8 N. Viswanadham and A. Samvedi

where X~ ij is the rating of the kth decision-maker for ith alternative with respect to jth criterion (Chen 2000).
K

Obtaining weights of the criteria and fuzzy ratings of alternatives with respect to each criterion, the fuzzy
multi-criteria decision-making problem can be expressed in matrix format as
2 3
X~ ij X~ ij ... X~ ij
6 . .. .. 7
D ¼ 4 .. . ... . 5;
X~ ij X~ ij  X~ ij

W ¼ ½w1 ; w2 ; . . . ; wn ; j ¼ 1; 2; . . . ; n;

where X~ ij is the rating of the alternative Ai with respect to criterion j (i.e. Cj ) and wj denotes the importance weight of
Cj . These linguistic variables can be described by triangular fuzzy numbers: X~ ij ¼ (aij ; bij ; cij ). To avoid the
Downloaded by [University of Missouri Columbia] at 12:50 22 September 2013

complicated normalisation formula used in classical TOPSIS, the linear scale transformation is used here to transform
the various criteria scales into a comparable scale. Therefore, we can obtain the normalised fuzzy decision matrix
denoted by R ~

~ ¼ ½~rij 
R mxn

where B and C are the set of benefit criteria and cost criteria, respectively, and
!
~aij ~bij ~cij
~r ¼ ; ; ; j 2 B;
cj cj cj

   
aj bj c j
~r ¼ ; ; ; j 2 C;
cij bij aij

cj ¼ max cij if j 2 B;

i

a
j ¼ min aij if j 2 C:
i

The normalisation method mentioned above is to preserve the property that the ranges of normalised triangular fuzzy
numbers belong to [0, 1].
Considering the different importance of each criterion, we can construct the weighted normalised fuzzy decision
matrix as

V~ ¼ ½~vij mxn i ¼ 1; 2; . . . ; m; j ¼ 1; 2; . . . ; n

where

~vij ¼ ~rij (:)d(Cj ):

According to the weighted normalised fuzzy decision matrix, we know that the elements ~vij 8i; j are normalised
positive triangular fuzzy numbers and their ranges belong to the closed interval [0, 1]. Then, we can define the fuzzy
positive-ideal solution (FPIS, A ) and fuzzy negative-ideal solution (FPIS, A ) as
 International Journal of Production Research 9

A ¼ (~v1 ; ~v2 ; . . . ; ~vn );

A ¼ (~v v
1 ;~ v
2 ;...;~n );

where

~vj ¼ (1; 1; 1) and ~v

j ¼ (0; 0; 0); j ¼ 1; 2; . . .

The distance of each alternative from A and A can be currently calculated as

X
n
di ¼ d(~vij ; ~vj ); i ¼ 1; 2; . . . ; m
Downloaded by [University of Missouri Columbia] at 12:50 22 September 2013

j¼1

X
n
di ¼ d(~vij ; ~v
j ); i ¼ 1; 2; . . . ; m
j¼1

where d(:; :) is the distance measurement between two fuzzy numbers calculating with the following formula:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 
q; s~) ¼
d(~ (q1  s1 )2 þ (q2  s2 )2 þ (q3  s3 )2
3

where q~ ¼ (q1 ; q2 ; q3 ) and s~ ¼ (s1 ; s2 ; s3 ) are two triangular fuzzy numbers. A closeness coefficient is defined to
 
determine the ranking order of all alternatives once the d~j and d~j of each alternative Ai (i ¼ 1; 2; . . . ; m) are calculated.
The closeness coefficient of each alternative is calculated as

d~j
CCi ¼   ; i ¼ 1; 2; . . . ; m
d~j þ d~j

Obviously, an alternative Ai is closer to the (FPIS, A ) and farther from (FPIS, A ) as CCi approaches to one.
Therefore, according to the closeness coefficient, we can determine the ranking order of all alternatives and select the
best one from among a set of feasible alternatives.

3.3 Proposed methodology

This section proposes an integrated methodology using fuzzy AHP and fuzzy TOPSIS for supplier risk assessment while
doing supplier selection. The methodology consists of steps as given in Figure 4. The reason behind choosing these
techniques is the popularity and wide acceptance they enjoy, when compared to other multi-criteria decision
methodology (MCDM) methods. A lot of other methods such as ANP, ELECTRE, PROMETHEE, etc. can also be used.
The comparative study though is not possible due to the nature of the output of these methods. This is because it is not
feasible to compare two ranks, unless we have the best rank available, which is normally not the case.
As can be seen from the Figure 4, there are two paths at the start wherein the supplier risk assessment is separated
from the supplier performance evaluation. The first path uses the standard fuzzy AHP procedure, whereas the second
path draws upon the methodology proposed by Samvedi and Jain (2012). The first step in both the paths though
requires the firm to come up with a comprehensive hierarchy of all the criteria on which the performance of suppliers is
tested or the risks are evaluated. This is done by thoroughly studying the considered chain and identifying potential
loopholes. These are then analysed for overlaps and categorised using similar characteristics. This exercise should be
repeated whenever a major change is made in the chain. The second step in the process involves assigning weights to
 10 N. Viswanadham and A. Samvedi

Supplier criteria classification Risk Classification

Form fuzzy pairwise Form fuzzy pairwise

comparison matrices comparison matrices for
higher levels and get
weights

For the lowest level in

hierarchy form the risk
table
Expert Inputs Required
Downloaded by [University of Missouri Columbia] at 12:50 22 September 2013

Apply fuzzy AHP to get Apply fuzzy TOPSIS to

supplier ranks get risk scores

Consolidate the supplier ranks and risk scores in one table

Select the right supplier

Figure 4. Proposed methodology for supplier selection.

the criteria according to their importance. Fuzzy AHP is used for this purpose and expert views are taken as input. For
the path one that is the performance evaluation, this step also provides with the performance scores for the alternatives.
But for the path two, there are two extra steps involved. The first of them requires expert inputs for the risk assessment
done for four criteria namely their probability of occurrence, their impact on the performance of supply chain, the effort
and time required in recovering from the impact and at what level does the risk affect. This is because, as can be seen
from the literature, the risk affecting the strategic level is much more dangerous than one affecting the operational level.
The last step in path two does the aggregation of these inputs using fuzzy TOPSIS.
The results from the two paths are then aggregated to come up with a decision table which contains supplier
alternative ranks and also the individual risk scores for these alternatives under different risk types. Also, the aggregated
risk score for every alternative is displayed in this table. This helps the managers to make an informed decisionon which
supplier to choose. The breakup score for each risk type is provided because sometimes the managers want to pay
particular attention to a type of risk. This can be because of several reasons such as that the said risk type is already
present in the supply chain in large and managers do not want it to be increased any further. Also, the computational
complexity of proposed methodology is quite low. This is because solutions like AHP are based on subjective
judgement of experts and the value of solution increases with sensitivity analysis of some of these opinions. The
complexity thus may not be computational, but the procedure may involve multiple iterations and thus complex. Thus,
after gathering the expert opinion analysing it is not computationally complex and hence the resources needed for it is
not significant.

4. Example
This section gives an illustrative example, to explain the workings of the methodology proposed, and also real time
scenarios where such a method can be useful. The Figure 2 depicts the supplier selection hierarchy, which has been
proposed in this study. As can be seen from the Figure 4, there are two major paths. One path evaluates the suppliers
on their performance criteria and the other evaluates them on their risk assessment. Most of the studies, which also
consider risk, do so by adding risk as a performance criterion. But with added emphasis given these days on risk
management, due to high vulnerability of businesses these days, it is better to treat risks separately. This helps in risks
getting the importance which they deserve.
When doing the performance evaluation any multi-criteria method can be used. Whereas, for selection through risk
assessment, this study uses the approach as explained in the previous section and illustrated through an example here.
 International Journal of Production Research 11

The approach involves two major steps, namely assigning weights to all the criteria and determining the scores of all
the risks at the lowest level in the hierarchy. These two values are then consolidated into one single-risk index value.
Here, we detail out the functioning of methodology proposed to handle risk assessment part of the process. The
performance evaluation part is dealt by using Fuzzy AHP, similar to the way first half of described method is solved.
The calculation for this part has not been provided here because of the shortage of space. This is also why only those
calculations which are necessary for the understanding of the method have been provided here.
The inputs come in the form of linguistic values. The expert inputs for the fuzzy AHP part are linguistic variables
as given in the Table 3. Normally, whenever such subjectivity is involved in judgements it is advised to have more than
one source of inputs. These inputs can be later aggregated for a better analysis of the system. In this study, inputs from
three experts are considered. In total, there will be five fuzzy pairwise comparison tables per expert. These are one for
criteria comparison and one each for comparison of sub-criteria under a given criterion. The calculations for sub-criteria
comparison under the criteria delivery infrastructure failure is shown in Table 4. The calculation is provided for the
pairwise comparison matrix of one expert. The remaining pairwise comparisons are solved in the similar way.
As seen from Table 4, the two risks are compared only once and the reverse comparison is supposed to take the
Downloaded by [University of Missouri Columbia] at 12:50 22 September 2013

reverse value automatically. When these linguistic inputs are converted to the fuzzy triangular numbers, we get Table 5.
The synthetic values are then calculated as shown in step 1 of Section 5.2. These synthetic values are then used to
reach the final weights. The calculations are done using the step 2, 3 and 4 of the same section. The results are shown
in Table 6. Similarly, the weights for all the criteria and sub-criteria are determined. The weights from different experts
are then averaged to get the mean weights. Now, the process moves on the second part namely risk assessment inputs.
Each risk is measured against four parameters, namely low importance, low probability of occurrence, low impact of
the risk on the supply chain if it occurred and less difficulty to mitigate that risk. The criteria are chosen in such a way
so that higher value is desired. This helps us in directly adding up the scores to the performance ones. Also, this
approach goes with the popular one wherein higher values for better alternatives are desired.
The inputs for the values of these parameters are taken from experts again in the form of linguistic expressions
which have earlier been defined as fuzzy intervals, as shown in the Figure 5. The linguistic expressions are randomly

Table 4. Pair wise comparison matrix for sub criteria under critical delivery infrastructure.

R14 R15 R16 R17

R14 1 Little importance – Strong importance

R15 – 1 – –
R16 Strong Importance Very strong imp 1 –
R17 – Little importance Very strong imp 1

Table 5. Pair wise comparison matrix with fuzzy triangular numbers.

R14 R15 R16 R17

R14 1 (1, 3, 5) – (3, 5, 7)

R15 – 1 – –
R16 (3, 5, 7) (5, 7, 9) 1 –
R17 – (1, 3, 5) (5, 7, 9) 1

Table 6. Synthetic values and corresponding weights.

Criteira Synthetic values Weights

R1 (0.0741, 0.4286, 1.1538) 0.3275

R2 (0.0296, 0.1429, 0.3846) 0.1706
R3 (0.1296, 0.3214, 1.3462) 0.3201
R4 (0.0556, 0.1071, 0.5769) 0.1998
 12 N. Viswanadham and A. Samvedi

generated and the values from three experts are averaged as done in the previous step. The resulting values are shown
in Table 7. Each risk input parameter is divided into five linguistic expressions with membership values as shown in
Figure 5.
The Table 7 shows the risk input matrix with the expert inputs entered. These inputs are then converted to risk
scores using fuzzy TOPSIS method as given in Section 5.3.
These scores are then consolidated using the weights assigned to all the risks. These scores are multiplied by the
weights assigned to the relative risks. The values obtained are then added up for the first-level risks. For example, the
values for first-five risks are added to give a score for the planning and product related risks. The scores obtained for
the first-level risks are then again multiplied by the weights assigned to these first-level risks and the resulting values
summed up to get the final-risk index value. The two scores are then consolidated into one. These values are shown in
Table 8. Thus, it can be seen that supplier 3 is the best in consolidated score and overall risk category. But, it ranks sec-
ond in performance. Also when individual risk categories are broken down, we see that supplier three is best for MR1
and MR3 category, whereas it ranks third for MR2 and last for MR4. Such a detailed examination is most of the times
very useful. Importance of detailing out the values in such a way is that the managers have the data in front of them
and are in a position to make a better informed decision. Sometimes giving only the final value can be a little mislead-
Downloaded by [University of Missouri Columbia] at 12:50 22 September 2013

ing. This can be explained by considering supplier three. As we can see that the total value of risk assessment is highest
for this supplier. That means this supplier is least risky overall. But suppose that the existing supply chain has a lot of
risk from MR4 category and the managers do not want that risk to increase anymore, then giving only the total value
can be misleading. Supplier three is actually the most risky in MR4 category.

µm
Low Mild High V High Extreme

0.2 0.4 0.6 0.8 1.0

Figure 5. Membership functions of the linguistic expressions.

Table 7. Averaged risk expert input matrix.

Risks Type of risk Probability Impact Mitigation Score

R1 (0.13, 0.33, 0.53) (0.47, 0.67, 0.80) (0.13, 0.33, 0.53) (0.13, 0.33, 0.53) 0.2955
R2 (0.40, 0.60, 0.73) (0.40, 0.60, 0.80) (0.20, 0.40, 0.60) (0.13, 0.33, 0.53) 0.3864
R3 (0.33, 0.53, 0.73) (0.20, 0.40, 0.60) (0.40, 0.60, 0.80) (0.40, 0.60, 0.73) 0.4545
R4 (0.33, 0.53, 0.73) (0.33, 0.53, 0.73) (0.20, 0.40, 0.60) (0.20, 0.40, 0.60) 0.3636
R5 (0.67, 0.87, 1.00) (0.00, 0.20, 0.40) (0.73, 0.93, 1.00) (0.27, 0.47, 0.67) 0.5682
R6 (0.33, 0.53, 0.73) (0.27, 0.47, 0.67) (0.33, 0.53, 0.73) (0.47, 0.67, 0.87) 0.4773
R7 (0.20, 0.40, 0.60) (0.47, 0.67, 0.87) (0.27, 0.47, 0.67) (0.27, 0.47, 0.60) 0.4091
R8 (0.13, 0.33, 0.53) (0.27, 0.47, 0.67) (0.00, 0.20, 0.40) (0.13, 0.33, 0.53) 0.1818
R9 (0.20, 0.40, 0.60) (0.40, 0.60, 0.80) (0.33, 0.53, 0.73) (0.60, 0.80, 1.00) 0.5227
R10 (0.27, 0.47, 0.67) (0.47, 0.67, 0.80) (0.07, 0.27, 0.47) (0.27, 0.47, 0.67) 0.3636
R11 (0.53, 0.73, 0.87) (0.33, 0.53, 0.73) (0.47, 0.67, 0.80) (0.40, 0.60, 0.80) 0.5909
R12 (0.33, 0.53, 0.73) (0.27, 0.47, 0.67) (0.53, 0.73, 0.93) (0.47, 0.67, 0.80) 0.5455
R13 (0.40, 0.60, 0.80) (0.07, 0.27, 0.47) (0.40, 0.60, 0.80) (0.53, 0.73, 0.93) 0.4773
R14 (0.47, 0.67, 0.80) (0.33, 0.53, 0.73) (0.27, 0.47, 0.67) (0.67, 0.87, 1.00) 0.5909
R15 (0.20, 0.40, 0.60) (0.27, 0.47, 0.60) (0.13, 0.33, 0.53) (0.33, 0.53, 0.73) 0.3182
R16 (0.20, 0.40, 0.60) (0.53, 0.73, 0.80) (0.40, 0.60, 0.80) (0.53, 0.73, 0.93) 0.5682
R17 (0.67, 0.87, 1.00) (0.00, 0.20, 0.40) (0.73, 0.93, 1.00) (0.40, 0.60, 0.80) 0.6136
 International Journal of Production Research 13

Table 8. Consolidated table with all the scores.

Risk scores
Suppliers Performance Scores MR1 MR2 MR3 MR4 Total Consolidated Scores

S1 0.3279 0.3025 0.5337 0.4926 0.4556 0.4328 0.7607

S2 0.1708 0.5571 0.5052 0.3738 0.5119 0.4681 0.6389
S3 0.2976 0.6600 0.4874 0.5341 0.4404 0.5791 0.8767
S4 0.2037 0.5253 0.4453 0.4951 0.4545 0.4892 0.6929

4.1 Extension to multi-tier supplier selection

When a supplier is selected in a supply chain then it is not just that supplier but also its entire sub-chain comes into the
system. Most of the times this sub-chain selection is ignored and the focus is only on the front supplier. This is risky as
Downloaded by [University of Missouri Columbia] at 12:50 22 September 2013

the entire ecosystem of the sub-chain is now the part of the overall supply chain and the risks can also emanate from
here. For example, Mattel recalled millions of toys in 2007 because high quantity of lead was found in the paint which
was used. The problem occurred from one of the sub-suppliers of a Chinese supplier to which the work was outsourced.
This shows the importance of keeping watch on the sub-chains of the selected suppliers and if possible better selects
the entire sub-chain.
The method given above can be easily extended to multi-tier supplier selection. The entire process is rerun for the
possible supplier alternatives at every tier in the chain. The numerical example here has three tiers overall with four
supplier alternatives in the front tier, five in the next upstream tier and three for the last tier. The calculations were
demonstrated for the front tier suppliers and these are now extended to the other two tiers. The details of calculations
are similar to the ones above but the hierarchy of criteria can be changed if needed. It is sometimes possible that the
importance of criteria is different for different tiers and also in some cases the list of criteria can change even. Tables 9
and 10 tabulate the values obtained for these tiers. Table 9 shows the values for the second tier in the upstream direction
and Table 10 shows the last tier in the upstream direction.
In total, then, there can be 4  5  3 = 60 possible chains involving these alternatives. But almost always there are
other constraints like compatibility issues between different firms, logistical connectivity issues, cultural differences,
regional problems, etc. Due to these, the number of possible alternative chains is always much lower than the total
possible chains. In this case, this number comes out to be nine feasible chains and they are:

Table 9. Consolidated table with all the scores for second tier upstream.

Risk scores
Suppliers Performance Scores MR1 MR2 MR3 MR4 Total Consolidated scores

SS1 0.1932 0.5878 0.6297 0.5002 0.4694 0.5698 0.7630

SS2 0.1477 0.5455 0.6159 0.5411 0.4861 0.5513 0.6990
SS3 0.2713 0.3025 0.5337 0.4926 0.4556 0.4328 0.7041
SS4 0.2264 0.5571 0.5052 0.3738 0.5119 0.4681 0.6945
SS5 0.1614 0.6600 0.4874 0.5341 0.4404 0.5791 0.7405

Table 10. Consolidated table with all the scores for the last tier upstream.

Risk scores
Suppliers Performance scores MR1 MR2 MR3 MR4 Total Consolidated scores

SSS1 0.4182 0.5253 0.4453 0.4951 0.4545 0.4892 0.9074

SSS2 0.3567 0.4741 0.6667 0.5700 0.6125 0.5896 0.9463
SSS3 0.2251 0.3515 0.4537 0.5607 0.4541 0.4428 0.6679
 14 N. Viswanadham and A. Samvedi

C1. S1 – SS2 – SSS1

C2. S1 – SS5 – SSS1
C3. S2 – SS1 – SSS3
C4. S2 – SS4 – SSS2
C5. S3 – SS3 – SSS2
C6. S4 – SS1 – SSS1
C7. S4 – SS3 – SSS3
C8. S4 – SS4 – SSS3
C9. S4 – SS5 – SSS2
The combined overall scores for these chains are given in Table 11 below.

Table 11. Consolidated table with total scores for sub-chains.

Downloaded by [University of Missouri Columbia] at 12:50 22 September 2013

Risk scores
Chains Performance scores MR1 MR2 MR3 MR4 Total Consolidated scores

C1 0.8938 1.3733 1.5949 1.5288 1.3962 1.4733 2.3671

C2 0.9075 1.4878 1.4664 1.5218 1.3505 1.5011 2.4086
C3 0.5891 1.4964 1.5886 1.4347 1.4354 1.4807 2.0698
C4 0.7539 1.5883 1.6771 1.3176 1.6363 1.5258 2.2797
C5 0.9256 1.4366 1.6878 1.5967 1.5085 1.6015 2.5271
C6 0.8483 1.6077 1.3958 1.3640 1.4209 1.4465 2.2948
C7 0.7001 1.1793 1.4327 1.5484 1.3642 1.3648 2.0649
C8 0.6552 1.4339 1.4042 1.4296 1.4205 1.4001 2.0553
C9 0.7218 1.6594 1.5994 1.5992 1.5074 1.6579 2.3797

These scores are just the additions of the values from the previous three tables. It can be seen from this table that
C5 is the best chain followed by C2 and C9. Also, the table shows that although C5 is the best overall, C9 scores the
highest in risk and thus is a better chain with respect to handling risks. C5 also scores the best in performance category.
The scores for all the chains are provided to the managers, who can then take an informed decision by taking all the
tradeoffs into consideration and also the current scenario.

5. Conclusion and future research work

Global supply chain networks are frequently subjected to severe disruptions. Keeping the performance at acceptable
levels during disruptions is one of the top most concerns of the managers in a supply chain today. Some do so in an ad
hoc or reactive fashion, responding to risks as they appear, while others are proactive, planning in advance. Thus,
supplier selection giving due importance to risk factors has become the need of the hour. This makes the study unique
where performance and risk criteria are considered together and with equal importance. Also, the use of ecosystem
approach in defining the criteria makes this paper more useful in practice.
The ecosystem approach used for the classification of risks seems to be the right way to move forward as it not only
considers the risks emanating from a supply chain but also from the entire ecosystem which affects a chain. All the four
elements of the ecosystem may not be important for all verticals at all times. The weights assigning process uses expert
knowledge to these criteria based on the perceived affect of these risks to the operations of the supply chain. The
resulting consolidated table provides an opportunity to the supply chain managers to select a better supplier. The risk
score table should be the starting point for the risk management team as it details the weights given to the different risk
types and also the degree of vulnerability for a particular risk.
This method is generic in nature and can easily be applied to any practical situation. Also, the method can easily be
extended to other similar selection situations. Multi-tier supplier selection is being pursued in recent times and our
method can be extended to this case. In case of longer supply chains with a number of options for suppliers at each
stage, the alternatives multiply requiring metaheuristics. Also, any study comparing the proposed methodology with
other existing methodologies would a very promising study on its own.
 International Journal of Production Research 15

Acknowledgement
The first author would like to thank INAE for all its support. Second author thanks Prof. Y. Narahari (Chairman CSA Dept., IISc)
and DST for their support and funds, which helped in successful completion of this study.

References

Asar, A. U., M. Zhou, R. J. Caudill, and S. U. Asar. 2006. “Modeling Risks in Supply Chains Using Petri Net Approach.”
International Journal of Services Operations and Informatics 1 (3): 273–285.
Chan, F. T. S., and N. Kumar. 2007. “Global Supplier Development Considering Risk Factors Using Fuzzy Extended AHP-based
Approach.” Omega 35 (4): 417–431.
Chan, F. T. S., N. Kumar, M. K. Tiwari, H. C. W. Lau, and K. L. Choy. 2008. “Global Supplier Selection: A Fuzzy-AHP Approach.”
International Journal of Production Research 46 (14): 3825–3857.
Chen, C. 2000. “Extensions of the TOPSIS for Group Decision-making under Fuzzy Environment.” Fuzzy Sets and Systems 114 (1):
1–9.
Chena, C. T., C. T. Lin, and S. F. Huang. 2006. “A Fuzzy Approach for Supplier Evaluation and Selection in Supply Chain
Downloaded by [University of Missouri Columbia] at 12:50 22 September 2013

Management.” International Journal of Production Economics 102 (2): 289–301.

Dotoli, M., M. P. Fanti, C. Meloni, and M. Zhou. 2006. “Design and Optimization of Integrated E-supply Chain for Agile and
Environmentally Conscious Manufacturing.” IEEE Transactions on Systems, Man, and Cybernetics: Part A 36 (1): 62–75.
Gaonkar, R., and N. Viswanadham. 2007. “An Analytical Framework for the Management of Risk in Supply Chains.” IEEE
Transactions on Automation Sciences and Engineering 4 (2): 265–273.
Kaya, T., and C. Kahraman. 2011. “Multicriteria Decision Making in Energy Planning Using a Modified Fuzzy TOPSIS
Methodology.” Expert Systems with Applications 38: 6577–6585.
Kutlu, A. C., and M. Ekmekçioğlu. 2012. “Fuzzy Failure Modes and Effects Analysis by Using Fuzzy TOPSIS Based Fuzzy AHP.”
Expert Systems with Applications 39 (1): 61–67.
Lu, L. Y. Y., C. H. Wu, and T. C. Kuo. 2007. “Environmental Principles Applicable to Green Supplier Evaluation by Using
Multi-Objective Decision Analysis.” International Journal of Production Research 45 (18/19): 4317–4331.
Luo, Y., M. Zhou, and R. J. Caudill. 2001. “An Integrated E-supply Chain Model for Agile and Environmentally Conscious
Manufacturing.” IEEE/ASME Transactions on Mechatronics 6 (4): 377–386.
Paksoy, T., N. Y. Pehlivan, and C. Kahraman. 2012. “Organizational Strategy Development in Distribution Channel Management
Using Fuzzy AHP and Hierarchical Fuzzy TOPSIS.” Expert Systems with Application 39 (12): 2822–2841.
Samvedi, A., and V. Jain. 2012. “Quantifying Risks in a Supply Chain through Integration of Fuzzy AHP and Fuzzy TOPSIS.”
International Journal of Production Research 51 (8): 2433–2442.
Viswanadham, N., and S. Kameshwaran. 2013. Ecosystem Aware Global Supply Chain Management. Singapore: World Scientific
Publishing.
Wang, T. Y., and Y. H. Yang. 2009. “A Fuzzy Model for Supplier Selection in Quantity Discount Environments.” Expert Systems with
Applications 36 (10): 12179–12187.