Mathematical and Computer Modelling 58 (2013) 1679–1695

Contents lists available at ScienceDirect

Mathematical and Computer Modelling

journal homepage: www.elsevier.com/locate/mcm

A fuzzy logic based decision support system for evaluation of

suppliers in supply chain management practices
Darshan Kumar a,∗ , Jagdev Singh a,1 , Om Pal Singh a,2 , Seema b,3
a
Department of Mechanical Engineering, Beant College of Engineering and Technology, Gurdaspur-143521, Punjab, India
b
Department of Chemical Engineering, Beant College of Engineering and Technology, Gurdaspur-143521, Punjab, India

article info abstract

Article history: Supply chain management is an increasingly important organizational concern, and proper
Received 14 September 2010 evaluation of suppliers constitutes one essential element of supply chain success. Continu-
Accepted 27 July 2013 ous evaluation of a particular supplier becomes more important considering the fact that in
most industries the cost of raw materials and component parts constitutes the main cost
Keywords: of a product, such that in some cases it can account for up to 70%. However, there is lit-
Supplier evaluation
tle research that has helped the organizations in continuous evaluation of their suppliers.
Quality
Cost
We propose a new model, based on fuzzy logic to handle the various attributes, associated
Time with supplier evaluation problems. Four multi-input single output (MISO) mamdani fuzzy
Fuzzy inference systems have been proposed for supplier evaluation. The proposed model has
also been illustrated through a case study.
© 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Since the introduction of the term supply chain management (SCM) by consultants in the early 1980s, it has gained
the attention of researchers and academicians. After the 1990s, academics have attempted to find more and more aspects
about SCM [1,2]. With these attempts of academicians and researchers, the concept arose from a number of changes in the
different stages of the supply chain. These changes effected the rising costs of manufacturing, the shrinking resources of
manufacturing bases apart from shortening product life cycles. As a result, the concept has undergone tremendous changes
over a few years. It has entirely replaced the traditional terms used to describe the management of material and service
flows [3].
The intensive global competition among manufacturers to co-ordinate with and respond quickly the industry value chain
from suppliers to customers has made the customer–supplier relationship in SCM important in the new business era. In such
circumstances, decision making in each business plays a key role in the cost reduction, and supplier evaluation is one of the
important functions in supplier relationship management because doing business with appropriate suppliers is beneficial
for the organization to provide a sufficient production volume with good quality [4]. This function becomes more important
considering the fact that in most industries the cost of raw materials and component parts constitutes the main cost of a
product, such that in some cases it can account for up to 70%. In such circumstances the purchasing department can play a

∗
Corresponding author. Tel.: +91 9463211090; fax: +91 1874221463.
E-mail addresses: darshanjindal@yahoo.com (D. Kumar), bks19701967@yahoo.co.in (J. Singh), ompal65@yahoo.co.in (O.P. Singh),
seemadarshan2007@gmail.com (Seema).
1 Tel.: +91 9855521560; fax: +91 1874221463.
2 Tel.: +91 9855472059; fax: +91 1874221463.
3 Tel.: +91 8872109032; fax: +91 1874221463.

0895-7177/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.mcm.2013.07.003
1680 D. Kumar et al. / Mathematical and Computer Modelling 58 (2013) 1679–1695

key role in cost reduction, and supplier selection and continuous evaluation of the supplier becomes some of the important
functions of purchasing management [5].
The increasing proportion of raw materials and work-in-process of manufactured products, where sourced globally
by multinational manufacturers, is a trend in today’s industries. The way to evaluate supplier capability is the main
scope of supplier selection. A multi-national manufacturer cannot have direct control over the capability and performance
of its hundreds or even thousands of suppliers. However, the evaluation of its suppliers’ capabilities to provide raw
materials/component parts is a crucial issue to a multi-national manufacturer. For example, it is common for suppliers,
after receiving an order to sub-contract to satisfy the demand, due to the tightness of their own production schedule.
Furthermore, internal re-scheduling of production by the supplier may have an impact on their performance level. Their
quality assurance and on-time delivery would be in doubt. Therefore, a manufacturer should analyze and evaluate the
potential threats when continuing business with suitable supplier(s) resulting from a systematic evaluation process and
its corresponding attributes [6]. After selection of an appropriate supplier, therefore, continuous evaluation of supplier(s) is
very essential. In [7], Kumar et al. study the single decision supplier selection process in the Indian textile industry, whereas
the current paper studies the continuous process of supplier evaluation.

2. Review of related work

The literature available on supplier selection and evaluation can broadly be considered having two components:
(1) supplier attributes (2) tool used in study.
Various attributes for supplier evaluation have been studied by various researchers. In general, supplier attributes from
purchasing professionals, such as a purchasing manager or buyer, should be first identified for supplier evaluation. These
attributes are fundamental when making purchasing decisions. While identifying these attributes, it should be clear that
all criteria, rules, and priorities are identified and systematically classified. This helps in evaluating the effectiveness of the
purchasing performance. Chao et al. [8] concluded that quality and on-time delivery are the most important attributes of
purchasing performance evaluation from his survey results in a number of industries. Similar work of supplier evaluation
was also considered by Talluri and Sarkis [9] taking price, quality, delivery and flexibility as variables apart from just-
in-time supply and by Talluri and Baker [10] taking cost, product variety, quality and lead time as main attributes. Wei
et al. [11] suggested that the purchasing factors usually considered should include a supplier’s history of supply, product
price, technical ability, and transportation cost. Ghodsypour and O’Brien [5] agreed that cost, quality, and service are the
three main categories when deciding supplier selection parameters. Yang et al. [12] took technological innovation as an
attribute for supplier evaluation. Arntzen et al. [13] considered strategic decisions, such as location of customers and
suppliers, location and availability of inexpensive skilled labor, cost of various transportation modes, export regulations
etc. for supplier selection. This revealed that the supplier evaluation process, usually made on the basis of cost and quality
has been recognized as a major decision making process. However, Briggs [14] stated that joint development, culture, supply
chain management, quality and communication are the key attributes to be considered, apart from optimum cost. Braglia
and Petroni [15] evaluated the relative performance of suppliers that have multiple outputs and inputs, based on capabilities
relating to management, production facilities, technology, price, quality, and delivery compliance for a bottling industry.
Ng et al. [16] considered supplier relationships, purchasing strategy, transportation cycle time, packaging for supplier
evaluation model development. The type of relationship between the supplier and buyer has also been studied by Toni
and Nassimbeni [17] by examining the role of supplier development in establishing and managing efficient buyer–supplier
operational links from this study of the 50 plants. Fynes et al. [18] also studied the different characteristics of the relationship
between the supplier and buyer and their effect on the supply chain.
The second part considered during the literature review for supplier evaluation is the tool used for evaluation. A
combination of analytic hierarchy process (AHP) and linear programming has been applied by Ghodsypour and O’Brien [5]
to study the problems related to supplier evaluation, including both qualitative and quantitative factors. But, in this method,
errors can creep in due to perception or biased behavior of the decision making managers and independent nature of
attributes used. This problem can be solved by getting the rating done by a group of decision makers for continuous
evaluation of vendors, following the principle of anonymity and integrating the method with a managerial tool—Delphi
method [19]. Handfield et al. [20] illustrated the use of AHP as a decision support model that included relevant environmental
criteria. Bhutta and Huq [21] applied AHP in the supplier evaluation process and compared it with the total cost ownership
method. Wang et al. [22] developed an integrated AHP and pre-emptive goal programming based multi-criteria decision-
making methodology to take into account both qualitative and quantitative factors in supplier evaluation. Till the early
years of 2000, no feedback from the customers was incorporated in the process and the attributes considered were largely
independent. These problems can be overcome by using more general form of AHP, called the Analytic Network Process
(ANP). Agarwal and Shankar [23] used ANP, which incorporates feedback and interdependent relationships among decision
attributes and alternatives. Lee et al. [24] also used ANP for selecting the appropriate acquisition mode for a required
technology during supplier selection and the evaluation process.
A new and general decision making method for evaluating suppliers of weapons using fuzzy AHP based on entropy
weight has been used [25]. However, the method used is very subjective and calculations are very complicated. A simpler
way to solve the same problem of evaluating suppliers of weapons has also been used [26,27]. Ruoning and Zhai [28] used
yet another method, the fuzzy logarithmic least square method, to solve a theoretical supplier evaluation problem. Ruoning
 D. Kumar et al. / Mathematical and Computer Modelling 58 (2013) 1679–1695 1681

and Zhai [29] again concluded that the manufacturer can induce the retailer to order the right quantity, and therefore realize
the maximum supply chain profit using fuzzy logic. Kumar et al. [7] used fuzzy logic to analyze various issues pertaining to
textile organizations.
Korpela et al. [30] used mixed integer programming in their study of supplier evaluation. Talluri et al. [31] and Weber
et al. [32] used a combination of data envelopment analysis and multi-objective programming. Vidal et al. [33] presented a
mathematical model, using a linear programming technique for supplier evaluation. Chung et al. [34] developed a closed-
loop supply chain inventory model to maximize the joint profits of the supplier, manufacturer and the retailer under
contractual design.
From the above discussion, it is clear that supplier evaluation is a multi-criteria problem which includes both qualitative
and quantitative factors. In order to select the best suppliers it is necessary to make a trade off between these tangible
and intangible factors, some of which may conflict. As the techniques used for supplier evaluation discussed above are
mathematical, they have significant problems in considering qualitative factors which are very important in supplier
evaluation [5]. The involvement of linguistic variables makes the problem more difficult. Moreover, most of the studies
discussed took two–three variables at a time, which actually are not sufficient in practical life. The present work focuses on
the application of a non-conventional technique as a methodology. This technique is capable of handling both qualitative
and quantitative factors for modeling and evaluation of suppliers in SCM practices.

3. Basics of fuzzy logic

Literature reveals that fuzzy logic was first developed by Lukasiewicz in the 1930s. Fuzzy logic techniques found usage in
engineering applications only after the 1960s through the concept of fuzzy sets and in 1968 through the fuzzy algorithm [35].
Fuzzy logic has found importance in engineering as it incorporates imprecision and subjectively into the model formulation
and solution process. It attains even more importance to model systems that are hard to define precisely, such as supplier
evaluation in SCM. Due to this distinct property of fuzzy logic, it has been able to aid research in production management
when the dynamics of the environment limits the specification of model objectives, constraints and measurements. Its
ability to quantitatively and qualitatively model problems involving vagueness and imprecision has made its use more
interesting [36].
A lot of applications can be found in literature, where fuzzy set theory has been used in the various fields of the supply
chain, including production management, quality and cost benefit analysis etc. The existence of imprecision and vagueness
of such problems is the main reason for the application of fuzzy set theory in these areas of research. Also, non-availability
of complete information, precise references and reliable data on such issues makes the application of fuzzy logic in such
fields more interesting. Such problems are handled by using the structured approach of scoring methods because of these
problems [5]. However, this approach is not effective in supplier evaluation decisions. In such decisions, the application of
fuzzy inference technique can be very effective, as proposed in the present work.
Fuzzy logic deals successfully with models that have fuzziness or vagueness, unlike other methods, like classical theory.
Elements of a fuzzy set are taken from a universe of discourse or universe for short. Fuzzy set theory deals with the
imprecision associated with many variables by permitting a grade of membership to be defined over the interval [0, 1],
called the membership function.
To understand the mathematical definition of fuzzy sets, let us consider a finite set of objects X , described as:
X = x1 , x2 , x3 , . . . , xn
where xi are the elements in the set X . Each element xi has a membership function (u). A fuzzy set A can be represented
as a linear combination of the following forms:
A = u1 (x1 ), u2 (x2 ), u3 (x3 ), . . . , un (xn ).
The fuzzy rule base consists of fuzzy ‘‘if–then’’ (also called Antecedent–Consequence) rules. These rules are generated
based on the concept of the domain rule of a data sample. For example, we take a rule:
‘‘if x1 is A and x2 is B, then y is C ’’
where x1 and x2 are variables, y is the solution variable and A, B, and C are fuzzy terms.

4. Design of fuzzy inference system model

In a fuzzy system, certain inputs are given to rules, depending upon the system under consideration. The outputs given
by the system are then utilized to reach the final decision. The whole process is clear from a typical fuzzy inference system
shown in Fig. 1.

4.1. Data collection

The population considered for the present study was Indian textile organizations. A survey was conducted through a
drafted questionnaire. It consisted of existing measurement scales for the research constructs. To check its content validity
and terminology, a group of 15 executives from different Indian textile organizations were interviewed at this pre-test
stage. As per their suggestions, a number of modifications were made to the questionnaire. As a result, the questionnaire
1682 D. Kumar et al. / Mathematical and Computer Modelling 58 (2013) 1679–1695

Fig. 1. Fuzzy inference systems.

became more meaningful and clear for the targeted population. User friendliness and ease of completion was kept in mind
while preparing the modified questionnaire. This structured questionnaire was used to collect responses from executives
of various Indian textile organizations. The questions of the questionnaire were framed on a five-point Likert scale. A
total number of 525 organizations were selected from the directory of public sector, private sector and government
organizations to collect information through this questionnaire. Out of the 525 questionnaires mailed, 36 questionnaires
returned undelivered, 2 units were reported to be closed and 14 refused to participate in this survey without mentioning
any reason. A total of 66 responses were collected, making an effective response rate of 13.95%. Though we understand that
the response rate is somewhat small, it does assist in digging and tracing the interesting and important aspects of the SCM
practices related to supplier evaluation, as being followed in the Indian textile organizations.
As we believe that organizations do not communicate; people do, most of the constructs used in the study are individual
level constructs. However, in order to translate these behavioral constructs at the dyadic, inter-organizational level (which is
the unit of analysis), we presume that individual views on the issues surveyed will be a function of their organizational roles.
This is also because, it was found at pre-testing of questionnaire level that individuals who occupy strategic positions in their
organizations would be more knowledgeable about the strategic aspects of inter-organizational exchange relationships.
Thus, we tapped responses to the questionnaire from key informants of purchase and material management departments
in Indian textile organizations. Many examples are available in the literature to prove usage of this practice in strategic
management research [17]. For example, the approach of surveying the buying firms’ executives to study the supplier
selection and evaluation problems has been widely adopted [37,38].
Respondents were requested to rate the importance of factors considered for supplier evaluation, using a five-point scale
with endpoints labeled ‘‘least important’’ (=1) and ‘‘most important’’ (=5). Factors, that were rated an average score less
than 2, were omitted. Results for the importance scores for the rest of the factors are provided in Fig. 2.

4.2. Identification of input and output variables

Based on the findings of Section 4.1, we select nine variables. These variables are: consistency in product quality
(CIPQ), improvement in incoming components (IIIC), reduction in damaged components in transit (RIDCIT), inventory level
reduction (ILR), lot size reduction (LSR), reduction in plant stoppage due to shortage of material (RIPS), on-time delivery
(OTD), reduction in order lead time (RIOLT), reduction in product development cycle time (RIPDCT). To avoid confusion
among the decision maker’s judgments, each variable along with its corresponding membership function is fully described
in the next section. Mathematically, the final decision represents a function, whose domain is a set of the above mentioned
ten variables. These variables are given a proper value, depending upon their importance for a particular organization. So, if
we assume y as the final supplier evaluation decision, then
y = f (x1 , x2 , x3 , x4 , x5 , x6 , x7 , x8 , x9 ).
These nine variables are further divided into three different groups as intermediate variables. This division is based on
the nature of variables. These groups are processed to give one output each. The three outputs so achieved are treated as
input to give a single output which is treated as a final recommendation. This way, a modular approach is created to convert
the multi-objective problem to a single one, which is capable of delivering a final decision. The characteristics of the groups
are discussed below.
Group 1 contains the variables that are directly concerned with the production quality. We term these variables as Quality
Characteristics. The inputs to this group are: CIPQ, IIIC and RIDCIT. We name the output of this group as the ‘‘Quality Effect’’.
Group 2 consists of the variables that are directly concerned with the cost of the final product. We name these variables
as Cost Characteristics. The inputs to this group are: ILR, LSR and RIPS. We name the output of this group as the ‘‘Cost Effect’’.
Group 3 contains the variables that reflect the effect of performance/seriousness of the supplier on production lead time.
We term these variables as Time Characteristics. The inputs to this group are: OTD, RIOLT and RIPDCT. We call the output of
this group ‘‘Time Effect’’.
Group 4 takes the outputs of above three groups as inputs and processes them to give single output. This single output is
named ‘‘Final Recommendation’’.
 D. Kumar et al. / Mathematical and Computer Modelling 58 (2013) 1679–1695 1683

Fig. 2. Importance scores of factors used to evaluate the supplier (1 = ‘‘least important’’ and 5 = ‘‘most important’’).

This way, in this modular approach we reduce nine variables to three and finally these three variables are transformed
into a single factor. We take these three variables (outputs of group 1, 2 and 3) as a, b and c; and output of the group 4 as d,
as shown in Fig. 3. Mathematically:
a = f1 (x1 , x2 , x3 ), b = f2 (x4 , x5 , x6 ), c = f3 (x7 , x8 , x9 ) and d = f4 (f1 (a), f2 (b), f3 (c )).

4.3. Design of membership functions

A fuzzy inference system, as shown in Fig. 1 is developed by using MatlabTM software. The nine variables, mentioned in
Section 4.2 are defined and for each of these variables, universe of discourse and degree of membership function are defined.
Gaussian and Bell membership functions are used in the proposed fuzzy inference system to define the shape of both
input and output variables. We shall draw the values of the vectors a, b, c and d, for each value of the inputs from x1 to x9 .
Analyzing vector a, we have three input variables (x1 , x2 , x3 ) and one output, defined as follows:
• CIPQ (x1 ): This defines the product image of the buyer expected from the supplier each time they make a purchase. It is
evaluated as ‘‘low’’, ‘‘medium’’ or ‘‘high’’ as compared to the mean quality of the same material being offered by other
suppliers. Fig. 4 shows the distribution of membership functions within the universe of discourse.
• IIIC (x2 ): This defines the ability of the supplier to maintain or enhance the quality level of products each time a purchase
is made by the buyer. Evaluation of this ability of supplier can be divided into three categories: ‘‘negligible’’, ‘‘some’’ or
‘‘appreciable’’. Fig. 5 shows the distribution of membership functions within the universe of discourse.
• RIDCIT (x3 ): This includes the packing material used and care taken during shipment by the supplier while supplying
material to the buyer. As seen in Fig. 6, such a reduction can be divided into ‘‘none’’, ‘‘few’’ or ‘‘many’’ following the levels
determined by membership functions.
• The output ‘‘Quality Effect’’ variable, as shown in Fig. 7, illustrates the three possible different outcomes about the
suitability of a particular supplier, based on inputs defined above.
Vector b, has three input variables (x4 , x5 , x6 , x7 ) and one output, defined as follows:
• ILR (x4 ): Inventory refers to the stock of resources that possess economic value, held by an organization at any point of
time. This variable defines a reduction in such stock as a result of service provided by the supplier. As shown in Fig. 8,
such a reduction is defined as ‘‘low’’, ‘‘medium’’ or ‘‘high’’ following the levels determined by membership functions.
• LSR (x5 ): Lot size refers to the measure or quantity increment acceptable to the buyer. LSR is a result of acceptable service
provided by the supplier in case of emergency. As shown in Fig. 9, this characteristic is defined as ‘‘negligible’’, ‘‘some’’
or ‘‘appreciable’’ following the levels determined by membership functions.
1684 D. Kumar et al. / Mathematical and Computer Modelling 58 (2013) 1679–1695

Fig. 3. Modular fuzzy inference system.

Fig. 4. Membership functions of input variable x1 ‘‘CIPQ’’.

 D. Kumar et al. / Mathematical and Computer Modelling 58 (2013) 1679–1695 1685

Fig. 5. Membership functions of input variable x2 ‘‘IIIC’’.

Fig. 6. Membership functions of input variable x3 ‘‘RIDCIT’’.

Fig. 7. Membership functions of output variable ‘‘Quality Effect’’.

• RIPS (x6 ): Delay in supply of ordered material or supply of low quality products by the supplier can be a cause of plant
stoppage. This variable considers the reduction in the number of such stoppages in a specified period. It is evaluated as
‘‘negligible’’, ‘‘some’’ or ‘‘a lot’’. Fig. 10 shows the distribution of membership functions within the universe of discourse.

The output variable ‘‘Cost Effect’’, as shown in Fig. 11, illustrates the three possible different outcomes, based on inputs
given as per actual situations, as explained above.
1686 D. Kumar et al. / Mathematical and Computer Modelling 58 (2013) 1679–1695

Fig. 8. Membership functions of input variable x4 ‘‘ILR’’.

Fig. 9. Membership functions of input variable x5 ‘‘LSR’’.

Fig. 10. Membership functions of input variable x6 ‘‘RIPS’’.

Vector c has three input variables (x7 , x8 , x9 ) and one output, defined as follows:

• OTD (x7 ): This variable refers to the number of orders delivered as per delivery schedule promised by the supplier. Fig. 12
shows three membership functions for this characteristic, as ‘‘reduced’’, ‘‘same’’ or ‘‘improved’’ service.
• RIOLT (x8 ): Order lead time is the period between placing an order and receiving the ordered item. Input for RIOLT to the
system is defined as ‘‘increased’’, ‘‘no difference’’ and ‘‘reduced’’, as seen in Fig. 13.
 D. Kumar et al. / Mathematical and Computer Modelling 58 (2013) 1679–1695 1687

Fig. 11. Membership functions of output variable ‘‘Cost Effect’’.

Fig. 12. Membership functions of input variable x7 ‘‘OTD’’.

Fig. 13. Membership functions of input variable x8 ‘‘RIOLT’’.

• RIPDCT (x9 ): This variable defines the help offered by the supplier to the buyer during the development/introduction of a
new product in the market by the organization. As seen in Fig. 14, it can be divided into ‘‘no interest’’, ‘‘some’’ or ‘‘helping’’
following the levels determined by membership functions.
• The output variable ‘‘Time Effect’’, as shown in Fig. 15, illustrates the three possible different outcomes for recommending
or not a particular supplier, based on inputs given as per actual situations, as explained above.
1688 D. Kumar et al. / Mathematical and Computer Modelling 58 (2013) 1679–1695

Fig. 14. Membership functions of input variable x9 ‘‘RIPDCT’’.

Fig. 15. Membership functions of output variable ‘‘Time Effect’’.

Fig. 16. Membership functions of output variable ‘‘Final Recommendation’’.

As explained in Section 4.2, vectors a, b and c are taken as inputs to group 4 to give vector d as output. This output variable
‘‘final recommendation’’, as shown in Fig. 16, illustrates the four possible different outcomes about the final recommendation
for a particular supplier, based on inputs given as per actual situations.

4.4. Design of rule base and rule viewers

After designing the membership functions for different input and output variables, the next step is to define the rules
that will govern these variables.
 D. Kumar et al. / Mathematical and Computer Modelling 58 (2013) 1679–1695 1689

Table 1
Rules for group 1.
If and and then
x1 (CIPQ) x2 (IIIC) x3 (RIDCIT) a (Quality Effect)

Low Negligible None Negligible

Medium Some Few Some
High Appreciable Many Appreciable
High Some None Some
Medium Appreciable None Some
Low Some Many Some
Medium Negligible Few Some
Low Appreciable Many Appreciable
Low Negligible Few Negligible

Fig. 17. Algorithm for finalizing rules.

For vector a, having three inputs, initially 27 rules were defined. After going through different iterations, non-contributing
rules were deleted. The algorithm used for deleting the non-contributing rules is shown in Fig. 17. Finally, 09 rules were left.
These rules are shown in Table 1.
Fig. 18 indicates rule viewers that show values of the various inputs (CIPQ, IIIC and RIDCIT) to the model and
corresponding computed output (Quality Effect).
Following the similar approach, rules for vector b, c and d were defined. These are shown in Tables 2–4. Rule viewers and
the corresponding computed outputs are shown in Figs. 19–21.

5. Testing of fuzzy inference system

As explained in Section 4, each component of each vector can be considered as a function of all components of the
preceding vector. The fuzzy inference approach is useful in most of the production systems because of the involvement
1690 D. Kumar et al. / Mathematical and Computer Modelling 58 (2013) 1679–1695

Fig. 18. Rule viewers of group 1.

Table 2
Rules for group 2.
If and and then
x4 (ILR) x5 (LSR) x6 (RIPS) b (Cost Effect)

Low Negligible Negligible Negligible

Medium Some Some To some extent
High Appreciable A lot Appreciable
High Some A lot Appreciable
Medium Some A lot Appreciable
Low Appreciable Some To some extent
Medium Appreciable Some To some extent
Low Some A lot To some extent

Fig. 19. Rule viewers of group 2.

of a number of variables and the ability of this system to reduce these variables in each successive step, until the final value
is achieved.
 D. Kumar et al. / Mathematical and Computer Modelling 58 (2013) 1679–1695 1691

Table 3
Rules for group 3.
If and and then
x7 (OTD) x8 (RIOLT) x9 (RIPDCT) c (Time Effect)

Reduced Increased No interest Fair

Same No difference Some Good
Improved Reduced Helping Excellent
Same No difference Helping Good
Reduced No difference No interest Fair
Improved No difference Some Excellent
Same Increased No interest Fair
Improved No difference Helping Excellent

Fig. 20. Rule viewers of group 3.

Table 4
Rules for ‘‘Final Recommendation’’ group.
If and and then
a (Quality Effect) b (Cost Effect) c (Time Effect) d (Final Recommendation)

Negligible Negligible Fair Change supplier

Some To some extent Fair Reduce orders
Some To some extent Good Continue
Appreciable Appreciable Excellent Increase orders
Appreciable Appreciable Good Increase orders
Some Appreciable Good Continue
Negligible To some extent Excellent Reduce orders
Appreciable To some extent Fair Continue

The involvement of most of the production cost in a raw material makes the supplier evaluation process in any industry
strategic and the decision makers have to make decisions repeatedly to ensure full effectiveness of the system adopted. Fuzzy
inference systems are best suited in such domains. The beauty of such systems is their ability to provide the same output
that a decision maker would provide in any given situation, from a set of given inputs. To further enhance the reliability of
the proposed system, a series of simulations were carried out by varying one input at a time or many inputs simultaneously.
After carrying out the above mentioned simulations, several evaluations were made in the final stage of the proposed
fuzzy inference systems. For each input, a specific value of x was given to define the intermediate vectors, using the rules
defined above. It was ensured that the values of every input were increased and decreased from almost its minimum to its
maximum. It makes the working of proposed system more understandable to get the final output y in a wide range of values.
The various simulations are illustrated below, by using different input values. The results obtained from the system are
presented in numerical and linguistic formats, as shown in Tables 5–8. These simulations demonstrate not only the working
of system, but also the variations in the values obtained for intermediate vectors a, b and c. Taking these intermediate vectors
as inputs, values for the vector d, i.e., ‘‘Final Recommendation’’ were determined, as shown in Table 8.
1692 D. Kumar et al. / Mathematical and Computer Modelling 58 (2013) 1679–1695

Fig. 21. Rule viewers of group 4.

Table 5
Group 1 results.
Inputs Output
CIPQ IIIC RIDCIT Quality Effect

Low (10) Negligible (10) None (10) Negligible (9.56)

Medium (50) Some (50) Few (50) Some (48)
High (95) Appreciable (90) Many (95) Appreciable (93.5)

Table 6
Group 2 results.
Inputs Output
ILR LSR RIPS Cost Effect

Low (10) Negligible (10) Negligible (10) Negligible (7.13)

Medium (55) Some (50) Some (50) To some extent (48.4)
High (95) Appreciable (95) A lot (95) Appreciable (93.7)

Table 7
Group 3 results.
Inputs Output
OTD RIOLT RIPDCT Time Effect

Reduced (15) Increased (15) No interest (15) Fair (8.47)

Same (50) No difference (50) Some (50) Good (48.4)
Improved (90) Reduced (90) Helping (90) Excellent (93.3)

Table 8
‘‘Final Recommendation’’ group results.
Inputs Output
Quality Effect Cost Effect Time Effect Final Recommendation

Negligible (10) Negligible (10) Fair (10) Change supplier (14.1)

Some (40) To some extent (40) Fair (20) Reduce orders (48.7)
Some (70) To some extent (70) Good (70) Continue (70.8)
Appreciable (95) Appreciable (95) Excellent (95) Increase orders (91.6)

6. Illustrative example through a case study

The proposed model can be illustrated by taking an example. We collected the data from a reputed man-made fiber
manufacturing organization in north India (not included in the survey). Managers/engineers of the organization (from
purchase, production and quality control departments) were requested to respond to a questionnaire to collect information
 D. Kumar et al. / Mathematical and Computer Modelling 58 (2013) 1679–1695 1693

Table 9
Values of the variables.
x1 x2 x3 a x4 x5 x6 b x7 x8 x9 c d

Average value 80 90 78.5 87.9 75.7 82.5 92.8 88.5 92.1 80.8 78 92.5 88.6

for the model. The questionnaire used is given in the Appendix. To make the terms used in the questionnaire clear to the
respondents, details of definitions of terms, as explained in Section 4.2 were attached with the questionnaire. An average
value of 28 responses collected was taken.
The organization chosen procures raw material (caprolactum) from three suppliers. The respondents were requested
to respond to the questions as per their judgment about one particular supplier. To convert the scale used (1–5) to a new
scale of 1–100 (i.e., the scale used for designing membership functions), average values of variables were multiplied by 20.
The values obtained are used in Figs. 7, 11 and 15 to get the value of ‘‘Quality Effect’’, ‘‘Cost Effect’’ and ‘‘Time Effect’’ as
87.9, 88.5 and 92.5 respectively. These values were used in Fig. 21 (by sliding the vertical lines) to get the value for ‘‘Final
Recommendation’’. This value comes out to be 88.6. So, the final recommendation to the organization is to increase the
orders of raw material to this particular supplier.
The calculated values of different variables are shown in Table 9.

7. Managerial implication of the study

The importance of the proposed system can be understood from the fact that most of the production cost consists of
raw material cost, for which effective evaluation of suppliers is essential. Purchasing professionals can demonstrate the
contribution they make to the firm through utilization of the proposed system for the same and make their claim more
meaningful for their involvement in strategic issues that are critical for the firm [37]. This system enables them to easily
evaluate their suppliers continuously. Using the proposed model, the purchasing professionals can evaluate the suppliers
from time to time, by just calculating the membership functions of the characteristics. Purchasing personnel can reach a
decision regarding the placement of order with a particular supplier, out of the various alternatives available with the help
of the proposed model easily and effectively. The proposed system further makes the purchasing managers and buyers have
a better idea of their supply base characteristics.
It is evident from the Sections 5 and 6 that the proposed fuzzy inference system can be applied by production managers
to various stages of the supply chain, viz, supply, production, distribution, evaluation etc. using a language to model
problems, which contain fuzzy relationships. The ability of the proposed model to handle indefinite and imprecise factors,
that managers come across in their practical life, makes it practically more usable.

8. Conclusions and directions for future research

In this paper, an effort has been made to develop a model for supplier evaluation through data collection from survey of
66 Indian textile organizations. A questionnaire has been used to determine the factors, which these organizations consider
while evaluating a particular supplier. A multi-input single output (MISO) mamdani fuzzy inference system has been
proposed for supplier evaluation taking the most important nine factors, opined by the respondents. As the literature reveals,
no logical method is being employed at present to continuously evaluate a supplier in the industry under study. Considering
the extent of production cost involved in raw material, the proposed system can be very helpful to the companies for making
decisions about supplier evaluation. Apart from supplier evaluation, the proposed methodology can be translated to model
the decision making processes of facilities and service purchases by just changing the variables. As illustrated in the above
sections, the proposed system is very simple for the engineers/managers of the related fields to use and they will find the
implementation of this technique very easy and suitable.
Like any other system, the fuzzy logic scheme has its own drawbacks. While developing fuzzy rules for the system, it
requires experience from field experts, experimental results and theoretical derivation to make the system effective [39].
Even in some cases, experts may need to be put on site to be sure about the characteristics that may effect the whole
system and to adjust the fuzzy rules at the beginning of the stage. This exercise may sometimes increase the cost of system
development. Fuzzy reasoning also sacrifices some explanation for accuracy, reliability, and compactness, and lack of the
concept of justification for fact as in a rule based system [40].
Model development through usage of a particular sample population can be another limitation of this research work. We
feel that a mixed population of respondents from multiple sources can increase the scope of generalizability of the model
and we propose the inclusion of this aspect in future research.
Though the system has some limitations as listed above, the potential benefits of the implementation of the proposed
system in Indian textile organizations are not in doubt. An effective mechanism for the art of designing the right system
using the proposed approach is debatable. Therefore, issues related to these concerns are worthy of future research. This
study is a modest contribution toward that end.
1694 D. Kumar et al. / Mathematical and Computer Modelling 58 (2013) 1679–1695

Acknowledgments

We are thankful to the organizations, who participated in the survey and also to those executives who spared valuable
time to share their experience with us through personal interaction and otherwise.

Appendix

Questionnaire
Least Most
important important
1 2 3 4 5
(i) Consistency in product quality [] [] [] [] []
(ii) Improvement in incoming [] [] [] [] []
components
(iii) Reduction in damaged [] [] [] [] []
components in transit
(iv) Inventory level reduction [] [] [] [] []
(v) Lot size reduction [] [] [] [] []
(vi) Reduction in plant stoppage due to [] [] [] [] []
shortage of material
(vii) On-time delivery [] [] [] [] []
(viii) Reduction in order lead time [] [] [] [] []
(ix) Reduction in product development [] [] [] [] []
cycle time

References

[1] R.A. Lancioni, New developments in supply chain management for the millennium, International Journal of Industrial Marketing Management 29
(2000) 1–6.
[2] Douglas M. Lambert, Martha C. Cooper, Issues in supply chain management, International Journal of Industrial Marketing Management 29 (2000)
65–83.
[3] R. Ballou, S. Gilbert, A. Mukherjee, New managerial challenges from supply chain opportunities, International Journal of International Marketing
Management 29 (2000) 7–18.
[4] K.L. Choy, W.B. Lee, V. Lo, Development of a case based intelligent customer–supplier relationship management system, International Journal of Expert
Systems with Applications 23 (2002) 281–297.
[5] S.H. Ghodsypour, C. O’Brian, A decision system for supplier selection using an integrated analytic hierarchy process and linear programming,
International Journal of Production Economics 56–57 (1998) 199–212.
[6] K.L. Choy, W.B. Lee, A generic supplier management tool for outsourcing manufacturing, International Journal of Supply Chain Management 8 (2003)
140–154.
[7] D. Kumar, J. Singh, O. Singh, Analysis of supplier related issues with implementation of fuzzy logic for Indian textile organizations, International Journal
of Logistics Systems and Management 10 (2011) 313–339.
[8] C. Chao, E.E. Scheuing, W.A. Ruch, Purchasing performance evaluation: an investigation of different perspectives, International Journal of Supply Chain
Management 29 (1993) 32–39.
[9] S. Talluri, J. Sarkis, A model for performance monitoring of suppliers, International Journal of Production Research 40 (2002) 4257–4269.
[10] S. Talluri, R.C. Baker, A multi-phase mathematical programming approach for effective supply chain design, European Journal of Operational Research
141 (2002) 544–558.
[11] S. Wei, J. Zhang, Z. Li, A supplier-selecting system using a neural network, IEEE International Conference on Intelligent Processing Systems 1 (1997)
468–471.
[12] P.C. Yang, H.M. Wee, S.L. Chung, P.C. Ho, Sequential and global optimization for a closed-loop deteriorating inventory supply chain, International
Journal of Mathematical and Computer Modelling 52 (2010) 161–176.
[13] B.C. Arntzen, G.G. Brown, T.P. Harrison, L.L. Trafton, Global supply chain management at Digital Equipment Corporation, Interfaces 25 (1995) 69–93.
[14] P. Briggs, Case study: vendor assessment for partners in supply, European Journal of Purchasing and Supply Management 1 (1994) 49–59.
[15] M. Braglia, A. Petroni, A quality assurance-oriented methodology for handling trade-offs in supplier selection, International journal of Physical
Distribution & Logistics Management 30 (2000) 96–111.
[16] B. Ng, B.G. Ferrin, J.N. Pearson, The role of purchasing/transportation in cycle time reduction, International Journal of Operations and Production
Management 17 (1997) 574–591.
[17] A.D. Toni, G. Nassimbeni, Just-in-time purchasing: an empirical study of operational practices, supplier development and performance, International
Journal of Management Science 28 (2000) 631–651.
[18] B. Fynes, S. Burca, J. Mangan, The effect of relationship characteristics on relationship quality and performance, International Journal of Production
Economics 111 (2008) 56–69.
[19] C. Murlidharan, N. Anantharaman, S.G. Deshmukh, Vendor rating in purchasing scenario: a confidence interval approach, International Journal of
Operations and Production Management 21 (2001) 1305–1326.
[20] R. Handfield, S.V. Walton, R. Sroufe, S.A. Melnyk, Applying environmental criteria in the application of the Analytic Hierarchy Process, European Journal
of Operational Research 141 (2002) 70–87.
[21] K.S. Bhutta, F. Huq, Supplier selection problem: a comparison of the total cost of ownership and analytic hierarchy process approaches, International
Journal of Supply Chain Management 7 (2000) 126–135.
[22] G. Wang, S.H. Huang, J.P. Dismukes, Product-driven supply chain selection using integrated multi-criteria decision-making methodology, International
Journal of Production Economics 91 (2004) 1–15.
[23] A. Agarwal, R. Shankar, Analyzing alternatives for improvement in supply chain performance, International Journal of Work Study 51 (2002) 32–37.
 D. Kumar et al. / Mathematical and Computer Modelling 58 (2013) 1679–1695 1695

[24] H. Lee, S. Lee, Y. Park, Selection of technology acquisition mode using the analytic network process, International Journal of Mathematical and
Computer Modeling 49 (2009) 1274–1282.
[25] D. Mon, Evaluating weapon system using fuzzy analytic hierarchy process based on entropy weight, International Journal of Fuzzy Sets and Systems
62 (1994) 127–134.
[26] S. Chen, Evaluating weapon systems using fuzzy arithmetic operations, International Journal of Fuzzy Sets and Systems 77 (1996) 265–276.
[27] C. Cheng, Evaluating weapon systems using ranking fuzzy numbers, International Journal of Fuzzy Sets and Systems 107 (1999) 25–35.
[28] X. Ruoning, X. Zhai, Fuzzy logarithmic least squares ranking method in analytic hierarchy process, International Journal of Fuzzy Sets and Systems 77
(1996) 175–190.
[29] X. Ruoning, X. Zhai, Manufacturer’s coordination mechanism for single-period supply chain problems with fuzzy demand, International Journal of
Mathematical and Computer Modeling 51 (2010) 693–699.
[30] J. Korpela, K. Kylaheiko, A. Lehmusvaara, M. Tuominen, An analytic approach to production capacity allocation and supply chain design, International
Journal of Production Economics 78 (2002) 187–195.
[31] S. Talluri, R.C. Baker, J. Sarkis, A framework for designing efficient value chain networks, International Journal of Production Economics 62 (1999)
133–144.
[32] C.A. Weber, J. Current, A. Desai, An optimization approach to determining the number of vendors to employ, International Journal of Supply Chain
Management 5 (2000) 90–98.
[33] C.J. Vidal, M. Goetschalckx, Theory and methodology: A global supply chain model with transfer pricing and transportation cost allocation, European
Journal of Operational Research 129 (2001) 134–158.
[34] S.L. Chung, H.M. Wee, P.C. Yang, Optimal policy for a closed-loop supply chain inventory system with remanufacturing, International Journal of
Mathematical and Computer Modeling 48 (2008) 867–881.
[35] J. Singh, N. Singh, J.K. Sharma, Fuzzy modeling and control of HVAC systems — a review, International Journal of Scientific and Industrial Research 65
(2006) 470–476.
[36] D.A. Carrera, R.V. Mayorga, Supply chain management: a modular fuzzy inference system approach in supplier selection for new product development,
International Journal of Intelligent Manufacturing 19 (2008) 1–12.
[37] A.S. Carr, J.N. Pearson, Strategically managed buyer–supplier relationships performance outcomes, International Journal of Operations Management
17 (1999) 497–519.
[38] A. Paulraj, A.A. Lado, I.J. Chen, Inter-organizational communication as a relational competency: Antecedents and performance outcomes in
collaborative buyer–supplier relationships, International Journal of Operations Management 26 (2008) 45–64.
[39] K.S. Leung, W. Lam, Fuzzy concepts in expert systems, IEEE 21 (1988) 43–56.
[40] V. Dhar, R. Stein, Seven Methods for Transforming Corporate Data into Business Intelligence, Prentice-Hall, NJ, 1997.