Beruflich Dokumente
Kultur Dokumente
overcome this error, Cheng [7] proposed a new model developed. Weber and Current [18] used
model that utilized fuzzy ratio scales to indicate the multi-objective linear programming in supplier
relative strength of factors used in the evaluation of selection. They developed a model to tradeoff among
suppliers of weapons. AHP approach may suffer several conflicting criteria. Many such linear
from some drawbacks including bias in estimation programming models have been developed for
process. To overcome this drawback, Muralidharan et minimization of cost of production and lot size to be
al. [8] proposed a methodology in their paper which ordered. Liu et al. [19], for example, proposed a
made use of estimation of the rating by a group on an linear programming model for optimization of lot
individual basis following the principle of anonymity. sizes in consequential production periods. Ustun and
They carried out a statistical analysis to determine the Demirtas [20] used an integrated approach of analytic
confidence intervals for the estimates of the network process (ANP) and multi-objective mixed
composite rating of the suppliers. To overcome the integer linear programming (MOMILP) to considers
same problem, Bhutta and Huq [9] used AHP to cope both tangible and intangible factors in choosing the
with multi criteria situations involving supplier best suppliers and defines the optimum quantities
selection and compared it with another approach, among selected suppliers to maximize the total value
called total cost of ownership that looked beyond just of purchasing (TVP), and to minimize the total cost
the price of purchase to better understand and and total defect rate and to balance the total cost
manage costs in selecting and maintaining among periods. Lee et al. [21] also used ANP for
relationships with suppliers. Chan and Kumar [10] selecting the appropriate acquisition mode for a
used fuzzy based AHP to tackle the different decision required technology during supplier selection and
criteria like cost, quality, service performance and evaluation process. Chung et al. [22] developed a
supplier's profile including the risk factors involved closed-loop supply chain inventory model to
in the selection of global supplier in the current maximize the joint profits of the supplier,
business scenario. Ruoning and Zhai [11] also used manufacturer and the retailer under contractual
fuzzy logic for the retailer to order the right quantity design.
to maximize supply chain profit. As discussed above, researchers and practitioners
Many researchers and practitioners used combination have focused their attention on selecting the most
of more than one method for solution of supplier suitable supplier for the different purposes using
selection problems. Ghodsypour and OBrien [12] complex mathematical tools. The models discussed
proposed a decision support system for supplier provide systematic approaches for purchasing
selection using an integration of AHP and linear managers to evaluate and score suppliers with multi-
programming to consider both tangible and intangible criteria. Nevertheless, these models are not easy to
factors in choosing the best supplier and placing the implement. Models based on multi-objective
optimum order quantities among them. Korpela et. al optimization require the decision makers to
[13] proposed a framework by which risks related to exogenously specify the exact values of weights of
a customer-supplier relationship can be included in individual criteria. It is however difficult to obtain
production capacity allocation, based on integrated precise weight values. The objective of the present
AHP and mixed integer programming (MIP) model. paper is to enhance evaluation and selection
Wang et. al [14] developed a multi-criteria decision- methodology and to recommend an alternative
making methodology using integrated AHP and pre- approach by applying fuzzy based multi-objective
emptive goal programming (PGP) to take into optimization on the basis of ratio analysis (MOORA)
account both quantitative and qualitative factors in method to solve supplier selection problems in
supplier selection. Ng [15] proposed a weighted manufacturing environment.
linear program for the multi-criteria supplier Rest of the paper is organized as follows. Section 3
selection problem. The proposed model allowed gives brief description of fuzzy set theory. Section 4
involvement of the decision maker in ranking the describes basics of MOORA and fuzzy MOORA
relativity of importance of criteria. In addition, the methods. Section 5 illustrates an example to validate
decision makers role was not as subjective as other the proposed model and concluding remarks are
approaches in AHP models. He presented a provided in section 6.
transformation technique which enabled the weighted 3. FUZZY SET THEORY
linear program to be solved without optimization. Xia Fuzzy set theory, pioneered by Zadeh [23] has been
and Wu [16] also considered an integrated approach widely used to model systems that are hard to define
of AHP improved by rough sets theory and multi- precisely. Therefore, fuzzy logic represents an
objective mixed integer programming to alternative tool to aid research in production
simultaneously determine the number of suppliers to management when the dynamics of the environment
employ and the order quantity allocated to these limits the specification of model objectives,
suppliers in the case of multiple sourcing, multiple constraints and measurements. The main interest of
products, with multiple criteria and with supplier's fuzzy set theory pertained to representing uncertainty
capacity constraints. in human cognitive processes in the early years of its
Another approach to solve supplier selection use. But, it is now applied to problems in
problems is mathematical programming. This engineering, businesses, medical and other sciences.
approach has been used by researchers and The use of this methodology for modeling and
practitioners in many ways. One category which is analyzing decision systems is of particular interest in
used in many problems in literature is multi-objective problems related to supplier selection due to fuzzy
optimization models. Gaballa [17] was the first theorys ability to quantitatively and qualitatively
researcher to utilize mathematical programming. He model problems which involve vagueness and
formulated a MIP for Australian post office. He tried imprecision [24]. Literature reveals many
to minimize suppliers discounted prices through the applications of fuzzy set theory in the fields of
Int J Adv Engg Tech/Vol. V/Issue I/Jan.-March.,2014/16-24
Seema et al., International Journal of Advanced Engineering Technology E-ISSN 0976-3945
production management, new product development, Table 1: A general decision matrix for supplier
production scheduling, inventory management, selection problem
facility layout, facility location, quality and cost
benefit analysis etc. The main reasons why fuzzy set
theory application is successful in these areas of
research are:
Imprecision and vagueness of such where, xij is the performance measure of the ith
problems. alternative on jth attribute, m is the number of
Most of the times, the information required alternatives and n is the number of attributes.
to develop a model is vague or not precisely In the next step, a ratio system is developed, where
measureable. each performance of an alternative on an attribute is
to be compared with the representative for all the
Imprecision and subjective opinion of the
alternatives. Brauers et al. [31] proposed the ratio,
people concerned may further dampen the
shown in equation 1, for desired ratio system
quality and quantity of available
(j=1,2,..,n) (1)
information.
In a universe of discourse X, a fuzzy sub-set is
xij = xij m 2
characterised by membership function (x) that x ij
i =1
maps each element x in X to a real number in the Where xij is a dimensionless number that belongs to
interval [0, 1]. The function value of (x) is termed the interval [0, 1] representing the normalized
as the grade of membership of x in . Generally, performance of the ith alternative on jth criterion.
triangular fuzzy numbers and trapezoidal fuzzy Next, the overall performance of each alternative is
numbers are used. Because of their calculation calculated as the difference between sums of its
easiness and added features, triangular fuzzy numbers normalized performances for beneficial attributes and
find their applications in engineering and non-beneficial attributes using the following
manufacturing environment. A triangular fuzzy equation.
variable is a triplet composed of crisp numbers, (a, b,
- .(2)
c), whose membership function is
xa Where g is the number of attributes to be maximized
A( x ) = , axb
ba and (n-g) is the number of attributes to be minimized,
=
xc
, bxc
and yi is the assessment value of ith alternative with
bc respect to all the attributes. In some cases, it
= 0, others is often observed that some attributes are more
important than the others. In order to give more
Figure 1 depicts membership function of triangular importance to an attribute, it could be multiplied with
fuzzy variable. its corresponding weight (significance coefficient)
[32]. When these attribute weights are taken into
consideration, equation (2) becomes as follows:
- (j = 1, 2, .., n)
(3)
where wj is the weight of jth attribute. The value of yi
so found can be positive or negative. This depends on
the total of its maxima (beneficial attributes) and
Figure 1: Membership function of triangular fuzzy minima (non-beneficial attributes) in the decision
variable matrix. An ordinal ranking of yi shows the final
4. THE MOORA METHOD preference. Thus, the best alternative has the highest
Multi-objective optimization, also known as multi- yi value, while the worst alternative has the lowest yi
criteria or multi-attribute optimization, is the process value.
of simultaneously optimizing two or more conflicting In the present paper, a new variant of MOORA
attributes. Such techniques, therefore, seem to be an method, i.e., fuzzy MOORA is proposed in order to
appropriate tool for ranking or selecting best take full advantage of MOORA method for solving
alternative from a set of available options. The the MCDM problems related to supplier selection,
MOORA method, first introduced by Brauers [25] is which generally have vague and imprecise data. The
such a multi-objective optimization technique that proposed method consists of following steps.
can be successfully applied to solve various types of Step 1: For developing the fuzzy decision matrix,
complex decision making problems in the opinions of the key decision makers are taken. In this,
manufacturing environment subject to certain each criterion is measured using triangular
constraints. This method has been successfully membership function. A typical fuzzy decision
applied in various engineering/management fields, matrix is shown below as equation (4).
for example selection of best intelligent X=
manufacturing system [26], site selection for setting
up a wind farm [27], project selection [28],
evaluation of inner climate [29], parametric
optimization of milling process [30] etc.
In MOORA method, first step is to develop a . (4)
decision matrix showing the pertinent evaluation where , and respectively are the lower,
attributes for different alternatives. A general middle and upper values of a triangular membership
decision matrix is shown in table 1.
Int J Adv Engg Tech/Vol. V/Issue I/Jan.-March.,2014/16-24
Seema et al., International Journal of Advanced Engineering Technology E-ISSN 0976-3945
function for alternative with respect to The required data has been collected from chemical
and biotechnology industries of India through a
criterion. survey. A questionnaire based on existing
Step 2: Fuzzy decision matrix, created in step 1 is measurement scales for the research constructs was
normalized using vector normalization procedure. developed for determining the factors that these
For normalization, equations (5), (6) and (7), given industries consider for supplier selection and
below are used. evaluate suppliers (see annexure A). These
a
x ij questions were framed on a five-point Likert scale. A
a
r ij = total of 625 questionnaires were sent to different
( ) ( ) ( )
m
a + x b + x c
2 2 2 industries throughout the country from the directory
x ij ij ij of public sector, private sector and government
i = 1 ............ (5)
organizations. A total of 88 responses were
xb
ij collected. This gives an effective response rate of
rb
ij = 14.08%. Though the response rate is somewhat less,
( ) ( ) ( )
m
a + x b + x c
2 2 2 nevertheless it does assist in digging and tracing the
x ij ij ij
i = 1
important factors related to supplier selection and
... (6) evaluation, as being followed in the Indian chemical
c
x ij and biotechnology organizations.
c=
r ij Most of the constructs used in the study are
( ) ( ) ( )
m
a + x b + x c
2 2 2 individual level constructs. We believe that
x ij ij ij
i = 1 .(7) organizations do not communicate; people do.
However, in order to translate these behavioural
Step 3: In this step, weighted normalized fuzzy constructs at the dyadic, inter-organizational level
decision matrix is determined by using equations (8), (which is the unit of analysis), we presume that
(9) and (10). individual views on the issues surveyed will be a
a = w ra
v ij function of their organizational roles. Also,
j ij individuals who occupy strategic positions in their
..(8)
organizations would be more knowledgeable about
vb b
ij = w j r ij the strategic aspects of inter-organizational exchange
..(9) relationships. Thus, we tapped responses to the
questionnaire from key informants of purchase and
..(10) material management departments of these
Step 4: In this step, overall ratings of beneficial and organizations. The usage of data collected this way is
non-beneficial criteria for each alternative is standard practice in strategic management research
calculated. For beneficial criteria, the overall ratings [33, 34].
of an alternative for lower, middle and upper values 5.2 Factors for development of model
of the triangular function are computed by using Respondents were asked to rate the importance of
equations (11), (12) and (13). factors considered for supplier evaluation, using a
= . (11) five-point scale with endpoints labeled least
important (=1) and most important (=5). Results
= (12) for the importance scores of the factors are provided
in figure 2.
= (13)
It can be seen from figure 1 that the personnel from
For non-beneficial criteria, the overall ratings of an the organizations surveyed consider service, quality,
alternative for lower, middle and upper values of the on-time delivery, pollution causing factor and total
triangular function are computed by using equations cost as much more important factors than other
(14), (15) and (16). factors while selecting a supplier. We, therefore,
= (14) consider these five factors for ranking the alternative
suppliers. These five factors are defined below.
= (15) Total Cost is the money that buyer has to
= (16) pay for delivery of material from buyers
site to his organization
Step 5: Overall performance index (Si) is determined Quality will take care of consistency in
for each alternative. For this, defuzzified values of quality of raw material being supplied and
the overall ratings for beneficial and non-beneficial percentage rejection of incoming material
criteria for each alternative are calculated using defected by the buyers quality and
equation (17) given below. production departments.
S ( s ,s ) = (
1 +a a 2
) ( )( +c c
) Service factor has been evaluated
2 2
+b b
i
+
i
i +
3 si si si si
si si considering supplier technical ability,
service provided by supplier in response to
(17) the order placed, at the time of any
Step 6: The overall performance indices are arranged complaint, when refund/replacement of
in the descending order and ranked best to worst, i.e., defective material is claimed and in case of
the alternative with the highest overall performance emergency.
index is the best choice. On-time delivery ensures that the required
5. Illustrative example raw material is delivered at the production
5.1 Data collection site before it is to be consumed. This ensures
that there are no losses because of shortage substances or energy, such as noise, heat or
of raw material. light. Pollutants, the components of
Pollution causing factors consider pollution, can be either foreign
contaminants into the natural environment substances/energies or naturally occurring
that cause adverse change. Pollution can contaminants.
take the form of chemical
Figure 2: Importance scores of factors used to select and evaluate the supplier
5.3 Implementation of fuzzy MOORA method (see annexure B). Using triangular fuzzy
AHP method was employed to determine the weights membership function, the fuzzy decision matrix for
of these five criteria as = 0.364, = 0.260, supplier selection was developed, as shown in table
2.
0.099, 0.209, 0.067 respectively
Table 2: Fuzzy supplier selection decision matrix
The fuzzy decision matrix is then normalized using Normalization is done to make the members of
equations (5), (6) and (7). This normalized fuzzy decision matrix dimensionless and comparable.
decision matrix is shown in table 3 below.
Table 3: Normalized fuzzy decision matrix
After normalization, weighted normalized fuzzy causing factors are the non-beneficial attributes as
decision matrix is obtained after multiplying the lower values for these criteria are desired. Now,
normalized fuzzy criteria values with the using equations (11)-(13), the overall ratings of
corresponding values of crisp criteria weights using beneficial criteria are calculated and using equations
equations (8)-(10). Among the five supplier selection (14)-(16), the overall ratings of non-beneficial criteria
criteria taken, quality and service are beneficial are calculated. These overall ratings for both
attributes as higher values for these criteria are beneficial and non-beneficial attributes for the
desired, whereas cost, on-time delivery and pollution alternative suppliers are shown in table 4.
Table 4: Weighted normalized fuzzy decision matrix
Now, the vertex method is used to defuzzify the alternative suppliers. It is observed that highest value
overall ratings for both beneficial and non-beneficial of overall performance index occurs for supplier 1,
attributes for the alternative suppliers so as to derive signifying that supplier 1 is the best choice. Supplier
the values of overall performance index using 4, having minimum value of overall performance
equation (17). Table 5 shows the ranking of index will be the worst choice.