Middle-East Journal of Scientific Research 8 (3): 699-706, 2011

ISSN 1990-9233
© IDOSI Publications, 2011

A Fuzzy TOPSIS-Based Approach to Maintenance Strategy Selection: A Case Study

1
Mansour Momeni, 2Mohammad Reza Fathi, 2Mohammad Karimi Zarchi and 3Sirous Azizollahi

Department of Management, University of Tehran, Tehran, Iran

1

Candidate of Industrial Management, University of Tehran, Tehran, Iran

2

3
Candidate of business Management, University of Tehran, Tehran, Iran

Abstract: In this paper, the selection of maintenance strategies in Electerofan Company is studied. The
evaluation of maintenance strategies for each piece of equipment is a multiple criteria decision-making (MCDM)
problem. To deal with the uncertain judgment of decision makers, a fuzzy TOPSIS method is applied as an
evaluation tool, where uncertain and imprecise judgments of decision makers are translated into fuzzy numbers.
A specific example of selection of maintenance strategies in this company with the application of the proposed
fuzzy TOPSIS method is given, showing that the Preventive maintenance strategy is the most suitable for
equipment.

Key words: Maintenance strategies Multiple criteria decision-making Fuzzy logic TOPSIS

INTRODUCTION application of the proposed evaluation method for the

selection of maintenance strategies in Electrofan
The importance of maintenance function has Company and conclusion finally.
increased due to its role in keeping and improving the
availability, product quality, safety requirements and plant Alternative Maintenance Strategies: Five alternative
cost-effectiveness levels. Maintenance costs constitute maintenance policies are evaluated in this case study
an important part of the operating budget of according to Bevilacqua et al. [1]. They are the following
manufacturing firms. One of the main expenditure items for as:
these firms is maintenance cost which can reach 15–70%
of production costs, varying according to the type of Preventive maintenance. Preventive maintenance is
industry [1]. On the other hand, maintenance plays an based on component reliability characteristics. This
important role in keeping availability and reliability levels, data makes it possible to analyze the behavior of the
product quality and safety requirements. Unfortunately, element in question and allows the maintenance
unlike production and manufacturing problems which engineer to define a periodic maintenance program
have received tremendous interest from researchers and for the machine. Preventive maintenance is effective
practitioners, maintenance received little attention in the in overcoming the problems associated with the
past. This is one of the reasons that results in low wearing of components. It is evident that, after a
maintenance efficiency in industry at present but today, check, it is not always necessary to substitute the
research in this area is on the rise and research on component: maintenance is often sufficient.
maintenance represents an opportunity for making Opportunistic maintenance. The possibility of using
significant contribution by academics. This paper is opportunistic maintenance is determined by the
organized as follows. Section 2 describes the possible nearness or concurrence of control or substitution
alternative maintenance strategies in this study. In times for different components on the same machine
Section 3, the comparing criteria for the selection of or plant. This type of maintenance can lead to the
maintenance strategies are presented. Section 4 whole plant being shut down at set times to perform
introduces fuzzy set theory. Section 5 describes the basic all relevant maintenance interventions at the same
concept of fuzzy TOPSIS. Section 6 describes the time.

Corresponding Author: Mohammad Reza Fathi, University of Tehran, Tehran, Iran.

699
 Middle-East J. Sci. Res., 8 (3): 699-706, 2011

Corrective maintenance. The main feature of Access to Equipment and Technology: One of the
corrective maintenance is that actions are only effective factors in selecting a strategy is access to
performed when a machine breaks down. There are equipment and technology needed to implement.
no interventions until a failure has occurred.
Predictive maintenance. Unlike the condition-based Technique Reliability: Still under development,
maintenance policy, in predictive maintenance the condition-based maintenance and predictive maintenance
acquired controlled parameters data are analyzed to may be inapplicable for some complicated production
find a possible temporal trend. This makes it possible facilities.
to predict when the controlled quantity value will
reach or exceed the threshold values. Safety: Safety levels required are often high in many
Condition-based maintenance. A requisite for the manufacturing factories, especially in industry companies.
application of condition-based maintenance is the The relevant factors describing the Safety are:
availability of a set of measurements and data
acquisition systems to monitor the machine Facilities: For example, the sudden breakdown of pump
performance in real time. The continuous survey of can result in serious damage in this plant.
working conditions can easily and clearly point out
an abnormal situation (e.g. the exceeding of a Personnel: The failure of many machines can lead to
controlled parameter threshold level), allowing the serious damage of personnel on site.
process administrator to punctually perform the
necessary controls and, if necessary, stop the Environment: The failure of equipment with poisonous
machine before a failure can occur. liquid or gas can damage the environment.

Comparing Criteria: Different manufacturing companies Cost: Different maintenance strategies have different
may have different maintenance goals.Accoding to Wang expenditure of hardware, software and Specialist
et al. [2], Bevilacqua et al. [1] and expert’s opinion; these employee.
goals can be divided into four aspects analyzed as
follows: Hardware: For condition-based maintenance and
predictive maintenance, a number of sensors and some
Added-value: A good maintenance program can induce computers are indispensable.
added-value, including low inventories of spare parts,
small production waste and Product quality. Specialist Employee Required: To implement each
strategy, the specialist force is required that the number
Product Quality: Equipment failure can affect the quality and type according to the chosen strategy will change.
of products which is produced. In fact, when the machin These forces can be absorbed from outside organizations,
is in better condition and with greater reliability work, the or individuals within the organization with the training
quality of products will increase. they provide.

Production Waste: The failure of more important Software: Software is needed for analyzing measured
machines in the production line often leads to higher parameters data when using condition-based maintenance
production loss cost. Selecting a suitable maintenance and predictive maintenance strategies.
strategy for such machines may reduce production waste.
Fuzzy Sets and Fuzzy Numbers: Fuzzy set theory, which
Spare Parts Inventories: Generally, corrective was introduced by Zadeh [3] to deal with problems in
maintenance need more spare parts than other which a source of vagueness is involved, has been
maintenance strategies. utilized for incorporating imprecise data into the decision
framework. A fuzzy set A can be defined mathematically
Applicability: Applicability refers to the appropriate by a membership function µA(X)  , which assigns each
conditions for implementing the strategy. element x in the universe of discourse X a real number in

700
 Middle-East J. Sci. Res., 8 (3): 699-706, 2011

decision problems including personnel selection (e.g.

rating for creativity, personality, leadership, etc.). The
primary reason for using triangular fuzzy numbers can be
stated as their intuitive and computational-efficient
representation [5].
Fig 1. A triangular fuzzy number A A linguistic variable is defined as a variable whose
values are not numbers, but words or sentences in natural
the interval [0,1]. A triangular fuzzy number A can be or artificial language. The concept of a linguistic variable
defined by a triplet (a, b, c) as illustrated in Fig 1. appears as a useful means for providing approximate
The membership function µA(X)  is defined as characterization of phenomena that are too complex or ill
defined to be described in conventional quantitative
x − a terms [6].
b − a a ≤ x ≤ b

 x−c The Fuzzy Topsis Method: This study uses this method
µA(x)=  b≤ x≤c
b−c (1) to select the best maintenance strategy. TOPSIS views a
0 oterwise MADM problem with m alternatives as a geometric

 system with m points in the n-dimensional space. The
method is based on the concept that the chosen
Basic arithmetic operations on triangular fuzzy alternative should have the shortest distance from the
numbers A1 (a1,b1,c1), where a 1 b 1 c 1 and A 2 positive-ideal solution and the longest distance from the
(a2,b2,c2), where a2 b2 c2,can be shown as follows: negative-ideal solution. TOPSIS defines an index called
similarity to the positive-ideal solution and the
Additions: remoteness from the negative-ideal solution. Then the
A1 A2 = (a1 + a2, b1 – b2, c1 + c2) (2) method chooses an alternative with the maximum
similarity to the positive-ideal solution [7]. It is often
Subtraction: difficult for a decision-maker to assign a precise
A1 A2 = (a1 – c2, b1 - b2, c1 – a2) (3) performance rating to an alternative for the attributes
under consideration. The merit of using a fuzzy approach
Multiplication: if k is a scalar is to assign the relative importance of attributes using
fuzzy numbers instead of precise numbers. This section
(ka1 , kb1 , kc1), k > 0 extends the TOPSIS to the fuzzy environment [8]. This
k ⊗ A1 =

(kc1, kb1, ka1), k < 0
method is particularly suitable for solving the group
decision-making problem under fuzzy environment. We
A1 A2 (a1a2, b1b2, c1c2) briefly review the rationale of fuzzy theory before the
development of fuzzy TOPSIS. The mathematics concept
if a1 0, a2 0 (4) borrowed from Ashtiani [7, 9, 10].

a b c  Step 1: Determine the weighting of evaluation criteria.

Division: A1 ∅ A 2 ≈  1 , 1 , 1  ,
 c2 b2 a2 
A systematic approach to extend the TOPSIS is
proposed to selecting best maintenance strategy under a
if a1 0, a2 0 (5) fuzzy environment in this section. In this paper the
importance weights of various criteria and the ratings of
Although multiplication and division operations on qualitative criteria are considered as linguistic variables
triangular fuzzy numbers do not necessarily yield a (Table 1) [11].
triangular fuzzy number, triangular fuzzy number
approximations can be used for many practical Step 2: Construct the fuzzy decision matrix and choose
applications [4]. Triangular fuzzy numbers are appropriate the appropriate linguistic variables for the alternatives
for quantifying the vague information about most with respect to criteria.

701
 Middle-East J. Sci. Res., 8 (3): 699-706, 2011

Table 1: Linguistic scales for the importance of each criterion Step 4: Determine the fuzzy positive-ideal solution (FPIS)
Linguistic variable Corresponding triangular fuzzy number and fuzzy negative-ideal solution (FNIS).
Very low (VL) (0.0, 0.1, 0.3) According to the weighted normalized fuzzy decision
Low (L) (0.1, 0.3, 0.5) matrix, we know that the elements V  are normalized
ij
Medium (M) (0.3, 0.5, 0.7) positive TFNs and their ranges belong to the closed
High (H) (0.5, 0.7, 0.9) interval [0, 1]. Then, we can define the FPIS A+ and FNIS
Very high (VH) (0.7, 0.9, 1.0) A– as following formula:

C1 C 2  C N A+ = (V1+ , V2+ ,...,Vn+ ) (10)

A1  x 11 x 12  x 1n 
 x n1 
A 2  x 21 x 22 
A− = (V1− , V2− ,...,Vn− ) (11)

D=      
  
 x m1 x m1  x mn 
AM
Where V + =(1,1,1) and Vj− =(0,0,0) j=1,2,…,n
j
i = 1,2,...,m; j=1,2,...,n
Step 5: Calculate the distance of each alternative from
1
xij= ( xij + xij2 + ...xijk ) FPIS and FNIS
2 (6)
The distances ( d + and d − ) of each alternative A+
Where xijk is the rating of alternative Ai with respect to i i

criterion Cj evaluated by K expert and from and A- can be currently calculated.

xijk = (aijk , bijk , cijk )
n
di+ = ∑ j=1d(v ij,V j+ ),
Step 3: Normalize the fuzzy decision matrix. i = 1,2,.....,m j = 1,2,.....n (12)
The normalized fuzzy decision matrix denoted by R n
is shown as following formula: di+ = ∑ j=1d(v ij,V j- ), (13)
i = 1,2,.....,m j = 1,2,.....n
  r 
R= ,
 ij  m×n
Step 6: Obtain the closeness coefficient (CC) and rank
the order of alternatives
i = 1,2,...,m; j=1,2,...,n (7)
The CCi is defined to determine the ranking order of
Then the normalization process can be performed by
all alternatives once the d + and d − of each alternative
following formula: i i
have been calculated. Calculate similarities to ideal
a ij b ij cij + solution. This step solves the similarities to an ideal
Where rij = ( , , ) c = max i cij
cj cj cj j solution by formula:

di−
The normalized rij are still triangular fuzzy numbers. CCi =
di+ + di− (14)
For trapezoidal fuzzy numbers, the normalization process
can be conducted in the same way. The weighted fuzzy i = 1,2,....m
normalized decision matrix is shown as following matrix
V :
According to the CCi, we can determine the ranking
order of all alternatives and select the best one from
 vij ]m× n ,
v=[ among a set of feasible alternatives.
In the last years, some fuzzy TOPSIS methods were
i = 1,2,...,m; j=1,2,...,n (8) developed in the different applied field. Lin and Chang
[12] adopted fuzzy TOPSIS for order selection and pricing
v=
ij rij ⊗ w j (9) of manufacturer (supplier) with make-to-order basis when

702
 Middle-East J. Sci. Res., 8 (3): 699-706, 2011

Fig. 2: Research framework

orders exceed production capacity. Chen and Tsao [13] TOPSIS approach for evaluating dynamically the service
are to extend the TOPSIS method based on interval- quality of three hotels of an important corporation in Gran
valued fuzzy sets in decision analysis. Ashtiani et al. [9] Canaria island via surveys. Wang and Elhag [20]
used interval-valued fuzzy TOPSIS method is aiming at proposed a fuzzy TOPSIS method based on alpha level
solving MCDM problems in which the weights of criteria sets and presents a non-linear programming solution
are unequal, using interval-valued fuzzy sets concepts. procedure. Chen et al. [11] applied fuzzy TOPSIS
Mahdavi et al. [14] designed a model of TOPSIS for the approach to deal with the supplier selection problem in
fuzzy environment with the introduction of appropriate supply chain system.
negations for obtaining ideal solutions. Büyüközkan et al.
[10] identified the strategic main and sub-criteria of Case Study: The Electrofan Company is a large, well
alliance partner selection that companies consider the known manufacturer that Working in LPG and CNG
most important through Fuzzy AHP and Fuzzy TOPSIS industry in Iran. In this research, 12 experts and managers
model and achieved the final partner-ranking results. Abo- were invited to survey five alternatives using the research
Sinna et al. [15] focused on multi-objective large-scale framework shown in Fig 2. This research framework
non-linear programming problems with block angular includes 11 evaluation criteria, such as Product Quality
structure and extended the technique for order preference (C 1), Production waste (C2), Spare part inventories (C 3),
by similarity ideal solution to solve them. Wang and Access to Equipment and Technology (C4), Technique
Chang [7] applied fuzzy TOPSIS to help the Air Force reliability (C5), Facilities (C6), Personnel (C7), Environment
Academy in Taiwan choose optimal initial training aircraft (C8), Hardware (C9), Specialist employee required (C10) and
in a fuzzy environment. Li [16] developed a compromise Software (C11). In addition, there are five alternatives.
ratio (CR) methodology for fuzzy multi-attribute group After the construction of the hierarchy the different
decision making (FMAGDM), which is an important part priority weights of each criteria, attributes and alternatives
of decision support system. Wang and Lee [17] are calculated using the fuzzy TOPSIS approach. The
generalized TOPSIS to fuzzy multiple-criteria group comparison of the importance or preference of one
decision-making (FMCGDM) in a fuzzy environment. criterion, attribute or alternative over another can be done
Kahraman et al. [18] proposed a fuzzy hierarchical TOPSIS with the help of the questionnaire. The method of
model for the multi-criteria evaluation of the industrial calculating priority weights of the different decision
robotic systems. Ben ´tez et al. [19] presented a fuzzy alternatives is discussed following part.

703
 Middle-East J. Sci. Res., 8 (3): 699-706, 2011

Table 2: Weights of each criterion BNPw1= [(Uw1-Lw1) + (Mw1-Lw1)] /3 +Lw1

BNP Rank = [(0.93-0.58)+(0.78-0.58)]/3+0.58=0.763 (15)
C1 (0.58, 0.78, 0.93) 0.763 2
C2 (0.55, 0.75, 0.90) 0.734 4 Then, the weights for the remaining dimensions can
C3 (0.32, 0.51, 0.70) 0.510 11 be found as shown in Table 2. Table 2 shows the relative
C4 (0.64, 0.83, 0.95) 0.806 1 weight of criteria, which obtained by fuzzy TOPSIS
C5 (0.54, 0.76, 0.83) 0.710 7 method. The weights for each criterion are: C1 (0.763), C2
C6 (0.38, 0.62, 0.80) 0.600 8 (0.734), C3 (0.710), C4 (0.806), C5 (0.710), C6 (0.600), C7
C7 (0.54, 0.72, 0.89) 0.716 5 (0.716), C8 (0.530), C9 (0.713), C10 (0.753) and C11 (0.536).
C8 (0.32, 0.53, 0.74) 0.530 10 From the fuzzy TOPSIS results, we can understand the
C9 (0.53, 0.73, 0.88) 0.713 6 first two important factors for selecting maintenance
C10 (0.58, 0.79, 0.89) 0.753 3 strategy are C4 (0.806) and C1 (0.763). Moreover, the less
C11 (0.34, 0.55, 0.72) 0.536 9 important factor is C3 (0.510).

Table 3: Linguistic scales for the rating of each cluster policy

Step 2: Estimating the performance
Linguistic variable Corresponding triangular fuzzy number
This paper focus on determining the best
Very poor (VP) (0, 1, 3)
maintenance strategy; so, we assume that
Poor (P) (1, 3, 5)
questionnaire have collected completely and will
Fair (F) (3, 5, 7)
start with building dataset that are collected. The
Good (G) (5, 7, 9)
evaluators have their own range for the linguistic
Very good (VG) (7, 9, 10)
variables employed in this study according to their
Very poor (VP) (0, 1, 3)
subjective judgments [9].
For each evaluator with the same importance, this
Step 1: Determine the linguistic weighting of each criteria study employs the method of average value to
integrate the fuzzy/vague judgment values of different
We adopt fuzzy TOPSIS method to evaluate the evaluators regarding the same evaluation dimensions. The
weights of different criteria for selecting the most efficient evaluators then adopted linguistic terms (Table 3),
maintenance strategy. Following the construction of fuzzy including “very poor”, “poor”, “fair”, “good” and “very
TOPSIS model, it is extremely important that experts fill the good” to express their opinions about the rating of every
judgment matrix. From the viewpoint of expert validity, the person, based on the fuzzy data of the four person listed
buildup of most of the operationalizations was based on in Table 4.
the literature that caused them to have expert validity.
This research applies the COA method to compute Step 3: Normalize the fuzzy decision matrix
the BNP value of the fuzzy weights of each dimension:
To take the BNP value of the weight of C1 as an example, Using Eq. (7), we can normalize the fuzzy decision
the calculation process is as follows: matrix as Table 5.

Table 4: Subjective cognition results of evaluators towards the five levels of linguistic variables
A1 A2 A3 A4 A5
C1 (4.21, 6.21, 8.21) (3.56, 6.45, 7.46) (2.92, 4.84, 6.84) (3.00, 5.00, 7.00) (2.00, 4.00, 6.00)
C2 (5.17, 7.17, 9.00) (2.45, 4.45, 6.45) (1.80, 3.65, 5.65) (3.48, 5.76, 7.76) (4.02, 6.00, 7.96)
C3 (5.11, 7.11, 9.16) (4.52, 6.50, 8.56) (3.50, 5.50, 7.50) (3.86, 5.86, 7.86) (3.18, 5.18, 7.18)
C4 (4.68, 6.68, 8.56) (4.65, 6.65, 8.65) (4.00, 6.00, 8.00) (2.06, 4.00, 6.00) (2.19, 4.19, 6.19)
C5 (4.33, 6.35, 8.35) (4.11, 6.00, 7.89) (3.86, 5.86, 7.86) (2.44, 4.46, 6.48) (2.53, 4.53, 6.53)
C6 (4.23, 6.32, 8.46) (3.00, 5.00, 7.00) (3.85, 5.85, 7.64) (4.50, 6.50, 8.50) (3.17, 5.17, 7.17)
C7 (4.46, 6.52, 8.58) (2.08, 4.00, 6.00) (2.45, 4.45, 6.45) (3.50, 5.50, 7.50) (1.75, 3.50, 5.50)
C8 (4.33, 6.33, 8.33) (1.70, 3.52, 5.52) (1.67, 3.56, 5.56) (1.67, 3.50, 5.50) (4.08, 6.00, 7.83)
C9 (2.36, 4.42, 6.42) (4.50, 6.50, 8.42) (2.67, 4.67, 6.67) (3.00, 5.00, 7.00) (3.24, 5.24, 7.38)
C10 (2.46, 4.38, 6.38) (4.86, 6.84, 8.78) (2.67, 4.67, 6.67) (3.50, 5.50, 7.50) (3.20, 5.20, 7.20)
C11 (1.67, 3.56, 5.56) (5.17, 7.17, 9.00) (4.33, 6.33, 8.25) (4.00, 6.00, 8.00) (1.67, 3.56, 5.56)

704
 Middle-East J. Sci. Res., 8 (3): 699-706, 2011

Table 5: Normalized fuzzy decision matrix

A1 A2 A3 A4 A5
C1 (0.46, 0.68, 0.90) (0.39, 0.70, 0.81) (0.32, 0.53, 0.75) (0.33, 0.55, 0.76) (0.22, 0.44, 0.66)
C2 (0.56, 0.78, 0.98) (0.27, 0.49, 0.70) (0.20, 0.40, 0.62) (0.42, 0.63, 0.85) (0.44, 0.66, 0.87)
C3 (0.56, 0.78, 1.00) (0.49, 0.71, 0.93) (0.33, 0.60, 0.82) (0.42, 0.64, 0.86) (0.35, 0.57, 0.78)
C4 (0.51, 0.73, 0.93) (0.51, 0.73, 0.94) (0.44, 0.66, 0.87) (0.22, 0.44, 0.66) (0.24, 0.46, 0.68)
C5 (0.47, 0.69, 0.91) (0.45, 0.66, 0.86) (0.42, 0.64, 0.86) (0.27, 0.49, 0.71) (0.28, 0.49, 0.71)
C6 (0.46, 0.69, 0.92) (0.33, 0.55, 0.76) (0.42, 0.64, 0.83) (0.49, 0.71, 0.93) (0.35, 0.56, 0.78)
C7 (0.49, 0.71, 0.94) (0.23, 0.44, 0.66) (0.27, 0.49, 0.70) (0.38, 0.60, 0.82) (0.19, 0.38, 0.60)
C8 (0.47, 0.69, 0.91) (0.19, 0.38, 0.60) (0.18, 0.39, 0.61) (0.18, 0.38, 0.60) (0.45, 0.66, 0.85)
C9 (0.26, 0.48, 0.70) (0.49, 0.71, 0.92) (0.29, 0.51, 0.73) (0.33, 0.55, 0.76) (0.35, 0.57, 0.81)
C10 (0.27, 0.48, 0.70) (0.53, 0.75, 0.96) (0.29, 0.51, 0.73) (0.38, 0.60, 0.82) (0.35, 0.57, 0.79)
C11 (0.18, 0.39, 0.61) (0.56, 0.78, 0.98) (0.47, 0.69, 0.90) (0.44, 0.66, 0.87) (0.18, 0.39, 0.61)

Table 6: Weighted normalized fuzzy decision matrix

A1 A2 A3 A4 A5
C1 (0.27, 0.53, 0.83) (0.23, 0.55, 0.76) (0.18, 0.41, 0.69) (0.19, 0.43, 0.71) (0.13, 0.34, 0.61)
C2 (0.31, 0.59, 0.88) (0.19, 0.36, 0.63) (0.11, 0.30, 0.56) (0.23, 0.47, 0.76) (0.24, 0.49, 0.78)
C3 (0.18, 0.40, 0.70) (0.16, 0.36, 0.65) (0.12, 0.31, 0.57) (0.13, 0.33, 0.60) (0.11, 0.29, 0.55)
C4 (0.33, 0.61, 0.89) (0.32, 0.60, 0.90) (0.28, 0.54, 0.83) (0.14, 0.36, 0.62) (0.15, 0.38, 0.64)
C5 (0.26, 0.53, 0.77) (0.24, 0.50, 0.72) (0.23, 0.49, 0.72) (0.14, 0.37, 0.59) (0.15, 0.38, 0.60)
C6 (0.18, 0.43, 0.74) (0.20, 0.34, 0.61) (0.16, 0.40, 0.67) (0.19, 0.44, 0.74) (0.13, 0.35, 0.63)
C7 (0.26, 0.51, 0.83) (0.12, 0.31, 0.58) (0.14, 0.35, 0.63) (0.21, 0.43, 0.73) (0.10, 0.28, 0.53)
C8 (0.15, 0.37, 0.67) (0.60, 0.20, 0.45) (0.60, 0.21, 0.45) (0.60, 0.20, 0.44) (0.14, 0.35, 0.63)
C9 (0.14, 0.35, 0.62) (0.26, 0.52, 0.81) (0.15, 0.37, 0.64) (0.17, 0.40, 0.67) (0.19, 0.42, 0.71)
C10 (0.16, 0.38, 0.62) (0.31, 0.59, 0.85) (0.17, 0.40, 0.65) (0.22, 0.47, 0.73) (0.20, 0.45, 0.70)
C11 (0.06, 0.21, 0.44) (0.19, 0.43, 0.71) (0.17, 0.38, 0.65) (0.15, 0.36, 0.63) (0.06, 0.21, 0.44)

Step 4: Establish the weighted normalized fuzzy decision Table 7: Closeness coefficients and ranking

matrix d i+ d i− Cci Rank

A1 11.02 9.74 0.47 1

The forth step in the analysis is to find the weighted A2 11.41 9.23 0.45 2
fuzzy decision matrix and the resulting A3 12.05 8.46 0.41 4
Fuzzy weighted decision matrix is shown as Table 6. A4 12.06 8.56 0.42 3
A5 12.27 8.30 0.40 5

Step 5: Determine the fuzzy positive and fuzzy negative-

ideal reference points Once the distances of cluster policy from FPIS and
FNIS are determined, the closeness coefficient can be
Then we can define the fuzzy positive-ideal solution obtained with Eq. (14). The index CC1 of first alternative is
(FPIS) and the fuzzy negative-ideal solution (FNIS) as: calculated as:
A+and A-. This is the fifth step of the fuzzy TOPSIS
analysis. di+ = 11.02 di− = 9.74
A+ = [ (1,1,1)]
A- = [ (0,0,0)] From the alternative evaluation results in Table 7, the
best person is P2.
Step 6: Ranking the alternatives
9.74
In order to calculate the closeness coefficients of =CCi = 0.47
11.02 + 9.74
each of the alternatives d + and d − calculation is used as
i i
an example as follows. CC1 > CC2 > CC4 > CC3 > CC5

705
 Middle-East J. Sci. Res., 8 (3): 699-706, 2011

CONCLUSION 10. Buyukozkan, G., O. Feyziog lu, and E. Nebol, 2007.

Selection of the strategic alliance partner in logistics
This paper aims to evaluate different maintenance value chain. International J. Production Economics.
strategies for different equipment. The selection of 113(1): 148-158.
maintenance strategies is a typical multiple criteria 11. Chen, C.T., C.T. Lin and S.F. Huang, 2006. A fuzzy
decision-making (MCDM) problem. Considering the approach for supplier evaluation and selection in
imprecise judgments of decision makers, the fuzzy TOPSIS supply chain management. International J.
is used for the evaluation of different maintenance Production Economics, 102(2): 289-301.
strategies. For making uniform consensus of the decision 12. Lin, H.T. and W.L. Chang, 2008. Order selection and
makers, we converted all pair-wise comparisons into pricing methods using flexible quantity and fuzzy
triangular fuzzy numbers to adjust fuzzy rating and fuzzy approach for buyer evaluation. European J.
attribute weight and used fuzzy operators to get to select Operational Res., 187(2): 415-428.
the best alternative. 13. Chen, T.Y. and C.Y. Tsao, 2008. The interval-valued
Finally, observing all these results, Fuzzy TOPSIS fuzzy TOPSIS method and experimental analysis.
approach proposes preventive maintenance strategy (A1) Fuzzy Sets and Sys., 159(11): 1410-1428.
as the best choice. 14. Mahdavi, I., N. Mahdavi-Amiri, A. Heidarzade and
R. Nourifar, 2008. Designing a model of fuzzy TOPSIS
REFERENCES in multiple criteria decision making. Applied
Mathematics and Computation. doi:10. 1016/j. amc.
1. Bevilacqua, M. and M. Braglia, 2000. The analytic 2008. 05.047.
hierarchy process applied to maintenance strategy 15. Abo-Sinna, M.A., A.H. Amer and A.S. Ibrahim, 2008.
Extensions of TOPSIS for large scale multi-objective
selection. Reliability Engineering and System Safety.
non-linear programming problems with block
70: 71-83.
angular structure. Applied Mathematical
2. Wang, L., J. Chu and J. Wu, 2007. Selection of
Modelling, 32(3): 292-302.
optimum maintenance strategies based on a fuzzy
16. Li, D.F., 2007. Compromise ratio method for fuzzy
analytic hierarchy process. Int. J. Production
multi-attribute group decision making. Applied Soft
Economics. 107, 151-163.
Computing, 7(3): 807-817.
3. Zadeh, L.A., 1975. The concept of a linguistic
17. Wang, Y.J. and H.S. Lee, 2007. Generalizing TOPSIS
variable and its application to approximate
for fuzzy multiple-criteria group decision-making.
reasoning-I. Information Sci., 8(3): 199-249.
Computers and Mathematics with Applications,
4. Kaufmann, A. and M.M. Gupta, 1988. Fuzzy
53(11): 1762-1772.
mathematical models in engineering and management 18. Kahraman, C., S. Cevik, N.Y. Ates and M. Gulbay,
science. Amsterdam: North-Holland. 2007. Fuzzy multi-criteria evaluation of industrial
5. Karsak, E.E., 2002. Distance-based fuzzy MCDM robotic systems. Computers and Industrial
approach for evaluating flexible manufacturing Engineering. 52(4): 414-433.
system alternatives. International J. Production Res., 19. Ben ´tez, J.M., J.C. Mart ´n and C. Roman, 2007.
40(13): 3167-3181. Using fuzzy number for measuring quality of
6. Zadeh, L.A., 1965. Fuzzy sets. Information and service in the hotel industry. Tourism Manage.,
Control., 8(3): 338-353. 28(2): 544-555.
7. Wang, T.C. and T.H. Chang, 2007. Application of 20. Wang, Y.M. and T.M.S. Elhag, 2006. Fuzzy TOPSIS
TOPSIS in evaluating initial training aircraft under a method based on alpha level sets with an application
fuzzy environment. Expert Systems with to bridge risk assessment. Expert Systems with
Applications, 33: 870-880. Applications, 31(2): 309-319.
8. Yang, T. and C.C. Hung, 2007. Multiple-attribute 21. Hsieh, T.Y., S.T. Lu and G.H. Tzeng, 2004. Fuzzy
decision making methods for plant layout design MCDM approach for planning and design tenders
problem. Robotics and Computer-Integrated selection in public office buildings. International J.
Manufacturing, 23(1): 126-137. Project Manage., 22(7): 573-584.
9. Ashtiani, B., F. Haghighirad, A. Makui and
G.A. Montazer, 2008. Extension of fuzzy TOPSIS
method based on interval-valued fuzzy sets. Applied
Soft Computing. doi:10.1016/j.asoc.2008.05.005.

706