A Comparison of Multi-Criteria Decision Making Approaches For Maintenance Strategy Selection (A Case Study) 1 PDF

International Journal of Strategic Decision Sciences
Volume 7 • Issue 3 • July-September 2016
A Comparison of Multi-Criteria Decision

Making Approaches for Maintenance
Strategy Selection (A Case Study)
Malek Tajadod, Department of Industrial Engineering, Shahid Bahonar University, Kerman, Iran
Mohammadali Abedini, Department of Industrial Engineering, Iran University of Science and Technology (IUST), Tehran,
Iran
Ali Rategari, School of Innovation, Design and Engineering, Mälardalen University, Eskilstuna, Sweden & Volvo GTO,
Köping, Sweden
Mohammadsadegh Mobin, Department of Industrial Engineering and Engineering Management, Western New England
University, Springfield, MA, USA
ABSTRACT
The growth of world-class manufacturing companies and global competition caused significant
changes in the manufacturing companies operations. These changes have affected maintenance and
made its role even more crucial to stay ahead of the competition. Maintenance strategy selection is
one of the strategic decision-making issues that manufacturing companies in the current competitive
world are facing. In this paper, a comparison between different Multiple Criteria Decision Making
(MCDM) approaches is conducted in a dairy manufacturing factory to rank the maintenance strategies.
The aim is to suggest an appropriate approach for the best selection of the maintenance strategy. The
decision-making elements including evaluation criteria/sub-criteria and problem alternatives, i.e.,
maintenance strategies are determined and a group of experts from the case-study factory are asked
to make their pair-wise comparisons. The pair-wise comparison matrix is constructed by using the
crisp and triangular fuzzy numbers, while the aggregation of individual priorities (AIP) approach is
utilized to aggregate the decision-makers’ judgments. The priority vectors of decision elements are
calculated by Mikhailov’s fuzzy preference programming (FPP) methods and the final weights of the
decision elements are found. Results show that when the effectiveness of one element on the other
elements is higher, it will have greater weights; and therefore, the results from the analytic network
process (ANP) method is completely different from those of the analytic hierarchy process (AHP).
The reason for the differences between the AHP and Fuzzy AHP (FAHP) with the ANP and Fuzzy
ANP (FANP) is that both AHP and FAHP evaluate the criteria only based on the level of importance
and do not consider the interdependencies and interactions among the evaluation elements. In this
research, a predictive maintenance is selected as the most appropriate strategy in the case company
and the preventive strategies outperformed the corrective strategies. The results of this research are
consistent with the results of previous studies found in the literature.
Keywords
Analytical Hierarchy Process (AHP), Analytical Network Process (ANP), Fuzzy Set Theory, Maintenance
Strategy Selection, Multi-Criteria Decision Making
DOI: 10.4018/IJSDS.2016070103
Copyright © 2016, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
51
1. INTRODUCTION
In the past, many companies considered maintenance as an inevitable source of cost. For these
companies, maintenance operations have a corrective function (fix it on failure) and are only executed
in emergency conditions. Today, this form of intervention is not acceptable because of certain critical
elements such as product quality, plant safety, and increased costs of maintenance departments
(Bevilacqua and Braglia, 2000). Furthermore, in the current competitive world, manufacturing firms
attempt to improve their performance in terms of cost, quality, and flexibility, in an effort to compete
with other firms of the global marketplace (Ertugrul Karsak and Tolga, 2001). In the manufacturing
firms, various problems such as manufacturing technology selection, maintenance strategy selection,
machine location, and evaluation of quality function would influence production cost, product quality,
and product delivery-time (Bashiri, Badri, and Hejazi, 2011). As indicated by Mobley (2002), one
third of all of the maintenance costs are wasted as a result of unnecessary maintenance. The use
of inefficient maintenance policies considerably increases the direct maintenance costs (Rastegari
and Bengtsson, 2014). On the other hand, the manufacturing firms are under a great pressure to
continuously reduce their production costs. One of the main expenditure items for these firms is their
maintenance costs, which can comprise up to 15%-70% of the overall production costs according to
the type of industry (Bevilacqua and Braglia, 2000; Rastegari and Bengtsson, 2015).
In recent years, the importance of maintenance strategy selection has been increased due to the
critical role it has for increasing the availability, the safety, and the mean time to failures, improvement
of system reliability and product quality, reducing shutdown time of factory, and preventing
unnecessary investments. Various methodologies have been used for selecting a suitable maintenance
strategy in the literature as presented as follows. Triantaphyllou, et al. (1997) have suggested a method
to address criticality of the criteria related to the problem of maintenance strategy selection. They first
prioritized the maintenance strategies using the AHP and then performed a sensitivity analysis on
different criteria to identify the most important one. The maintenance strategy selection was presented
by Luce (1999), Okumura and Okino (2003) based on different production loss and maintenance
costs incurred by different maintenance strategies. Azadivar and Shu (1999) proposed a method for
selecting an appropriate strategy for each class of systems in a just in time environment exploring
16 characteristic factors that could play a role in the maintenance strategy selection. Using the AHP
method for the maintenance strategy selection was suggested by Bevilacqua and Braglia (2000). Al-
Najjar and Alsyouf (2003) and Sharma, et al. (2005) evaluated the maintenance strategies through the
fuzzy interference system (FIS) and the fuzzy multi-criteria decision-making methods. Mechefske
and Wang (2003) used the fuzzy linguistic variables to evaluate and select the optimum maintenance
strategy and condition monitoring technique. Bertolini and Bevilacqua (2006) presented a combined
model based on the AHP and the goal programming (GP) to identify the best maintenance strategy
for the critical centrifugal pumps in an Italian oil refinery.
Wang, et al. (2007) evaluated different maintenance strategies based on the FAHP. In their study,
a new method is also proposed for obtaining the priorities from the fuzzy pair-wise comparisons
matrix. A combination of the FAHP and TOPSIS methods was utilized by Shyjith et al. (2008) to
select an optimum maintenance strategy for textile industries. Jafari, et al. (2008) applied a fuzzy
Delphi method in simple additive weighting (SAW) for the maintenance strategy selection which
could determine the best maintenance strategy by considering the uncertainty levels and considering
various maintenance criteria and their importance. Arunraj and Maiti (2010) used the combination
of AHP and GP for the maintenance strategy selection based on the risk of equipment failure and
the maintenance cost. They applied this approach in a case study in a benzene extraction unit of a
chemical plant. Cheng and Tsao (2010) suggested the ANP method to select an optimum maintenance
strategy for rolling stock in a railway system.
Bashiri et al. (2011) developed a new interactive method based on the fuzzy linear assignment
method for an efficient maintenance strategy selection which uses the quantitative and qualitative data
52
for ranking the maintenance strategies. Jajimoggala et al. (2011) presented an integrated approach
for the maintenance policy selection based on fuzzy Analytic Network Process (ANP) and the Goal
Programming. Fouladgar et al. (2012) proposed a new fuzzy MCDM method based on the concepts
of COPRAS (Complex Proportional Assessment) and AHP (Analytical Hierarchy Process) to evaluate
the feasible maintenance strategy. In their study, Fuzzy AHP was utilized to calculate the weights of
the evaluation criteria; and COPRAS was applied to rank the alternatives. Zaim, et al. (2012) used
AHP and ANP methods in a case study to select best maintenance strategy in a local newspaper
printing facility in Turkey. We can mention Ferdousmakan at al. (2014) in recent studies in which
they integrated risk management and fuzzy analytical hierarchy process in order to select the optimal
maintenance strategy. They made a risk priority matrix that includes five different risk levels and
four criteria to investigate four possible maintenance strategies.
The main objective of the present study is to determine an appropriate approach for the
maintenance strategy selection. The rest of this paper is organized as follows. The methodology of this
study is introduced in Section 2. Section 3 describes implementing the proposed model in a case study.
The results of this study are also presented in Section 3. Finally, Section 4 will conclude the paper.
2. RESEARCH METHODOLOGY
According to Saaty and Vargas (2006), there are two known ways to analyze cause and effect relations:
one is the use of traditional deductive logic that is initiated with assumptions and precisely deduces
an outcome from them. The other is the holistic approach in which all the factors and criteria that
involved are laid out in advance in a hierarchy or a network system that allows for dependencies. All
possible outcomes are joined together in these structures and then both judgment and logic are used
to estimate the relative influence from which the overall answer is derived. By using this approach
in the real-world problems, we will generally reach to a sound overall outcome (Saaty and Vargas,
2006). If the decision-making elements involved in a problem are independent of each other, these
elements will be laid out in a hierarchy structure and the AHP presented by (Saaty, 1980) will be
used to solve it. In the case of dependence and interaction among the elements, a network of decision-
making elements will be formed and the ANP presented by (Saaty, 2001) will be used.
2.1. AHP Method

The analytic hierarchy process is a well-known technique that decomposes the decision-making
problem into several levels in hierarchical manner with assumption of being unidirectional, and
considering hierarchical relations between the levels (Meade, 1997). The highest level of this hierarchy
structure is the main goal of the decision problem. The subsequent levels are tangible and intangible
criteria and sub-criteria which contribute to the goal. The bottom level includes alternatives that
must be evaluated in terms of the criteria. This process uses pair-wise comparisons to extract weights
of the elements through Saaty’s nine-point scale, and finally, the overall weights are calculated for
evaluation of the bottom level. This method also calculates consistency ratios (CR) to verify coherency
of the judgments, which must be about 0.10 or less to be acceptable. For studying the mathematics
foundations of AHP, the interested readers are referred to (Saaty, 1990; Saaty, 2001). The AHP
is one the most practical multi-criteria decision-making methods. The reason for the widespread
application of the AHP is its simplicity, ease of use and high flexibility (Aragonés-Beltrán, et al.,
2010). Nevertheless, the AHP is not an efficient tool for resolving the decision-making issues, which
require modelling the interactions, interdependencies and feedback within different levels of elements
(Saaty, 2001). As a resolution for solving these kinds of issues, Saaty (2001) developed the ANP as
a generalization of the AHP. In fact, if all the criteria and alternatives of a problem are linked to each
other in a network system that allows different types of interdependencies, then an overall approach
like ANP would be necessary for resolving it (Ayağ and Özdemir, 2009). A comprehensive study for
comparing the ANP and AHP has been conducted by (Taslicali and Ercan, 2006).
53
Table 1. Linguistic variables and related fuzzy numbers
Fuzzy Numbers Linguistic Variables Fuzzy Number’s Scale
1 equally important (1,1,3)
3 weakly important (1,3,5)
5 essentially important (3,5,7)
7 very strongly important (5,7,9)
9 absolutely important (7,9,9)
2.2. ANP Method

The ANP approach considers interdependency, feedback and interactions among decision-making
elements, therefore it can be used for modelling complexities of many real-world problems (Saaty,
2004; Saaty and Vargas, 2006). The main steps of the ANP method are stated as the following: Step
1) identifying elements and network clusters and their relationships; Step 2) conducting pair-wise
comparisons on the elements and clusters, and obtaining priorities from the pair-wise comparisons
matrix: the pair-wise comparisons process in ANP, same as AHP is done through the Saaty’s nine-
point scale and the weights are determined by the eigenvector method; Step 3) establishing the
unweighted super matrix: the obtained weights (priorities) from the pair-wise comparisons matrices,
are placed in the unweighted super matrix; Step 4) establishing the weighted super matrix: it means
the sum of each column of the unweighted super matrix must be equal to one; Step 5) establishing
the limit super matrix: raising the weighted super matrix into the limiting powers until the weights
converge and remain stable.
2.3. FAHP and FANP Methods

The data are not crisp in many real-world problems because of incomplete or unspecified information.
Due to the ambiguity in the decision-makers’ (DMs’) judgments, fuzzy pair-wise comparisons are
preferred to their crisp form in these issues. In addition, it is more realistic for the DMs to present
their preferences with the fuzzy judgments instead of the crisp judgments. In this study, the fuzzy
sets theory has been used and the modified approaches are called the fuzzy AHP (FAHP) and the
fuzzy ANP (FANP). In these processes, pair-wise comparisons are performed based on linguistic
variables. The linguistic variables are transformed into fuzzy numbers, and triangular fuzzy numbers
are mostly used for this area (Tseng, 2010). Five triangular fuzzy numbers are considered with their
corresponding membership function adopted from Mon, et al. (1994), Mobin et al. (2015), and
Saeedpoor et al. (2015). The five fuzzy numbers are presented in Table 1.
Determining the CRs is necessary to identify the consistent matrices. Many methods are presented
to determine the priorities from the matrixes of pair-wise comparison. Due to the advantages of the
method proposed by Mikhailov (2000); fuzzy preferences programming (FPP), this method was
used to calculate the priority vector from the pair-wise comparisons matrices. During the conducted
reviews, it has been specified that this method: 1) Suggests an appropriate index for calculating the
inconsistency; 2) Obtains priorities from judgments of the fuzzy pair-wise comparisons with no need
to create a fuzzy comparison matrix from the reversible elements; 3) Uses a Max-Min optimization
approach; 4) Extracts priorities in a crisp manner without any additional ranking process; 5) Is easily
used in group decision-making problems with good accuracy and strong ranking protection; and 6)
54
Acts better than other methods when inconsistency increases. These methods are described in the
following sub-sections in details.
2.4. Fuzzy Preferences Programming

The fuzzy preferences programming presented by Mikhailov (2000), shows the process of determining
the weight as the interaction among the fuzzy sets, which describes human’s uncertain preference
and determines the crisp priority vector by maximizing the degree of decision-makers’ satisfaction.
In this method the fuzzy numbers are converted to interval numbers through the concept of α-cut to
perform the calculations. In fact, if a triangular fuzzy number (aij = 1ij, mij, uij) indicates the importance
(or influence) of element i towards j in the fuzzy pair-wise comparisons matrix, then, it’s α -cut will
be equal to 1:
= l α , u α  = (m − l ) α + l , − (u − m ) α + u 
∀α ∈ 0, 1 M (1)
a  ij ij   ij ij ij ij ij ij 
where, α represents the confidence level of the decision-makers.

{ ( ) ( )}
Consider a set of m ≤ n (n-1)/2 judgments of interval pair-wise comparisons F1 = lij al , uij al
at level α = α1. A solution vector is good enough when satisfies Equation (2) in all the interval
judgments and in this case we have tried to find such vectors:
wi ≤
lij (α) ≤ u (α) (2)
ij
wj
The symbol ≤ denotes the statement “fuzzy less or equal”. In order to handle the above
inequalities easily, one can represent them as a set of single-side fuzzy constraints (3):
wi − w j uij (α) ≤
0
−wi + w jlij (α) ≤

0 (3)
The above set of 2m fuzzy constraints can be given in a matrix as (4):
0; R ∈ R 2m×n
Rw ≤ (4)
0, k = 1, 2,..., 2m , is a fuzzy linear constraint

The kth row of matrix Rw that is shown by Rk w ≤
and may be characterized with a linear membership function (5):
 Rw
1 − k Rk w ≤ dk
µk (Rk w ) =  dk (5)

 0 Rk w ≥ dk
55
where, dk is a tolerance parameter, showing the admissible interval of approximate satisfaction of

Rkw≤0. The membership function (5) represents the decision-maker’s satisfaction with a specify
priority vector, with respect to the kth single-side constraint (3).
( )
Let µk Rk w ,k = 1, 2, …, 2m be membership functions of the fuzzy constraints Rk w ≤ 0
on the n-1 dimension simplex (6).
Q n −1 = {(w , …, w
1 n
|wi > 0, w1 + … + wn = 1 } (6)
Definition 1: The fuzzy feasible area P on simplex Qn-1 is a fuzzy set, described by the membership
function (7).
 { }
µp (w ) = min µ1 (R1w ), …, µm (Rm w ) | w1 + … + wn = 1

(7)
The convex fuzzy feasible area P shows the overall satisfaction of the decision-makers with a
specific crisp priority vector w. Assuming that the decision-makers are interested in the best possible
solution, it is reasonable to determine a priority vector that maximizes his overall degree of satisfaction.
Definition 2: The maximization solution is a crisp vector w* which corresponds to the maximum
fuzzy feasible area (8):
( ) { }
µp w * = min µ1 (R1w ), …, µm (Rm w ) | w1 + … + wn = 1
 
(8)
By introducing a new variable called λ that measures the membership degree of a given priority
vector in the feasible area P and by using (5) and (8), the problem of finding the maximizing solution
can be represented under a linear program (9):
MAX λ
s.t d k λ + Rk w ≤ dk ,
n
(9)
∑wi = 1,wi > 0,
i =1
i = 1, 2, …, n k = 1, 2, …, 2m
The optimal solution to the above problem is the (W*, λ*) vector, where W* represents the
priority vector that has a maximum degree of membership in the fuzzy feasible area and λ* shows
the maximum value for the membership degree. Meanwhile, the value of λ* measures the satisfaction
degree as a natural indicator for inconsistency of the decision-makers’ judgments and that is why it
is called the consistency index. When the human’s interval judgments are consistence, this index is
larger or equal to unit. For inconsistent judgments, the consistency index λ* takes a value between
56
0 and 1 depending on the degree of inconsistency and the values of the tolerance parameters dk
(Mikhailov and Singh Madan, 2003).
3. CASE STUDY
In this study AHP, ANP, FAHP and FANP are compared with each other for the maintenance strategy
selection problem. To test the applicability of the presented research methodology, a dairy factory in
Iran was selected as the case study. For implementation of the above approach, high-risk equipment
must be identified at first. In this research, a sterile machine was selected as one of the high-risk
equipment with respect to the factory conditions and the related experts’ comments. The reason of
selecting the sterile machine is the key position it incorporates in the production line that in the case
of any failure, the production line will stop working. Therefore, the recommended approaches for
the sterile machine were executed and its results are stated in this section.
3.1. Determination of Decision-Making Elements

The first step to implement this methodology is to identify and introduce the decision-making elements
in the maintenance strategy selection issue. In this section, the evaluation criteria and sub-criteria are
introduced along with the maintenance strategies (as alternatives).
3.1.1. Determination of Evaluation Criteria and Sub-Criteria

To determine the evaluation criteria in this research, a comprehensive study was first done in the
literature and then a meeting was held with the maintenance experts of the factory. After collecting
and reviewing the decision-makers’ viewpoints, three groups of criteria were determined for the
maintenance strategy selection: namely profit, cost, and risk. The evaluation criteria and sub-criteria
in this research are as follows:
Criteria 1: The final product quality (C1): different strategies have different influences on quality
of the final product. For example, when a strategy is used for identification of the faults and
increasing the machine availability, the quality of the final product will be influenced accordingly.
Criteria 2: Efficiency (C2): the maintenance strategy efficiency can be noted from two aspects: 1)
Improvement of staff work efficiency (C21): its means using more appropriate and efficient human
resources related to the maintenance department; 2) Fault identification and reduction of failure
rate (C22): the aim of fault identification and failure reduction, involved in the condition-based
and predictive maintenance strategies, is to inform the maintenance engineers where and why a
failure occurs. As the result: the time of maintenance can be reduced and the availability of the
manufacturing system may be improved.
Criteria 3: Cost (C3): different maintenance strategies have different expenditure of hardware,
software and personnel training. The sub-criteria for criteria 3 include: 1) The hardware and
software cost (C31): for the condition-based and predictive maintenance, it is necessary to have
a number of sensors and computers. Software is needed to analyse the measured parameter data
in the condition-based and predictive strategies; 2) The production loss (C32): failure of the
more important equipment in the production line often leads to higher costs of production loss.
Selecting an appropriate maintenance strategy for such machines would reduce the production
loss; 3) The personnel training cost (C33): Only after sufficient training, the maintenance staff
can make full use of the related tools and techniques, and reach the maintenance goals.
Criteria 4: Time (C4): This criterion consists of two sub-criteria: shut-down time and maintenance
time. The sub-criteria for criteria 4 include: 1) The shutdown time (C41): the time that the
maintenance is postponed due to lack of spare parts, maintenance staff or etc. and the production
57
line is thus stopped. 2) The maintenance time (C42): the required time for the maintenance activities
are different in various strategies.
Criteria 5: Availability (C5): The sub-criteria for criteria 5 include: 1) Maintain facility in good
conditions (C51): if the availability level of a machine is less than a certain limit, it is often
impossible to produce a high quality product. Different strategies will cause different availability
levels for the machines; 2) Maintain appropriate usable spare part (C52): Generally, corrective
maintenance needs more spare parts than the other maintenance strategies. The spare parts for
some machines are really expensive.
Criteria 6: Feasibility and reliability (C6): feasibility of the maintenance strategies is divided into
acceptance by the labours and the technique reliability. The sub-criteria for criteria 6 include: 1)
Technique’s reliability (C61): the condition-based and predictive maintenance strategies may be
inapplicable for some complicated production facilities; 2) Acceptance by labours (C62): managers
and maintenance staff prefer maintenance strategies that are easily implemented and understood.
3.1.2. Determination of Maintenance Strategies

Maintenance strategy involves identification, investigation, and implementation of several repairs,
replaces and/or inspect decisions (Rastegari and Mobin, 2016; Al-Najjar,1997; Kelly, 1997). Different
classifications have been proposed for the maintenance strategies. In this research, according to the
other previous researches, e.g., Li, et al. (2006); Waeyenbergh and Pintelon (2004); Rastegari and
Salonen (2013); and Wang et al., 2007, maintenance strategies are classified into two main types:
preventive and corrective.
According to MIL-STD-721B, Preventive Maintenance (PM) means that all the actions are
performed before the system failure in order to retain an item in specified condition by providing
systematic inspections, detection, and prevention of incipient failures (Wang et al., 2007). Based on
the development of preventive maintenance techniques, this strategy is divided into three classes here
in this paper: time-based preventive maintenance (TBM), condition-based preventive maintenance
(CBM), predictive maintenance (PM).
Alternative 1: Time-Based Preventive Maintenance (TBM): According to reliability
characteristics of equipment, maintenance is planned and performed periodically, to reduce the
frequent and sudden failure. In TBM, maintenance performed in specified intervals, regardless of the
equipment’s condition (Al-Najjar and Alsyouf, 2003; Gits, 1992; Herbaty, 1990; Wang et al., 2007).
Periodic times are often optimized in order to minimize costs. It should be noted that the term “time”
may refer to calendar time, operating time, or age.
Alternative 2: The Condition-Based Preventive Maintenance (CBM): Condition-based
maintenance (CBM) was proposed based on development of machine diagnostic techniques in
the 1970s (Luce, 1999). In the condition-based maintenance strategy, a diagnostic tool is used to
measure the physical condition of the equipment, such as temperature, vibration, noise, lubrication
and corrosion (Eade, 1997). In this method, preventive activities are implemented in response to the
meaningful deterioration of a unit or system that is identified through the change in the monitored
parameter during the monitoring or fault diagnosis process (Gits, 1992; Vanzile and Otis, 1992).
Alternative 3: Predicative Maintenance (PM): The predicative maintenance is a maintenance
policy in which the selected physical parameters associated with an operating machine are periodically
or continuously censored, measured, and recorded, in order to reduce, analyse, compare and display
the extracted data and information for the support decisions related to operation and maintenance of
the machine (Rao, 1996).
Alternative 4: Corrective Maintenance (CM): Some researchers refer to CM as repairs. This
strategy is also known as a failure-based maintenance. According to MIL-STD-721B, CM is used to
address all actions performed as a result of failure and to return a facility to a specified condition. In
CM, no action is made to detect and prevent the failures (Swanson, 2001). In other words, CM is a
reactive and unplanned event in order to face with the equipment’s failure.
58
3.1.3. Determination of the Decision-Making Structure of the Proposed Model

The next step for implementation of the proposed model is to determine the relations among the
decision-making elements. As mentioned before, in this study the two methods of AHP and ANP
have been used to model the problem. Generally, the hierarchy of a decision consists of four different
levels: goal, criteria, sub-criteria and alternatives (Figure 1) and this is the same as the hierarchy
structure of AHP, but the most important step in execution of the ANP method is determining the
network structure of decision-making.
To determine the decision-making structure and to establish a network of decision elements,
the interactions and feedbacks among the evaluation criteria and sub-criteria should be addressed
properly. In this research, in order to identify mutual independence of the elements such as criteria,
the following process has been used.
Seven decision-makers were asked to construct a 0-1 matrix of the decision elements regarding
decision elements, in this matrix, 1 refers to the dependence of one element on another and 0 represents
no dependence, so this matrix is called influence matrix. It should be noted that the unit value represents
the influence of a column element on a row element. After showing the decision hierarchy (AHP)
and network (ANP) structures, the priority vectors of the criteria and sub-criteria were calculated.
3.2. Determination of Priority Vectors for Evaluation of Criteria and Sub-Criteria

Determination of the priority vectors of criteria and sub-criteria are discussed in this section. Therefore,
two types of relations between the criteria and the sub-criteria are defined: 1) The importance of
criteria with respect to the higher-level elements on the hierarchical structure, such as AHP relations;
2) The interactions and feedbacks among the criteria and the sub-criteria.
3.2.1. Determination of Importance of Decision-Making Elements

In order to address importance of the decision elements, questionnaire number 1 was designed and
given to the seven experts. Questionnaire number 1 is consisted of three parts: in the first part, the
DMs were asked to conduct the pair-wise comparisons among the criteria with respect to the goal of
maintenance strategy selection without considering the interactions among criteria. In this section,
some questions are asked as listed below: with respect to the goal of maintenance strategy selection,
which of the criteria (C1 or C2) are more important and to what extent? In the second part, the pair-wise
comparisons are used for evaluating the importance of all the sub-criteria below a given criterion with
respect to that criteria, without considering the interdependencies. In this part, the decision-makers
were asked to answer some questions including: which of the sub-criteria, hardware, software, or
training costs are more important with respect to the cost criteria and to what extent? In the third part,
the importance vectors of the alternatives were calculated with respect to the evaluation sub-criteria.
Figure 1. Hierarchy and network of the decision elements
59
In this section, some questions are asked such as below: which of the alternatives are more important
with respect to the hardware cost sub-criteria and to what extent? It should be mentioned that both
the AHP and FAHP are modelled based on the priority vectors of importance of the evaluation
elements, while the global weights of the criteria, sub-criteria and alternatives are calculated. In order
to model the ANP and the FANP, in addition to the importance priority vectors of the elements, the
interdependence and interactions priority vectors must be calculated.
3.2.2. Determination of the Interdependencies and Interactions

of Evaluation Criteria and Sub-Criteria
Questionnaire No. 2 was designed to show the interactions among the evaluation elements by using
the influences matrix mentioned before in sub-Section 2.4. The questionnaire was submitted to seven
experts, to investigate influence of the criteria and sub-criteria on each other from their own point
of views. This questionnaire consists of three main parts: in the first part, the influence levels are
examined in interdependences among the criteria. The DMs are asked to compare the influence level
of two of the criteria on a given controlling criterion. The following questions were asked in this
respect: which criterion has more influence on the quality of the final product: cost or efficiency?
and to what extent each? In the second part, the influence levels in interdependencies among the sub-
criteria are evaluated. The DMs are asked to compare the influence level of two of the sub-criteria on
a given controlling sub-criterion. In this section, below questions are asked: which of the sub-criteria
have more influence on fault identification and reduction of the failure rate: Hardware or training
cost? and each to what extent? In the third part, the feedbacks from the sub-criteria to the criteria
are considered. The DMs are asked to compare the level of influence the two criteria have under a
certain sub-criterion. They respond to the questions such as: which criterion is more influenced by
the maintenance time: cost or time? And each one is influenced to what extent? The priority vectors
are derived from the pair-wise comparison matrix and the un-weighted super matrix is constructed
by using these priorities. The general form of the un-weighted super matrix used for this study is
summarized in Table 2.
X denotes the column priorities vector of the criteria with respect to the goal of maintenance
strategy selection, whereas Y denotes the column priorities vector for the sub-criteria with respect
to each criterion they belong to and M is the column priorities vector of the alternatives with respect
to the evaluation sub-criteria. Z and H represent the matrices for the column priorities vectors of
interdependence among the criteria and the matrix for the column priorities vectors of interdependence
among the sub-criteria, respectively. Finally, F stands for the matrix of the column priorities vectors
of the feedbacks from the sub-criteria to the criteria.
3.3. Application and Comparison of MCDM Methods in the

Maintenance Decision Making in a Case Factory
The aim of this research is to compare the AHP, ANP, FAHP and FANP in the problems of maintenance
strategy selection. The AHP and FAHP in this study have a hierarchy structure including four levels:
goal, criteria, sub-criteria, and alternatives (maintenance strategies). In order to model the ANP and
Table 2. Unweighted super matrix
Goal Criteria Sub-Criteria Alternatives

goal 0 0 0 0
criteria X Z F 0
Sub-criteria 0 Y H 0
alternatives 0 0 M 0
60
FANP, the influence matrix of the evaluation elements was first constructed as shown in Table 3,
and the questionnaire number 2 was designed accordingly.
After that, the questionnaires number 1 and 2 were distributed among seven factory experts to
obtain their pair-wise judgments on the level of importance and influence. Determining the level
of importance and influence among the criteria and sub-criteria requires a group decision-making
approach. For this purpose, the AIP method was used to aggregate the group opinions and to determine
the priority vectors from these opinions. Therefore, the priority vectors of the individual DM were
calculated through the Mikhailov’s method first and then the arithmetic mean of these vectors were
determined. In addition, in order to simplify the calculations in the FAHP and FANP, α-cuts of the
fuzzy numbers were calculated as:
if a = (l, m, u ) ∀α ∈ 0, 1 | M = l α , u α  = (m − l ) α + l, − (u − m ) α + u  .

, M
a a    
In the present study, values for α were selected as α= 0, 0.25, 0.5, 0.75 and 1. It should be noted
that the inconsistency indices for all the pair-wise comparison matrices were calculated and the DMs
were asked to revise their inconsistent comparison matrices. Followed by determination of the priority
vectors of elements’ importance, the AHP and the FAHP were run for different values of α in order
to calculate weights of the elements and then, the mean values of these weights were computed as
the final weights of criteria, sub-criteria and alternatives.
To calculate the weights of the decision elements in ANP and FANP, the priority vectors of
elements’ influence must be calculated as well as their priority vectors. In the ANP and FANP, the
vectors of importance and influence for different values of α were entered the unweighted super
matrix. For example, the unweighted super matrix for α = 0.5 in FANP is shown in Table 4. In the
next stage, the weighted super matrices were calculated and the limited super matrices were obtained
by using these super matrices. The limited super matrix in the FANP for α = 0.5 is shown in Table
5.
Table 3. Influence matrix
Goal Criteria Sub-Criteria

c1 c2 c3 c4 c5 c6 c21 c22 c31 c32 c33 c41 c42 c51 c52 c61 c62
Goal 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Criteria c1 1 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0
c2 1 1 0 0 1 1 1 1 1 1 1 0 0 0 0 0 1 1
c3 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 0 0
c4 1 0 0 1 1 0 0 1 0 0 1 0 1 1 1 1 1 1
c5 1 0 1 1 1 0 0 1 1 1 0 0 0 1 1 1 0 0
c6 1 0 1 1 1 1 0 0 0 0 1 0 0 0 0 0 1 1
Sub- c21 0 0 1 0 0 0 0 0 1 1 0 1 1 1 1 1 0 1
Criteria c22 0 0 1 0 0 0 0 0 0 1 0 0 1 1 1 0 0 1
c31 0 0 0 1 0 0 0 1 1 1 0 0 0 0 1 1 1 0
c32 0 0 0 1 0 0 0 1 1 1 0 0 1 1 1 0 0 1
c33 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0
c41 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 1
c42 0 0 0 0 1 0 0 0 0 0 0 1 0 0 0 0 0 0
c51 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 1 0 0
c52 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 1 0
c61 0 0 0 0 0 0 1 1 0 0 0 0 0 1 0 0 0 0
c62 0 0 0 0 0 0 1 1 0 0 0 0 1 1 0 0 0 0
61
62
Table 4. Unweighted super matrix for α = 0.5

Goal 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Criteria c1 0.2539 0 0 0 0 0 0.4365 0 0 0 0 0 0 0 0.3703 0 0 0
c2 0.1919 0.7277 0 0 0.4375 0.5962 0.2918 0.6292 0.4905 0.2914 0.3830 0 0 0 0 0 0.1413 0.4509
c3 0.0989 0.2723 0.3423 0.2075 0.2641 0.2716 0.3996 0.4761 0.1702 0 0.3182 0.4085 0 0.1641 0 0
c4 0.1398 0 0.6351 0 0 0.1896 0 0 0.1277 0 0.6818 0.3996 0.4482 0.2333 0.1634 0.1411
c5 0.1591 0 0.2664 0.2588 0.1603 0 0 0.1813 0.1098 0.2325 0 0 0 0.1918 0.1815 0.6025 0 0
c6 0.1564 0 0.3913 0.1061 0.1946 0.1397 0 0 0 0 0.3191 0 0 0 0 0 0.6954 0.4081
Sub- c21 0 0 0.5508 0 0 0 0 0 0.5176 0.4122 0 0.1902 0.1848 0.2072 0.2914 0.3000 0
Criteria c
22
0 0 0.4492 0 0 0 0 0 0 0.3191 0 0 0.2486 0.4121 0.4028 0 0 0.2575
c31 0 0 0 0.2690 0 0 0 0.1092 0.1637 0.0000 0 0 0 0 0.1435 0.2193 0.4277 0
c32 0 0 0 0.3936 0 0 0 0.3775 0.3186 0.2687 0 0 0.1591 0.0890 0.1623 0 0 0.4461
c33 0 0 0 0.3374 0 0 0 0 0 0 0 0 0 0 0 0 0 0
c41 0 0 0 0 0.5820 0 0 0 0 0 0 0 0 0 0 0 0 0.2964
c42 0 0 0 0 0.4180 0 0 0 0 0 0 0.5836 0 0 0 0 0 0
c51 0 0 0 0 0 0.4468 0 0 0 0 0 0.2262 0 0 0 0.4807 0 0
c52 0 0 0 0 0 0.5532 0 0 0 0 0 0 0 0.2917 0 0 0.5723 0
c61 0 0 0 0 0 0 0.7256 0.1718 0 0 0 0 0 0 0 0 0
c62 0 0 0 0 0 0 0.2744 0.3415 0 0 0 0 0.4076 0 0 0 0
Table 5. Limited super matrix for α=0

Goal 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Criteria c1 0.0252 0.0252 0.0252 0.0252 0.0252 0.0252 0.0252 0.0252 0.0252 0.0252 0.0252 0.0252 0.0252 0.0252 0.0252 0.0252 0.0252 0.0252
c2 0.1542 0.1542 0.1542 0.1542 0.1542 0.1542 0.1542 0.1542 0.1542 0.1542 0.1542 0.1542 0.1542 0.1542 0.1542 0.1542 0.1542 0.1542
c3 0.1097 0.1097 0.1097 0.1097 0.1097 0.1097 0.1097 0.1097 0.1097 0.1097 0.1097 0.1097 0.1097 0.1097 0.1097 0.1097 0.1097 0.1097
c4 0.0831 0.0831 0.0831 0.0831 0.0831 0.0831 0.0831 0.0831 0.0831 0.0831 0.0831 0.0831 0.0831 0.0831 0.0831 0.0831 0.0831 0.0831
c5 0.0731 0.0731 0.0731 0.0731 0.0731 0.0731 0.0731 0.0731 0.0731 0.0731 0.0731 0.0731 0.0731 0.0731 0.0731 0.0731 0.0731 0.0731
c6 0.0909 0.0909 0.0909 0.0909 0.0909 0.0909 0.0909 0.0909 0.0909 0.0909 0.0909 0.0909 0.0909 0.0909 0.0909 0.0909 0.0909 0.0909
Sub- c21 0.0849 0.0849 0.0849 0.0849 0.0849 0.0849 0.0849 0.0849 0.0849 0.0849 0.0849 0.0849 0.0849 0.0849 0.0849 0.0849 0.0849 0.0849
Criteria c
22
0.0604 0.0604 0.0604 0.0604 0.0604 0.0604 0.0604 0.0604 0.0604 0.0604 0.0604 0.0604 0.0604 0.0604 0.0604 0.0604 0.0604 0.0604
c31 0.0390 0.0390 0.0390 0.0390 0.0390 0.0390 0.0390 0.0390 0.0390 0.0390 0.0390 0.0390 0.0390 0.0390 0.0390 0.0390 0.0390 0.0390
c32 0.0657 0.0657 0.0657 0.0657 0.0657 0.0657 0.0657 0.0657 0.0657 0.0657 0.0657 0.0657 0.0657 0.0657 0.0657 0.0657 0.0657 0.0657
c33 0.0185 0.0185 0.0185 0.0185 0.0185 0.0185 0.0185 0.0185 0.0185 0.0185 0.0185 0.0185 0.0185 0.0185 0.0185 0.0185 0.0185 0.0185
c41 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291
c42 0.0282 0.0282 0.0282 0.0282 0.0282 0.0282 0.0282 0.0282 0.0282 0.0282 0.0282 0.0282 0.0282 0.0282 0.0282 0.0282 0.0282 0.0282
c51 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291 0.0291
c52 0.0358 0.0358 0.0358 0.0358 0.0358 0.0358 0.0358 0.0358 0.0358 0.0358 0.0358 0.0358 0.0358 0.0358 0.0358 0.0358 0.0358 0.0358
c61 0.0403 0.0403 0.0403 0.0403 0.0403 0.0403 0.0403 0.0403 0.0403 0.0403 0.0403 0.0403 0.0403 0.0403 0.0403 0.0403 0.0403 0.0403
c62 0.0329 0.0329 0.0329 0.0329 0.0329 0.0329 0.0329 0.0329 0.0329 0.0329 0.0329 0.0329 0.0329 0.0329 0.0329 0.0329 0.0329 0.0329
63
The normalized numbers of the limited super matrix were considered as the weight of element on
each row. In the ANP and FANP method, to obtain the final weights of the decision-making elements,
the mean weights of different α values were calculated. The final weights of criteria, sub-criteria,
and alternatives are shown in Tables 6-8, respectively.
Table 6 shows the final weights for the evaluation criteria with their rankings in the four mentioned
approaches. As the results show, the weights and rankings in the AHP and FAHP approaches are
similar to each other and the same similarity exists for the ANP and FANP methods. The results of
AHP and FAHP show that the greatest weight of the evaluation criteria is related to the quality of
the final product and after that the feasibility and reliability criteria are the most important, though
the time and cost criteria show the smallest weights. While in the ANP and FANP methods, the
greatest weight is associated to the efficiency and cost criteria and the smallest weight is related to
the time and quality of the final product criteria. The reason for these differences between the AHP
and FAHP with the ANP and FANP is that the methods of AHP and FAHP evaluate the criteria only
based on the level of importance and do not consider the interdependencies and interactions among
the evaluation elements. However, in most of the real-world problems, the evaluation elements have
interdependencies and interactions. For example, in the current study, the quality of the final product
is the most important criteria for the company in order to compete with the other companies and also
to reach the highest level of customer satisfaction, but this criterion in the ANP structure is the sink
node and does not affect the other criteria, yet it is still strongly influenced by the other criteria. For
example, if the maintenance cost rises, it substantially influences the product quality and leads to
a higher product quality. On the other hand, although in this problem the cost criterion in the AHP
and FAHP methods have the smallest weight, but due to its significant impact on the other criteria
(influence on all the criteria according to the influence matrix which is depicted in Figure 1), in
the ANP and FANP, this criterion is ranked in the second order. As another instance, the efficiency
criterion in the AHP and FAHP methods are ranked as being the 4th and 3rd, but this criterion is
ranked as the 1st one in the ANP and FANP methods. This matter is mainly attributed to the influence
of this criterion on the other criteria, except the cost.
Table 7 lists the sub-criteria weights and their rankings. The results of Table 7, similar to Table
6, indicate similarity of the weights and rankings in the AHP and FAHP methods and also in the ANP
and FANP methods. The results of Table 7 show that in the AHP and FAHP methods the greatest
weight is dedicated to the technique reliability, maintain appropriate usable spare part, shut-down
time and improvement of staff work efficiency sub-criteria, while in the ANP and FANP methods the
sub-criteria of improvement of the staff work efficiency, fault identification, and personnel training
cost have the greatest weights. Moreover, Table 7 shows that in the AHP and FAHP, the smallest
weights are related to the personnel training cost, production loss and hardware and software cost sub-
criteria, while in ANP and FANP, the lowest weights sub-criteria are shut-down time, maintenance
time, production loss, and maintain facility in good condition. The reason for the different rankings
Table 6. Weights and rankings of different criteria on mentioned approaches
Criteria AHP Rank ANP Rank FAHP Rank FANP Rank

Product quality 0.2120 1 0.052 6 0.2627 1 0.0585 6
EFFICIENCY 0.1699 4 0.2487 1 0.1712 3 0.2697 1
Cost 0.1033 6 0.2093 2 0.0929 6 0.1930 2
Time 0.1619 5 0.1529 5 0.1256 5 0.1469 5
Availability 0.1713 3 0.1612 4 0.1676 4 0.1494 4
Feasibility & reliability 0.1813 2 0.1763 3 0.1800 2 0.1824 3
64
Table 7. Weights and rankings of different sub-criteria on mentioned approaches
Sub-Criteria AHP Rank ANP Rank FAHP Rank FANP Rank

Improvement of staff work efficiency 0.1095 4 0.1564 1 0.1275 2 0.1721 1
Fault identification and reduction of failure
0.1064 5 0.1299 2 0.1043 4 0.1262 3
rate
Hardware and software 0.0356 11 0.0864 6 0.0281 11 0.0725 6
Personnel training 0.0488 9 0.1286 3 0.0635 9 0.1421 2
Production loss 0.0469 10 0.0499 11 0.0344 10 0.0416 11
Shut down time 0.1156 3 0.0653 9 0.0991 6 0.0652 8
Maintenance time 0.0892 7 0.0604 10 0.0709 7 0.0605 10
Maintain facility in good condition 0.0969 6 0.0731 7 0.1008 5 0.0648 9
Maintain appropriate usable spare part 0.1208 2 0.0878 5 0.1268 3 0.0861 5
Technique reliability 0.1541 1 0.0918 4 0.1767 1 0.0979 4
Acceptance by labor 0.0763 8 0.0704 8 0.0679 8 0.0709 7
of the sub-criteria between these methods is the influence of the sub-criteria on each other and on
the criteria itself (feedback from the ANP system).
Table 8 depicts the weights and rankings of the alternatives (the maintenance strategies). It can be
observed from this table that the alternative’s rankings in all the four methods are completely similar
with only their weights being different. In addition, the obtained results show that the predictive
maintenance is the most appropriate strategy for the mentioned problem and after that; the time-based
preventive maintenance strategy has the most suitability.
4. CONCLUSION
This paper presents a comparison of ANP, AHP, FAHP and FANP in maintenance decision making.
Alternatives, including different maintenance strategies are compared using the mentioned MCDM
method, under certain and uncertain numbers, and considering different criteria. The results of this
research show that determining the local priorities of the elements requires a group process and an
individual process is not accurate enough. The proposed model is very flexible to allow the changes in
the decision elements such as inclusion of the new criteria or sub-criteria, removal of the inappropriate
criteria or sub-criteria, and changes in the pair-wise comparison judgments. In this research the
evaluation criteria and sub-criteria are dedicated to a dairy factory, while it can even be adapted or
extended for the other manufacturing environments by making the necessary modifications on the
elements such as adding a safety criterion with its related sub-criteria. At the end, it can be seen that
Table 8. Weights and rankings of maintenance strategy (alternative) on mentioned approaches
Maintenance Strategy AHP Rank ANP Rank FAHP Rank FANP Rank
Predictive maintenance 0.3030 1 0.2939 1 0.3275 1 0.3214 1
Preventive time-based maintenance 0.2432 3 0.2377 3 0.2354 3 0.2176 3
Preventive condition-based maintenance 0.2563 2 0.2800 2 0.2713 2 0.3067 2
Corrective maintenance 0.1976 4 0.1883 4 0.1657 4 0.1543 4
65
as higher as the effectiveness of one element on the other elements is, this element will have greater
weights and therefore, the results from the ANP method is completely different from those of the
AHP. In this research, a predictive maintenance was selected as the most appropriate strategy and the
preventive strategies outperformed the corrective strategies. The results of this research is consistent
with the results of previous studies done by Nakagawa (1989), Huang, et al. (1995), Chelbi and Aı̈t-
Kadi (2001), Wang et al. (2007), Pariazar, et al. (2008), and Cheng and Tsao (2010) on the industrial
equipment. The following items were also found interesting for the future researches in this area:
Different importance weights of DMs, which can be obtained from pair-wise comparisons among
DMs, are necessary for running the aggregation of individual priorities (AIP) approach. In addition,
a sensitivity analysis can be performed for the pair-wise comparison of DMs.
ACKNOWLEDGMENT
The authors would like to thank the anonymous reviewers and the editor for their insightful comments
and suggestions. We also would like to appreciate the supports from the managers of the case company
who provide appropriate information to accomplish this research.
66
REFERENCES
Al-Najjar, B. (1997). Condition-Based Maintenance: Selection and Improvement of a Cost-Effective Vibration-

Based Maintenance Policy for Rolling element Bearings. Lund University.
Al-Najjar, B., & Alsyouf, I. (2003). Selecting the most efficient maintenance approach using fuzzy multiple
criteria decision making. International Journal of Production Economics, 84(1), 85–100. doi:10.1016/S0925-
5273(02)00380-8
Aragonés-Beltrán, P., Chaparro-González, F., Pastor-Ferrando, J., & Rodríguez-Pozo, F. (2010). An ANP-based
approach for the selection of photovoltaic solar power plant investment projects. Renewable & Sustainable Energy
Reviews, 14(1), 249–264. doi:10.1016/j.rser.2009.07.012
Arunraj, N., & Maiti, J. (2010). Risk-based maintenance policy selection using AHP and goal programming.
Safety Science, 48(2), 238–247. doi:10.1016/j.ssci.2009.09.005
Ayağ, Z., & Özdemir, R. G. (2009). A hybrid approach to concept selection through fuzzy analytic network
process. Computers & Industrial Engineering, 56(1), 368–379. doi:10.1016/j.cie.2008.06.011
Azadivar, F., & Shu, V. (1999). Maintenance policy selection for JIT production systems. International Journal
of Production Research, 37(16), 3725–3738. doi:10.1080/002075499190013
Bashiri, M., Badri, H., & Hejazi, T. H. (2011). Selecting optimum maintenance strategy by fuzzy interactive
linear assignment method. Applied Mathematical Modelling, 35(1), 152–164. doi:10.1016/j.apm.2010.05.014
Bertolini, M., & Bevilacqua, M. (2006). A combined goal programming—AHP approach to maintenance
selection problem. Reliability Engineering & System Safety, 91(7), 839–848. doi:10.1016/j.ress.2005.08.006
Bevilacqua, M., & Braglia, M. (2000). The analytic hierarchy process applied to maintenance strategy selection.
Reliability Engineering & System Safety, 70(1), 71–83. doi:10.1016/S0951-8320(00)00047-8
Chelbi, A., & Ait-Kadi, D. (2001). Spare provisioning strategy for preventively replaced systems subjected
to random failure. International Journal of Production Economics, 74(1-3), 183–189. doi:10.1016/S0925-
5273(01)00125-6
Cheng, Y.-H., & Tsao, H.-L. (2010). Rolling stock maintenance strategy selection, spares parts’ estimation,
and replacements’ interval calculation. International Journal of Production Economics, 128(1), 404–412.
doi:10.1016/j.ijpe.2010.07.038
Eade, R. (1997). The importance of predictive maintenance. Iron Age New Steel, 9(13), 68-72.
Ertugrul Karsak, E., & Tolga, E. (2001). Fuzzy multi-criteria decision-making procedure for evaluating advanced
manufacturing system investments. International Journal of Production Economics, 69(1), 49–64. doi:10.1016/
S0925-5273(00)00081-5
Ferdousmakan, M., Vasili, M., Vasili, M., Tang, S., and Lim, N. T. (2014). Selection of Appropriate Risk-based
Maintenance Strategy by Using Fuzzy Analytical Hierarchy Process. Main Thematic Areas, 77.
Fouladgar, M. M., Yazdani-Chamzini, A., Lashgari, A., Zavadskas, E. K., and Turskis, Z. (2012). Maintenance
strategy selection using AHP and COPRAS under fuzzy environment. International journal of strategic property
management, 16(1), 85-104.
Gits, C. (1992). Design of maintenance concepts. International Journal of Production Economics, 24(3), 217–226.
doi:10.1016/0925-5273(92)90133-R
Herbaty, F. (1990). Handbook of maintenance management: Cost-effective practices. NJ: Noyes Publications
Park Ridge.
Huang, J., Miller, C., & Okogbaa, O. (1995). Optimal preventive-replacement intervals for the Weibull life
distribution: solutions and applications. Paper presented at the Reliability and Maintainability Symposium ‘95.
doi:10.1109/RAMS.1995.513271
Jafari, A., Jafarian, M., Zareei, A., & Zaerpour, F. (2008). Using fuzzy Delphi method in maintenance strategy
selection problem. Journal of Uncertain Systems, 2(4), 289–298.
67
Jajimoggala, S., Rao, V., & Satyanarayana, B. (2011). Maintenance strategy evaluation using ANP and goal
programming. International Journal of Strategic Decision Sciences, 2(2), 56–77. doi:10.4018/jsds.2011040104
Kelly, A. (1997). Maintenance strategy. Elsevier.
Li, J. R., Khoo, L. P., & Tor, S. B. (2006). Generation of possible multiple components disassembly sequence
for maintenance using a disassembly constraint graph. International Journal of Production Economics, 102(1),
51–65. doi:10.1016/j.ijpe.2005.01.012
Luce, S. (1999). Choice criteria in conditional preventive maintenance. Mechanical Systems and Signal
Processing, 13(1), 163–168. doi:10.1006/mssp.1998.0176
Meade, L. M. (1997). A methodology for the formulation of agile critical business processes. University of
Texas at Arlington.
Mechefske, C. K., & Wang, Z. (2003). Using fuzzy linguistics to select optimum maintenance and condition
monitoring strategies. Mechanical Systems and Signal Processing, 17(2), 305–316. doi:10.1006/mssp.2001.1395
Mikhailov, L. (2000). A fuzzy programming method for deriving priorities in the analytic hierarchy process. The
Journal of the Operational Research Society, 51(3), 341–349. doi:10.1057/palgrave.jors.2600899
Mikhailov, L., & Singh Madan, G. (2003). Fuzzy analytic network process and its application to the development
of decision support systems. IEEE Transactions on Systems, Man, and Cybernetics, Part C: Applications and
Reviews, 33(1), 33–41.
Mobin, M., Roshani, A., Saeedpoor, M., & Mozaffari, M. M. (2015). Integrating FAHP with COPRAS-G Method
for Supplier Selection (Case Study: an Iranian Manufacturing Company).Proceedings of the International Annual
Conference of the American Society for Engineering Management, Indianapolis, Indiana, USA
Mobley, R. K. (2002). An introduction to predictive maintenance. Butterworth-Heinemann.
Mon, D.-L., Cheng, C.-H., & Lin, J.-C. (1994). Evaluating weapon system using fuzzy analytic hierarchy process
based on entropy weight. Fuzzy Sets and Systems, 62(2), 127–134. doi:10.1016/0165-0114(94)90052-3
Nakagawa, T. (1989). A replacement policy maximizing MTTF of a system with several spare units.
Okumura, S., & Okino, N. (2003). A maintenance policy selection method for a critical single-unit itme in
each workstation composing a FMS with CBM optimisation. International Journal of COMADEM, 6(2), 3–9.
Pariazar, M., Shahrabi, J., Zaeri, M., & Parhizi, S. (2008). A combined approach for maintenance strategy
selection. Journal of Applied Sciences, 8(23), 4321–4329. doi:10.3923/jas.2008.4321.4329
Rao, B. (1996). Handbook of condition monitoring. Elsevier.
Rastegari, A., & Bengtsson, M. (2014). Implementation of Condition Based Maintenance in Manufacturing
Industry. Proceedings of theIEEE International Conference on Prognostics and Health Management, Washington,
USA.
Rastegari, A., & Bengtsson, M. (2015). Cost Effectiveness of Condition Based Maintenance in Manufacturing.
Proceedings of the IEEE 61st Annual Reliability and Maintainability Symposium, Florida, USA. doi:10.1109/
RAMS.2015.7105079
Rastegari, A., & Mobin, M. (2016). Maintenance Decision Making, Supported by Computerized Maintenance
Management System. Proceedings of theAnnual Reliability and Maintainability Symposium (RAMS). IEEE.
Tucson, Arizona, USA. doi:10.1109/RAMS.2016.7448086
Rastegari, A., & Salonen, A. (2013). Strategic Maintenance Management - Formulating Maintenance Strategy.
Proceedings of the26th International Conference of Condition Monitoring and Diagnostic Engineering
Management, Helsinki, Finland.
Saaty, T. L. (1980). The analytic hierarchy process: planning, priority setting, resources allocation. New York:
McGraw.
Saaty, T. L. (1990). How to make a decision: The analytic hierarchy process. European Journal of Operational
Research, 48(1), 9–26. doi:10.1016/0377-2217(90)90057-I
68
Saaty, T. L. (2001). Decision Making with Dependence and Feedback: The Analytic Network Process: the
Organization and Prioritization of Complexity. Rws Publications.
Saaty, T. L. (2004). Fundamentals of the analytic network process—Dependence and feedback in decision-making
with a single network. Journal of Systems Science and Systems Engineering, 13(2), 129–157. doi:10.1007/
s11518-006-0158-y
Saaty, T. L., & Vargas, L. G. (2006). Decision Making with the Analytic Network Process: Economic, Political,
Social and Technological Applications with Benefits, Opportunities, Costs and Risks. Springer.
Saeedpoor, M., Vafadarnikjoo, A., Mobin, M., & Rastegari, A. (2015). A SERVQUAL Model Approach
Integrated with Fuzzy AHP And Fuzzy TOPSIS Methodologies To Rank Life Insurance Firms. Proceedings
of the International Annual Conference of the American Society for Engineering Management, Indianapolis,
Indiana, USA.
Sharma, R. K., Kumar, D., & Kumar, P. (2005). FLM to select suitable maintenance strategy in process
industries using MISO model. Journal of Quality in Maintenance Engineering, 11(4), 359–374.
doi:10.1108/13552510510626981
Shyjith, K., Ilangkumaran, M., & Kumanan, S. (2008). Multi-criteria decision-making approach to evaluate
optimum maintenance strategy in textile industry. Journal of Quality in Maintenance Engineering, 14(4),
375–386. doi:10.1108/13552510810909975
Swanson, L. (2001). Linking maintenance strategies to performance. International Journal of Production
Economics, 70(3), 237–244. doi:10.1016/S0925-5273(00)00067-0
Taslicali, A. K., & Ercan, S. (2006). The analytic hierarchy and the analytic network processes in multicriteria
decision making: a comparative study. Citeseer.
Triantaphyllou, E., Kovalerchuk, B., Mann, L., & Knapp, G. M. (1997). Determining the most important
criteria in maintenance decision making. Journal of Quality in Maintenance Engineering, 3(1), 16–28.
doi:10.1108/13552519710161517
Tseng, M.-L. (2010). Implementation and performance evaluation using the fuzzy network balanced scorecard.
Computers & Education, 55(1), 188–201. doi:10.1016/j.compedu.2010.01.004
Vafadarnikjoo, A., Mobin, M., Allahi, S., & Rastegari, A. (2015). A Hybrid Approach of Intuitionistic Fuzzy
Set Theory and DEMATEL Method to Prioritize Selection Criteria of Bank Branches Locations.Proceedings
of the International Annual Conference of the American Society for Engineering Management. Indianapolis,
Indiana, USA.
Vanzile, D., & Otis, I. (1992). Measuring and controlling machine performance. Handbook of Industrial
Engineering (pp. 1575–1584). New York: Wiley.
Waeyenbergh, G., & Pintelon, L. (2004). Maintenance concept development: A case study. International Journal
of Production Economics, 89(3), 395–405. doi:10.1016/j.ijpe.2003.09.008
Wang, L., Chu, J., & Wu, J. (2007). Selection of optimum maintenance strategies based on a fuzzy analytic hierarchy
process. International Journal of Production Economics, 107(1), 151–163. doi:10.1016/j.ijpe.2006.08.005
Zaim, S., Turkyilmaz, A., Acar, M. F., Al-Turki, U., & Demirel, O. F. (2012). Maintenance strategy selection
using AHP and ANP algorithms: A case study. Journal of Quality in Maintenance Engineering, 18(1), 16–29.
doi:10.1108/13552511211226166
69

A Comparison of Multi-Criteria Decision Making Approaches For Maintenance Strategy Selection (A Case Study) 1 PDF

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

A Comparison of Multi-Criteria Decision Making Approaches For Maintenance Strategy Selection (A Case Study) 1 PDF

Încărcat de

Drepturi de autor:

Formate disponibile

International Journal of Strategic Decision Sciences

Volume 7 • Issue 3 • July-September 2016

A Comparison of Multi-Criteria Decision

2.1. AHP Method

Table 1. Linguistic variables and related fuzzy numbers

Fuzzy Numbers Linguistic Variables Fuzzy Number’s Scale

1 equally important (1,1,3)

3 weakly important (1,3,5)

5 essentially important (3,5,7)

7 very strongly important (5,7,9)

9 absolutely important (7,9,9)

2.2. ANP Method

2.3. FAHP and FANP Methods

2.4. Fuzzy Preferences Programming

where, α represents the confidence level of the decision-makers.

−wi + w jlij (α) ≤

The above set of 2m fuzzy constraints can be given in a matrix as (4):

 0, k = 1, 2,..., 2m , is a fuzzy linear constraint

where, dk is a tolerance parameter, showing the admissible interval of approximate satisfaction of

3.1. Determination of Decision-Making Elements

3.1.1. Determination of Evaluation Criteria and Sub-Criteria

3.1.2. Determination of Maintenance Strategies

3.1.3. Determination of the Decision-Making Structure of the Proposed Model

3.2. Determination of Priority Vectors for Evaluation of Criteria and Sub-Criteria

3.2.1. Determination of Importance of Decision-Making Elements

Figure 1. Hierarchy and network of the decision elements

3.2.2. Determination of the Interdependencies and Interactions

3.3. Application and Comparison of MCDM Methods in the

Table 2. Unweighted super matrix

Goal Criteria Sub-Criteria Alternatives

if a = (l, m, u ) ∀α ∈ 0, 1 | M  = l α , u α  = (m − l ) α + l, − (u − m ) α + u  .

Table 3. Influence matrix

Goal Criteria Sub-Criteria

Goal Criteria Sub-Criteria

Goal Criteria Sub-Criteria

Table 6. Weights and rankings of different criteria on mentioned approaches

Criteria AHP Rank ANP Rank FAHP Rank FANP Rank

Table 7. Weights and rankings of different sub-criteria on mentioned approaches

Sub-Criteria AHP Rank ANP Rank FAHP Rank FANP Rank

Table 8. Weights and rankings of maintenance strategy (alternative) on mentioned approaches

Al-Najjar, B. (1997). Condition-Based Maintenance: Selection and Improvement of a Cost-Effective Vibration-

S-ar putea să vă placă și

1 equally important (1,1,3)

3 weakly important (1,3,5)

5 essentially important (3,5,7)

7 very strongly important (5,7,9)

9 absolutely important (7,9,9)

0, k = 1, 2,..., 2m , is a fuzzy linear constraint

if a = (l, m, u ) ∀α ∈ 0, 1 | M = l α , u α  = (m − l ) α + l, − (u − m ) α + u  .