You are on page 1of 12

The current issue and full text archive of this journal is available on Emerald Insight at:

www.emeraldinsight.com/2043-9377.htm

Preventive
An uncertain programming maintenance
model for preventive scheduling

maintenance scheduling
Hamed Maleki 111
Department of Industrial Engineering, Yazd University, Yazd, Iran, and
Received 29 July 2016
Yingjie Yang Revised 24 August 2016
Accepted 26 August 2016
Centre for Computational Intelligence, De Montfort University, Leicester, UK and
Institute for Grey System Studies, Nanjing University of Aeronautics and Astronautics,
Nanjing, China

Abstract
Purpose – The purpose of this paper is to illustrate an uncertain programming model for scheduling of
preventive maintenance (PM) actions. The PM scheduling, in which PM actions are performed under fixed
intervals, is solved by grey systems theory.
Design/methodology/approach – The paper applied the grey evaluation method based on triangular
whitenization weight functions which includes two classes: endpoint evaluation method and center-point
evaluation method.
Findings – Two methods give the same results based on endpoint and center-point triangular whitenization
weight functions. For validation, the results were compared by Cassady’s method.
Originality/value – The scheduling of PM is crucial in reliability and maintenance engineering. Hundreds of
parts compose complex machines that require replacement and/or repairing. It is helpful to reduce the outage
loss on frequent repair/replacement parts and avoid lack of maintenance of the equipment by controlling the
equipment maintenance frequency.
Keywords Uncertainty, Failure, Grey systems theory, Preventive maintenance scheduling
Paper type Research paper

Nomenclature
β Weibull shape parameter for N(τ) number of machine failures in τ time
probability distribution of T units of machine operation
η Weibull scale parameter for z(t) hazard function of T
probability distribution of T tr time required to repair the machine
τ PM interval for the machine tp time required to perform PM on the
A(τ) steady-state machine availability machine
m(τ) E(N(τ))

1. Introduction
Timely and cost-effective production is becoming increasingly important in today’s global
competitive markets. For the manufacturers, it is vital to optimize the machine resource
utilization by ensuring an efficient and stabilized schedule at the operational level.
Traditional literatures on scheduling assume that machines are available at all times.
However, in many realistic situations, machines may be unavailable during the scheduling
horizon for different reasons, such as breakdowns and scheduled maintenances in typical
Grey Systems: Theory and
industrial settings. This availability consideration adds complexity to any scheduling Application
problem, even in a single machine environment (Lu et al., 2015). Vol. 7 No. 1, 2017
pp. 111-122
In a long-term scheduling, production system is always subject to deterioration with © Emerald Publishing Limited
2043-9377
usage, which will affect the availability of machines and the stability of the system. DOI 10.1108/GS-07-2016-0015
GS Preventive maintenance (PM) has proved its effectiveness in industry. A machine should be
7,1 maintained after it continuously works for a period of time. Thus a long scheduling horizon
should consist of several PM periods (Lu et al., 2015).
Literature classifies maintenance planning and scheduling into two major categories: the
scheduled maintenance and the unscheduled maintenance. The first includes preventive and
routine maintenance, and the scheduled overhauls and corrective maintenance. The second
112 deals with emergency breakdowns. The unscheduled maintenance is stochastic in nature
(Manzini et al., 2015). According to Duffuaa and Al-Sultan (1999) “this stochastic nature
makes maintenance scheduling a challenging problem.”
The maintenance tasks refer to the set of activities necessary to replace a component or a
group of components subjected to wear and tear within a generic plant or machine.
The group of maintenance tasks including all the repairing and/or replacing activities that a
generic machine or a plant requires over its own life-cycle is named task plan. Each task to
be scheduled usually involves spare parts, personnel, resources and equipment.
The frequency of each task is generally determined by the failure rates (i.e. the curve of
failure probability to the machine up time) of the most critical component of the task
(Manzini et al., 2015).
This paper presents an uncertain programming model based on grey theory systems for
the determination of the maintenance schedule, being complex and uncertain, that gives
desired results to the task plan (Maleki and Taghavi Fard, 2015).
In recent times effective management of maintenance operations have been of paramount
importance to decision makers in the industry. It is realized that significant saving could be
made with the development and implementation of techniques and models for efficient
maintenance scheduling operations (Oke and Charles-Owaba, 2006). Here, a novel model for
scheduling of PM actions will be presented using technologies in grey systems. The paper is
organized as follows. A literature review is presented in Section 2 and then Section 3
describes the methodology. Section 4 gives a numerical example and the computational
results. Finally, Section 6 concludes the paper and Section 7 discusses future research.

2. Literature review
PM is defined as a set of activities aimed at improving the overall reliability and availability
of a system. While planning the PM schedule according to the defined activities, the time
between two PM actions is an important aspect of actions adopted and it would affect the
maintenance economics. In term of PM interval, PM can be categorized into two kinds:
periodic PM (fixed interval) and non-periodic PM (dynamic interval) (Martorell et al., 2002;
Ahmad and Kamaruddin, 2012).
In the former approach, the planning horizon is segmented into discrete equal intervals.
For each PM stage, a decision must be made to perform one of the three PM actions
(inspection, repair and replacement) on each component by maximizing system benefit in
PM. Instead of searching for equally spaced PM actions, non-periodic PM approach search
for the flexible interval in which PM interventions should be performed (Tsai et al., 2004).
During the last few decades, numerous papers have been published on periodic PM
modeling and optimization. Cassady and Kutanoglu (2005) developed an integrated
mathematical model for a single-machine problem with periodic PM interval. They defined
the total weighted expected completion time as the objective function. Their model allowed
multiple maintenance activities and explicitly captures the risk of not performing
maintenance. Bartholomew-Biggs et al. (2006) proposed a new PM formulation which allows
the optimal number of occurrences of PM to be determined along with their optimal timings.
The formulation involved the global minimization of a non-smooth performance function.
Lim and Park (2007) proposed a periodic PM policy, which keeping the pattern of hazard
rate unchanged by PM. They evaluated the expected cost rate per unit time based on
computing the expected number of failures depending on the hazard rate of the underlying Preventive
life distribution of the system. Liao et al. (2010) developed a reliability-centered sequential maintenance
PM model for monitored repairable deteriorating system. They supposed that system’s scheduling
reliability could be monitored continuously and perfectly, whenever it reaches the threshold,
the imperfect repair must be performed to restore the system. Wang and Tsai (2012)
established a bi-objective imperfect PM model of a series-parallel system. They developed a
unit-cost cumulative reliability expectation measure to evaluate the extent to which 113
maintaining each individual component benefits the total maintenance cost and system
reliability over the operational life time. El-Ferik and Ben-Daya (2006) developed a hybrid
age-based model for imperfect PM involving maintainable and non-maintainable failure
modes. They determined the number of PM actions and the length of PM intervals that
minimize the total long-term expected cost per unit time. Ebrahimipour et al. (2013)
developed a multi-objective PM scheduling model in a multiple production line. They
defined reliability of production lines, costs of maintaining, failure and down time of system
as multiple objectives, and different thresholds were applied for available manpower, spare
part inventory and periods under maintenance.
Levitin and Lisnianski (2000) presented an optimization model for PM scheduling in
multi-state series-parallel systems. They considered the cost of unsupplied demand due to
failures of components as an important part of the PM activities cost. Schutz et al. (2011)
proposed and modeled periodic and sequential PM policies for a system. The objective of
periodic PM policy was to determine the optimal number of PMs to achieve, and the
objective of sequential PM policy was to determine the optimal number of PM intervals and
the duration of these different intervals. Wang and Lin (2011) proposed an improved particle
swam optimization of periodic PM. The optimal maintenance periods for all components in
the system were determined according to the importance of components on system
reliability, to minimize the periodic PM cost for a series-parallel system. Tam et al. (2006)
analyzed the affection of reliability, budget and breakdown outages cost to the calculation of
optimal maintenance intervals. Three models were proposed to calculate optimal
maintenance intervals for multi-component system in a factory subjected to minimum
required reliability, maximum allowable budget and minimum total cost. Alardhi et al.
(2007) presented an optimization method for scheduling the PM tasks in separate and linked
cogeneration plants and satisfying the maintenance and production constraints.
Shirmohammadi et al. (2007) presented an optimization method for scheduling the PM of
a system that is subject to random failures, and investigated the decision rule for PM. They
defined the time between preventive replacements and cut-off age as decision variables to
determine the optimal maintenance policy. Bartholomew-Biggs et al. (2006) proposed a
model in which each action of the PM reduces the equipment’s effective age.
The optimization process involved minimizing a performance function that allows for
the costs of the minimal repairs and eventual system replacement as well as for the costs of
the PM during equipment is operating life time. Harrou et al. (2010) formulated the model of
imperfect maintenance optimization for series-parallel transmission system structure. They
improved the availability of transmission system through selecting the optimal sequence of
intervals to perform PM actions. Moghaddam and Usher (2011) presented mathematical
models and a solution approach to determine the optimal PM schedules for a repairable and
maintainable series system with equally sized periods. Lin and Wang (2012)
identified important components and determined their maintenance priorities of a series-
parallel system. The optimal maintenance periods of these important components were
determined to minimize total maintenance cost given the allowable worst reliability of a
repairable system.
Todosijević et al. (2016) proposed a new mixed integer programming formulation and a
new heuristic method based on general variable neighborhood search called nested general
GS variable neighborhood search for the periodic maintenance problem. It consists to find an
7,1 optimal maintenance schedule of M machine over a cyclic horizon composed of τ periods.
Mjirda et al. (2016) proposed a new model for the joint optimization of the periodic PM and
spare parts inventory problem.
According to Tsai et al. (2004), the interval and the type of PM action are the main decision
variables of non-periodic PM. These make the optimization problem of non-periodic PM more
114 complex. However, the interval of non-periodic PM is more flexible and the schedules of
non-periodic PM can be better adjusted, reducing the outage loss on frequent repair parts
through the interval (PM frequency) control (Nakagawa, 1986). In view of the potential
important applications of non-periodic PM, some improvement ideas have been proposed. Wang
(2002) summarizes, classifies, and compares various existing maintenance policies for both
single-unit and multi-unit systems. Relationships among different maintenance policies are also
addressed. He found that non-periodic approach provided gains in the overall system average
availability. Percy and Kobbacy (2000) investigated two principal types of general models for
determining the optimal maintenance intervals. The first model considered fixed PM intervals,
while the second one was adaptable, allowing variable PM intervals. But they did not give clear
comparison to help decide between these two approaches. Lapa et al. (2003) defined inspections
as special PM actions and the PM was said to be non-periodic. They developed a surveillance
test optimization methodology based on genetic algorithms. Later, Lapa et al. (2006) presented a
methodology for PM policy evaluation based upon a cost-reliability model. They used flexible
intervals between maintenance interventions. However, the proposed method defined
maintenance action as PM action and replacement action was neglected. The method selected
components to undergo maintenance at different time, while avoiding this for components in
parallel. Lin and Wang (2010) proposed a non-periodic PM policy method for series-parallel
systems, which is based on failure limit policy. First, they identified the parallel sub-system
required to be maintained. Second, they identified the maintained parallel sub-system by the
measure of unit-cost extended life. But it was a time-consuming task for searching the optimum
values of the parameter and the replacement action was ignored in the method. The detail of the
solving method was not mentioned. At first, Fitouhi and Nourelfath (2012) proposed an
integrated non-cyclical PM method with tactical production planning for a single machine. Next,
they introduced an integrated non-cyclical PM method with tactical production planning in
multi-state systems (Fitouhi and Nourelfath, 2014). They suggested non-cyclical preventive
replacements of components, and minimal repair on failed components. However, only
preventive replacements were considered in their method for the sake of simplicity.
Yu and Seif (2016) presented a lower-bound-based genetic algorithm for minimizing
tardiness and maintenance costs in flow shop scheduling as a mixed-integer linear program
after incorporating flexible interval time.
The above models either need objective functions with constraints or are based on
complicated mathematical equations which need cost rate, hazard rate and, etc. to be solved,
but the proposed model does not. The main contribution of the paper is to define a general
model for PM scheduling problem, in which PM actions are performed under fixed intervals,
and then develop a solving framework to determine optimal periodic PM schedules.
Most studies paid attention on periodic PM modeling and optimization. PM planning
with dynamic interval is a complex decision-making process, which joint PM intervals and
PM actions optimization and it is a NP-hard problem (Gao et al., 2015). So the mathematical
models give exact solutions but they are time consuming. Moreover, meta-heuristic and
heuristic models need to adjust parameters to avoid local optima.

3. Methodology
It is helpful to reduce the outage loss on frequent repair/replacement parts and avoid lack
of maintenance of the equipment by controlling the equipment maintenance frequency
(Gao et al., 2015). It makes PM scheduling be complex and uncertain, and it is very rare for Preventive
such systems to be exactly determined in all their complexity. According to grey systems maintenance
theory, the maintenance period can be considered as the object that extension is definite but scheduling
intension is uncertain.
The grey systems theory is mainly utilized to study systems that model uncertainty,
analyze system relations, establish models, and make forecasts and decisions (Tsai et al.,
2005; Wu et al., 2012). Grey system theory can be applied to analyze relationships between 115
discrete quantitative and qualitative series, whose components are existent, countable,
extensible and independent (Zhou and Thai, 2016). Since scheduling problems have all of
these properties, therefore, grey theory can be applied to scheduling.
The focus of grey systems theory, established by Julong Deng in 1982, is on the
uncertainty problems of small samples and poor information that are difficult for
probability and fuzzy mathematics to handle. Big Data is prone to vagueness and
uncertainties. Fuzzy mathematics has proven itself to be useful for handling these
vaguenesses and subjective uncertainties (Lewis and Martin, 2015).
Fuzzy mathematics emphasizes on the investigation of problems with cognitive
uncertainty, where the research objects possess the characteristic of clear intension and
unclear extension. For instance, “young man” is a fuzzy concept, because each person knows
the intension of “young man.” However, if you are going to determine the exact range within
which everybody is young and outside which each person is not young, then you will find
yourself in a great difficulty. That is because the concept of young man does not have a clear
extension. For this kind of problem of cognitive uncertainty with clear intension and unclear
extension, the situation is dealt with in fuzzy mathematics by making use of experience and
the so-called membership function. The clear difference of fuzzy mathematics is that grey
systems theory emphasizes on the investigation of such objects that process clear extension
and unclear intension (Liu and Lin, 2010).
Grey systems theory uses a black-grey-white color spectrum to describe a complex
system whose characteristics are only partially known or known with uncertainty.
It explores and uncovers the realistic laws of evolution and motion of events and materials
through information coverage and through the works of sequence operators.
Grey clustering is a method developed for classifying observation indices or observation
objects into definable classes using grey incidences matrices or grey whitenization weight
functions. Each cluster can be seen as a set consisting of all observational objects of
the same kind.
The paper applied the grey evaluation method based on triangular whitenization weight
functions which includes two classes: endpoint evaluation method and center-point
evaluation method (Liu and Lin, 2010).

4. A general model
For scheduling with variable maintenance, the starting time is a decision variable, which is
determined by the scheduler, and the maintenance duration is a positive and non-decreasing
function of its starting time. In scheduling with fixed maintenance, the starting time and the
duration of maintenance are both determined in advance (Luo and Ji, 2015).
This paper considers the PM scheduling problem with one job to be processed in a single
machine. The job is available at time zero and no preemption is allowed. Figure 1 shows
an interval in which distances are equal. The paper applies two following methods
to solve the problem.

4.1 Evaluation model using endpoint triangular whitenization functions


Assume that n objects are to be clustered into s different grey classes. The observed value of
object i is i ¼ 1, 2, …, n. Object i needs to be evaluated and diagnosed based on the value xi.
GS The particular computational steps of the grey evaluation model based on endpoint triangular
7,1 whitenization functions are the following.
Step 1: based on the predetermined number s of grey classes for the planned evaluation,
divide the individual ranges of the alternative into s grey classes.
For example, suppose each distance is ten days (m ¼ 10). Hereafter we use
the abbreviation of Ds instead of days. We have partitioned this interval into four
116 grey classes.
Let us expand the given grey interval [2, 10] to 0 on the left and 12 on the right so that the
corresponding sequence of subintervals is showed in Table I.
Step 2: let the value of the whitenization weight function for λk ¼ (ak + ak +1)/2 to belong
to the kth grey class is 4. For an observed value x, its degree f k(x) of membership in the kth
grey class, k ¼ 1, 2, …, s, can be computed out of the following formula:
0 1
0 x2= ½ak1 ; ak þ 2 
B C
f k ðxÞ ¼ @ ðxak1 Þ=ðlk ak1 Þ x A ½ak1 ; lk  A
ðak þ 2 xÞ=ðak þ 2 lk Þ x A ðlk ; ak þ 2 

where ak(k ¼ 1,2, …, s, s + 1) can be generally determined based on specific requirements or


relevant qualitative analysis. For the example, we have:
0 1 0 1
0 x2= ½0; 6 0 x2= ½2; 8
B x A ½0; 3 C B C
f 1 ðxÞ ¼ @ x=3 A f ðxÞ ¼ @ x2=3 x A ½2; 5 A
2

6x=3 x A ð3; 6 8x=3 x A ð5; 8

0 1 0 1
0 x2= ½4; 10 0 x2= ½6; 11
B C B C
f 3 ðxÞ ¼ @ x4=3 x A ½4; 7 A f 4 ðxÞ ¼ @ x6=3 x A ½6; 9 A
10x=3 x A ð7; 10 11x=2 x A ð9; 11
n
Step 3: from Maxf f k g ¼ f k , it follows that object i belongs to grey class k*. When several
objects belong to the same k* grey class, one can further determine the order of preference of
these objects in grey class k* by using the magnitudes of f k.

Figure 1.
Fixed interval events
that are uniform a0 a1 a2 a3 a4 an
at m spans
m m m m m

Grey class Subintervals Description of grey class

– [0, 2) –
1 [2, 4) Poor
2 [4, 6) Fine
3 [6, 8) Good
Table I. 4 [8, 10) Excellent
Grey class – [10, 11] –
Table II shows 8Ds, 9Ds and 10Ds are “Excellent,” because d83 ¼ 2=3 is equal to d84 ¼ 2=3, it Preventive
can be further concluded that the achievements and efficiency of 8Ds is really somewhere maintenance
between “Excellent” and “Good” instead of a satisfactory “Excellent.” 9Ds is selected scheduling
because of the magnitudes of f k.

4.2 Evaluation model using center-point triangular whitenization functions


When deciding on the grey classes, the point within a grey class with the maximum degree 117
of greyness is referred to as the center of the class. The specific steps of employing the
grey evaluation model based on center-point triangular whitenization weight functions are
given as follows.
Step 1: based on the number s of grey classes required by the evaluation task,
respectively, determine the grey classes 1, 2, …, s’s centers λ1, λ2, …, λs, which stand for the
particular points for the observed values to belong to particular grey classes (they could be
either their individual intervals centers or not). Also correspondingly partition the field of
each alternative into s grey classes, which are represented, respectively, by using their
centers λ0, λ1, λ2, …, λs, λs + 1.
Step 2: expand the grey classes in two different directions by adding 0 and (s+1) grey
classes with their centers λ0 and λs+1 determined. So, we have a new sequences of centers:
λ0, λ1, λ2, …, λs, λs+1. For an observed value x, we can employ the formula:
0 1
0 x2= ½lk1 ; lk þ 1 
B C
f k ðxÞ ¼ @ ðxlk1 Þ=ðlk lk1 Þ x A ½lk1 ; lk  A
ðlk þ 1 xÞ=ðlk þ 1 lk Þ x A ðlk ; lk þ 1 

To compute its degree of membership f k(x) in grey class k, k ¼ 1, 2, …, s. For the example, so
we have:
0 1 0 1
0 = ½1; 5
x2 0 = ½3; 7
x2
B C B C
f 1 ðxÞ ¼ @ x1=2 x A ½1; 3 A f 2 ðxÞ ¼ @ x3=2 x A ½3; 5 A
5x=2 x A ð3; 5 7x=2 x A ð5; 7

0 1 0 1
0 x2= ½5; 9 0 x2= ½7; 10:5
B x A ½5; 7 C B x A ½7; 9 C
f 3 ðxÞ ¼ @ x5=2 4
A f ðxÞ ¼ @ x7=2 A
9x=2 x A ð7; 9 10:5x=1:5 x A ð9; 10:5
n
Step 3: from Maxf f k g ¼ f k , it is decided that object i belongs to grey class k*. When there
are several objects in grey class k*, those objects can be ordered according to the
magnitudes of f k.

Days
Grey class 2 4 6 8 9 10 11 12

1 2/3 2/3 0 0 0 0 0 0 Table II.


2 0 2/3 2/3 0 0 0 0 0 Assign class to
3 0 0 2/3 2/3 1/3 0 0 0 objects in endpoint
4 0 0 0 2/3 1 1/2 0 0 triangular
Description of grey Poor Poor and Fine and Good and Excellent Excellent – – whitenization
class fine good excellent functions
GS Table III shows 8Ds, 9Ds and 10Ds are excellent, because d83 ¼ 1=2 is equal to d84 ¼ 1=2, it can
7,1 be further concluded that the achievements and efficiency of 8Ds is really somewhere between
“Excellent” and “Good” instead of a satisfactory “Excellent,” but 9Ds is selected because of the
magnitudes of f k. The conclusions are exactly the same as those obtained earlier using the
grey evaluation model based on endpoint triangular whitenization weight functions.

118 5. Cassady’s method


5.1 The PM planning problem
Suppose the machine used to process the jobs is subject to failure, and the time to failure for
the machine is governed by a Weibull probability distribution having shape parameter
greater than 1. When the machine fails, we assume it is minimally repaired, i.e. the machine
is restored to an operating condition, but machine age is not altered. This implies that, upon
machine failure, the machine operator does just enough maintenance to resume machine
function. Because βW 1, it may be practical to perform PM on the machine in order to reduce
the increasing risk of machine failure. We assume that PM restores the machine to a “good
as new” condition, such that the machine’s age becomes 0. This implies that PM is a more
comprehensive action than repair, perhaps corresponding to the replacement of several key
components in the machine (Cassady and Kutanoglu, 2005).
We assume an age-based PM policy is applied, i.e. PM is performed on the machine after
τ time units of operation. Assuming our objective is to maximize machine availability, we
can use mathematical modeling to determine an optimal value for τ:
(1) the initiation of machine operation; and
(2) the end of each PM activity.
Because we assume repair is minimal, we can model the occurrence of failures during each
“cycle” of the renewal process using a nonhomogeneous Poisson process. Then, the
expected value of N(τ) is given by:
Z t Z t
 b  b
mðtÞ ¼ zðt Þd ðt Þ ¼ b=Z t b1 dt ¼ t=Z
0 0
where z(t) corresponds to the hazard function of the underlying Weibull probability
distribution. So, the “average” cycle consists of an “uptime” period of τ time units of
operation; and a “downtime” period of m(τ) repairs of length tr, and a PM action of length
tp. Therefore, the resulting steady-state availability of the machine, expressed as a function
of the PM interval, is given by:
 
AðtÞ ¼ t= t þmðtÞt r þt p
Differentiation and algebraic analysis yields an optimal PM interval of:
 1=b
tn ¼ Z t p =t r ðb1Þ (1)

Days
Grey class 2 4 6 8 9 10 11 12

Table III. 1 1/2 1/2 0 0 0 0 0 0


Assign class to 2 0 1/2 1/2 0 0 0 0 0
objects in center 3 0 0 1/2 1/2 0 0 0 0
point triangular 4 0 0 0 1/2 1 1/3 0 0
whitenization Description of grey Poor Poor and Fine and Good and Excellent Excellent – –
functions class fine good excellent
For the above example we have: Preventive
maintenance
9:91; 10:30; 8:60; 9:82; 7:89; 10:54; 10:29; 10:67; 10:71; 8:21
scheduling
The data have Weibull distribution with theses parameters η ¼ 10, β ¼ 10.
Other characteristics are also tp ¼ 0.12, tr ¼ 0.25. Evaluation of (1) indicates that PM
should be performed on this machine after 7.45 days of operation.
Grey systems theory gives three answers eight, nine and ten days, but the best one is 9. 119
The Cassady’s method presents 7.45 days, we have eight days if we round up it. So the
result is the same in two methods.
Failure pattern is divided to three categories in a machine life cycle: premature, natural
life and erosion. The first one follows exponential or normal distribution, the second follows
exponential distribution, and the third follows normal or Weibull distribution
(Shirmohammadi, 2010). As you can see in Figure 2, the Weibull distribution occurs in
the erosion step of a machine life cycle. In a case of erosion step, the Cassady’s method is
true but the grey method does not require that.

6. Conclusion
This paper studies PM with periodic interval for a single machine. The time of PM period in
the proposed periodic interval model is a fixed constant, which equals from interval-down to
interval-up.
Finding PM period is complex and uncertain, and it is very rare for such systems to be
exactly determined in all their complexity. According to grey systems theory, the
maintenance period can be considered as the object that extension is definite but intension is
uncertain. The aim is to define the best maintenance schedule that reduces the outage loss
on frequent repair/replacement parts and avoid lack of maintenance of the equipment by
controlling the equipment maintenance frequency.
The paper studies the grey evaluation method based on triangular whitenization weight
functions which includes two classes: endpoint triangular whitenization functions, and
center-point triangular whitenization functions. Finally, the paper compares the proposed
method with Cassady’s method.

7. Future research directions


All resources (machine, tool, mold) are subject to deterioration and need PM activities to
restore the working conditions. For the human resource, leisure time or vacation are also
necessary. Maleki and Taghavi Fard (2015) proposed a model solved by grey systems
theory to determine job scheduling. An uncertain programming model solved by grey

r (t )

Premature Natural life Erosion

Figure 2.
Failure in a machine
life cycle
t0 t1 t2 t3 Time
GS systems theory is presented to determine maintenance scheduling. Future research
7,1 could be improved by integrating other resources as noted by researchers. The paper also
gives a numerical example to evaluate the validity of the model because of generalizing
results. Future research can apply the model into practice and compare other methods
mentioned in Section 2.

120 References
Ahmad, R. and Kamaruddin, S. (2012), “An overview of time-based and condition-based maintenance in
industrial application”, Computers & Industrial Engineering, Vol. 63 No. 1, pp. 135-149.
Alardhi, M., Hannam, R.G. and Labib, A.W. (2007), “Preventive maintenance scheduling for
multi-cogeneration plants with production constraints”, Journal of Quality in Maintenance
Engineering, Vol. 13 No. 3, pp. 276-292.
Bartholomew-Biggs, M., Christianson, B. and Zuo, M. (2006), “Optimizing preventive maintenance
models”, Computational Optimization and Application, Vol. 35 No. 2, pp. 261-279.
Cassady, C.R. and Kutanoglu, E. (2005), “Integrating preventive maintenance planning and production
scheduling for a single machine”, IEEE Transactions on Reliability, Vol. 54 No. 2, pp. 304-309.
Duffuaa, S.O. and Al-Sultan, K.S. (1999), “A stochastic programming model for scheduling maintenance
personnel”, Applied Mathematical Modelling, Vol. 25 No. 5, pp. 385-397.
Ebrahimipour, V., Najjarbashi, A. and Sheikhalishahi, M. (2013), “Multi-objective modeling for
preventive maintenance scheduling in a multiple production line”, Journal of Intelligent
Manufacturing, Vol. 26 No. 1, pp. 1-12.
El-Ferik, S. and Ben-Daya, M. (2006), “Age-based hybrid model for imperfect preventive maintenance”,
IIE Transactions, Vol. 38 No. 4, pp. 365-375.
Fitouhi, M.C. and Nourelfath, M. (2012), “Integrating noncyclical preventive maintenance scheduling
and production planning for a single machine”, International Journal of Production Economics,
Vol. 136 No. 2, pp. 344-351.
Fitouhi, M.C. and Nourelfath, M. (2014), “Integrating noncyclical preventive maintenance scheduling
and production planning for multi-state systems”, Reliability Engineering & System Safety,
Vol. 121, pp. 175-186.
Gao, Y., Feng, Y., Zhang, Z. and Tan, J. (2015), “An optimal dynamic interval preventive maintenance
scheduling for series systems”, Reliability Engineering and System Safety, Vol. 142, pp. 19-30.
Harrou, F., Tassadit, A., Bouyeddou, B. et al. (2010), “Efficient optimization algorithm for preventive-
maintenance in transmission systems”, Journal of Modeling, Simulation, and Scientific, Vol. 1
No. 1, pp. 59-67.
Lapa, C.M.F., Pereira, C.M.N.A. and Emelo, P.F.F. (2003), “Surveillance test policy optimization through
genetic algorithms using non-periodic intervention frequencies and considering seasonal
constraints”, Reliability Engineering and System Safety, Vol. 81 No. 1, pp. 103-109.
Lapa, C.M.F., Pereira, C.M.N.A. and DeBarros, M.P. (2006), “A model for preventive maintenance
planning by genetic algorithms based in cost and reliability”, Reliability Engineering and System
Safety, Vol. 91 No. 2, pp. 233-240.
Levitin, G. and Lisnianski, A. (2000), “Short communication optimal replacement scheduling in multi-
state series-parallel systems”, Quality and Reliability Engineering International, Vol. 16 No. 2,
pp. 157-162.
Lewis, D.J. and Martin, T.P. (2015), “Managing vagueness with fuzzy in hierarchical Big Data”,
Procedia Computer Science, Vol. 53, pp. 19-28.
Liao, W., Pan, E. and Xi, L. (2010), “Preventive maintenance scheduling for repairable system with
deterioration”, Journal of Intelligent Manufacturing, Vol. 21 No. 6, pp. 875-884.
Lim, J.H. and Park, D.H. (2007), “Optimal periodic preventive maintenance schedules with improvement
factors depending on number of preventive maintenances”, Asia Pacific Journal of Operational
Research, Vol. 24 No. 1, pp. 111-124.
Lin, T.W. and Wang, C.H. (2010), “A new approach to minimize non-periodic preventive maintenance Preventive
cost using importance measures of components”, Journal of Scientific and Industrial Research, maintenance
Vol. 69 No. 9, pp. 667-671.
scheduling
Lin, T.W. and Wang, C.H. (2012), “A hybrid genetic algorithm to minimize the periodic preventive
maintenance cost in a series-parallel system”, Journal of Intelligent Manufacturing, Vol. 23 No. 4,
pp. 1225-1236.
Liu, S. and Lin, Y. (2010), Grey Systems: Theory and Applications, Springer, Milton Keynes. 121
Lu, Z., Cui, W. and Han, X. (2015), “Integrated production and preventive maintenance scheduling for a
single machine with failure uncertainty”, Computers & Industrial Engineering, Vol. 80, pp. 236-244.
Luo, W. and Ji, M. (2015), “Scheduling a variable maintenance and linear deteriorating jobs on a single
machine”, Information Processing Letters, Vol. 115 No. 1, pp. 33-39.
Maleki, H. and Taghavi Fard, M.T. (2015), “Evaluation of scheduling using triangular whitenization
functions”, Journal of Manufacturing Technology Management, Vol. 26 No. 2, pp. 297-310.
Manzini, R., Accorsi, R., Cennerazzo, T., Ferrari, E. and Maranesi, F. (2015), “The scheduling of
maintenance. A resource-constraints mixed integer linear programming model”, Computers &
Industrial Engineering, Vol. 87, pp. 561-568.
Martorell, S., Sánchez, A., Carlos, S. and Serradell, V. (2002), “Comparing effectiveness and efficiency in
technical specifications and maintenance optimization”, Reliability Engineering and System
Safety, Vol. 77 No. 3, pp. 281-289.
Mjirda, A., Benmansour, R., Allaoui, H. and Goncalves, G. (2016), “On the joint optimization of the
periodic preventive maintenance and the spare parts inventory problem”, 8th IFAC Conference
on Manufacturing Modelling, Management and Control MIM, Troyes, 28-30 June.
Moghaddam, K.S. and Usher, J.S. (2011), “Preventive maintenance and replacement scheduling for
repairable and maintainable systems using dynamic programming”, Computers & Industrial
Engineering, Vol. 60 No. 4, pp. 654-665.
Nakagawa, T. (1986), “Periodic and sequential preventive maintenance policies”, Journal of Applied
Probability, Vol. 26 No. 4, pp. 536-542.
Oke, S.A. and Charles-Owaba, O.E. (2006), “An approach for evaluating preventive maintenance
scheduling cost”, International Journal of Quality & Reliability Management, Vol. 23 No. 7,
pp. 847-879.
Percy, D.F. and Kobbacy, K.A.H. (2000), “Determining economical maintenance intervals”, International
Journal of Production Economics, Vol. 67 No. 1, pp. 87-94.
Schutz, J., Rezg, N. and Léger, J.B. (2011), “Periodic and sequential preventive maintenance policies over
a finite planning horizon with a dynamic failure law”, Journal of Intelligent Manufacturing,
Vol. 22 No. 4, pp. 523-532.
Shirmohammadi, A.H. (2010), Industrial Maintenance Planning and Control, Arkandanesh, Isfahan.
Shirmohammadi, A.H., Zhang, Z.G. and Love, E. (2007), “A computational model for determining
the optimal preventive maintenance policy with random breakdowns and imperfect repairs”,
IEEE Transactions on Reliability, Vol. 56 No. 2, pp. 332-339.
Tam, A.S.B., Chan, W.M. and Price, J.W.H. (2006), “Optimal maintenance intervals for a multi-component
system”, Production Planning and Control, Vol. 17 No. 8, pp. 769-779.
Todosijević, R., Benmansour, R., Hanafi, S., Mladenović, N. and Artiba, A. (2016), “Nested general
variable neighborhood search for the periodic maintenance problem”, European Journal of
Operational Research, Vol. 252 No. 2, pp. 385-396.
Tsai, M.T., Hsiao, S.W. and Liang, W.K. (2005), “Using grey theory to develop a model for forecasting
the demand for telecommunications”, Journal of Information and Optimization Sciences, Vol. 26
No. 3, pp. 535-547.
Tsai, Y.T., Wang, K.S. and Tsai, L.C. (2004), “A study of availability-centered preventive maintenance
for multi-component systems”, Reliability Engineering and System Safety, Vol. 84 No. 3,
pp. 261-270.
GS Wang, C.H. and Lin, T.W. (2011), “Improved particle swarm optimization to minimize periodic
7,1 preventive maintenance cost for series-parallel systems”, Expert Systems with Applications,
Vol. 38 No. 7, pp. 8963-8969.
Wang, C.H. and Tsai, S.W. (2012), “Optimizing bi-objective imperfect preventive maintenance model
for series-parallel system using established hybrid genetic algorithm”, Journal of Intelligent
Manufacturing, Vol. 25 No. 3, pp. 1-14.
Wang, H. (2002), “A survey of maintenance policies of deteriorating systems”, The European Journal of
122 Operational Research, Vol. 139 No. 3, pp. 469-489.
Wu, W.H., Lin, C.T., Peng, K.H. and Huang, C.C. (2012), “Applying hierarchical grey relation clustering
analysis to geographical information systems – a case study of the hospitals in Taipei city”,
Expert Systems with Application, Vol. 39 No. 8, pp. 7247-7254.
Yu, A.J. and Seif, J. (2016), “Minimizing tardiness and maintenance costs in flow shop scheduling by a
lower-bound-based GA”, Computers & Industrial Engineering, Vol. 97, pp. 26-40.
Zhou, Q. and Thai, V.V. (2016), “Fuzzy and grey theories in failure mode and effect analysis for tanker
equipment failure prediction”, Safety Science, Vol. 83, pp. 74-79.

Further reading
Goren, S. and Sabuncuoglu, I. (2008), “Robustness and stability measures forscheduling single-machine
environment”, IIE Transactions, Vol. 40 No. 1, pp. 66-83.
Tellis, W. (1997), “Introduction to case study”, The Qualitative Report, Vol. 3 No. 2, pp. 5-6, available at:
www.nova.edu/ssss/QR/QR3-2/tellis1.html

Corresponding author
Hamed Maleki can be contacted at: h.maleki@stu.yazd.ac.ir

For instructions on how to order reprints of this article, please visit our website:
www.emeraldgrouppublishing.com/licensing/reprints.htm
Or contact us for further details: permissions@emeraldinsight.com

You might also like