Anda di halaman 1dari 21

The current issue and full text archive of this journal is available at

www.emeraldinsight.com/1741-038X.htm

Maintenance
Maintenance policy selection policy selection
model a case study model
in the palm oil industry
415
Siew-Hong Ding, Shahrul Kamaruddin and Ishak Abdul Azid
School of Mechanical Engineering, Received 8 March 2012
Universiti Sains Malaysia, Nibong Tebal, Malaysia Revised 13 September 2012
8 November 2012
6 December 2012
Abstract 9 December 2012
Purpose An optimal maintenance policy is key to the improvement of the availability and Accepted 12 December 2012
reliability of a system at an acceptable level without a significant increase in investment. However,
the selection process is a complicated task because it requires in-depth knowledge on maintenance
policies and on the technical requirements of maintenance. The difficulties and complexity of the
selection process arise from the combination of conflicting maintenance constraints such as available
spares, size of workforce, and maintenance skills. The paper aims to discuss these issues.
Design/methodology/approach The proposed maintenance policy selection (MPS) model is
separated into three major phases. The first phase identifies the critical system (CS) based on failure
frequency. The failure mechanism in the CS is then analyzed by using a failure mode and effect
analysis in the second phase. In the third phase, a multi-criteria decision making method, called the
technique for order of preference by similarity to ideal solution, is adopted to identify an optimal
maintenance policy that can minimize the failures.
Findings Through a case study, preventive maintenance was selected as the optimal maintenance
policy for the reduction of system failures. The results obtained from the case study not only provide
evidence of the feasibility and practicability of the developed model, but also test the acceptability and
rationale of the developed model from the industry perspective. Valuable knowledge and experience
from employees were extracted and utilized through the proposed model to rank the optimal
maintenance policy based on the capability to reduce failure.
Originality/value The practicality of the MPS model is justified through an implementation in the
palm oil industry. The application of the MPS model can also be extended to other manufacturing
industries.
Keywords Decision making, Maintenance, Critical analysis, Maintenance policy selection
Paper type Research paper

1. Introduction
A machine is the primary component of any manufacturing industry. However,
a machine is incapable of preventing failure. The occurrence of failure, whether serious
or not, results in uncertain losses in terms of money, time and life. Maintenance is thus
necessary to reduce losses. Proper maintenance, which keeps life cycle costs down,
not only helps extend the system lifetime, but also positively contributes to the
overall performance of the company (Jou et al., 2009). However, maintenance also
contributes significantly to total production cost. The maintenance cost of a firm can
Journal of Manufacturing Technology
Management
The authors wish to acknowledge the support of the Fundamental Research Grant Scheme Vol. 25 No. 3, 2014
(FRGS) from the Ministry of Higher Education (MOHE) for funding this research. The authors pp. 415-435
q Emerald Group Publishing Limited
are gratefully appreciated to the anonymous referees for their constructive comments that 1741-038X
enabled the improvement of the paper. DOI 10.1108/JMTM-03-2012-0032
JMTM vary from 15 to 70 per cent of the production costs, depending on the industry type
25,3 (Waeyenbergh and Pintelon, 2004; Savsar, 2011). This cost often forms the basis for the
performance improvement demands on the maintenance department. Over the years,
several maintenance policies have been introduced, including preventive maintenance
(PM) and predictive maintenance (PdM), to improve system performance by enhancing
system availability and reliability. However, not every maintenance policy is suitable
416 for implementation in a system because each maintenance policy has its own attributes
and produces different effects when implemented. Therefore, a selection model is
needed to identify the optimal maintenance policy. The optimal maintenance policy
is necessary for the improvement of system performance by increasing the availability
and reliability levels of the system without a significant increase in investment
(Wang et al., 2007).
Given the significance and difficulty of selecting the optimal maintenance policy,
different models have been developed for this goal. For instance, Bevilacqua and
Braglia (2000) presented a multi-criteria decision making (MCDM) model for the
selection of the optimal maintenance policy in an Italian oil refinery processing plant.
The optimal maintenance policy was selected by using the analytical hierarchy process
(AHP) according to various features such as economics, applicability, cost and safety.
A similar model was also proposed by Bertolini and Bevilacqua (2006). Moreover,
a model that employs an integration of the weighted sum method with fuzzy logic was
proposed by Al-Najjar and Alsyouf (2003) for the selection of the most cost-effective
maintenance policy. Li and Xu (2007) also proposed an MCDM model by integrating
the elimination and choice translating reality with fuzzy logic in identifying the
optimal maintenance policy for a compressor. Safety, cost, added value and information
were the five criteria used to evaluate the maintenance policy. The technique for order
of preference by similarity to ideal solution (TOPSIS) was also adopted by Shyjith et al.
(2008) for the selection of the optimal maintenance policy in the textile industry.
Different aspects including environment, machine, workforce and the maintenance
policy itself were considered in the selection process. Moreover, an integration of fuzzy
AHP with TOPSIS was also developed by Ilangkumaran and Kumanan (2009) to
determine the optimal maintenance policy for the textile industry.
Majority of the studies agreed that most problems associated with system reliability,
availability and maintainability can be solved by implementing the optimal maintenance
policy. The developed maintenance selection models typically focus on determining
how much maintenance must be conducted on the components and how frequent the
components must be replaced. These maintenance selection models do not focus on
which system must be improved and on what maintenance policy is required. Thus,
the studies were directly focused on a specific maintenance action without performing
any analysis on the suitability of the maintenance policy for the system.
In addition, the practicability of the model in the industry must also be considered.
Therefore, a model that integrates failure analysis and the MCDM method on the
system is proposed for the selection of the optimal maintenance policy. Furthermore,
the MCDM method, which utilizes expert knowledge in collecting data for analysis,
is more practical. This practicality stems from the fact that experts with the most
knowledge and experience can be easily identified. In addition, the MCDM method can
consider a large number of criteria in selection process, thus improving the overall
reliability of the selected maintenance policy. Through a selection process that uses
the suggested model, the maintenance policy is expected to reduce the failure impact Maintenance
to a minimum level. By using this model, management can plan for and implement a policy selection
more effective maintenance process at a minimal amount of time.
The remainder of this paper is organized as follows: Section 2 describes the general model
principle of the maintenance policy selection (MPS) model. Section 3 describes the
application of the MPS model in the palm oil industry. The results of the MPS model
are discussed in Section 4. Section 5 draws the conclusion of this paper. 417
2. MPS model
The basic principle of the MPS model is that the selection of the optimal maintenance
policy should be based on the causes of system failure. However, most industries lack a
complete data recording system. This inadequacy may be attributed to the lack of
awareness of the management regarding the importance of data recording. Moreover,
setting up a complete data recording system requires a certain amount of investment,
which is why numerous companies are not keen on this aspect. The lack of a complete
data recording system has increased the difficulty of collecting accurate and precise
data, which adversely affects the accuracy of the final result obtained from the model.
Technically, these data reside within the workers under the form of skills, know-how,
and capabilities. Therefore, expert judgment is the most suitable methodology that can
be used to collect data from workers.
The MPS model proposed in this study was developed according to three principles:
problem identification, analysis, and solution finding. Based on these three principles,
the model was developed by integrating different approaches, including tally chart,
functional block diagram (FBD), functional failure mode and effect analysis (FMEA),
expert judgment and TOPSIS. The model is generally divided into three modules: scope
identification, functional analysis and maintenance policy evaluation, as shown in Figure 1.
As shown in Figure 1, the objective of Module I is to specify the scope of analysis in
order to have effective analysis by focusing on the fundamental area. In Module II,
a series of analysis is conducted to investigate and relate the failure causes and
associate effects on the critical system (CS). An optimal maintenance policy that is able
to reduce the failure effect is selected in Module III by referring to the results obtained
in Module II. The details of each module are explained in following paragraph.

Production process description


Module I
Scope Identification System separation
Critical system selection

Module II
Critical System Functional identification
Assessmment Functional failure analysis

Module III
Maintenance policy proposition
Maintenance Policy Maiintenance ppolicy rankking Figure 1.
Evaluattion MPS model
JMTM Module I: scope identification
25,3 Module I aims to study and to separate the entire production line into several systems
according to the main functions of the system. The operational flow and the functions
of the related system are studied through actual operation observations, production
handbooks and operation manuals. These functions are illustrated by using FBD.
The adoption of FBD provides a diagrammatical and structural breakdown of a
418 complex system in functional terms. The knowledge acquired is then used to separate
the production line into several systems to simplify the analysis process.
In a CS identification process, failure frequency is used to prioritize the criticality level
of each system on the production line for effective analysis for a limited duration. Failure
frequency was collected from historical failure records in maintenance documents and
then recorded in a tally chart. A tally chart is an easy and efficient method for the
collection of occurrences such as statistical information to show the relative occurrence
frequency (Martin, 2008; NSW Health Department, 2002). The final goal of Module I
is to determine which system has the highest failure frequency. This system then
becomes the analysis subject in Module II.

Module II: CS assessment


Activities in Module II can be separated into two major steps including functional
identification and functional failure analysis. Second-level FBD broken down from the
FBD that was produced in Module I is used to illustrate functional flow of CS. The lists
of functional modes are identified according to the second-level FBD. Functional mode
used in this identification process is defined as the activities assigned to a system to
accomplish specific function. Once the functions are identified, a functional failure
analysis is conducted to identify the failure mode and the failure causes of the CS by
using functional FMEA. Functional FMEA implemented in this model is to provide
better understanding about failure mechanism including how the failure occurs,
why the failure occurs and what are the impacts when the failure occurs.
The functional failure modes are identified based on the results obtained in functional
identification. In this identification process, the functional failure mode is described as
the sub-system state when it lost its function. Whereas functional failure causes are
defined as the physical processes that directly cause a system failure. As a functional
failure mode may have more than one failure causes; all probable causes for each
functional failure mode shall be identified and described. Once the failure causes are
recognized, failure impact computation of each failure cause will be conducted.
Functional failure impact is a relative measure of the consequences of a functional
failure cause. Each functional failure cause creates different impact on the CS.
The quantification of the failure impact is required in order to integrate the value of failure
impact into the maintenance policies selection process. Risk priority number (RPN) is used
to quantify the failure impact of the functional failure cause (Yang et al., 2008). The
indication process is based on three criteria which are occurrence, severity and detection.
In order to perform an overall evaluation from failure severity to the effectiveness of
existing failure detection possibility, a RPN rating scale that suit to the evaluation
scenario are developed from the discussion with FMEA team. Table I tabulated
the evaluation rating of occurrence. The definition of occurrence in this model is the
duration between two related functional failure causes occurred that named as mean
time between failures (MTBF).
Maintenance
Rating MTBF
policy selection
10 Less than 1 hour model
9 1-8 hours
8 9-16 hours
7 Two to six days
6 One to three weeks 419
5 One to two months
4 Three to five months
3 Six to nine months
2 Ten to 12 months Table I.
1 Greater than one years Occurrence rating criteria

The developed severity rating is tabulated in Table II. The severity is defined as the
seriousness of the functional failure cause on the CS when it occurs. In a continuous
production flow with preponderance of unrefined product that is in fluid form, any
nature of component failure will cause certain duration of stoppage. Therefore, severity
of the functional failure effect is justified according to the duration of stoppage when
the functional failure cause occurs.
The third criterion in the criticality evaluation is detection where it is the possibility
of the functional failure cause being detected before failure. The rating scale is
presented in Table III. The percentage of detection is based on average ten times
failure, how many percent the failure could be detected before failure out of ten times
failures.
The RPN is computed according to equation (1):
RPN OFJ SFJ DFJ 1

where OFJ represents the occurrence of the failure cause, Fj; SFJ represents the severity of
Fj; and DFJ represents the possibility of detecting Fj.
The list of failure causes serves as the evaluation criterion during the maintenance
policy evaluation conducted in Module III. The RPN is the criterion that is weighted in
the TOPSIS computation.

Rating Description (evaluation criteria)

1 Cause delay less than 15 minutes. No defective component


2 Cause delay less than 15 minutes, parts defect
3 Systems down less than 30 minutes, parts defect
4 Systems down for 30 minutes to 1 hour, parts defect
5 Systems down for 1-2 hours, parts defect
6 Systems down for 3-5 hours, parts defect
7 Systems down for 6-8 hours, parts defect
8 Systems down more than 9 hours, parts defect
9 Failure mode affects operator safety or involves non-compliance with
government regulation with warning
10 Failure mode affects operator safety or involves non-compliance with Table II.
government regulation without warning Severity rating criteria
JMTM
Rating Description (%)
25,3
1 More than 90
2 80-89
3 70-79
4 60-69
420 5 50-59
6 40-49
7 30-39
8 20-29
Table III. 9 10-19
Detection rating criteria 10 Less than 10

Module III: maintenance policy evaluation


Discussion with experts in the corresponding organization is performed to identify the
feasible maintenance policies that could be implemented according to the capabilities
and constraints of the organization. With the same group of experts, evaluations are
conducted to rate the potential performance of the proposed maintenance policies in
terms of reducing the failure impact from related failure causes. TOPSIS is then used
in the MPS model to rank the proposed maintenance policies. The equations used in
TOPSIS can be found in the study by Yang and Hung (2007). The rating is then
organized into a matrix, as shown in equation (2):
F1 F2 ... Fj . . . Fn
2 3
A1p11 p 12 ... p 1j ... p 1 n
6 7
A2 6 p p 22 ... p 2 j ... p 2 n 7
6 21 7
6
.. 6 .. .. .. .. 77
. 6 . . ... . ... . 7 2
6 7
D 6 7
Ai 6 p i1 p i2 ... p i j ... p i n 7
6 7
.. 6 . .. .. .. 7
. 6
6 .. . ... . ...
7
. 7
4 5
Am p m1 p m2 ... p mj ... p m n

where Ai denotes the proposed maintenance policies stated in Section 3.3.1 for i 1, 2,
3, . . . , m; Fj represents the functional failure causes identified in Module II for j 1, 2,
3, . . . , n; and p F j Ai is the average maintenance performance rating of each maintenance
policy Ai with respect to each functional failure cause Fj.
Once the matrix representation p has been completed, it will be normalized to
transform the different scales and units among the functional failure causes into
common measurable units to enable comparisons across the criteria (Chen, 2004).
The normalized value rij is calculated as follows:
pij
r ij q
Pn 2 ; j 1; . . . ; n; i 1; . . . ; m 3
j1 pij
where rij is the normalized preference measure of the ith maintenance policy in terms Maintenance
of the jth functional failure cause.
Each functional failure cause produces a different failure impact to the system and
policy selection
must have a different priority in the MPS process. The functional failure cause that has model
a higher failure impact must be the main concern in the selection process. Thus, the
RPN computed in Module II is converted to represent the weight (w) of each functional
failure cause to form a normalized weighted decision matrix. The weight is given in the 421
following equation:
wFj RPNFj 4
where wFj represents the weight for each functional failure cause Fj, for j 1, 2, 3, . . . ,
n; and RPN is the risk priority number obtained from equation (1).
The normalized weighted decision matrix V is then formed as equation (5):
2 3
wF1 r 11 wF2 r 12 . . . wFj r 1i . . . wFn r 1n
6 7
6 wF1 r 21 wF2 r 22 . . . wFj r 2i . . . wFn r 2n 7
6 7
V RW 6 6 .. .. .. .. 7 7 5
6 . . ... . ... . 7
4 5
wF1 r m1 wF2 r m2 . . . wFj r ji . . . wFn r mn

where w represents the weight of the functional failure cause, Fj, for j 1, 2, 3, . . . , n.
After V is formed, the ideal (A* ) and negative ideal (A*2 ) maintenance solutions of
each functional failure cause are determined, denoted as:
A* {max vij j j [ J 1 ; i 1; 2; 3; . . . ; m} {v1 ; v2 ; . . . ; vn }; 6
A*2 {min vij j j [ J 1 ; i 1; 2; 3; . . . ; m} {v1 ; v2 ; . . . ; vn 2 } 7
where vij is the weighted normalized value that indicates the average performance
rating of each maintenance policy Ai with respect to each functional failure cause Fj, for
J1 {j 1, 2, 3, . . . , n}.
Based on equations (6) and (7), the most preferable and the least preferable
maintenance solutions for each functional failure cause are identified. Once these
maintenance solutions are computed, the overall performance of the maintenance policy
is then determined by using the n-dimensional Euclidean distance to the ideal and
negative ideal maintenance solutions. A is the distance of each maintenance policy
from the ideal maintenance solution and is defined as:
v
uX
u n  2
A t
vij 2 v j for i 1; 2; 3; . . . ; m: 8
j1

The distance from the negative ideal maintenance solution (A2 ) is defined as:
v
uX
u n  2
A2 t vij 2 v2 j ; for i 1; 2; 3; . . . ; m: 9
j1
JMTM After obtaining each of the maintenance policy separation distances, the relative
closeness of each maintenance policy to the ideal maintenance solution is calculated to
25,3 rank the proposed maintenance policies. The relative closeness to the ideal maintenance
solution is defined as:
A2
i
C ; 10
422 A
i A2
i

where C denotes relative closeness, with 0 , C , 1 for i 1, 2, 3, . . . , m.


The maintenance policies can then be ranked by preference according to the
descending order of C (equation (10)). A larger index value denotes better performance
of the maintenance policy in preventing failure modes. The maintenance policy that
obtains the highest performance index is selected and then proposed as the optimal
maintenance policy for the CS.

3. Case study
A case study of a processing company in Malaysia was conducted to verify and to
validate the MPS model. The company is in the agriculture industry and is in the
business of extracting crude palm oil (CPO) from palm fruits. The company is small,
having approximately 40 employees with production running in two shifts.
The company management has set a target to achieve the maximum throughput of
the production line at 30 tons of palm fruits processed each hour. The whole production
line in the palm oil mill (POM) is constructed in a serial configuration. When the
continuous production line is in a serial configuration, the occurrence of failure causes
a stoppage in the whole production line. The stoppage has serious effects either on the
maintenance department or on the production department. Before the case study,
the high failure frequency that occurred because of unplanned maintenance affected the
production line throughput. The failure occurrences were largely attributed to the failure
on the part of the maintenance management to have proper guidelines to analyze failures
and subsequently identify a suitable solution either at the system level or component
level. Appropriate maintenance must be identified at the system level before proceeding
to an analysis for planning maintenance actions at the component level. Therefore,
the case study focused on selecting an optimal maintenance policy for the CS in the
production line based on the developed selection model. The application of the selection
model is discussed in detail in the following section.

3.1 Module I: scope identification


Referring to the MPS model shown in Figure 1 the most important CS must be
identified at the initial stage. Module I primarily aims to obtain information on the
major functions of the production line. The production line is then separated into
several systems based on their functions. Failure frequencies are collected and then
used to identify the CS as the subject of analysis in Module II. Module I is initiated by
process descriptions, followed by system separation, and then CS identification.
3.1.1 Production process description. The verification and validation process is
initiated by studying the POM production line. Fresh fruit bunches (FFBs), which are the
unprocessed fruit that are in bunches or in loose form, are generally derived from Elaeis
guineensis. FFBs are harvested when ripe and then transported to the POM by lorry. The
FFBs are tipped into a chute at the loading ramp. The FFBs are then loaded into the cages
beneath the chute and pushed into horizontal sterilizers for cooking. The sterilization Maintenance
process aims to inactivate the enzymes that produce free fatty acids, loosen the fruits policy selection
from bunches, soften the fruits, condition the nuts, and to coagulate the proteins.
Once the sterilization process is completed, the cages are hoisted up and then emptied model
into the feed hopper of the bunch stripper. The stripping process comprises a horizontal
rotating drum, in which each bunch is lifted and dropped several times to shake out the
fruits. The separated fruits are then transferred for an oil-extracting process. The empty 423
bunches that are produced from the separation process are conveyed to the dump site.
The fruits are elevated to digesters, which consist of a steam-jacketed vessel and are
fitted with stirring arms. The action of the arms breaks up the fruit, particularly the
oil-containing cells. The digester is linked to a continuous double-screw press. The press
squeezes out crude oil and press cake, which contain fruit fiber and nuts. The extracted
crude oil that comprises oil, water and dirt is clarified in the oil room to produce pure
crude oil.
The extracted crude oil flows through a vibrating screen into a settling tank to
separate the clean oil and the sludge layer. Dirt and moisture are removed from the oil
layer in a hermetically sealed purifying centrifuge, and the oil is dried in a vacuum
drier before storage. The sludge layer is passed through a small de-sanding cyclone
before going into a sludge centrifuge. The recovered oil from the centrifugal stage is
returned to the settling tank and the final sludge is sent to a waste treatment system.
Fiber, shell and solids that are removed by the screen are recycled to the digester.
The cake from the screw press passes down to the cake breaker conveyor.
The conveyor is specially designed to break up and to dry the cake by evaporation.
The broken-up cake, nuts and fiber are then fed into moving air columns that suck
most of the fiber. The nuts are dropped into a nut-polishing drum, which frees the rest
of the fibers so that these fibers can also be removed in the air stream. The nuts are
stored in a nut silo to reduce their moisture content by using a stream of heated air
before they are cracked. The nuts are cracked in a nut cracker for kernels extraction.
The mixture of kernels and shell fragments is first separated in a pneumatic column and
then in a hydro cyclone. The separated shell fragments are then sent to a dump site, and
the kernels are dried in a silo by using hot air before being sent for further processing.
Through these steps, all functions in the production line were studied and charted,
as discussed in Section 2. The knowledge gained regarding these functions is vital for
the preparation of the system separation process, which is conducted through the steps
outlined below.
3.1.2 System separation. According to the process description in the previous section,
the whole production line is divided into six systems based on their main functions.
The FFB are processed through six systems, which include sterilization, stripping, crude
oil extraction, crude oil clarification, nut cracking and kernel separation. These systems
are each numbered by using a functional code. A functional code is a numbering system
that is used for the breakdown order of the functional system. This code is essential to
provide traceability through each level of indenture. The functions that are identified in
the FBD at each level are numbered in a manner that preserves the continuity of the
functions and provides information on the function origin throughout the system.
The complete flow of the process is represented in the FBD depicted in Figure 2.
Referring to Figure 2, the function of the sterilization system is to sterilize the FFB.
Then, the sterilized fruits are transferred to the stripping system for separating
JMTM Fresh fruit bunches 1.0
25,3 (FFB) Sterilization

2.0
Empty bunches
Stripping
424
4.0 3.0 5.0
Crude oil clarification Crude oil extraction Nut cracking

Crude palm oil 6.0


(CPO) Kernel separation
Figure 2.
FBD of palm oil
production line Palm kernel

the fruits from bunches. Once the fruits have been completely separated, it will be
move to crude oil extraction system for extracting the crude oil. The extracted crude oil
is clarified in the crude oil clarification system to obtain a pure CPO. The nut extracted
from the fruits during crude oil extraction process is transferred to the nut cracking
system. The nut cracking system is used to extract the kernels from nuts by cracking
the shells. The final process is to separate the mixture of kernel and shells and this is
carried out in the kernel separation system.
The most critical issue in adopting FBD is defining the boundary between systems.
The reason is that the production line involved in this case study is a continuous process
and grouping machines that are located between systems is difficult. This approach
usually involves a transport machine such as a conveyor or an elevator. In this case
study, however, the grouping of these kinds of machines was determined based on the
processing stage and the function of the production line. When the production line had
been separated into systems, the process of critical identification is performed during the
CS selection stage.
3.1.3 CS selection. System failure frequency is the main reference for the CS
selection process. Failure frequency is defined as the number of failure occurrences
within a specific period. The failure frequency of each system was collected based on
the maintenance records from the maintenance department. To achieve better
accuracy in terms of collected failure frequency, failure records from the production
department were referenced. The total failure frequency of each system is shown in
Figure 3.
Based on the collected failure data shown in Figure 3, the stripping system had the
highest failure frequency recorded that is 26 times per month, followed by nut
cracking, kernel separation, sterilization, crude oil clarification. Crude oil
extraction system has a only seven times failure occurrence per month. This is the
lowest failure frequency among these systems. Thus, the stripping system which
had the highest failure frequency was assigned as CS and will be the focused in
Module II.
30 Maintenance
26 policy selection
25 model
22

20 18
16 425
15
11
10
7

0 Figure 3.
Failure frequency of
Sterilization Stripping Crude oil Crude oil Nut Kernel separated systems
extraction clarification cracking separation

3.2 Module II: CS assessment


CS assessment is the second module in the MPS model that will be validated. This
module comprises two sections: identification of operational flow and evaluation of
failure occurrence by using functional FMEA and expert judgment. The first step in
Module II is the detailing of the operational flow in the stripping system. When
operational flow is studied, critical analysis is performed to identify the failure
mechanisms involved in the stripping system. The failure impact is then quantified.
3.2.1 Functional identification. The sub-functional systems of the stripping system
are studied in this section. Each sub-functional system was given a code to simplify the
reorganization of each functional system in the stripping system. The sub-functional
systems were coded based on the FBD coding system. The coded functional flow is
shown in Figure 4.
As shown in Figure 4, the stripping system comprises six sub-functional systems.
The cage that contains the sterilized FFBs is lifted and discharged into the feeder. The
sterilized FFBs are then dispensed into the stripping drum for the separation process
through an auto feeder. Separating the fruits from the FFB bunches in the stripping
drum is the third sub-function. Then, the separated fruits are conveyed into the fruit
elevator using a screw conveyor. Finally, the separated fruits are transferred to the crude
oil extraction system through a fruit elevator. The empty bunches produced during the
separation process are then moved to the dumpsite by the empty bunches conveyor.
When the FBD system is developed, functional failure analysis can be performed
systematically by referring to the identified sub-function. Thus, the failure mechanism
underlying the sub-functions, which includes failure modes, causes, and failure effects,
can be investigated.
3.2.2 Functional failure analysis. Functional failure analysis is the second section of
Module II in terms of selection model validation. Functional FMEA was adopted
in this step to perform failure analysis on the CS. The first step in functional FMEA is
the selection of experts. The process comprises two stages that will be elaborated in
detail below.
JMTM 2.1
Cage with
25,3 sterilized FFB
Lifting the cage consists of sterilized
FFB and discharging into hopper
feeder

426 2.2
Feeding the sterilized FFB into the
stripping drum for separation process
feeder

2.3
Separating the fruits in the stripper
drum

2.4 2.6
Transfer the fruits into Transferring the empty Empty bunches
the fruit elevator bunches the dump site

2.5
Transferring the separated fruits Separated fruits
Figure 4.
FBD of stripping system to the crude oil extraction system

(a) Selection of experts. Qualified experts were identified based on the developed
criteria. The three criteria were: minimum of ten years of working experience in
workshop and maintenance with a minimum of seven years of industrial experience in
a related industry, capable of dedicating the required time to perform evaluations and
committed to participate, as required. Six candidates were interviewed. These
candidates held different positions, including mill manager, mill assistant (responsible
for the entire production process), and maintenance planner. Four technical personnel
in the maintenance department were also interviewed. Eventually, a team of five
qualified experts comprising engineers and foremen were invited to participate in the
functional FMEA elicitation. When the qualified experts were identified, a meeting was
conducted to introduce the project, to familiarize them with the expert judgment and
the elicitation process and to foster critical discussions of key evidence relevant to
the questions posed. This step aimed at reducing potential bias in judgment
(Roman et al., 2008).
(b) Functional FMEA elicitation. Functional FMEA elicitation was conducted at this
stage. This step comprised two stages. The first stage was in the form of a group
discussion aimed at collecting information on functional failure modes and failure
causes in the stripping system. The second stage involved individual face-to-face
interviews aimed at quantifying the criticality of each functional failure cause that was
identified during the group discussion.
During the discussion, the first priority was to determine the functional failure modes Maintenance
in each of the functional subsystems identified in Section 3.2.1. When the functional policy selection
failure modes were known, the causes for each were also identified. During the
discussion, 19 functional failure modes were identified. Considering that the occurrence model
of each functional failure cause has different impacts on the system, a second section
of elicitation was conducted to quantify these impacts. As previously mentioned,
this step was an individual activity. 427
The second section of the functional failure cause impact evaluation process involved
quantification based on the experts experience and knowledge. As generally known,
human judgment is very subjective and the answers provided will have a significant
effect on the final results. Thus, a simple statistical analysis was performed to test the
inter-rater reliability among the five experts. The inter-rater reliability was used to test
the degree of consistency on the answer provided by different experts in terms of scale
and to determine the consistency and reliability of the given results.
In the reliability testing, Cronbachs a was calculated based on three criteria:
occurrence, severity and detection. The consistency in evaluation among the five experts
for each criterion can be judged through this test. The values of the coefficients are
tabulated in Table IV.
The Cronbachs a coefficients for the three criteria were 0.953, 0.913 and 0.954.
The Cronbachs a coefficients were higher than 0.7. From the analysis, the rating given
by the experts were rationally explainable and did not show significant bias. This
result indicates that, by referring to these data, the final outcome is reliable.
(c) RPN computation. When the rating of each criterion was determined as reliable,
these data were then used in the RPN computation process. When identification and
calculation were completed, data were documented by using an FMEA worksheet.
This data set comprises the results of the functional failure analysis stage in Module II.
The functional FMEA data that were collected from the maintenance experts were
recorded in the FMEA worksheet that is attached in the Appendix.
The determination of functional failure causes and their RPN values is essential
for the maintenance policy evaluation in Module III because the performance of the
maintenance policies will be rated based on these failure causes. The RPN values of the
failure causes were converted into weighted values during the maintenance policy
ranking in Module III.

3.3 Module III: maintenance policy evaluation


Maintenance policy evaluation is the final module in the MPS model. This module is
divided into two stages. The first stage is the proposal of maintenance policies that are
applicable to the selection. The second stage is the evaluation and ranking (by using
TOPSIS) of the performance of these maintenance policies to reduce the failure impact
of the failure causes identified in Module II.

Criteria Number of expert Number of question Cronbachs a Judgment

Occurrence 5 128 0.953 Good


Severity 5 128 0.913 Good Table IV.
Detection 5 128 0.954 Good Reliability statistics
JMTM 3.3.1 Proposal of maintenance policies. Four maintenance policies (i.e. the autonomous
25,3 maintenance policy, (AM, A1), corrective maintenance policy, (CM, A2), PdM policy,
(A3) and PM (A4)] were proposed as options in this research.
3.3.2 Ranking of maintenance policies. After maintenance policies were proposed,
the performances of these proposed maintenance policies were evaluated by a selected
expert (identified in Module II). The process was conducted in two stages; the first
428 stage was the maintenance policy performance elicitation, and the second stage was
the performance analysis.
(a) Maintenance policy performance elicitation. The five experts who participated in
functional FMEA were asked to give their respective performance ratings on the
proposed maintenance policy. During the interview, the experts were asked to assess
the performance of each of the maintenance policies in reducing the failure effect
according to three criteria: occurrence (O), severity (S) and detection (D). At the end of
the elicitation process, the performances of the maintenance policies corresponding to
the three maintenance criteria were rated and the analyzed by using TOPSIS.
To identify the reliability and bias of the rating, the two-way mixed effects model
type with 95 percent confidence interval intra class correlation was used to compute the
Cronbachs a coefficient. In the reliability computations, the Cronbachs a coefficient
was calculated based on three criteria: failure frequency, failure duration and detection.
The consistency in the evaluation of each criterion among the five experts can be judged
through this analysis. The values of the coefficients are tabulated in Table V.
All three criteria had Cronbachs a values of more than 0.7. This result suggests the
high reliability of the experts. From the analysis, the ratings given by the experts were
rationally explainable and did not show significant bias. By referring to these data,
the final outcome will be reliable. When the first stage elicitation of the proposed
maintenance policy is computed, performance analysis can be conducted.
(b) Maintenance policy performance analysis. Referring to the first stage, five sets of
data on the maintenance policy ratings were obtained from five maintenance experts.
Among the four maintenance policies, all experts gave the lowest rating to
maintenance policy A2 (CM), followed by A1 (AM). A3 (PdM) and A4 (PM) had almost
similar and higher ratings compared with A1 and A2. Every functional cause had a
different significance to system function. Thus, identifying the weight of the
significance of each functional failure cause and incorporating such significance into
the normalized decision matrix is necessary. The maximum and minimum values of
each functional failure cause in the weighted normalized decision matrix were
identified. The maximum value is the positive ideal solution, and the minimum value is
the negative ideal maintenance. The results are tabulated in Table VI.
Referring to the results in Table VI, A2 had the farthest distance (578) from the
positive ideal solution. A1 had been separated by 262 from positive ideal distance.
A3 and A4 had similar distances between their A and A2 . The relative closeness of
each maintenance policy was determined. When the calculation of relative closeness

Criteria Number of experts Number of question Cronbachs a Judgment

Occurrence 5 128 0.911 Good


Table V. Severity 5 128 0.890 Good
Reliability statistics Detection 5 128 0.901 Good
was completed, the maintenance policies were ranked based on a descending value of Maintenance
relative closeness. From the results shown in Table VII, A4 had the largest relative policy selection
closeness distance to ideal maintenance, followed by A3 and A1. A2 had the farthest
maintenance policy from ideal maintenance. model
PM had the highest ranking, although it had a similar value as PdM (A3). However,
the objective of this model is to select the optimal maintenance policy. Thus, the PM
policy was chosen as the optimal maintenance policy for the stripping system. 429
4. Discussion
The MPS model was validated by using a case study from a certain industry. The
purpose of model validation is to make the model useful in the sense that the model
addresses the appropriate problem, provides accurate information about the system
being modelled and ensures that the model is applicable in the industry. Validation by
using a case study not only provides evidence of the feasibility and practicability of the
developed model but also tests the acceptability and the rationale of the developed
model from the industry perspective. In addition, by conducting a case study, a clear
and complete picture of the industrial environment including a companys needs and
expectations as well as the constraints that are actually faced in the industry can also
be considered.
In this paper, model validation focused on identifying the optimal maintenance
policy for a system to demonstrate that the developed model is applicable in actual
conditions. Effectiveness remains an important issue in any industry. Everything must
be fast and effective, with the best outcome obtained within the shortest time. One of
the ways to achieve the highest effectiveness is by placing high importance on the
critical issue. In this case study, the system that caused the most problems and has
become the bottle-neck in the production line was identified before attempting to find
the solution. According to Krishnan (1992), different criteria such as downtime, failure
frequency and spare parts may be referenced as identification guidelines. A similar
criterion, failure frequency, has also been adopted. Consequently, the CS can be easily
and clearly identified. Through the selection process, a CS was chosen, which proves
the robustness of Module I of the MPS model.

Maintenance polices Positive distance, A Negative distance, A2

A1 262 378 Table VI.


A2 578 0 Positive distance and
A3 162 484 negative distance of
A4 162 503 maintenance policy

Maintenance polices Relative closeness, C Ranking

A1 0.410 3 Table VII.


A2 0.000 4 Relative closeness of
A3 0.749 2 maintenance policy to the
A4 0.756 1 ideal maintenance
JMTM Technical analysis in Module II was performed to obtain useful insight into the failure
25,3 mechanism underlying the CS. The obtained results show that the model is valid in
terms of credibility and plausibility. The result of the discussion among experts with an
average of 19 years of working experience was reliable, which is consistent with
Dimattia et al. (2005), who stated that experts with more than ten years working
experience in a related industry could provide reliable judgment. However,
430 all measurements and judgments were subject to experimental and judgment errors
(Park and Lee, 2008). Relative to error, one of the crucial problems is that error can result
in inconsistency. A built-in consistency checking mechanism will improve the accuracy
of a judgment, in addition to the overall consistency.
In the model, the consistency of the information provided as well as the reliability of
the experts was confirmed by using intra-judgment consistency. According to Sekaran
(2002), a Cronbachs a coefficient higher than 0.7 shows good inter-rater reliability,
which is consistent with the results obtained from individual judgments (referring to
Table IV). Thus, these results provide evidence that the judgments provided by the five
experts are consistent and precise, thereby supporting the validity and reliability of
expert judgments and of the results generated by the model. The robustness of the model
was also proven.
When the failure mechanisms were known, a suitable maintenance policy was
selected in Module III. Similar to Module II, an expert judgment was adopted to
evaluate the maintenance policy performance. The data were validated and found
reliable because the Cronbachs a coefficient was more than 0.7. This result also proves
the trustworthiness of the experts. The final results obtained from the developed model
clearly show the ranking of each proposed maintenance policy. The results were
consistent with those of several others, such as those of Bevilacqua and Braglia (2000),
Al-Najjar and Alsyouf (2003) and Waeyenbergh and Pintelon (2004), who performed
case studies of different manufacturing industries aimed at reducing the failure
impacts in these companies. These results prove that similar results were obtained
with greater ease and simplicity by using the MPS model compared with the model
developed by the authors.
The good results obtained in this case study can largely be attributed to the
integration of different methods, including tally chart, FMEA and TOPSIS, thus making
the MPS model unique. Moreover, the integration of quantitative expert judgment to
quantify subjective value increased the accuracy of our results. This scheme is
supported by Spurgin (2002), who stated that high-quality results depend on experts in
related fields and on feedback based on their experience. Waeyenbergh and Pintelon
(2002) also agreed that experienced and knowledgeable personnel have adequate ability
to provide useful judgment. The reason is that such personnel are the most familiar with
the actual operation of the production line and thus have adequate knowledge in the
maintenance aspect.
Another significant aspect of the MPS model is the integration of the quantitative
aspect that enabled the expression of the MPS outcome to be in a numerical form.
Quantitative expression is not only easier to understand compared with qualitative
expression, but also brings more accurate meaning that will increase the accuracy of
judgment. This hypothesis is supported by Keeney and Winterfeldt (1989), who stated that
qualitative terms, such as small chance, have large ranges of interpretation, depending
on who had been asked. Moreover, evaluations performed by using quantification
values can also be analyzed based on consistency and reliability. The analysis will bolster Maintenance
the reliability of the results that were obtained from the model. policy selection
Hence, the developed model is reliable, valid and applicable in the manufacturing
industry. The MPS model can be used as a new MPS method with full assurance. model
Through the discussion of the significant aspects of the MPS model, the developed
model was proven robust based on validity, accuracy, reliability and practicability.
In other words, the MPS model can be applied in the selection of a maintenance policy 431
in the manufacturing industry.
Optimal maintenance policy is a key decision in the manufacturing industry for the
improvement of productivity performance (Lu and Sy, 2009). With the appropriate
maintenance policy, maintenance management becomes more effective and agile by
having better planning on maintenance, human resources and spare parts arrangement
without disrupting the production schedule. The optimal maintenance policy that was
determined by using the MPS model is expected to reduce the failure impact of the
system based on product, system, people and environment. Apparently, malfunction
could affect productivity as well as product quality. Failures eventually reduce the
useful life of the system. Notably, possible employee injury caused by these failures
can be reduced by the maintenance policy that was determined from the MPS model.
The policy can further enhance productivity, which heavily relies on employee
performance. On the environmental aspect, a maintenance policy is expected to prevent
catastrophic failures that may cause pollution to the environment and against
government regulations. Particularly, the optimal maintenance policy determined
from the MPS model can reduce not only the failure impact on the system but also
company losses. This effect will enhance competitiveness among manufacturing
industries.

5. Conclusion
An example of a successful implementation of a model for the selection of an optimal
maintenance policy was demonstrated in this paper. The MPS model can be effectively
used for quantifying the performance of each maintenance policy, thereby reducing the
failure impact on the system and selecting the optimal maintenance policy according to
quantified performance values. On the basis of the obtained results, this model can aid
companies in selecting the most suitable maintenance policy for the system and in
achieving more effective maintenance planning. The ultimate aim is to enhance the
companys maintenance effectiveness by implementing the optimal maintenance
policy. Moreover, the developed model is capable of providing accurate results within
a short time, which is suitable for industries that require effective decisions within a
short time.
Regarding future studies and developments, a computerized integration with the
developed maintenance selection model should be developed to attain better results
within a significantly shorter time. In actual industries, everything must be fast and
accurate, and the maintenance team must be capable of adapting well. Developing a
simple but effective information collection system must also be considered. During
analysis, the common problems found were the lack of necessary information and the
accuracy of information. Thus, a systematic procedure for recording information for
analysis is necessary. The developed information collection system must be user
friendly to enhance its practical applicability.
JMTM References
25,3 Al-Najjar, B. and Alsyouf, I. (2003), Selecting the most efficient maintenance approach using
fuzzy multiple criteria decision making, International Journal of Production Economics,
Vol. 84 No. 1, pp. 85-100.
Bertolini, M. and Bevilacqua, M. (2006), A combined goal programming AHP approach to
maintenance selection problem, Reliability Engineering & System Safety, Vol. 91 No. 7,
432 pp. 839-848.
Bevilacqua, B. and Braglia, M. (2000), The analytic process applied to maintenance strategy
selection, Reliability Engineering & System Safety, Vol. 70 No. 1, pp. 71-83.
Chen, M.F. (2004), Combining grey relation and TOPSIS concepts for selecting an expatriate
host country, Mathematical and Computer Modelling, Vol. 40 No. 13, pp. 1473-1490.
DiMattia, D.G., Khan, F.I. and Amyotte, P.R. (2005), Determination of human error probabilities
for offshore platform musters, Journal of Loss Prevention in the Process Industries, Vol. 18,
pp. 488-501.
Ilangkumaran, M. and Kumanan, S. (2009), Selection of maintenance policy for textile industry
using hybrid multi-criteria decision making approach, Journal of Manufacturing
Technology Management, Vol. 20 No. 7, pp. 1009-1022.
Jou, Y.T., Wee, H.M., Chen, H.C., Hsieh, Y.H. and Wang, L. (2009), A neural network forecasting
model for consumable parts in semiconductor manufacturing, Journal of Manufacturing
Technology Management, Vol. 20 No. 3, pp. 404-412.
Keeney, R.L. and Winterfeldt, D.V. (1989), On the uses of expert judgment on complex technical
problems, IEEE Transactions on Engineering Management, Vol. 36 No. 2, pp. 83-86.
Krishnan, N.T.B. (1992), A simulation model for maintenance planning, Reliability and
Maintainability Symposium 1992 Proceedings of the International Conference, Las Vegas,
NV, USA, pp. 109-118.
Li, C. and Xu, M. (2007), ELECTRE III based on ranking fuzzy numbers for deterministic and
fuzzy maintenance strategy decision problems, Automation and Logistics Proceedings of
the IEEE International Conference, Jinan, China, pp. 309-312.
Lu, K.Y. and Sy, C.C. (2009), A real-time decision-making of maintenance using fuzzy agent,
Expert Systems with Applications, Vol. 36 No. 2, pp. 2691-2698.
Martin, F. (2008), A performance technologists approach to process performance
improvement, International Society for Performance Improvement, Vol. 47 No. 2, pp. 30-40.
Park, K.S. and Lee, J.I. (2008), A new method for estimating human probabilities: AHP-SLIM,
Reliability Engineering & System Safety, Vol. 93 No. 4, pp. 578-587.
Roman, H.A., Walker, K.D., Richmond, H.M., Hubbell, B.J. and Kinney, P.L. (2008), Expert
judgment assessment of the morality impact of changes in ambient fine particulate matter
in the US, Environmental Science & Technology, Vol. 42 No. 7, pp. 2268-2274.
Savsar, M. (2011), Analysis and modeling of maintenance operations in the context of an oil
filling plant, Journal of Manufacturing Technology Management, Vol. 22 No. 5,
pp. 679-697.
Sekaran, U. (2002), Research Methods for Business: A Skill Building Approach, Wiley,
Hoboken, NJ.
Shyjith, K., Ilangkumaran, M. and Kumanan, S. (2008), Multi-criteria decision-making approach
to evaluate optimum maintenance strategy in textile industry, Journal of Quality in
Maintenance Engineering, Vol. 14 No. 4, pp. 375-386.
Spurgin, A.J. (2002), Critique of current human reliability analysis methods, Human Factor and
Power Plant 2002 Proceedings of the International Conference, pp. 12-18.
Waeyenbergh, G. and Pintelon, L. (2002), A framework for maintenance concept development, Maintenance
International Journal of Production Economics, Vol. 77 No. 3, pp. 299-313.
Waeyenbergh, G. and Pintelon, L. (2004), Maintenance concept development: a case study,
policy selection
International Journal of Production Economics, Vol. 89 No. 3, pp. 395-405. model
Wang, L., Chu, J. and Wu, J. (2007), Selection of optimum maintenance strategies based on a
fuzzy analytical hierarchy process, International Journal of Production Economics,
Vol. 107 No. 1, pp. 151-163. 433
Yang, T.H. and Hung, C.C. (2007), Multiple-attribute decision making methods for plant layout
design problem, Robotics and Computer Integrated Manufacturing, Vol. 23 No. 1,
pp. 126-137.
Yang, Z., Bonsall, S. and Wang, J. (2008), Fuzzy rule-based Bayesian reasoning approach for
prioritization of failures in FMEA, IEEE Transactions on Reliability, Vol. 57 No. 3,
pp. 517-528.

Further reading
Bogonovo, E., Marseguerra, M. and Zio, E. (2000), A Monte Carlo methodological approach to
plant availability modelling with maintenance, aging and obsolescence, Reliability
Engineering & System Safety, Vol. 67 No. 1, pp. 61-73.

About the authors


Siew-Hong Ding received the BEng degree from Universiti Sains Malaysia, Malaysia in 2007, the
MSc degree also from Universiti Sains Malaysia, Malaysia, in 2010. Currently, Siew-Hong Ding is
a PhD candidate in the School of Mechanical Engineering at the University Sains Malaysia.
Her research interest is on industrial maintenance management.
Shahrul Kamaruddin received the BEng (Hons) degree from University of Strathclyde,
Glasgow, Scotland in 1996, the MSc degree from University of Birmingham, UK, in 1998, and the
PhD from University of Birmingham, in 2003. Currently, Shahrul Kamaruddin is an Associate
Professor with the School of Mechanical Engineering (under the manufacturing engineering with
management programme), Universiti Sains Malaysia. He has various past experiences with
manufacturing industries from heavy to electronics industries especially in the field of industrial
engineering, manufacturing processes and product design. He has more than 20 publications in
reputed international and national journals/conferences. His current research interests include
simulation and modelling of manufacturing systems, production planning and control,
maintenance management and application of artificial intelligence techniques in manufacturing.
Shahrul Kamaruddin is the corresponding author and can be contacted at: meshah@eng.usm.my
Ishak Abdul Azid is an Associate Professor in the School of Mechanical Engineering at the
University Sains Malaysia. His research interests included structural analysis, finite element
method (FEM) and genetic algorithm in optimization.

(The Appendix follows overleaf.)

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com


Or visit our web site for further details: www.emeraldinsight.com/reprints
25,3

434
JMTM

Table AI.
FMEA worksheet
Functional
code Functional mode Functional failure mode Functional failure causes O S D RPN
Appendix

2.1 Lift the cages and pour the sterilized FFB Cannot move forward/backward or Brake jammed 4.0 6.4 8.0 204
into hopper feeder tilting/boom or up/down Coupling broken 4.0 3.4 8.0 109
Electrical fault 4.8 5.2 9.2 230
Gear box failed 4.0 6.0 7.0 168
Motor burnt 5.0 4.4 9.8 216
Shaft broken 4.6 5.8 8.0 213
Liner broken Wear and tear 4.2 9.0 4.2 159
2.2 Feed the sterilized FFB into stripper Hopper cannot move Coupling broken 1.6 4.0 6.0 38
drum Electrical fault 5.6 4.4 8.8 217
Gear box failed 2.8 3.2 6.0 54
Motor burnt 3.0 6.0 7.2 130
Shaft broken 1.4 5.6 8.4 66
2.3 Separate fruits from the bunches Jammed and tripped Electrical fault 3.8 6.4 8.6 209
Gear box failed 4.8 3.6 6.0 104
Motor burnt 4.2 5.4 9.4 213
Shaft broken 1.6 5.4 8.4 73
Vee belt loose 3.2 6.0 3.6 69
Stripper drum chocked Insufficient clearance 2.0 7.4 3.4 50
2.4 Transfer the fruits into the fruit elevator Jammed and tripped Coupling broken 6.6 5.6 6.0 222
Drive chain failed 3.2 2.0 9.0 58
Electrical fault 1.4 6.0 8.0 67
Gear box failed 4.2 3.2 4.6 62
Hanger bearing failed 3.0 5.4 9.4 152
Motor burnt 4.2 3.6 6.0 91
Shaft broken 2.6 5.4 6.8 95
Ribbon broken Ribbon wear and tear 3.0 5.8 5.4 94
2.5 Transport the fruits to crude oil Chain came out from sprocket gear Bottom adjusting bearing 4.8 6.6 1.4 44
extraction system failed
Chain elongated one side 1.6 5.2 3.2 27
(continued)
Functional
code Functional mode Functional failure mode Functional failure causes O S D RPN

Sprocket gear teeth 3.4 7.0 3.8 90


broken
Stuck by empty bunches 3.2 7.4 3.6 85
Chain snapped Chain came out from 3.8 6.4 6.2 151
sprocket gear
Chain link failed 3.4 6.2 6.2 131
Liner failed 4.8 5.4 3.0 78
Stuck by empty bunches 6.6 4.2 3.8 105
Jammed and tripped Coupling broken 1.6 8.0 8.0 102
Drive chain failed 3.0 3.2 4.8 46
Electrical fault 4.4 4.2 9.0 166
Gear box failed 3.6 4.4 6.8 108
Motor burnt 3.0 5.2 8.6 134
2.6 Transfer the empty bunches to dump site Chain came out from sprocket gear Bottom adjusting bearing 4.8 7.0 1.4 47
failed
Chain elongated one side 4.6 4.4 4.0 81
Sprocket gear teeth 3.4 7.0 3.8 90
broken
Stuck by empty bunches 3.2 7.4 3.8 90
Chain snapped Chain came out from 5.2 6.2 6.2 200
sprocket gear
Chain link failed 4.4 6.2 6.2 169
Liner failed 5.6 6.6 3.2 118
Stuck by empty bunches 6.6 4.2 3.8 105
Jammed and tripped Coupling broken 3.2 6.0 8.0 154
Drive chain failed 2.4 4.0 4.8 46
Electrical fault 3.4 6.4 9.2 200
Gear box failed 3.8 4.4 6.6 110
Motor burnt 3.0 5.4 9.2 149
Maintenance
policy selection
model

Table AI.
435

Anda mungkin juga menyukai