You are on page 1of 15

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/330464112

Spare parts’ criticality assessment and prioritization for enhancing


manufacturing systems’ availability and reliability

Article in Journal of Manufacturing Systems · January 2019


DOI: 10.1016/j.jmsy.2019.01.003

CITATIONS READS

30 12,534

2 authors:

Katarzyna Antosz R.M. Chandima Ratnayake


Rzeszów University of Technology University of Stavanger (UiS)
79 PUBLICATIONS 789 CITATIONS 235 PUBLICATIONS 1,822 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by R.M. Chandima Ratnayake on 26 February 2019.

The user has requested enhancement of the downloaded file.


Journal of Manufacturing Systems 50 (2019) 212–225

Contents lists available at ScienceDirect

Journal of Manufacturing Systems


journal homepage: www.elsevier.com/locate/jmansys

Spare parts’ criticality assessment and prioritization for enhancing T


manufacturing systems’ availability and reliability
Katarzyna Antosza, , R.M. Chandima Ratnayakeb

a
Rzeszow University of Technology, Faculty of Mechanical Engineering and Aeronautics, Department of Manufacturing and Production Engineering, Al. Powstancow
Warszawy 8, 35-959, Rzeszow, Poland
b
Department of Mechanical and Structural Engineering and Materials Science, University of Stavanger, PO. Box 8600 Forus, N-4036, Stavanger, Norway

ARTICLE INFO ABSTRACT

Keywords: Manufacturing systems (MSs) have become more and more complex, due to global competition. Optimal spare
Manufacturing systems parts provisioning plays a critical role in sustaining an anticipated operational competitiveness level, via effi-
Availability cient and effective maintenance of machinery. The forecasting of intermittent demand for spare parts is a
Reliability challenge, as it is not always possible to avoid random unforeseen breakdowns, which reduce availability and
Spare parts
increase the unreliability of manufacturing systems. The aforementioned requires a systemic perspective, in
Maintenance
Prioritization
order to perform a criticality analysis and prioritization of the spare parts needed to increase a manufacturing
Empirical model system’s availability and reliability. This paper first demonstrates the development of an empirical model (EM),
Analytic hierarchy process using a case study MS. The EM enables a criticality analysis to be performed, considering the system perspective
of spare part management, by taking maintenance-related and logistics-related factors into account. After that,
the machineries are categorized into groups, considering the factors related to the maintenance, logistics and
criticality levels. The second part presents how to perform spare part prioritization within a selected group, via
an analytic hierarchy process (AHP), to minimize ad hoc suboptimal assessments, together with sensitivity
analyses. Finally, it presents the spare part prioritization and the subsequent sensitivity analysis results.

1. Introduction inventories such as work-in process and finished products” [2]. There
are two main differences (i.e. in relation to the functionality and the
It is important to maintain spare parts inventories for serving the policy for managing the inventory) between these two types of in-
needs of the maintenance and replacement of operating machinery in ventories [8]. In this context, the spare parts inventory levels are pri-
manufacturing systems (MSs) [1,2], in order to enhance the availability marily dependent on the ‘function’: how machineries are utilized and
of expensive and technologically advanced/complex systems [3]. Many how they are maintained [9], while the key decision variable (i.e. op-
industries, such as manufacturing, oil and gas production, aerospace timal stocking level) remains the same for spare parts, as well as for
and defense, transportation, telecommunications and information work-in-progress and final finished-product inventories [2]. However,
technology, utilities and the supply of durable goods, rely on the ef- the demand for spare parts is ‘lumpy’. This means that the demand for
fective management of spare parts [4]. Spare part management is a spare part items is typically intermittent, with orders arriving spor-
significantly challenging task, particularly due to the fact that “The adically, and the demand can also be highly variable [10]. What is
parts can be expensive, their demand is highly erratic and intermittent, more, the demand only arises at system breakdowns. Hence, the in-
yet their shortage costs can be very large” [5]. The aforementioned has termittent, sporadic and highly variable nature of spare part demand
been further exacerbated by the high obsolescence risk, due to spare diminishes the use of forecasting techniques with the same quality,
parts’ specific functionalities. Hence, it is vital to find the right balance which unable to estimate demand for the finished products [11].
between the cost of spare part acquisition, inventory holding, stock-out Spare part assessment and control has acquired great interest in the
and obsolescence costs, while focusing on minimizing downtime costs literature to date. The studies cover a wide range of relevant research
(i.e. opportunity cost in the case of lost production, liability costs, and/ areas such as spare part inventory control, supply chain management,
or loss of goodwill) [6,7]. maintenance, demand pattern, item value, system reliability and per-
The “management of spare parts differs from other manufacturing formance, etc. For instance, [12] demonstrated the management of


Corresponding author.
E-mail addresses: katarzyna.antosz@prz.edu.pl (K. Antosz), chandima.ratnayake@uis.no (R.M.C. Ratnayake).

https://doi.org/10.1016/j.jmsy.2019.01.003
Received 12 October 2018; Received in revised form 21 December 2018; Accepted 10 January 2019
Available online 17 January 2019
0278-6125/ © 2019 Published by Elsevier Ltd on behalf of The Society of Manufacturing Engineers.
K. Antosz, R.M.C. Ratnayake Journal of Manufacturing Systems 50 (2019) 212–225

spare parts in order to reduce the holding costs, focusing on the spare
part distributors’ point of view, by adopting a probabilistic approach.
Based on real-life spare part networks, the work of [13] introduced a
spare part inventory model with lateral transhipment. Ref [14].’s work
discussed how to handle complex circumstances in relation to main-
tenance processes, focusing on the spare parts logistics. Durán, Macchi
et al. [15] adopted a simulation for testing a specific spare part in-
ventory policy that focuses on a continuous review system in order to
assess the effects on operational performance of an industrial commi-
nution plant on its total cost of ownership. Hellingrath, Pereira et al.
[16] demonstrated how to improve the effectiveness and efficiency of
managing the spare part supply chains associated with complex man-
ufacturing systems. Behfard, van der Heijden et al. [17] developed a
heuristic method to investigate the near-optimal last time buy quantity
in the presence of an imperfect repair option for failed spare parts
which are returned from the field. The study in [18] demonstrated the Fig. 1. Spare part criticality categories [19].
use of the Bayesian model as a decision support tool for the choice of a
policy and procurement planning of spare parts, focusing on the
maintenance of machinery in operational conditions and avoiding associated with the consequences of a failure or shortage of an item/a
stock-outs. Molenaers et al. [19] showed a criticality classification of spare part, whereas control criticality is associated with the possibility
spare parts, focusing on stock-out effects, item value and demand pat- of controlling the availability of a spare part [19].
tern. However, none of these authors managed to take into account the Hence, it is evident that the concept of criticality depends on a
resultant influence of spare part demand on different machinery in number of factors, making the criticality assessment significantly dif-
relation to: the physical model (design, technology, etc.); the logistic ficult in a formal and uniform way. However, approaches for the ca-
model (internal organization, support processes, resources, etc.); the tegorization of spare parts, by assigning appropriate criticality levels,
behavioral model (robustness, aging, obsolescence, etc.); and the eco- were demonstrated in the standards and research studies [5,9]. Many
nomic model (cost reduction, optimization of resources, etc.). models have been developed, which focus on: What to stock? Where to
Kennedy et al. [8] provided a classification process to categorize the stock? How much to stock? However, due to the inherent variable
spare parts into useful classes, based on one or more relevant attributes. nature of spare parts, it is vital to have an overall systemic perspective
ABC analysis has widely/traditionally been employed as a classification for spare part assessment and control, based on an item’s criticality and
method in industrial organizations to determine the service requirements a subsequent prioritization analysis. Hence, it is necessary to investigate
of spare parts, in which the items are ranked in terms of a single cri- practical and realistic approaches, such as those presented by Mole-
terion, such as a demand value or a consumption pattern. For instance, naers et al. [19], Braglia et al. [31], Syntetos et al. [20], Wang et al.
Syntetos et al. [20] demonstrated the use of ABC analysis in order to [35] and Chen et al. [36], rather than re-inventing approaches which
characterize spare parts, based on a demand value for an electronics are purely theoretical in nature.
manufacturer. ABC analysis is not considered the most suitable method, Such models were developed in works [37–39]. In [37], condition-
as it is based on one parameter, while the assortment of industrial spare based spares’ ordering for critical components is proposed. In this
parts is significantly heterogeneous [21–23]. Hence, the traditional ABC paper, the ordering decision depends on the remaining useful life (RUL)
classification was extended [e.g. weighted linear optimization, Artificial estimation, obtained through the assessment of component age and
Neural Networks (ANN), etc.] to a multi-criteria ABC analysis, including condition indicators (covariates), which are indicative of the state of
other parameters like unit cost, critical factor and lead-time, by different health of the component, at every inspection time. The authors consider
researchers [21,22,24–26]. However, there are inherent drawbacks a random lead time for spares and a single-component, single-spare
present in the multi-criteria classification models. For instance, ANN configuration, which is not uncommon for very expensive and highly
models are not suitable for incorporating qualitative parameters, critical equipment. In [38], the authors focus on condition managed
whereas weighted linear models are not suitable for incorporating cate- critical spares (CMCS): namely, spares which are expensive, highly re-
gorical measurements. Moreover, none of them is capable of entirely liable, with higher lead times, and not available in store. The aim of the
replacing professional judgments [22]. In this context, the multi-criteria paper is to orient the decision time for CMCS ordering or to just con-
analysis approach, AHP, enables both qualitative and quantitative cri- tinue the production. Ref [39]. presents an inventory pooling model for
teria to be considered in a classification/prioritization scheme, at the spare units of critical systems that serve multi-companies. Companies
same time enabling the assignment of weights to the different parameters located in the same geographical area can benefit from a coalition to
[27–29]. For instance, Gajpal et al. [30] and Braglia et al. [31] demon- share the storage cost of such spare units, which may be stored as one
strated the use of the AHP approach in criticality-based spare part clas- unit at only one company's location. In the presented model, the storage
sification. However, the AHP approach has inherent limitations, as it cost, replacement costs and downtime cost are amalgamated into a
undergoes a high level of subjectivity in performing pairwise compar- unified cost function, giving the expected cost per company per unit
isons. Moreover, the case study performed by Botter and Fortuin [32] time as our objective function. The optimal solution, with regard to
(i.e. to distinguish important spare parts from others, based on several these two decision variables under a certain cost structure, is presented,
criteria) revealed that the AHP approach is too theoretical and not ac- and a simulation method verifies the correctness of the model. A series
ceptable by the management. Hence, it was modified to a simplified of sensitivity analyses are also conducted to show which variable has
spare part functionality-based VED (Vital, Essential, Desirable) approach. the most influence on the expected cost. The aforementioned models
It is possible to classify the spare parts from a maintenance view- can be fulfilled by the development of an empirical model, based on a
point, using classification criteria such as: machine failure, lead times, criticality analysis and the subsequent prioritization practice, using a
supplier’s reliability and item criticality. In this context, the criticality multi-criteria analysis that enables easy and uniform use as an assess-
of a spare part is expressed as “the impact a shortage can have on ment and control tool for reinforcing the practitioners’ decision-making
production, safety and environment” [19]. The aspects related to cri- process. This enhances the transparency and minimizes the ad hoc
ticality are broadly divided into two categories: process criticality and decision making, resulting in less variability in the overall performance.
control criticality [33,34]. Fig. 1 illustrates that process criticality is Additionally, the spare part criticality assessment will help to increase

213
K. Antosz, R.M.C. Ratnayake Journal of Manufacturing Systems 50 (2019) 212–225

the machine’s system availability and reliability. Therefore, it is vital to the event of a failure in a given time period (Eq. (5)) [43].
develop system perspective criticality and multi-criteria [e.g. the ana-
Total maintenance time (TMT )
lytic hierarchy process (AHP)] analysis approaches for spare part in- MTTR =
Number of repairs (5)
ventory assessments and subsequent prioritizations.
This work starts by developing an empirical model to perform a cri- The Mean Time to Repair (MTTR) is the average time needed to solve
ticality analysis, considering the system perspective of spare part man- the problem and repair the damaged equipment and restore it to normal
agement, by taking maintenance-related (i.e. machine categories, spare working conditions. Maintenance time (MT) is defined as the time from
part replacement time, complexity of the replacement process, failure the beginning of the event until the system is restored to production. MT
type, failure frequency and employee qualification) and logistics-related includes notification time, waiting time for service, diagnostic time, re-
(i.e. part cost, lead time, storage, and number of potential suppliers) pair time, calibration, test time, return to production, etc. The average
factors into account. Then, the machinery is categorized into groups, repair time allows how an organization can respond to a problem and
considering factors related to the maintenance, logistics and criticality eliminate it to be assessed. MTTR provides a wealth of information that
levels. Finally, it demonstrates how to perform spare part prioritization can help reliability engineers make informed decisions such as repair or
within a selected group (i.e. using AHP) in order to minimize the ad hoc replacement, rental, optimization of maintenance schedules, storage of
suboptimal assessments and associated sensitivity analysis. on-site spare parts or change of spare part management strategy. This is
especially important in keeping systems in operation for a long time. The
2. Influence of spare part prioritization on availability and longer the MTTR, the worse the system works. If the failure elimination
reliability of the machines’ system time is extended, the system will have less availability.
The Mean Logistic Delay Time (MLDT) refers to delays of the sup-
Availability of the machines’ system means the system's ability to port resources and the administrative delay time of the equipment.
relate the required functions in specific conditions, a specific moment MLDT is a main parameter influencing Ao, describing the support re-
or a specific time interval, under the assumption that external condi- source, including spare part, support equipment, technical data and
tions are provided. We distinguish two types of accessibility: Ao (op- worker. After transforming Eq. (1), MLDT is expressed as Eq. (6):
erational availability) and Ai (inherent availability). Ao is a parameter MTBF
describing the relation between uptime and downtime. In the design MLDT = MTBF MTTR
Ao (6)
process, Ai is determined. Ai takes into account downtime and repair
time, ignoring the delay of preventive actions, administration and lo- Fig. 2 presents selected elements that affect the value of the MLDT
gistics. Ao is considered the system`s reliability, maintainability, sup- indicator. Delays are mainly related to the flow of information and the
portability and testability. Ao is a parameter describing the readiness supply of resources such as: spare parts, equipment, tools and em-
capability and support [40]. Mathematically, Ao and Ai are expressed ployees [40].
by Eqs. (1) and (2). Analyzing Eqs. (1)–(4), it can be observed that increasing the value
of the MLDT indicator reduces Ao and MTBF and thus reduces the re-
MTBF liability of the system (Table 1).
Ao =
MTBF + MTTR + MLDT (1) In view of the above, it is important to take actions aimed at re-
MTBF ducing the MLDT value. Therefore, in this article, actions aimed at re-
Ai =
(2) ducing the MLDT value in the area related to the supply of spare parts
MTBF + MTTR
have been taken. Proposed spare part criticality assessment metho-
where: dology will help increase the availability of spare parts and decrease the
MLDT time, which, finally, will increase the machines’ system avail-
MTBF – Mean Time Between Failures,
ability and reliability.
MTTR – Mean Time to Repair,
3. Methodology
MLDT – Mean Logistic Delay Time.

The MTBF indicator measures the reliability of the system, i.e. its 3.1. Problem description
ability to perform the required function under certain conditions for a
given period of time [41]. This is an indicator often expressed in hours It is a challenge to determine the spare part inventory levels using
as the average time interval that elapses before the item fails and re- machinery criticality, as some machines in a plant are quite critical, while
quires servicing, for repairable systems. MTBF is defined by the mean of others are almost redundant. In addition, there is a possibility that similar
the arithmetic value of the reliability function R(t), which can be ex- parts are installed in a machine which has different criticality levels in
pressed as the expected value of the density function ƒ(t) time to failure terms of maintenance and logistics. In this context, the machinery criti-
(Eq. (3)) [42]. cality is assessed with regard to its importance in sustaining production in
a safe and efficient way without endangering the society and environ-
MTBF =
0
R (t ) dt =
0
tf (t ) dt (3) ment. The machinery’s criticality is also a function of its intended pur-
pose, rather than of itself (e.g. a hydraulic pump may, in one case, be
In practice, MTBF is calculated according to Eq. (4): critical and, in another, auxiliary). Hence, it is necessary to assess the
(Start of Downtime Start of Uptime ) spare parts’ criticality in a system perspective with regard to process and
MTBF = control criticalities. Furthermore, it is important to have formal ap-
Number of failures (4)
proaches for making a criticality analysis and subsequent prioritization, in
MTBF can also be useful in determining the frequency of controls or order to minimize ad hoc assessments. The aforementioned minimizes ad
preventive replacements. For unrecoverable systems, it is recommended hoc decisions that shall be made by experts and lead to suboptimal con-
to use the mean time to failure (MTTF) instead of MTBF, because MTBF is clusions, based on the existing data and information, as well as on the
the time between failures in a system that can be repaired [41]. The particular individual’s knowledge, experience, intuition or intentions.
related MTBF indicator is the average repair time (MTTR), which is the
average time interval (usually expressed in hours) needed to repair the 3.2. Criticality assessment and prioritization approach
damaged item. Mathematically, this is the total repair time of a break-
down divided by the total number of corrective maintenance actions in The methodology used in this work is illustrated in Fig. 3.

214
K. Antosz, R.M.C. Ratnayake Journal of Manufacturing Systems 50 (2019) 212–225

Fig. 2. MLDT – influencing factors [40].

Table 1
The impact of MLDT on system availability and reliability.
Indicator System availability System reliability

MLDT Increase (↑) Decrease (↓) Decrease (↓)


Decrease (↓) Increase (↑) Increase (↑)

Step I: Empirical spare parts’ classification model


The main goal of this step is the identification of the spare parts’
criticality. This work step consists of the following stages (refer to
Fig. 3):

• identification of the spare parts’ criticality assessment criteria, based


on the maintenance requirements,
• identification of the spare parts’ criticality assessment criteria, based
on the logistic requirements,
• assignment of points for each criterion, based on the maintenance
requirements,
• assignment of points for each criterion, based on the logistic re-
quirements,
• the spare parts’ assessment, according to the maintenance criteria
(i.e. calculation of a maintenance category – PC indicator),
m

• the spare parts’ assessment, according to the maintenance criteria


(i.e. calculation of a maintenance category – PC indicator),
l

• assignment of the spare parts to the final category,


• identification of spare parts’ criticality.
Based on the proposed empirical model, spare parts were evaluated
in relation to two main criteria: maintenance and logistics [44]. Within
the requirements of a maintenance discipline, the following criteria are
proposed: machine category (Cm), spare part replacement time (Rt),
complexity of the replacement process (CoR), failure type (Ft) and fre-
quency (Ff), qualifications required for an employee to replace a spare
part (Eq). The spare parts’ classification criteria specify different point Fig. 3. Overall criticality assessment and prioritization approach.
values for every alternative. Table 2 illustrates the possible alternatives,
with descriptions and values of points for each criterion in the main- points for each criterion in the logistics discipline: the cost of spare
tenance discipline. The established ranges of points for criteria are a parts, spare part lead time, storage, the number of potential suppliers.
proposal. The values can be adjusted according to the needs and re- The established ranges of points for the criteria are a proposal. The
quirements of the company. values can be adjusted according to the needs and requirements of the
Table 3 presents possible alternatives, descriptions and values of

215
Table 2
Maintenance discipline – the proposed criteria.
Area Criterion Criterion description Criterion characteristics Points

Maintenance Cm This criterion shows for which machine category the spare parts are mainly used. Machine category A means that this A 4
machine is very important for the production process performance (i.e. bottlenecks). The highest points are assigned to the B 3
spare part that is mainly used for very important machines (i.e. A) and vice versa. C 2
D 1
K. Antosz, R.M.C. Ratnayake

Rt This criterion defines the expected time needed for the replacement of the spare part in case of its failure. The replacement Up to 2 hrs. 1
time influences the total time for the elimination of the machine failure. The highest points are assigned to the spare part that 2-8 hrs. 2
has the longest replacement time (i.e. over 8 hrs) and vice versa. Over 8 hrs. 3
Co R This criterion shows how difficult/complex the replacement process of a spare part is and that there is a need to have special Low – no special devices and tools are required, we do not need to 1
devices and tools in the replacement process. The complexity of the replacement process will also have an influence on the disassemble additional parts
total failure elimination time and, finally, on the criticality of the spare parts. The highest points are assigned to spare parts Medium – particular devices and tools are required, only some additional 2
with high levels of complexity and the lowest points are assigned to the least complex spare parts. parts should be disassembled
High – special devices and tools are required; many additional parts should 3
be disassembled
Ft This criterion determines the type of failure: accidental or chronic. Parts whose failures are chronic are less important Accidental 2
because it is easier to foresee the failure. The part whose failure type is accidental is allocated the highest points the points Chronic 1
are assigned to the part whose failure type is chronic.
Ff This criterion shows the frequency level of the spare part’s failure. For determining the spare part’s criticality, the failure rate Up to 8 per year 1
is important. If it is very high, the spare parts are more important. Spare parts whose failure frequency is very high (i.e. over 8-16 per year 2
16 per year) have the highest points. Over 16 per year 3
Eq This criterion determines the level of qualifications, which an employee needs for the spare part’s replacement process. The Low – during the replacement process, some basic experience is required 1
highest points are assigned to the spare part that has to be outsourced and the lowest points are assigned to the spare part Medium – during the replacement process, an employee needs special 2
that does not need to be outsourced. permissions
Outsourcing (an outside company) is required 3

216
Table 3
Logistics discipline – the proposed criteria.
Area Criterion Criterion description Criterion characteristics Points

Logistics Part cost This criterion shows the cost in a spare part purchasing process. This will help to determine the possibility of a Low 1
Cp purchasing and warehousing process. The company sometimes does not wish to buy a critical part because the Medium 2
cost is very high. At the same time, there are many less important spare parts in the storage, which are not High 3
needed and the sum cost is higher than that of the critical spare part which is important.
The highest points are assigned to the spare part that has the highest cost (i.e. over 7 days) and the lowest
points are assigned to the spare part with the lowest cost.
Lead time Lt This criterion shows how much time is needed for a spare part purchasing process. The highest points are Short – up to 48 hrs 1
assigned to the spare part that has the longest lead time (i.e. over 7 days) and the lowest points are assigned to Medium – from 48 hrs-7 days 2
the spare part with the shortest lead time. Long – over 7 days 3
Storage The criterion shows whether the spare part requires any special conditions and additional space in the No special conditions are required in the storage (e.g. storage surface, humidity, temperature), 1
Ps storage. The highest points are assigned to the spare part that requires special conditions and additional space the spare parts do not need much space in the storage
and lead time and the lowest points are assigned to the spare part that does not requires special conditions Particular conditions are required in the storage (e.g. storage surface, humidity, temperature), 2
and additional space and lead time. the spare parts need more space in the storage
Some special conditions are required in the storage (e.g. storage surface, humidity, temperature), 3
the spare parts need additional/much space in the storage
Number of This criterion shows how many suppliers are available on the market. This information is very important for Only 1 3
potential spare part criticality. Moreover, in many cases, the number of suppliers affects the spare part’s lead time. The 1-4 2
suppliers highest points are assigned to the spare part that has the fewest suppliers (i.e. only 1) and the lowest points >4 1
Sn are assigned to the spare part with the highest numbers of suppliers.
Journal of Manufacturing Systems 50 (2019) 212–225
K. Antosz, R.M.C. Ratnayake Journal of Manufacturing Systems 50 (2019) 212–225

company. favorable and a fraction if the comparison is regarded as less favorable.


Evaluated spare parts are assigned to the appropriate spare part When the evaluations are made at different levels along the hierarchy,
category in two analyzed areas: maintenance and logistics. The as- and if no consensus is reached, then a geometric means (w̄ ) is used to
signment of the spare parts to the appropriate category is done ac- average the judgments (i.e. weights, wi). The geometric means is given
cording to the values of PCm and PCl indicators. by Eq. (9):
Eqs. (7) and (8) were established to calculate the values of PCm and
n
PCl indicators: w¯ = ( wi )1/ n (9)
i=1

PCm = w1Cm+w2Rt+w3CoR + w4Ft+w5Ff+w6Eq (7)


It is important to provide an introduction (e.g. workshop, examples)
PCl = w1Cp+w2Lt+w3Ps+w4Sn (8) regarding the evaluation approach prior to the distribution of the
questionnaire, in order to familiarize an evaluation team with the AHP.
The values of indicators, Cm, Rt, CoR, Ft, Ff, Eq, Cp, Lt, Ps, Sn, should be
In addition, an evaluation team shall consist of experienced experts
established according to the limitations assumed in Tables 1 and 2.
from different fields (e.g. lead engineer, senior engineer mechanical/
Nevertheless, the values’ importance wi can be assumed by companies,
structural, specialists in engineering control, senior engineer in design,
according to their requirements. The importance values’ sum wi for both
senior engineer in testing, etc.) [27,28]. The AHP approach provides
PCm and PCl should equal 1. The spare part category should be de-
the possibility to carry out interviews and analysis independently for
termined according to the values of parameters PCm and PCl (Table 4).
different layers, based on different expertise. Hence, it is possible, for
The maintenance and logistics are two areas in which the spare part
instance, to address reliability, legislative requirements, technical per-
categories will be identified. The possible spare part categories have
formance, cost, etc., based on a particular expert’s relevance. The
been developed, based on the authors’ experience and the requirements
aforementioned enables a multi-criteria analysis to be performed in an
of the analyzed company (Table 4).
effective and efficient manner. In this work, the authors used the re-
When the spare parts category for the part (in two areas: main-
lative importance of two factors for making comparisons, using the
tenance and logistic) is assigned, we can identify the final category for
possibilities provided by the ExpertChoice software.
the spare parts. Based on the final spare parts category the strategy for
It is necessary to establish priorities. After modelling, spare parts
this part as well as the recommended actions and criticality level will be
need to be prioritized in a hierarchy. Each node is evaluated against
determined (Table 5).
each of its peers in relation to its parent node. These evaluations are
Table 5 was developed on the basis of the authors’ and supervisors’
referred to as “pairwise comparisons”. The pairwise comparisons of the
experience from the case study company and taking into account the
elements at each level of the spare part prioritization hierarchical
good practices from lean manufacturing and maintenance management.
model are made in terms of either: Importance – when comparing goals,
If the part has a different category in these two proposed areas, the
criteria, sub-criteria or alternatives in relation to their relative im-
maintenance area is more important than the logistics, from the proper
portance; Preference – when comparing the preference alternatives in
realization of the maintenance management point of view, so the cri-
relation to goals, criteria, sub-criteria; Likelihood ‒ when comparing
ticality level of spare parts depends mainly on the maintenance area.
uncertain events or scenarios in relation to the probability of their oc-
This area should be taken into consideration first. For all spare parts
currence [47]. When comparing a pair of factors (i.e. criteria, sub-cri-
with A category in maintenance, regardless of the category in the lo-
teria or alternatives), a ratio of the relative importance, preference, or
gistics area, criticality was assumed to be H. For spare parts with B
likelihood of the factors is established [48]. In the pairwise comparison
category in maintenance, regardless of the category in the logistic areas,
process, it is also possible to take the probability and consequence of a
criticality is M and so on.
possible event (i.e. criticality aspect) into the evaluation.
The numerical values (i.e. weights) that represent the judgments of
Step II: AHP analysis for prioritization the comparisons are arranged in a matrix for the further calculations.
In the AHP approach, expert judgments provide the necessary in- The weights in the pairwise comparison matrices (PCMs) are composed
formation for the ranking of criteria. Fig. 4 illustrates the model of a using the numerical values in Table 6, while comparing all elements at
hierarchical structure developed for the prioritization of spare parts each level of the spare part prioritization hierarchical model. The PCM
with high criticality. “A_ ” for comparing “n” elements is given by:
In this context, both qualitative and quantitative criteria are com-
pared using informed judgments [with the help of a particular dis- 1
cipline expert(s)] to derive weights and priorities [45,46]. Using the A = [aij ] where, aji =
_ , aii = 1, (1 i n) and (1 j n
aij (10)
scale illustrated in Table 6, pairwise comparison matrices are developed
for the analysis [47].
The PCM (i.e. A
_ ) has a form as follows:
The evaluations are conducted by experts, using a questionnaire
similar to the sample (i.e. a part of the questionnaire) given in Fig. 5.
The judgment value is a full number if the comparison is regarded as

Table 4
Spare part categories.
Part category Value of A Critical part PCm ≥2.5, PCl ≥2.5 B Needed part 1.5 ≤ PCm < 2.5, 1.5 ≤ PCl < 2.5 C Other parts, not classified in
PCm and PCl indicators categories A or B
PCm < 1.5, PCl < 1.5

Maintenance Availability is required immediately, a failure rate is Availability is required in the shortest possible time, a Short purchasing time, failure rate
very high, employees need special skills, or failure rate is on the medium level, employees need is on the low level, employees do
outsourcing is required special skills, or outsourcing is required not require any special skills
Logistics Availability is required immediately, lead time is Availability is required in the shortest possible time, Short purchasing time – lead time is
very long, one supplier is available, cost is very high, lead time is on the medium level, more than one short, there are many suppliers, cost
parts require special conditions and space in the supplier is available, cost is on the medium level, part is on the low level
storage does not require special conditions and space in the
storage

217
K. Antosz, R.M.C. Ratnayake Journal of Manufacturing Systems 50 (2019) 212–225

Table 5
The recommended actions and strategy for final spare parts category.
Maintenance Logistics Spare parts Criticality level Strategy Recommended actions
category

A A AA Availability in the storage is required. Availability and part quality are under special
A B AB control, Kanban is required.
A C AC
B A BA
B B BB Availability in the storage is recommended dependent on provide Availability and part quality are under control,
B C BC financial and storage possibilities of the company. Kanban is recommended.
C B CB

C A CA Availability in the storage is not required – the company makes the decision, purchase on demand.

_
Aw_ = nw
_ (12)

The problem of solving Eq. (12) for a nonzero solution is an ei-


genvalue problem, which is given by:
Aw
_ _ = w
_ (13)

As matrix A_ has a special form (i.e. each row is a constant multiple


of the first row), the rank of the matrix is one, and, except for one
eigenvalue, all the other eigenvalues of matrix A_ are zero. However, as
the sum of the eigenvalues of a positive matrix is equal to the trace of
the matrix, the nonzero eigenvalue has a value of n (i.e. the size of the
matrix). This is referred to as max and used for a consistency analysis.
The consistency analysis is carried out in order to determine the
consistency of experts’ judgments. For instance, if matrix w _ is not
known, only the estimates of aij s in matrix A
_ are known, and no strong
consistency property (i.e. small errors and inconsistencies in judg-
ments) exists, then the eigenvalue problem for the inconsistent case is:
Aw
_ _ = max w
_ (14)

where max n and other s ˜0 . The estimates of weights for the ac-
Fig. 4. Hierarchical structure for prioritization (G: goal; Ci = ith criteria; SCi = tivities are calculated by normalizing the eigenvector corresponding to
ith sub-criteria; Ai = ith alternative). the largest eigenvalue in Eq. (14). The normalization is performed by
summing each column and then dividing each column entry by its re-
w1 w1 w1 w1 spective column sum. The normalization process is given by:
w1 w2 w3 wn
w2 w2 w2 w2 aij
w1 w2 w3 wn a, = [aij ] whereaij = n for (1 i n) and (1 j n
_A = w3
w1
w3
w2
w3
w3
w3
wn
a
k = 1 ik (15)
.

wn wn wn wn Then, the average value in each row of the normalized matrix is


w1 w2 w3 wn (11)
calculated to obtain the relative weights or eigenvector, which is de-
Once the weights of the PCM have been established, then the matrix termined by:
product of matrix A_ with vector w
_ = [ w1 w2 w3 wn ] enables the n
aija
calculation of the weights of w
_: W
_ = [wk ] wherewk = for (1 i n), (1 j n) and (1 k n
i=1
n
w1 w1 w1 w1
w1 w2 w3 wn
w1 nw1 (16)
w2 w2 w2 w2
w1 w2 w3 wn w2 nw2
w3 w3 w3 w3 * w3 = nw3 The notion here is that the closer the max is to n, the more con-
w1 w2 w3 wn
sistent the judgments are. Hence, it is possible to use the difference
wn wn wn wn
wn nwn ( max n ) as a measure of inconsistency (i.e. the perfect consistency
w1 w2 w3
refers to the zero difference). Hence, the consistency index (CI) is
wn

Table 6
Saaty’s scale for judgments.
Intensity Definition Explanation
of importance

1 Equal importance Two factors contribute equally to the objective


3 Slightly more important Experience and judgment slightly favor one over another
5 Much more important Experience and judgment strongly favor one over another
7 Very much more important Experience and judgment very strongly favor one over another
9 Definitely more important The evidence favoring one over another is of the highest possible validity
2,4,6,8 Intermediate values When a compromise is needed

218
K. Antosz, R.M.C. Ratnayake Journal of Manufacturing Systems 50 (2019) 212–225

Fig. 5. Excerpted sample questions of the AHP interviews.

defined (see [49]) as Eq. (17), as it represents the average of the re- failures occur in the machines. This causes a big problem for the con-
maining eigenvalues: tinuity of production, particularly when the spare part is mainly used in
a very important machine (i.e. bottleneck) with a very long lead time.
( max n)
CI = For the model evaluation, 24 parts were chosen randomly (i.e. 10%
(n 1) (17)
of the spare parts from the spare part list that have already been
In order to derive a meaningful interpretation of the CI, random identified as the most critical without performing a formal analysis).
pairwise comparisons were simulated at the inception of the AHP for These were spare parts such as: timing belts, filament filters, actuators,
different sizes of matrices. The calculations were performed for calcu- solenoid valves, transistors and others. They were evaluated according
lating CIs and arriving at an average CI for random judgments for each to the proposed empirical criteria (Tables 2 and 3).
size of the matrix [i.e. referred to as a random consistency index (RCI)]. For the analysis, the experts from the maintenance department of
The size of the matrix vs. RCI is illustrated in Table 7 [49]. the company set the wi importance values. Eqs. (19) and (20) show the
Hence, the consistency ratio (CR) is defined as the ratio of the CI for standard importance values for the maintenance and logistics areas in
a particular set of judgments to the average CI (or RCI) for random the case manufacturing company.
comparisons for a matrix of the same size [see Eq. (18)]. PCm = 0.3Cm+0.2Rt+0.1CoR+0.1Ft+0.2Ff+0.1Eq (19)
CI
CR = PCl = 0.2Cp+0.3Lt+0.1Ps+0.4Sn (20)
RCI (18)
Table 8 presents the number of spare parts that have been classified
where: if CR < 0.1 (or 10%) , then the judgments are considered to be
into each criticality level using the proposed model. Detailed calcula-
consistent.
tions for Table 8 are presented in Appendix A.
The chosen parts were mostly identified in the company as critical
4. Application of the methodology to a manufacturing company spare parts. The company accepted and confirmed that the obtained
results are proper. Table 8 shows that only 2 out of the analyzed spare
4.1. Empirical model based on criticality analysis and results parts are critical, and a significantly large number of spare parts were
categorized as criticality level M (19 spare parts). However, if there are
The main goal of this step was gathering the information about the spare parts (e.g. criticality levels H, L), then it is not necessary to make
correctness of the model and analyses of the received results. such a prioritization in the case of spare parts with criticality H it is
This step consists of the following stages: obligatory to have them in storage. When it comes to spare parts with
criticality L, the company can decide whether they want to have them
• select a random sample (e.g. 10% spare parts in this study) of spare in storage or not – it mainly depends on the company.
parts from the spare part list that have already been identified as the It is possible to prioritize the spare parts with ‘M’ criticality. This
most critical (i.e. without performing a formal analysis); prioritizing of the parts within each type is done in order to deploy
• use the empirical model and assess the level of the spare parts’ maintenance activities effectively. Therefore, it is necessary to improve
criticality, the methodology to further prioritize the spare parts in criticality M and
• prioritize the spare parts within an identified criticality class (i.e. L, to increase the efficiency of maintenance management activities.
M, H), based on the number of items in each category; Hence, criticality level M (i.e.19 spare parts) was analyzed. Table 9
• investigate the criticality level within each class, based on the shows the results. Within M criticality, we distinguish two groups of
maintenance and logistics criteria for making sub-groups. parts: with criticality BB and BC.
In this paper, the spare parts with criticality BB were chosen for
The case study manufacturing company produces parts for the au-
tomotive industry. This company uses the corrective maintenance for Table 8
its machines. It does not have a specific system for spare part evalua- The number of spare parts in each criticality level according to the proposed
tion. It buys spare parts when they are needed – especially when model.
Criticality level
Table 7
Size of the matrix vs. RCI.
Size of the 1 2 3 4 5 6 7 8 9 10
matrix

RCI 0.00 0.00 0.58 0.90 1.12 1.24 1.32 1.41 1.45 1.49 Number of spare parts 2 19 3

219
K. Antosz, R.M.C. Ratnayake Journal of Manufacturing Systems 50 (2019) 212–225

Table 9 and two sub-criteria. However, depending on a particular case, the


The number of spare parts in each category for M criticality. number of criteria, sub-criteria and alternatives can change. Fig. 6 il-
Criticality level Spare parts category lustrates a generic hierarchical structure that was established for per-
(M) forming the spare part prioritization in the case study manufacturing
BB BC company.
Number of spare parts 8 11
4.2.2. Data collection: pairwise comparisons for spare part prioritization
Using Saaty’s scale (refer to Table 3), pairwise comparisons were
made at different levels of the hierarchy (refer to Fig. 5).
Pairwise comparison of spare parts in relation to a goal and criteria
(i.e. logistics and maintenance) [CR = 0]:
Goal L M
L 1 1/7
M 7 1
Pairwise comparison of spare parts in relation to criteria and sub-
criteria (i.e. logistics vs. sub-criteria) [CR = 0.23]:
Logistics Cp LT S NoS
PC 1 1/5 7 1/5
LT 5 1 9 1/5
S 1/7 1/9 1 1/5
NoS 5 5 5 1
Pairwise comparison of spare parts in relation to criteria and sub-
criteria (i.e. maintenance vs. sub-criteria) [CR = 0.15]:
Fig. 6. A hierarchical structure for spare part prioritization.
Maintenance MC RT CoR FT FF EQ LT
MC 1 1/3 7 5 5 5 3
further analysis because this criticality is more important than BC cri- RT 3 1 5 5 7 3 3
ticality. The analytic hierarchy process (AHP) was used to investigate CoR 1/7 1/5 1 1 1/3 1/3 1/7
how the spare parts shall be prioritized within a certain spare part FT 1/5 1/5 1 1 5 1/5 1/5
category. FF 1/5 1/7 3 1/5 1 1/3 1/5
EQ 1/5 1/3 3 5 3 1 1/5
4.2. AHP-related analysis and results LT 1/3 1/3 7 5 5 5 1

The spare part assessment results that were obtained based on the
This work step consists of the following stages: empirical spare part classification model (see Table 9) were applied to
perform the AHP-related pairwise comparisons (i.e. to perform further
• develop a hierarchical structure for the spare part prioritization, prioritization) using Saaty’s scale. The part cost (Cp) criterion was
• use Saaty’s scale and develop the pairwise comparison matrices, chosen to demonstrate the pairwise comparisons between the spare
• perform the AHP analysis and calculate priority weights, parts. For instance, we can consider the comparison of spare parts SP2
• perform the sensitivity analysis and reflect on the results. to SP5 in relation to the Cp criterion. Based on the empirical modeling-
related analysis (see Table 10), the Cp criterion received 3 points (i.e.
4.2.1. A hierarchical structure for spare part prioritization maximum) for SP2 and 1 point (i.e. low) for SP5. Hence, during the
It is possible to prioritize spare parts based on different sets of cri- pairwise comparison, it was revealed that SP2 is of ‘absolutely more
teria (i.e. in relation to logistics and maintenance) and sub-criteria (i.e. importance’ compared to SP5. The relevant Saaty’s scale amounts to 9
logistics: part cost, lead time, storage and the number of potential (see Cp matrix). On the contrary, during the pairwise comparison be-
suppliers; and maintenance: machine category, spare parts replacement tween SP7 and SP6 in relation to Cp criterion, it was revealed that SP7 is
time, complexity of a replacement process, failure type, failure fre- of ‘much more importance’, compared to SP1, and the relevant Saaty’s
quency and employee qualifications and, additionally, the lead time). scale amounts to 5 (see Cp matrix). The aforementioned is established
For instance, in the current spare part prioritization case study, there on the fact that, based on the empirical modeling-related analysis (see
are eight alternatives: SP1, SP2… SP8 (see Fig. 6). It is necessary to Table 10), Cp criterion received 3 points (i.e. maximum) for SP7 and 2
perform the prioritization of these eight alternatives (i.e. spare parts points (i.e. moderate) for SP6. This demonstrates that SP7 is just ‘much
with high criticality) in this illustrative case in relation to six criteria more important’ than SP6. Similar arguments and the procedure were

Table 10
The results of the spare part assessment – criticality BB.
No. Spare part Cm Rt Co R Ft Ff Eq Cp Lt Ps Sn Spare part Spare part
category criticality

SP1 4 1 1 1 1 2 2 1 1 2 BB M
SP2 4 1 1 1 1 1 3 1 2 3 BB M
SP3 4 2 1 1 1 2 2 1 2 2 BB M
SP4 4 2 3 1 1 2 3 1 2 3 BB M
SP5 2 1 1 1 2 2 1 2 1 3 BB M
SP6 4 2 3 1 1 2 2 1 2 3 BB M
SP7 4 1 1 1 2 1 3 1 2 3 BB M
SP8 4 1 1 1 1 2 3 1 2 2 BB M

220
K. Antosz, R.M.C. Ratnayake Journal of Manufacturing Systems 50 (2019) 212–225

used to make pairwise comparisons between spare parts. 4.2.3. AHP analysis results and discussion
This methodology was used to perform pairwise comparisons for all In Fig. 7, the hierarchical structure for the prioritization of spare
spare parts in each criterion. The matrices were developed as follows: parts and the corresponding individual priority weights is presented.

The AHP analysis results reveal that the critical spare parts shall be
Expert Choice software was used to calculate the overall prior- given priority in the following order: SP6, SP4, SP3, SP7, SP5, SP8, and
itization weights. SP1. It is worth noting that spare parts SP5 and SP8 have almost equal

221
K. Antosz, R.M.C. Ratnayake Journal of Manufacturing Systems 50 (2019) 212–225

priority (i.e. 5th priority). Fig. 8 illustrates the synthesis (normalized how the spare parts were prioritized based on different responsible
values of priority weights) with respect to the goal (i.e. prioritization of parties (e.g. maintenance engineering and logistics). Moreover, it is
spare parts). possible to perform a similar analysis for all the other combinations
Overall inconsistency CR is 0.08 (i.e. < 0.10). Hence, it is revealed along the decision hierarchy.
that the judgments are consistent with the spare part prioritization The gradient sensitivity analysis enables spare part prioritization
process. with respect to one criterion at a time. For instance, Fig. 11 demon-
strates the machine category.
The red vertical line indicates the criterion priority (based on the
4.2.4. Sensitivity analysis decision-maker’s paired comparisons). Dragging the red bar to the left
The sensitivity analysis was performed to increase transparency in or right, it is possible to indicate how a criterion’s priority changes.
the prioritization process and also to enable trade-offs to be made be- Such an analysis enables a logical trade-off to be made between dif-
tween different criteria, sub-criteria and alternatives. Fig. 9 illustrates ferent alternatives, instead of depending on ad hoc conclusions.
the sensitivity analysis dashboard for the maintenance in connection
with the related prioritization criteria vs. prioritization of spare parts.
The synthesis with respect to the goal (i.e. the prioritization of spare 5. Conclusions
parts) is indicated on the right ‘y-axis’.
The sensitivity analysis approach enables a comparative overall The prioritization of spare parts is a challenge (although risk-based
estimation to be made of how overall spare part prioritization changes prioritizations are possible) among the engineering experts, as it re-
in relation to the change in a single criterion. For instance, if the quires the knowledge of the multi-criteria analysis approaches in order
priority of one sub-criterion has been changed, then it is possible to to minimize ad hoc assessments and recommendations after conducting
observe how spare part priority has been changed. The aforementioned a macro-level risk analysis. The multi-criteria analysis approaches en-
enables an effective trade-off analysis to be made before concluding able complex problems to be coped with (i.e. especially in the micro-
final priorities. Furthermore, it provides a high level of transparency, level analysis) by imposing a disciplined structure that directs attention
whilst enabling a holistic view of the overall prioritization process. It is to the criteria in proportion to the weight which they deserve. Having
possible to perform a similar analysis for each node along the hier- the formal multi-criteria analysis approaches increases the transparency
archical diagram, providing greater transparency and subsequent flex- and visualization of how different factors are influenced in the final
ibility to investigate all the potential scenarios in the spare part prior- prioritization. This work first demonstrated risk-based analysis at a
itization process. The priority value is illustrated by the overall priority macro level and then used the multi-criteria analysis approach – the
from the intersection of the ‘right y-axis’. AHP – to conduct a micro-level analysis in order to minimize ad hoc
It is also possible to perform two-dimensional analyses for each assessments and increase the transparency of the assessment and re-
node. Two-dimensional sensitivity analyses for nodes were carried out commendation process that supports system safety.
to illustrate the alternatives priorities with respect to two criteria at a The standard format of the AHP was used to derive weights and the
time. For instance, Fig. 10 demonstrates two-dimensional sensitivity scores that were realized by the alternatives which are based on pair-
analysis for nodes: logistics vs. maintenance. wise comparisons between the criteria and between the options. The
The area of the two-dimensional plot is basically divided into AHP offered a pairwise comparison form of the data input, providing a
quadrants here. The most favorable alternatives with respect to the straightforward and convenient approach for performing a multi-cri-
criteria on the two axes are those in the upper-right quadrant (i.e. the teria analysis. In addition, the AHP has provided the following benefits:
closer to the upper-right corner, the better the alternative). The least flexibility; intuitive appeal to analysts; ability to check inconsistencies;
favorable alternatives are shown in the lower-left quadrant (i.e. the possibility of decomposing a decision problem into its constituent parts;
closer to the lower-left corner, the less favorable the alternative). If the ability to build hierarchies of criteria; capability to capture both sub-
alternatives are located in the upper-left and lower-right quadrants, jective and objective evaluation measures; a mechanism for checking
they indicate key trade-offs. For instance, it is possible to observe that the consistency of the evaluation measures and alternatives; support for
SP6 & SP4 are in the upper-left quadrant and SP5 in the lower-right the group decision-making and the ability to help model situations of
quadrant. This also demonstrates where the descending priorities in the uncertainty and risk.
final synthesis are. In the spare part prioritization process, first a risk-based analysis
Two-dimensional sensitivity analysis for nodes enables the visuali- was performed at a macro level and then, to make a finer analysis
zation and interpretation of how each factor (i.e. in this case main- (micro level), the AHP approach was used. The AHP approach provided
tenance vs. logistics) is interrelated with a different spare part prior- a backbone to carry out such an analysis in a formal manner with a
itization. Such an analysis enables a detailed explanation to be made of defined mechanism to incorporate the criteria and sub-criteria for

Fig. 7. Hierarchical structure for the prioritization of spare parts and corresponding individual priority weights.

222
K. Antosz, R.M.C. Ratnayake Journal of Manufacturing Systems 50 (2019) 212–225

Fig. 8. Prioritization of spare parts: overall priority weights distribution: a) sorted by name and normalized, b) sorted by priority and normalized.

Fig. 9. Sensitivity analysis results (performance sensitivity for nodes) for maintenance in relation to related prioritization criteria vs. prioritization of spare parts.

Fig. 10. Two-dimensional sensitivity analysis for nodes: maintenance vs. logistics.

prioritizing the spare parts. In addition, the suggested approach enabled Furthermore, the work illustrates how to use a risk-based analysis
the spare part prioritization assessments to be visualized along a hier- and the subsequent AHP for performing a spare part prioritization and
archy, facilitating the interpretation of the whole set of criteria and sub- possible sensitivity analysis in order to estimate the trade-offs. An
criteria with the potential alternatives. overall synthesis provides possible priorities, based on the pairwise

223
K. Antosz, R.M.C. Ratnayake Journal of Manufacturing Systems 50 (2019) 212–225

Fig. 11. Gradient sensitivity analysis with respect to machine category.

comparisons. In addition, performance, gradient and two-dimensional (n-1)/2], while modelling and analysis become a lengthy task).
sensitivity analyses provide an alternative means to study how the final Future research shall be carried out in order to incorporate the
selection is made. This allows one to see how different criteria and sub- fuzziness present in pairwise comparisons of the spare part prioritiza-
criteria contribute to the final priorities. Hence, the suggested approach tion analysis. This will also allow the limitation of the use of a nine-
provides transparency in prioritizing spare parts in relation to the cri- point digital scale of the AHP in the spare-part-related prioritizations to
teria and sub-criteria. be overcome.
However, the drawbacks, such as ‘ranking irregularities’, ‘need to Additionally, the proposed work has some limitations, due to the
decompose machine criticality analysis problem into a number of fact that the method was verified in only one production company. That
subsystems’, and ‘requirement to perform a substantial number of is why, in further work, this method will also be verified and general-
comparisons within which and between which’, limit the use of the AHP ized in other companies: different types and branches. An analysis
in the spare-part-related prioritizations. Moreover, the AHP approach should also be made of costs or availability, which would show that the
has the limitation of being based on the number of pairwise compar- proposed methodology really has an influence on the effectiveness of
isons to be made (i.e. the number of comparisons becomes very large [n spare part management.

Appendix A

See Table A1.

Table A1
Detailed calculations for Table 8.
Spare parts Cm Rt Co R Ft Ff Eq Cp Lt Ps Sn Maintenance Logistics Spare parts category Criticality level

4 2 1 1 3 2 3 1 2 3 2.6 A 2.2 B AB H
4 2 3 1 2 2 3 2 2 3 2.6 A 2.5 B AB H
7 1 1 1 1 1 1 1 1 1 1.3 C 1 C CC L
2 1 1 1 1 1 1 1 1 1 1.3 C 1 C CC L
2 1 1 1 1 1 1 1 1 1 1.3 C 1 C CC L
4 1 1 1 1 2 2 1 1 2 2 B 1.5 B BB M
4 1 1 1 1 1 3 1 2 3 1.9 B 2.2 B BB M
4 1 1 1 1 2 1 1 1 1 2 B 1.3 C BC M
4 2 2 1 1 2 3 1 1 3 2.3 B 2 C BC M
4 1 1 1 3 1 1 1 1 1 2.3 B 1 C BC M
3 2 3 1 1 2 1 1 1 2 2.1 B 1.3 C BC M
3 1 1 1 1 1 1 1 1 1 1.6 B l C BC M
3 1 1 1 1 1 1 1 1 1 1.6 B 1 C BC M
4 2 1 1 1 2 2 1 2 2 2.2 B 1.7 B BB M
4 2 3 1 1 2 3 1 2 3 2.4 B 2.2 B BB M
2 1 1 1 2 2 1 2 1 3 1.6 B 1.9 B BB M
4 2 3 1 1 2 2 1 2 3 2.4 B 2 B BB M
4 1 1 1 1 1 1 1 1 1 1.9 B 1 C BC M
4 1 1 1 1 1 1 1 1 2 1.9 B 1.3 C BC M
4 1 1 1 1 1 1 1 1 2 1.9 B 1.3 C BC M
4 1 1 1 2 1 3 1 2 3 2.1 B 2.2 B BB M
4 1 1 1 1 1 1 1 1 1 1.9 B 1 C BC M
4 1 1 1 1 2 3 1 2 2 2 B 1.9 B BB M
4 1 1 1 2 1 1 1 1 1 2.1 B 1 C BC M

224
K. Antosz, R.M.C. Ratnayake Journal of Manufacturing Systems 50 (2019) 212–225

References [29] Ratnayake RMC. A decision model for executing plant strategy: maintaining the
technical integrity of petroleum flowlines. Int J Dec Sci Risk Man (IDJRSM)
2012;4(1/2):1–24.
[1] Hellingrath B, Cordes AK. Conceptual approach for integrating condition mon- [30] Gajpal PP, Ganesh LM, Chandrasekharan R. Criticality analysis of spare parts using
itoring information and spare parts forecasting methods. Prod Man Res the analytic hierarchy process. Int J Prod Econ 1994;35(1-3):293–7.
2014;2(1):725–37. [31] Braglia M, Grassi A, Montanari R. Multi-attribute classification method for spare
[2] Wang W. A stochastic model for joint spare parts inventory and planned main- parts inventory management. J Qual Main Eng (JQME) 2004;10(1):55–65.
tenance optimisation. Eur J Oper Res 2012;216(1):127–39. [32] Botter R, Fortuin L. Stocking strategy for service parts - a case study. Int J Op Prod
[3] Karsten F, Basten RJI. Pooling of spare parts between multiple users: how to share Man 2000;20(5-6):656–74.
the benefits. Eur J Oper Res 2014;233:94–104. [33] Huiskonen J. Maintenance spare parts logistics: special characteristics and strategic
[4] Lengu D, Syntetos AA, Babaic MZ. Spare parts management: linking distributional choices. Int J Prod Econ 2001;71(1–3):125–33.
assumptions to demand classification. Eur J Oper Res 2014;235(3):624–35. [34] Paakki J, Huiskonen J, Pirttilä T. Improving global spare parts distribution chain
[5] Dekker R, Pinçe C, Zuidwijkc R, Naiman MJ. On the use of installed base in- performance through part categorization: a case study. Int J Prod Econ
formation for spare parts logistics: a review of ideas and industry practice. J. Prod. 2011;133:164–71.
Econ. 2014;143(2):536–45. [35] Wang C, Xu J, Wang H, Zhang Z. A criticality importance-based spare ordering
[6] Grondys K. Economic and technical conditions of selection of spare parts suppliers policy for multi-component degraded systems. Eksploat I Niezawodn - Maint Reliab
of technical equipment. Proc. Econ. and Fin. 2015;27:85–92. 2018;20(4):662–70. https://doi.org/10.17531/ein.2018.4.17.
[7] Tiemessen HGH, Fleischmann M, van Houtum GJ, van Nunen JAEE, Pratsini E. [36] Chen X, Xu D, Xiao L. Joint optimization of replacement and spare ordering for
Dynamic demand fulfilment in spare parts networks with multiple customer classes. critical rotary component based on condition signal to date. Eksploat I Niezawodn -
Eur J Oper Res 2013;228(2):367–80. Maint Reliab 2017;19(1):76–85. https://doi.org/10.17531/ein.2017.1.11.
[8] Kennedy WJ, Patterson JW, Fredendall LD. An overview of recent literature on [37] Louit D, Pascual R, Banjevic D, Jardine AKS. Condition-based spares ordering for
spare parts inventories. Int J Prod Econ 2002;76(2):201–15. critical components. Mech Sys Sig Proc 2011;25:1837–48.
[9] NORSOK-Z008. Risk based maintenance and consequence classification Available [38] Godoy DR, Pascual R, Knights P. Critical spare parts ordering decisions using
from 2015http://www.mop.ir/portal/file/showfile.aspx?id=96b3dc75-14e4-4303- conditional reliability and stochastic lead time. Reliab Eng Syst Saf
99bd-9e3d2b5857dc. 2013;119:199–206.
[10] Boylan JE, Syntetos AA. Forecasting for inventory management of service parts. In: [39] Wang W, Wang SY. An inventory pooling model for spare units of critical systems
Kobbacy KAH, Murthy DNP, editors. Complex system maintenance handbook. New that serve multi-companies. Trans Res Part E 2015;76:34–44.
York: Springer Verlag; 2008. p. 479–506. [40] Na H, Yi L, Wang YG, Liu J, Bo Z, Lv XZ. Research on the mean logistic delay time of
[11] Pennings CLP, van Dalen J, van der Laan EA. Exploiting elapsed time for managing the development phrass. Phys Procedia 2012;33:375–9. https://doi.org/10.1016/j.
intermittent demand for spare parts. Eur J Oper Res 2017;258(3):958–69. phpro.2012.05.077.
[12] Ronzoni C, Ferrara A, Grassi A. A stochastic methodology for the optimal man- [41] Lienig J, Bruemmer H. Reliability analysis. Fundamentals of electronic systems
agement of infrequent demand spare parts in the automotive industry. IFAC- design. Springer International Publishing; 2017. p. 49–56. ISBN 978-3-319-
PapersOnLine 2015;48(3):1405–10. 55839-4.
[13] Kranenburg AA, van Houtum GJ. A new partial pooling structure for spare parts [42] Birolini A. Reliability engineering: theory and practice. Berlin: Springer; 2013. ISBN
networks. Eur J Oper Res 2009;199(3):908–21. 978-3-642-39534-5.
[14] Lewandowski M, Oelker S. Towards autonomous control in maintenance and spare [43] Institute for Telecommunications Sciences. Mean Time To Repair, Definition
part logistics – challenges and opportunities for preacting maintenance concepts. Archived 2008-09-25 at the Wayback Machine.
Proc Techn 2014;15:333–40. [44] Antosz K, Ratnayake RMC. Classification of spare parts as the element of a proper
[15] Durán O, Macchi M, Roda I. On the relationship of spare parts inventory policies realization of the machine maintenance process and logistics – case study. IFAC-
with total cost of ownership of industrial assets. IFAC-PapersOnLine PapersOnLine 2016. 49-12: 389-1393.
2016;49(28):19–24. [45] Ratnayake RMC. Modelling of asset integrity management process: a case study for
[16] Hellingrath B, Pereira C, Espíndola D, Frazzo EM, Cordes AK, Saalmann P, et al. On computing operational integrity preference weights. Int J Com Syst Eng
the integration of intelligent maintenance and spare parts supply chain manage- 2012;1(1):3–12.
ment. IFAC-PapersOnLine 2015;48(3):983–8. [46] Ruschel E, Alvers Portela Santos E. Freitas Rocha Loures E. Industrial maintenance
[17] Behfard S, van der Heijden MC, Al Hanbali A, Zijm WHM. Last time buy and repair decision – making: a systematic literature review. J Man Sys 2017;45:180–94.
decisions for spare parts. Eur J Oper Res 2015;244(2):498–510. https://doi.org/10.1016/j.jmsy.2017.09.003.
[18] Ghorbel N, Addouche SA, El Mhamedi A. Forward management of spare parts stock [47] Saaty TL. How to make a decision: the analytical hierarchy process. Eur J Oper Res
shortages via causal reasoning using reinforcement learning. IFAC-PapersOnLine 1990;48(1):9–26.
2015;48(3):1061–6. [48] Forman E, Sally M. Decision by objectives. World Scientific Press; 2002. ISBN: 981-
[19] Molenaers A, Baets H, Pintelon L, Waeyenbergh G. Criticality classification of spare 02-4142-9.
parts: a case study. J Prod Econ 2012;140(2):570–8. [49] Saaty TL. Decision making for leaders: the analytical hierarchy process for decisions
[20] Syntetos A, Keyes M, Babai M. Demand categorization in a European spare parts in a complex world. Belmont, CA: Lifetime Learning Publications; 1982.
logistics network. Int. J Op Prod Man 2009;29(34):292–316.
[21] Ramanathan R. ABC inventory classification with multiple-criteria using weighted
linear optimization. Com Op Res 2006;33(3):695–700. Katarzyna Antosz Dr., Assistant Professor, since 2002 has been working at Rzeszow
[22] Partovi FY, Anarajan M. Classifying inventory using an artificial neural network University of Technology, Faculty of Mechanical Engineering and Aeronautics,
approach. Com Ind Eng 2002;41(4):389–404. Department of Manufacturing and Production Engineering and is a member of the Lean
[23] Teunter R, Babai M, Syntetos A. ABC classification: service levels and inventory Learning Academy Poland. She is a specialist in the field of improving and increasing the
costs. PROZIMA 2010;19(3):343–52. efficiency of production in organizations, using the methods and tools of Lean
[24] Zhou P, Fan L. A note on multi-criteria ABC inventory classification using weighted Manufacturing and Six Sigma, and the reliability and operation of machines and main-
linear optimization. Eur J Oper Res 2007;182(3):1488–91. tenance services in enterprises. Author of many publications in this fields.
[25] Ng WL. A simple classifier for multiple criteria ABC analysis. Eur J Oper Res
2007;177(1):344–53. R.M. Chandima Ratnayake Dr., Professor, since 2007 has been working in the
[26] Chen J. Peer-estimation for multiple criteria ABC inventory classification. Com Department of Mechanical and Structural Engineering and Materials Science, University
Oper Res 2011;38(12):1784–91. of Stavanger. He is also Visiting Professor in the Department of Process Engineering,
[27] Ratnayake RMC, Markeset T. Implementing company policies in plant level asset Memorial University, St. John’s NL Canada, and the University of Petronas, Malaysia. He
operations: measuring organisational alignment. Eur J Ind Eng 2010;4(3):355–71. is a specialist in the fields of offshore engineering (risk based inspection and main-
[28] Ratnayake RMC, Markeset T. Technical integrity management: measuring HSE tenance), asset integrity assessment, control and management, and mechanical systems
awareness using AHP in selecting a maintenance strategy. J Qual Main Eng (JQME) engineering. Author of many publications in these fields.
2010;16(1):44–63.

225

View publication stats

You might also like