You are on page 1of 9

Int. J.

Production Economics 140 (2012) 570578

Contents lists available at SciVerse ScienceDirect

Int. J. Production Economics


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

Criticality classication of spare parts: A case study


An Molenaers a,b,n, Herman Baets c, Liliane Pintelon a, Geert Waeyenbergh a,b
a
b
c

Centre for Industrial Management (CIB), Katholieke Universiteit Leuven, Celestijnenlaan 300 A, B-3001 Heverlee, Belgium
Research Group Industrial Management, GROUP T International University College Leuven, Andreas Vesaliusstraat 13, B-3000 Leuven, Belgium
BASF ANTWERP NV, Scheldelaan 600, B-2040 Zandvliet, Belgium

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 25 March 2010
Accepted 16 August 2011
Available online 8 September 2011

This paper proposes a spare part classication method based on item criticality. Starting from a multicriteria analysis, the proposed model converts relevant criteria impacting item criticality into a single
score presenting the criticality level. The obtained criticality level is used to rationalize the efciency of
the spare parts inventory policy. The model presents the multi-criteria classication problem in a logic
decision diagram where AHP is used to solve the multi-criteria decision sub-problems at the different
decision nodes of the diagram. The method was tested and implemented in a petrochemical plant and
the results of this case study are presented.
& 2011 Elsevier B.V. All rights reserved.

Keywords:
Spare parts
Criticality
AHP
Multi-criteria classication

1. Introduction
For many asset-intensive industrial plants, classication of the
total spare parts assortment into relevant categories is a crucial
task in order to control the wide and highly varied assortment.
Spare parts differ strongly in terms of stock-out effects, item value
and demand pattern. Syntetos et al. (2009) states that classication enables managers to focus on the most important items
and facilitates the decision making process. However important
items from a maintenance perspective are rather different compared to important parts from an inventory or logistics viewpoint. Those parts for which its unavailability would result in
severe consequences for the plant are perceived as important
items for maintenance managers. From an inventory management
perspective, other parameters like holding costs and demand
pattern seem valuable classication criteria when dening appropriate stocking policies for the different classes. Although both
parties cope with different goals, the link between them is crucial
in the spare parts dilemma. The function of holding spare parts in
stock is supporting maintenance activities and ensuring continuity of safe and reliable operations.
Due to the high service requirements and nancial resources
involved, it seems obvious that spare parts classication is an
important area of research. The diverse set of spare parts
characteristics makes the classication process not an easy job
and sets pressure for a multi-criteria classication approach.
Although theoretical models such as AHP (Analytic Hierarchy
Process) and linear additive methods optimization are proposed

Corresponding author. Tel.: 32499/567731.


E-mail address: An.molenaers@mech.kuleuven.Be (A. Molenaers).

0925-5273/$ - see front matter & 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.ijpe.2011.08.013

in literature, real case studies are limited. Industrial implementations seem to lag behind the theoretical models since the most
popular spare part classication method is still the classical ABCapproach. In this research project, we will focus on bridging the
gap between theory and practice by developing and implementing a practical classication scheme based on item criticality in an
industrial environment.
The research work was initiated along with the Spare Parts
Management project in a large petrochemical organization and
has been conducted in close collaboration with a dedicated expert
team of asset, maintenance and supply chain managers of the site.
The remainder of this paper is organized as follows. In the next
section, a brief review on classication approaches in literature is
presented. Next, the business context and the ndings of the AS IS
analysis are presented. Section 4 describes the developed multicriteria classication approach for spare parts. Finally, the general
conclusions of the case study are included.

2. Literature review
Over the past decades, spare parts management has acquired
great interest in literature. The topic covers a wide range of
relevant research areas such as inventory control, maintenance
and reliability, supply chain management. Especially in the eld
of stocking strategies, many models are developed by researchers,
answering the basic questions: what to stock? Where to stock?
How much to stock? Due to the variable character of spare parts,
classication of the total assortment was put forward as the
solution for matching appropriate stocking policies to different
classes of items. It is beyond the scope of this paper to review the
huge amount of inventory models developed in literature.

A. Molenaers et al. / Int. J. Production Economics 140 (2012) 570578

Kennedy et al. (2002) provide a good overview of recent research


on spare parts inventories. The focus of this paper is on the
classication process of industrial spare parts. Here the major
challenge is to categorize the spare parts assortment into useful
classes based on one or more relevant attributes. The traditional
classication method is ABC-analysis, which is widely applied in
industries to determine service requirements of spare parts.
Typically items are ranked in terms of one single criterion, such
as demand value or consumption pattern. Class A items require
close managerial attention because they represent such a signicant proportion of the total inventory value. A recent case
study of Syntetos et al. (2009) used ABC-analysis to characterise
spare parts based on demand value for an electronics manufacturer. The researchers realized signicant improvements in terms
of service levels using their approach instead of classifying spares
in terms of demand frequency. Despite the fact that the classical
ABC-method is easy to understand and implement, it is only
successful when the assortment differs mainly in terms of one
single criterion. In the context of industrial spare parts, the
assortment is far more heterogeneous and ABC-classication
based on one parameter is therefore not considered as the most
suitable method by Ramanathan (2006) and Partovi and Anarajan
(2002). Teunter et al. (2010) have showed in their research that
the ABC ranking criteria as demand value and demand volume
can lead to cost inefcient solutions for inventory management.
This has led researchers to extend the traditional ABC-classication to a multi-criteria ABC-analysis including other parameters
like unit cost, critical factor and lead time. A number of authors
such as Ramanathan (2006), Partovi and Anarajan (2002), Zhou
and Fan (2007), Ng (2007) and Chen (2011) proposed the use of
ABC-analysis integrated with models like weighted linear optimization or Articial Neural Networks (ANN). Although these
multi-criteria classication models have several advantages, they
also have some drawbacks. For example it is difcult to incorporate qualitative parameters in ANN models and categorical
measurements in the weighted linear models. Moreover Partovi
and Anarajan (2002) suggest that these models should not
entirely replace professional judgement. Gajpal et al. (1994) and
Braglia et al. (2004) adopted the AHP for spare parts classication
based on criticality. A major advantage of the AHP approach is
that both qualitative and quantitative criteria can be included in
the classication scheme. In addition, the assignment of weights
to the different parameters is considered as a positive characteristic. On the other hand, the amount of subjectivity involved in
the process of pair-wise comparisons is often viewed as the main
limitation of this method. In a case study, Botter and Fortuin
(2000) presented the use of multi-criteria approach to distinguish
important parts from others based on several criteria. An interesting conclusion drawn from this research was that the best
solution to the problem, namely AHP, was considered as too
theoretical and not acceptable by management. The suggested
approach was modied to a simplied VED approach (Vital,
Essential, Desirable), distinguishing the assortment into vital,
essential and desirable parts based on its functionality.
In the eld of inventory control, spares are typically classied
based on attributes like demand pattern, unit price and inventory
costs. In a case study of a petrochemical company, Gelders and
Van Looy (1978) take into account value consumption but also
the functional role of an item in developing appropriate inventory
procedures for slow-moving items. In terms of classifying the
spares from a maintenance viewpoint, classication criteria
emerge such as machine failure, lead times, supplier reliability
and item criticality. The latter, item criticality, has been recognized in literature, but is not well dened and certainly no
consensus has been reached for measuring it. Usually criticality
of an item is expressed as the impact a shortage can have on

571

production, safety and environment. Huiskonen (2001) divides


the aspect of criticality into process criticality and control
criticality. Process criticality is related to consequences associated
with item failure or shortage in terms of production loss, safety
damage or environmental threat. Control criticality refers to the
possibility to control the availability of a spare part. In this issue,
research of Paakki et al. (2011) control criticality is referred to as
risk availability, which is measured in terms of lead-time variance
and accuracy of delivered quantities. Clearly understanding the
concept of criticality depends on a number of factors making it
difcult to measure criticality in a formal and uniform way.
Several research studies point out the need for categorizing
industrial spares by assigning appropriate criticality levels to
them. Dekker et al. (1998) classify the demand for spare parts
into critical and non-critical demand, depending on the criticality
level of the equipment in which it is installed. Parodi and Pintelon
(2007) combine two criteria item criticality and item specicity into a bi-dimensional matrix to categorize spare parts into
four categories: non-critical items, strategic items, critical items
and special items. Advantage of the matrix is its exibility and
clear visualization of the classes. However the drawback is that
criteria are assumed to be of equal importance.
In this paper we are primarily interested in practical and
realistic classication approaches such as those presented by
Braglia et al. (2004) and Syntetos et al. (2009) rather than
approaches of a purely theoretical nature. The purpose is to
develop a classication scheme based on item criticality that
can be easily and uniformly used as a management tool by
practitioners in their decision making.

3. Case study
3.1. The organization
The research work was executed at an international petrochemical company located in Belgium. The industrial plant disposes of more than 50 production installations; most of them are
built as completely integrated production lines. The complex
exists since 1967 and over the years many installations have
been renewed or expanded. As any asset-intensive industry, the
organization strongly depends on the availability of its technical
equipment, like pumps, compressors, heat exchangers and so
forth, as it ensures the continuity of production ow. In this
respect, the spare parts needed to support maintenance tasks,
new investments and modications are crucial to ensure the
minimum downtime possible. Each chemical plant is considered
as an internal client of the on-site logistics provider who is
responsible for the correct and on-time supply of spare parts to
the requestors. Recently the organization rearranged the logistics
network by building a central warehouse on-site, which offers
service to all business plants. Customer needs can be fullled via
three channels: the central warehouse, regional or local
warehouses.
The research project formed part of the Spare Parts Improvement project at BASF ANTWERP NV. The research work was
initiated in September 2008 starting with an in-depth analysis
of the current business situation in terms of spare parts
management.
3.2. Analysis phase
At the moment of the study, the complex held more than 100
thousand SKUs in stock representing a total stock value of about
100 million euros. An overall ABC analysis showed that 98% of the
total demand value was represented by merely 20% of the total

572

A. Molenaers et al. / Int. J. Production Economics 140 (2012) 570578

article assortment, comprising 46 percent of the total stock value.


It was highlighted that the organization disposed of an extensive
and highly valuable spare parts assortment, which for a distinct
proportion, 54%, could be classied as non-moving since four
years. The corresponding stock value of these items comprised
over 50 million euros. There was no information available of the
consumption pattern before 2004. The question that was automatically raised here was why did more than half of the total
article assortment not move? Are those parts critical in ensuring
required operational availability of equipment and therefore held
in stock? Surprisingly, half of the non-movers were not linked to
any bill of material in the system. If the link between spares and
equipment was not documented in the system, how could those
parts be identied when needed?
The total spares assortment is divided into two main categories: standard articles, which are used by more than two
different plants and business specic articles, which are for
designated use of one or two business plants. The majority of
stock articles, 62%, comprised business specic articles representing nearly 80% of total stock value and 35% of total demand
during the four year analysis period. Further investigation
revealed that the stock value of the business specic non-movers
covered 50 million euros Fig. 3.1.
While business specic items are high valuable slow moving
parts, standard articles are less valuable items with a higher
rotation rate. The decision to keep these business specic items
in stock and in what quantities is performed by the plant
maintenance manager based on personal experience and judgement independent of quantied data. Hereby, the most important
factor is the perceived risk of unavailability of the item in case of
breakdown or failure and the associated consequences. Due to lack
of a uniform quantitative method, this leads usually to either a
surplus of nancial resources locked up for holding inventory or to
risking the unavailability of important equipment. As the latter
could result in a considerable amount of hidden costs, related to
production loss, environmental or accidental threat, it is preferred
to hold spares in stock. Unfortunately other logistics specications
of the items, such as replenishment time, are currently not taken
into consideration. One certainty is that, in the case company,
maintenance knowledge and logistics specications are not shared
nor used as a combined information source to control spares
inventories more efciently. Moreover, the discrepancy of strategies between the two parties is extremely broad. While the main
goal of the on-site logistics provider is to control inventory levels
at minimum inventory investments, maintenance people strive for
immediate availability of parts and its prompt delivery.

3.3. Problem delineation


As it is typically seen, many of the problems form part of the
legacy decisions made in the past, which sometimes generate
Business specific versus standard stock articles
100%
35%

80%
60%

62%

79%

40%
20%

65%
38%

21%

0%
Stock articles

Stock Value
Business specific

Demand
(since 4 yrs)
Standard

Fig. 3.1. Analysis of business specic versus standard articles.

frustration with the current problem(s) owners. However, and


specically at an operational level, important organizational gaps
in the spare parts management were identied. Lack of a common
view and strategy for spare parts management between requestor
and provider is believed to be one of the major causes of the spare
parts management problems. These gaps led to problems in
Spare part transparency and visibility
Cooperation between users and logistics provider
Control responsibilities and information sharing between
users and logistics provider
Inventory assortment management (justied basis for holding
inventory)
The objective of the overall Spare Parts Improvement project is
to bridge the major gap in the supply chain on an organizational,
tactical and operational level. In cooperation with management it
was decided that this part of the research work would focus more
specically on the operational application of spare parts control
methods. The in-depth analysis of the current situation revealed the
need to develop a spare parts classication scheme, which exposed
the opportunity to identify and handle more efciently the critical
parts. It was decided that the focus hereby should be on business
specic items only since criticality is a signicant factor for further
decision making. The challenge was to increase the transparency
and visibility of critical spare parts and to provide a justied basis
for holding these items in stock. The classication scheme is
therefore specically developed for business specic items, typically characterized by a high unit price and low usage rate.

4. Spare parts classication based on item criticality


4.1. The model
The method proposed by Braglia et al. (2004) forms the basis
for development of the multi-criteria classication scheme. The
ruling criterion of our classication method is item criticality.
Evaluating the criticality of items is not an easy task because
various characteristics can have an impact on the degree of
criticality. In order to effectively deal with this multi-criteria
problem, a combined methodology is proposed. Two approaches
are used, namely logic decision diagrams and the multi-attribute
technique AHP. The model presents the multi-criteria classication problem in a logic decision diagram where AHP is proposed
to solve the multi-criteria decision sub-problems at decision
nodes of the diagram. The basic idea is to develop a decision
diagram, which guides the analyst towards the best criticality
class of a spare part. The choice for AHP lies in the fact that it is a
powerful and exible tool to integrate qualitative aspects and
assigning weights to different criteria.
The objective of the method is to evaluate the criticality of
spare parts in an efcient and detailed way and to classify the
parts based on their level of criticality. The method aims to create
a profound spare parts classication scheme, which can serve as a
transparent support tool for maintenance and asset managers.
Identication of critical spares is a fundamental attribute to
ensure reliable and safe operation of any modern industrial
system. Furthermore the level of criticality can serve as a valuable
parameter to justify the basis for holding inventory and to
determine service requirements.
4.2. Criticality criteria
The rst step of the multi-criteria classication process is the
identication of relevant criteria impacting item criticality. It was

A. Molenaers et al. / Int. J. Production Economics 140 (2012) 570578

decided to approach item criticality from two perspectives as


proposed by Huiskonen (2001): process criticality and control
criticality. A spare part is considered process critical if its failure
or malfunction can result in severe consequences for the plant, for
example consequences related to loss of lives, environmental
contamination or production loss. Control criticality is not related
to the consequences associated with failure or shortage. In terms
of control criticality, a spare part is considered critical if the
possibility to assure immediate availability of the part is difcult
to control. To assure the usability of the classication tool, a
careful selection of criteria was organized. Several factors were
selected based on their importance, as indicated by the expert
team through two brainstorming sessions. The number of item
characteristics was limited to the most signicant and generic
factors, keeping the multi-criteria problem manageable, exible
and generalizable. A total of six attributes, were selected as
presented in Table 4.1.
Two other criteria namely substitutability and commonality
were initially raised as important criticality parameters by the
experts. Substitutability was in this context dened as the possibility of providing a substitute for a specic spare or repairing a failed
item internally. Finally experts agreed that substitution or repair
was applied in practice only as a temporary solution and therefore
this parameter was excluded. The measure of how frequently the
same spare is used in different components was described as
commonality. Since several spares were presented in the material
breakdown structure of multiple systems, commonality was considered as a valuable parameter by experts. Nevertheless, the
results of the pair-wise comparisons of the AHP process performed
in a later stage, showed that the impact of commonality was
negligible in comparison with the other criteria. Therefore the
latter was also omitted from the classication process. Experts

573

considered service level, space and usage rate also as relevant


information to take into account. However it was pointed out that
these parameters should be included in a later stage when developing the appropriate inventory control strategy for each class.
Next the wide ranges of possible quantitative and qualitative
outcomes of the above criterion are categorized based on the VED
scale. In order to nd the cut-off values for these categories,
several discussions sessions were organized with the expert team.
The values were chosen in such a way that they are appropriate
and applicable for the grand majority of the spare parts assortment. The three categories namely desirable, essential or vital are
summarized in Table 4.2.
Category desirable indicates that the item disposes of the
most preferable outcome when evaluating the item in terms of
criticality for that specic criterion. The opposite is true for
category vital. For example, the most desirable situation in
terms of replenishment time is that the spare part can be
available for use within two days. The replenishment time
becomes essential if the item is available between two days and
one month. If it takes longer than one month to replenish the
item, this criterion is ranked as vital.
For criterion maintenance type, only two distinct alternatives
are considered as important namely corrective maintenance or
preventive/predictive maintenance. In case preventive/predictive
maintenance is executed on a part, the item is considered to be of
a lower criticality level in comparison with an item for which
corrective maintenance is considered. In case of preventive/
predictive maintenance, one is capable to prevent failure/deterioration or to measure indication of failure. Therefore the item is
assessed as less critical.

4.3. Multi-criteria classication


Table 4.1
List of criticality criteria.
Criticality criteria

Description

Equipment criticality

Equipment criticality refers to the criticality class of


equipment. The organization distinguishes between
six possible classes namely A, B, C, D, E and F. Based
on a risk matrix, the involved risk category of
equipment (equipment criticality level) is evaluated
as the frequency of a failure of the equipment and
the possible consequences of the failure.
The probability of failure is the likelihood of failure
or breakdown of the spare part.
The total elapsed time from when a material need is
communicated until the item has been received,
checked, binned and is available for use.
The number of potential suppliers (off-site) who are
able to deliver the specic spare part to the
requestor.
The availability of the technical specications (BOM,
CAD-CAM drawing and order text) of the article.
The type of maintenance performed on the
equipment.

Probability of item
failure
Replenishment time

Number of potential
suppliers
Availability of technical
specications
Maintenance type

4.3.1. Decision diagram


Our method suggests the inclusion of different parameters of
item criticality and presents those as decision nodes in a decision
diagram. The nodes refer to the identied criticality
criteriaequipment criticality, probability of failure, logistics
characteristics and maintenance type. Decision node logistics
characteristics integrates three logistics attributes, namely
replenishment time, number of potential suppliers and availability of technical specications. The decision path to be followed is
based on the outcome of the alternatives (desirable, essential or
vital) of each decision node. Since equipment criticality is considered as the most important factor in classifying spare parts
based on criticality level, the decision diagram starts with this
node. The rst question of the decision diagram says: to which
criticality class does the assigned equipment belong to? The
answer, the equipment criticality class, directs the analyst to
the next question, which considers the likelihood of failure or
breakdown of the item itself. Thirdly logistics characteristics are
analyzed. Since this decision node integrates three logistics
attributes, AHP is proposed to solve the multi-criteria problem

Table 4.2
Categorical measurements of criticality criteria.
Criticality criteria

Equipment criticality class


Probability of item failure
Replenishment time
Number of potential suppliers
Availability of technical specications

Categories
Vital

Essential

Desirable

Criticality classes A, B
Z 1/year
41 month
Only 1 supplier
Not available

Criticality classes C, D
Z 1/5 year and o1/year
42 days andr 1 month
41 and r 3 suppliers
General specications available

Criticality classes E, F
o1/5 year
r2 days
43 suppliers
Detailed specications available

574

A. Molenaers et al. / Int. J. Production Economics 140 (2012) 570578

at this decision node. AHP determines an overall alternative score,


being desirable, essential or vital, based on a combination of the
individual evaluations of the sub-criteria. The fourth decision
node, maintenance type, is included only in specic situations.
Taking into account the outcome of the latter, the diagram directs
the user to the appropriate criticality level of that specic item.
The developed decision diagram is presented in Fig. A1 in
Appendix.

4.3.2. Analytic hierarchy process


We identied three logistics characteristics as relevant parameters impacting item criticality. AHP is proposed to solve the
multi-criteria problem. The Analytic Hierarchy Process is based on
expert judgments and pair-wise comparisons, assigning relative
weights to the different attributes. The decision maker must
determine which are the most important criteria and alternatives.
Also the difference of importance between the elements should be
identied based on a ratio scale and are presented in judgment
matrices. In order to calculate the relative importance or weight,
these judgment matrices are converted into mathematical matrices.
The solution is given by the computation of the priority vector of
the matrices, which is the normalized eigenvector of the matrix.
Table 4.3 shows the AHP judgment matrix for the three sub-criteria.
The hierarchy in Fig. 4.1 presents the relative weights of the
criteria and the alternative scores are presented in the same.
Criterion replenishment time was attributed the largest
weight (0.669) since this parameter was considered as the most
important logistics attribute in terms of item criticality. The relative
weights of the number of potential suppliers and availability of
specications are, respectively, 0.064 and 0.267. The composite
weights are computed by multiplying the relative weights of the
attributes by those of the alternatives. Based on the composite
weights, the expert team indicates the upper and lower limits for
the alternatives of logistics characteristics presented in Table 4.4.
By computing the boundaries of each alternative class, every
possible combination of logistics factors is computed into a single

4.4. Criticality levels


The level or the degree of item criticality is determined by
following the decision path of the diagram. We can distinguish
four levels of criticality ranging from highly critical (criticality
level 1) to non-critical parts (criticality level 4). The levels high,
medium and low are assigned to items with a certain risk
involved (Porras and Dekker (2008)), whereas the risk of noncritical parts is controlled to a large extent. The different criticality levels are described in Table 4.5.
The model can be depicted in a graphical framework, the
criticality cube, as presented below in Fig. 4.2. The axes of the
cube present the categorical measurements of the different
criticality criteria. Along each axis, the three measurements
desirable, essential, vital are dened. The nine segments of this
cube are grouped according to the level of criticality ranging from
one to four. The upper right corner of the cube shows the highly
critical items characterized by high equipment criticality, high
probability for failure and long replenishment time.
An item build in vital equipment (equipment class A or B) can
still be classied as non-critical if the probability of failure and
logistics characteristics are both assessed as desirable. The
reason therefore is the impact replenishment time has on the
overall level of criticality. If the logistic conditions allow parts to
be available within two days, the criticality of the part can be
controlled and reduced to a high extent.

5. Implementation of the method


After development and validation of the multi-criteria classication approach, maintenance and asset managers from three
business plants were selected to test the multi-criteria classication approach. By choosing three different plants in terms of type
Table 4.4
Composite weight for logistics characteristics.

Table 4.3
AHP judgment matrix for sub-criteria.
Replenishment No. of
time
potential
suppliers
Replenishment time 1.000
No. of potential
0.111
suppliers
0.333
Technical
specications

score, which ts simply into one of the three alternative classes


for logistics characteristics.

Technical
specications

Criteria

Normalized
eigenvector

9.000
1.000

3.000
0.200

0.669
0.064

5.000

1.000

0.267

Composite weights of alternatives

Replenishment time
No. of potential suppliers
Availability of technical
specications
Logistics characteristics

Vital

Essential

Desirable

0.046
0.005
0.015

0.104
0.018
0.079

0.519
0.041
0.173

0.0660.260

0.2610.602

0.6030.734

Logistics
characteristics
Replenishment time
(0.669)

# of potential suppliers
(0.064)

Availability of technical
specifications
(0.267)

Desirable
(0.777)

Desirable
(0.643)

Desirable
(0.649)

Essential
(0.154)

Essential
(0.283)

Essential
(0.294)

Vital
(0.069)

Vital
(0.074)

Vital
(0.057)

Fig. 4.1. Hierarchy process.

A. Molenaers et al. / Int. J. Production Economics 140 (2012) 570578

was so far not involved in the project, and an asset manager, who
formed part of the expert team of the project, were asked to study
and evaluate thoroughly the assortment in the context of criticality. Based on expertise, they had to provide the appropriate
criticality ranking, rst independently from each other and nally
a combined score was given. In a second phase, the classication
tool was applied to compute the criticality ranking. The pilot
results showed that the outcome of the classication tool was
highly accurate. 95.4% of the total number of articles classied
with the tool gave a corresponding outcome of criticality level
compared to the manual score. Moreover they perceived the
procedure as a useful and objective tool for classication of spare
parts. For a small proportion of items, 4.6 percent, experts
proposed a higher level of criticality compared to the outcome
of the tool. Main reasons for deviating from the proposed
criticality level were specic requirements in terms of functionality, safety, repair time and lead time. An overview of the pilot
results is listed in Table 5.1.
Another advantage of the tool concerns the transparency and
systematic approach of the method. Item information is now
documented in a consistent and transparent manner ensuring
high visibility to all users. The criticality criteria are commonly
agreed upon and their impact is well understood, which improves

of installation, a representative sample of articles was extracted


to test the method in terms of correctness, reliability but also
functionality and user-friendliness. As the user-friendliness of the
tool was perceived as extremely important, Microsoft Excel was
put forward as the preferred program. Every user is familiar with
the program and importing data information from the ERPsystem can easily be executed. Users of the classication tool
are asked to ll in item attributes based on a drop-down list
where upon the appropriate criticality level is automatically
computed. After optimizing the tool based on feedback reports,
a pilot plant tested the classication method for 2685 business
specic spare parts. In a rst phase a maintenance manager, who
Table 4.5
Criticality levels.
Level
1: High

Description

 Unavailability of the item causes an unacceptable condition,





2: Medium

3: Low

either operational or with respect to safety or the


environment.
Immediate supply of the materials is required.
Risk in the process of ordering and stocking is not tolerated.

 Unavailability causes an unacceptable condition, either





operational, or with respect to safety or the environment but


the consequences can be corrected or controlled.
Material should be supplied within a short time period.
A calculated risk in the process of ordering and stocking.

Table 5.1
Summary pilot results.

 Unavailability causes an acceptable condition, either


operational, or with respect to safety or the environment.

 Supply of materials after a longer period of time.


 Risk in the process of ordering and stocking can be justied.
 Unavailability causes no effect on the processes or on the

Vital

Essential

Desirable

Essential

Total number of spare parts (Pilot plant)


Accurate outcome of the classication tool
Non-accurate outcome of the classication tool

2685
2562
123

95.4
4.6

Reasons for deviation:


Safety requirements
Functional requirements
Repair time
Lead time

44
36
25
18

1.6
1.4
0.9
0.7

Lo

gis

tic

sc

ha

Desirable
Vital

rac
ter

Vital

Essential

Absolute

ist
ics

safety of the people and the environment.


Supply of materials after a long period of time.
Risk in the process of ordering and stocking is normal.

Desirable




Probabiliity of failure

4: No

575

Equipment criticality

4
3

Fig. 4.2. Criticality cube.

3
3
Critical

2
3

2
3

576

A. Molenaers et al. / Int. J. Production Economics 140 (2012) 570578

managerial communication signicantly. The classication


method is documented as a formal procedure and was launched
in December 2009 in the company. All business plants will apply
the classication scheme to categorize their assortment of business specic spare parts in terms of criticality.
Users experience the outcome of the classication also as an
extremely valuable information source for optimizing bill of
materials and they will use the outcome as a justied basis for
holding an item in stock or not. As the function of spare parts
inventories is to assist maintenance staff in keeping equipment in
operating condition, dening inventory guidelines for each class
becomes a crucial task. The next objective in the Spare Parts
Improvement project can be summarized as follows: to ensure
that the requested spares are available in the correct quantity at
the correct time and price, in order to satisfy the desired service
level with minimum costs. Dependent on the level of criticality,
the appropriate inventory control policy is implemented.
Preliminary inventory guidelines were proposed by management:
items with criticality class 1 should be kept in stock requiring
immediate availability and a 100% service level, whereas items
with criticality class 2 should be kept in stock at the optimal
inventory level and a 98% service level. In theory, it is impossible
to reach a 100% service level since extreme outliers are unlikely,
but still possible. In practice, the assumption of 100% service level
is plausible, but this is difcult to translate into a method making
use of a statistical distribution. In a next phase of the Spare
Parts Improvement project, detailed inventory guidelines will be
developed for items with criticality classes 1 and 2. Hereby, rst
a comparison and evaluation of eight theoretical forecasting
methods will be carried out. Among others Holts method and
Willemain bootstrapping will be put to test based on a periodical
review (s, nQ) inventory model. The latter is decided to best suit
the case study as it means only a slight adjustment to the model.
An ex-post approach will be adopted using the available historic
dataset not only for the set-up of the methods, but also for the
performance evaluation of service level and total costs.
It should be noted that some issues were raised in terms of the
data collection process. It was agreed that the up-to-date item
information is a necessary condition in order to effectively use the
proposed tool. Especially in terms of probability of failure,
accurate information is nowadays not always available in the
organization. In this eld, it was proposed to obtain the required
information from an FMEA analysis, which is the preferred
approach when analyzing technical component criticality as
suggested by Waeyenbergh and Pintelon (2002).
The classication tool developed during this research project
contributes signicantly to the problem of providing a quantiable solution to a rather subjective problem, namely when is a
spare part critical? The theoretical model used is tested and
successfully applied in a real industrial environment. The insights
gained throughout the case study can be further used and
generalized towards a generic decision making tool in similar
industrial settings. By applying the information, which is

available, the spare parts assortment can be classied into


criticality classes. The tool allows incorporating even other signicant criticality characteristics, in case of other conditions.
Setting up the logic decision diagrams and performing pair-wise
comparisons should only be performed once. As many industrial
companies cope with an extensive and highly varied spare parts
assortment, generic categorization based on item criticality can
help in further decision making for e.g. inventory control but also
maintenance planning.

6. Conclusions
In the eld of the Spare Parts Management project at BASF
ANTWERP NV, we proposed to develop a multi-criteria classication method based on spare parts criticality. The different parameters inuencing spare parts criticality are equipment
criticality, probability of failure of the item, replenishment time,
number of potential suppliers, availability of technical specications and maintenance type. Based on these characteristics, spare
parts are classied into four classes representing different levels
of criticality (high, medium, low, no). The multi-criteria classication method is based on the AHP and the logic of decision
diagrams. By combining these two techniques, numerous potential attributes inuencing spares parts criticality are taken into
account in an easy and rational manner. The development of the
multi-criteria classication tool helped the organization to identify the critical stock items, which is considered as valuable
information for both asset and maintenance managers. Decisions
to hold those items in stock can now be based on quantitative and
objective information. The transparency and the user-friendliness
of the classication tool were perceived as the major advantages
of the procedure. The contribution of the research work was not
merely focused on the development of a theoretical classication
model but rather on actual implementation of a classication
method in an industrial environment. Moreover the insights
gained can be generalized towards a generic decision making
tool in similar industrial settings.

Acknowledgments
This research work was conducted in close cooperation with
BASF ANTWERP NV. The authors thank the experts of BASF
ANTWERP NV who participated in the brainstorming and validation sessions organized in the course of this research work.

Appendix
See Appendix Fig. A1.

A. Molenaers et al. / Int. J. Production Economics 140 (2012) 570578

577

Vital/Essential
Vital

Logistics
charact.

Desirable
2
Vital

Ess.

Probability
of failure

Logistics
charact.

Essential

Desirable
Vital

Desirable

Logistics
charact.

Vital/Essential

Maintenance type

2
3

Desirable

Vital

Equipment
criticality

Essential

Vital

Logistics
charact.

Ess./ Desirable

Vital/Essential

Ess.
Probability
of failure

Desirable

Logistics
charact.

Desirable

Logistics
charact.

Vital/Essential

Maintenance type

Desirable

2
3

Prev.
Corr.

Desirable
Vital/Essential
Probability
of failure

Vit./Ess.

Logistics
charact.

2
3
4
2

Desirable

Desirable

3
4

Fig. A.1. Decision diagram Criticality Classication.

References
Botter, R., Fortuin, L., 2000. Stocking strategy for service partsa case study.
International Journal of Operations & Production Management 20 (5-6),
656674.
Braglia, M., Grassi, A., Montanari, R., 2004. Multi-attribute classication method
for spare parts inventory management. Journal of Quality in Maintenance
Engineering 10 (1), 5565.
Chen, J., 2011. Peer-estimation for multiple criteria ABC inventory classication.
Computers & Operations Research 38 (12), 17841791.
Dekker, R., Kleijn, M., de Rooij, P., 1998. A spare parts stocking policy based on
equipment criticality. International Journal of Production Economics 56-7,
6977.
Gajpal, P.P., et al., 1994. Criticality analysis of spare parts using the analytic
hierarchy process. International Journal of Production Economics 35 (1-3),
293297.

Gelders, L.F., Van Looy, P.M., 1978. An inventory policy for slow and fast movers in
a petrochemical plant: a case study. The Journal of the Operational Research
Society 29 (9), 867874.
Huiskonen, J., 2001. Maintenance spare parts logistics: special characteristics and
strategic choices. International Journal of Production Economics 71 (13),
125133.
Kennedy, W., Patterson, J., Fredendall, L., 2002. An overview of recent literature on
spare parts inventories. International Journal of Production Economics 76 (2),
201215.
Ng, W.L., 2007. A simple classier for multiple criteria ABC analysis. European
Journal of Operational Research 177 (1), 344353.
T., 2011. Improving global spare parts distribution
Paakki, J., Huiskonen, J., Pirttila,
chain performance through part categorization: a case study. International
Journal of Production Economics 133 (1), 164171.
Parodi, A., Pintelon, L., 2007. Une procedure de classication des pie ces de
rechange. Internal study of Centre of Industrial Management, KULeuven.

578

A. Molenaers et al. / Int. J. Production Economics 140 (2012) 570578

Partovi, F.Y., Anarajan, M., 2002. Classifying inventory using an articial neural
network approach. Computers & Industrial Engineering 41 (4), 389404.
Porras, E., Dekker, R., 2008. An inventory control system for spare parts at a
renery: an empirical comparison of different re-order point methods.
European Journal of Operational Research 184 (1), 101132.
Ramanathan, R., 2006. ABC inventory classication with multiple-criteria using
weighted linear optimization. Computers & Operations Research 33 (3), 695700.
Syntetos, A., Keyes, M., Babai, M., 2009. Demand categorisation in a European
spare parts logistics network. International Journal of Operations & Production
Management 29 (34), 292316.

Teunter, R., Babai, M., Syntetos, A., 2010. ABC classication: service levels and
inventory costs. Production and Operations Management 19 (3), 343352.
Waeyenbergh, G., Pintelon, L., 2002. A framework for maintenance concept
development. International Journal of Production Economics 77 (3), 299313.
Zhou, P., Fan, L., 2007. A note on multi-criteria ABC inventory classication using
weighted linear optimization. European Journal of Operational Research 182
(3), 14881491.

You might also like