You are on page 1of 13

Journal of Quality in Maintenance Engineering

Multi‐attribute classification method for spare parts inventory management


Marcello Braglia Andrea Grassi Roberto Montanari
Article information:
To cite this document:
Marcello Braglia Andrea Grassi Roberto Montanari , (2004),"Multi#attribute classification method for spare parts inventory
management", Journal of Quality in Maintenance Engineering, Vol. 10 Iss 1 pp. 55 - 65
Permanent link to this document:
http://dx.doi.org/10.1108/13552510410526875
Downloaded on: 21 November 2014, At: 01:14 (PT)
References: this document contains references to 18 other documents.
To copy this document: permissions@emeraldinsight.com
The fulltext of this document has been downloaded 4289 times since 2006*
Downloaded by UNIVERSIDAD DE SEVILLA At 01:14 21 November 2014 (PT)

Users who downloaded this article also downloaded:


Adolfo Crespo Márquez, Beno^t Iung, Marco Macchi and Khairy Kobbacy, Irene Roda, Marco Macchi, Luca Fumagalli, Pablo
Viveros, (2014),"A review of multi-criteria classification of spare parts: From literature analysis to industrial evidences", Journal
of Manufacturing Technology Management, Vol. 25 Iss 4 pp. 528-549
A.A. Syntetos, M. Keyes, M.Z. Babai, (2009),"Demand categorisation in a European spare parts logistics network", International
Journal of Operations & Production Management, Vol. 29 Iss 3 pp. 292-316 http://dx.doi.org/10.1108/01443570910939005
Benito E. Flores, D. Clay Whybark, (1986),"Multiple Criteria ABC Analysis", International Journal of Operations &
Production Management, Vol. 6 Iss 3 pp. 38-46 http://dx.doi.org/10.1108/eb054765

Access to this document was granted through an Emerald subscription provided by 406779 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service
information about how to choose which publication to write for and submission guidelines are available for all. Please visit
www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company manages a portfolio of
more than 290 journals and over 2,350 books and book series volumes, as well as providing an extensive range of online
products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics
(COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation.

*Related content and download information correct at time of download.


Practical implications to the paper
Multi-attribute
This paper is motivated by the fact that, for several
classification method for production plants, the shortage cost incurred when a
spare part is needed but unavailable is frequently
spare parts inventory dramatic. At the same time, the existing spare parts
management inventory is excessive and could be reduced
significantly. Hence any improvement in the
management of this type of inventory is desirable
Marcello Braglia and useful in practice, leading to both improved
Andrea Grassi and factory performance and reduced investment in
inventories. In spite of these considerations and
Roberto Montanari because of the complexity of this problem, there has
not been a great deal of research on managing spare
The authors parts inventories, and few inventory models are
Marcello Braglia, is based in Dipartimento di Ingegneria
available in the literature. The goal of this study is
Meccanica, Nucleare e della Produzione, Università di Pisa, Pisa, therefore to develop a decision support tool for use
Downloaded by UNIVERSIDAD DE SEVILLA At 01:14 21 November 2014 (PT)

Italy. by maintenance manager/staff that could be


Andrea Grassi is based in Dipartimento di Scienze e Metodi implemented as an internal structured procedure of
dell’Ingegneria, Università degli Studi di Modena e Reggio the company and adopted as basic approach to revise
Emilia, Reggio Emilia, Italy. and validate the inventory policy used for each type
Roberto Montanari is based in Dipartimento di Ingegneria of spare part in an easy and fast way.
Industriale, Università degli Studi di Parma, Parma, Italy.

Keywords
Inventory control, Analytical hierarchy process Introduction
Abstract The reliability and availability of industrial plants
Inventory constraints, costs of lost production, safety and represents a critical aspect in many modern
environmental objectives, strategies of maintenance adopted, manufacturing and service organisations.
logistics aspects of spare parts are some of the criteria taken into Increased efficiency of production plants requires
account, and spare parts classification is thus defined with the minimisation of machines downtime. Spare
respect to multiple attributes. In virtue of the large number of the parts availability and its prompt accession is
potential operational characteristics to be considered, the among the major factors leading to a reduction of
decision diagram is integrated with a set of analytic hierarchy the protraction of downtime when a breakdown
process models used to solve the various multi-attribute decision
occurs. Thus, a logical approach to solving the
sub-problems at the different levels/nodes of the decision tree.
An inventory policy matrix is defined to link the different classes issue of spare parts availability lies in preserving
of spare parts with the possible inventory management policies ample sizes of inventories of spare parts for
so as to identify the “best” control strategy for the spare stocks. immediate disposition whenever needed. On the
The principles of the theory and an actual application in a other hand, stocking is limited by space and cost.
company operating in the paper industry are reported in the For these reasons, designing the reserve of spare
paper. parts in an optimal way represents a critical and
important task for each production manager.
Electronic access Briefly, the two main approaches followed to
The Emerald Research Register for this journal is develop a possible spares provisioning decision
available at model are (Huiskonen, 2001):
www.emeraldinsight.com/researchregister (1) Mathematical models.
The current issue and full text archive of this journal is available
(2) Classification approaches.
at The first approach concerns the development of
www.emeraldinsight.com/1355-2511.htm mathematical models based on linear
programming, dynamic programming, goal
programming, simulation, etc. Similar approaches
have been examined extensively in the past and a
vast number of inventory models have been
developed during the last decades. Starting from
Journal of Quality in Maintenance Engineering
Volume 10 · Number 1 · 2004 · pp. 55-65 the multi-echelon technique for recoverable item
q Emerald Group Publishing Limited · ISSN 1355-2511 control (METRIC) model of Sherbrooke (1968),
DOI 10.1108/13552510410526875 several researches have been conducted including
55
Multi-attribute classification method for inventory management Journal of Quality in Maintenance Engineering
Marcello Braglia, Andrea Grassi and Roberto Montanari Volume 10 · Number 1 · 2004 · 55-65

more and more different aspects of the spares of analytic hierarchy process (AHP) methodology
management problem (see, for examples, the for spare parts classification are illustrated in the
overview of Kennedy et al. (2002)). These works papers of Gajpal et al. (1994) and Sharaf and
are generally concentrated on the mathematical Helmy (2001). Attributes such as usage rate,
optimisation of the inventory costs and service standard characteristics of spare, lead time of
levels associated with a potential spares inventory supply, spare cost are considered in their models.
policy in terms of economic order quantity, reorder Becker (1987) and de Almeida (2001) propose the
point, safety stocks, and so on. Unfortunately, use of the multi-attribute utility theory (MAUT)
most of these methodologies are either too to this purpose. Risk of item non-supply and spares
complex, abstract or oversimplified, thus reducing cost are the two criteria considered.
their usefulness for a maintenance manager. In In this paper, a new MADM classification
addition, these models do not consider several model as a tool for spare parts inventory
(intangible) additional factors such as management is proposed. The operative scheme is
obsolescence, standard characteristics of the item, based on the application of two different
type and quality of suppliers, etc. techniques/methods: the reliability centered
The use of classification schemes as a spare maintenance (RCM), and the AHP.
parts management tool represents a popular It is known that RCM methodology provides a
Downloaded by UNIVERSIDAD DE SEVILLA At 01:14 21 November 2014 (PT)

approach in industrial world. ABC-classification practical and structured approach for arriving at a
according to the Pareto’s principle is the most well satisfactory maintenance strategy for each
known and used classification scheme to manage component/cause of failure of a given system. In
the spare parts inventory management problems. choosing a maintenance strategy, the methodology
A criticality classification of spare parts is generally takes into account several aspects such as safety
based on administrative efficiency considerations requirement, maintenance costs, costs of lost
(such as inventory costs, usage rates, etc.) derived production, quality problems, etc. Similar to the
from historical data of the company. The modern decision making step in RCM analysis, the basic
production planning software packages (e.g. ERP) idea of our procedure is to define a decision
are able to obtain similar analyses easily and with a diagram which guides the analyst towards the best
reduced time consuming operation. Based on criticality classification for each type of spare part.
these analyses, oversized inventories, obsolescence As in conventional RCM, for the choice of the best
aspects or sock-out problems for the different class (i.e. the leaves of the decision tree), the
items are recognised. finding of the right answer to the questions/nodes
Unfortunately, these approaches are based on a of the diagram is clearly crucial. This might be a
one-dimensional (e.g. the classical ABC-analysis) difficult and not trivial task in several industrial
or a two-dimensional (e.g., Duchessi et al. (1988)) applications because of questions without a unique
classification scheme that does not make it possible answer. Studies concerning the use to probabilistic
to discriminate all the potential control parameters (Eisinger and Rakowsky, 2001) or expert system
of different types of items. To overcome this (Fonseca and Knapp, 2000) approaches to
limitation, some authors have developed new overcome this problem confirm the intrinsic
multi-attribute classification models, which are
difficulties of RCM implementation.
able to manage multiple factors which conflict with When spare parts classification is considered,
each other and heterogeneous units (a unit of one the choice problem at each node of the diagram is
attribute cannot be compared directly with a unit
probably more complex and difficult than in
of another). As also reported in Huiskonen (2001),
conventional maintenance selection. This is due to
Fuller et al. (1993) use a classification scheme
the following considerations:
including six different criteria in their analysis. A .
the choice at each node is rarely binary;
classification expert system model is proposed by .
generally, more attributes must be considered;
Petrovic et al. (1992). Availability of required
and
system, essentiality, price, weight and volume of .
most of these attribute are intangible.
the item, market availability of the spares, and
efficiency of repair are the factors considered in To support the selection process of the analyst, and
their model. Cohen and Ernst (1988) present a to create a complete (i.e. able to manage several
general grouping method that can be used to potential spare control parameters), rational and
define group-based operational control policies. structured resolution approach, in our procedure
Few works (de Almeida, 2001; Becker, 1987; several AHP models are implemented and
Gajpal et al., 1994; Sharaf and Helmy, 2001) integrated with each node of the decision tree. In
concerning application of multi-attribute decision this way, numerous potential attributes influencing
making (MADM) techniques for spares the spares inventory management policy are taken
classification can be also considered. Applications into account in an easy and rational manner but
56
Multi-attribute classification method for inventory management Journal of Quality in Maintenance Engineering
Marcello Braglia, Andrea Grassi and Roberto Montanari Volume 10 · Number 1 · 2004 · 55-65

without using over-complex and unmanageable Table I “Consumption-inventory” matrix


decision diagrams or AHP models. Inventory level
Consumption A B C
A Just-in-time OK
B
The spare parts inventory management C Reduction or elimination of stocks Not important
problem
As shown in Table I, in the squares of the matrix
In industrial contexts, the spare parts inventory the corresponding inventory management policy is
management is generally based on the following reported. For example, for a spare characterised by
steps: low consumption rate (C-class) does not justify an
(1) A criticality analysis of the spare parts is high level of inventory (A-class). The management
conducted (normally through a failure mode policy of the stocks for these kinds of spare
effects and criticality analysis or FMECA). components is not optimised and must be revised.
This is generally necessary to obtain a first Obsolete items are typical of this situation.
reduction of the problem dimensions (i.e. only The best components are characterised by high
Downloaded by UNIVERSIDAD DE SEVILLA At 01:14 21 November 2014 (PT)

the most critical spares are analysed). use and low inventories. In this case changes in
(2) An ABC analysis according to Pareto’s management policy are not generally required.
principle of the most critical spares is carried Spares with low utilisation and low inventories
out to classify the different items. In this way a are considered to be of low importance. Only
second reduction of the items is obtained (i.e. obsolescence/deterioration problems must be
only the spare parts classified into the most carefully taken into account for this type of item.
critical class will be considered in the next The spares with high inventories and high
step). The conventional ABC analysis consumption rates represent the most important
adopted by many firms, classifies the plant case. A careful revision of the safety stocks, reorder
components into three classes of criticality: points and order quantities is generally required so
. very important (A-class); as to attempt to reduce the stocks. In particular, for
.
important (B-class); and the items that are already characterised by low
.
less important (C-class). safety stocks, improvements can be obtained
Only the components belonging to class A adopting the just-in-time paradigms.
require management attention. As one can note, only quantifiable factors are
(3) Finally, an accurate re-definition of the stock taken into account during these standard types of
levels for the different spare parts in function analysis: item costs, usage rate, and historical
of the results of step 2 is executed. inventory levels. Unfortunately, other attributes
concerning more intangible aspects such as safety
It is evident from the decision process that the objectives, provisioning characteristics, type of
criticality analysis/classification of the components maintenance adopted, loss of production, etc. are
represents a fundamental aspect of the study. As generally neglected due their intrinsic difficulties
noted by Sharaf and Helmy (2001), thousands of of quantification. For this reason, research
items may be potentially held in inventory by a addressing the development of more complete
typical company, but only a small portion of them methodologies and making it possible to take into
deserve management’s close attention and account more potential (and intangible) factors
accurate control. influencing the spare parts classification/inventory
The classification of spare parts is frequently strategies, is desirable and useful in practice.
based on a single criterion. The most frequently
adopted is the annual cost usage, which can be
calculated in the following manner:
Multi attribute spare tree analysis
annual cost usage ¼ annual usage rate
(MASTA)
£ dollar value per unit:
The MASTA approach is based on two successive
To support the final definition of the inventory steps. In the first MASTA recognises four spare
policies in real industrial situations, a crossed part classes of criticality and use a logic tree to
analysis is also conducted considering an screen provisioning methods. In the second step,
additional ABC analysis concerning the historical the four classes are crossed with possible different
stock levels recorded for the different items. In inventory management strategies, so as to obtain a
other words, a matrix “consumption/inventory “inventory management policy” (IMP) matrix.
level” can be realised (Table I). Based on this matrix, the most appropriate strategy
57
Multi-attribute classification method for inventory management Journal of Quality in Maintenance Engineering
Marcello Braglia, Andrea Grassi and Roberto Montanari Volume 10 · Number 1 · 2004 · 55-65

for each item can be derived. This two-step AHP structures, a mode of quantification is
approach ensures consistency in determining how proposed and shown below. Using these evaluation
to perform inventory management on all types of judgements as a basis, the pairwise comparisons of
facility equipment. the alternatives (e.g. critical, important and
desirable) with respect to each spare can be
conducted in an easy and quick manner:
Spare parts classification (1) Spare parts plant criticality
As in conventional RCM approach (Eisinger and .
Quality problem:
Rakowsky, 2001), many different decision diagrams
– Critical: quality yield [ [0% 4 85%]
can be potentially proposed for use in spare parts
classification. In this paper, a particular decision – Important: quality yield
diagram developed in a company operating in the [ [85% 4 95%]
paper industry context is presented. The – Desirable: quality yield
importance of the determination of the appropriate [ [95% 4 100%].
stocking level of spare parts in this field, and of the .
Production loss:
number of machine tools that should be carried as
– Critical: productive yield
spares in the operation of a large paper mill, it has
[ [0% 4 90%]
Downloaded by UNIVERSIDAD DE SEVILLA At 01:14 21 November 2014 (PT)

already been underlined by Dhakar et al. (1994).


Figures 1-3 show the logic tree used to – Important: productive yield
determine the best class of criticality for each spare [ [90% 4 97%]
of facility equipment. – Desirable: productive yield
The decision problem at each node of the [ [97% 4 100%].
diagram is supported/executed by using an AHP .
Domino effect:
model. A brief description of AHP functioning is – Critical: global productive effect (many
reported in Appendix 1 and a numerical example
productive parts are involved)
of AHP-node evaluation concerning a bearing
spare part is shown in Appendix 2. The block – Important: limited productive effect
diagrams of the four AHP models adopted are (limited productive parts are involved)
reported in Figure 4. As one can see, all the models – Desirable: no domino effect.
are quite simple but nevertheless make it possible .
Internal safety and environmental:
to consider each decision problem in a more – Critical: one or more fatality
complete and rational way. A total of 17 different
attributes are considered at the end of the selection – Important: one or more seriously
process. This is a more extended number than in injured
the previous researches cited above. – Desirable: one or more productive
As suggested in Gajpal et al. (1994) and Sharaf parts are damaged.
and Helmy (2001), for each criterion of the different .
External safety and environmental:

Figure 1 A sub-tree explosion

58
Multi-attribute classification method for inventory management Journal of Quality in Maintenance Engineering
Marcello Braglia, Andrea Grassi and Roberto Montanari Volume 10 · Number 1 · 2004 · 55-65

Figure 2 A sub-tree explosion


Downloaded by UNIVERSIDAD DE SEVILLA At 01:14 21 November 2014 (PT)

Figure 3 A sub-tree explosion

– Critical: large environmental – Important: setting time device [ [0.5


contamination (water table 4 1.0] lead time
contamination, ecosystem mutations, – Desirable: setting time device $ lead
etc.), ineffective environmental work time.
reclamation .
Cannibalism:
– Important: environmental restoration – Critical: it is not possible replace the
by work reclamation parts with a similar removed by the plant
– Desirable: no environmental – Important: it is possible replace the
contamination. parts with a similar removed by the
(2) Spare supply characteristics: plant but it is not advisable
.
Masked time: – Desirable: it is possible replace the
– Critical: setting time device , 0.5 lead parts with a similar removed by the
time plant without side effect.
59
Multi-attribute classification method for inventory management Journal of Quality in Maintenance Engineering
Marcello Braglia, Andrea Grassi and Roberto Montanari Volume 10 · Number 1 · 2004 · 55-65

Figure 4 AHP models adopted in decision diagram nodes


Downloaded by UNIVERSIDAD DE SEVILLA At 01:14 21 November 2014 (PT)

.
Lead time: – Desirable: price part , $1,000.
– Critical: .¼ four months
.
Space required:
– Important: . one month and , four – Critical: space required part . 10 per
cent available space
months
– Important: space required part [
– Desirable: , four months.
[1% 4 10%] available space
.
Number of potential suppliers:
– Desirable: space required part , 1 per
– Critical: only one supplier in the world
cent available space.
– Important: # five suppliers in your .
Obsolescence:
country
– Critical: MTBF . mean time between
– Desirable: . five suppliers in your new version
country.
– Important: MTBF , mean time
.
Internal repair possibility cost:
between new version
– Critical: internal repair cost .
– Desirable: no obsolescence.
substitution cost .
Deterioration problems:
– Important: internal repair cost [ [0.5 – Critical: ratio price/expiration time . 5
4 1.0] substitution cost per cent spare parts budget
– Desirable: internal repair cost , 0.5 – Important: ratio price/expiration time
substitution cost. [ [1% 4 5%] spare parts budget
(3) Inventory problems:
– Desirable: ratio price/expiration time
.
Price:
, 1% spare parts budget.
– Critical: price part . $25,000 (4) Usage rate:
– Important: price part [ [$1,000 4 .
Number of identical components in the
$25,000] plant:
60
Multi-attribute classification method for inventory management Journal of Quality in Maintenance Engineering
Marcello Braglia, Andrea Grassi and Roberto Montanari Volume 10 · Number 1 · 2004 · 55-65

– Critical: more than ten (2) Single item inventory – this is the most basic
– Important: between 2 and 10 approach. Considerations concerning the risk
of unavailability for the plant, the capacity to
– Desirable: Only one in the plant.
prevent failure time with appropriate
.
Redundancies:
maintenance policies, inventory problem and
– Critical: no redundancy supply characteristics lead toward the stock of
– Important: multiple redundancy a single unit of the spare part.
– Desirable: simple redundancy. (3) Just-in-time policy – this is the most desirable
.
Frequency of failure: approach. In spite of the potential critical
– Critical: high consequences for the plant of an unavailability
of the item, considerations about the easiness
– Important: moderate
and velocity of supplying, the good “control”
– Desirable: low. of failure phenomenon with appropriate
maintenance policies and the problems of
As reported previously, the structure of the decision storage (i.e. cost, space, etc.) lead toward the
diagram, the AHP models and the quantification absence of inventory. This is mainly due to the
modes presented in this sub-section arise from the high level of integration between customer
Downloaded by UNIVERSIDAD DE SEVILLA At 01:14 21 November 2014 (PT)

actual application described in the next paragraph and supplier which makes it possible to obtain
and concerning a paper industry case-study. This quick and reliable supplying in cases of failure.
means that to obtain an effective and useful decision (4) Multi item inventory – this expensive strategy is
support system (DSS), the framework of the mainly due to the high critical impact of an
method has been developed following suggestions unavailability of the spare on the performances
obtained from interviews/suggestions of (cost, quality) and safety of the plant.
maintenance staff and production managers of the Difficulties in spare provisioning do not allow a
firm. But as one can note, the approach philosophy just-in-time strategy of supplying to be used.
has a general validity for the utilisation in different
industrial contexts. Nevertheless, changes in the
proposed procedures are admissible to fit the A case study
particular user necessities and/or requests.
The results of the methodology application in a
paper industry is now presented. The equipment
Inventory management policy matrix analysed concerns a plant working in tissue
The IMP matrix represents a good support to plan production and converting. In particular, the plant
the inventory management policies since it makes is organised as follows:
it possible to identify the strategy in function of the .
a continuous processing plant for paper
criticality class of the item. Table II shows the IPM production directly from pure cellulose;
matrix developed in our case study following the .
three converting lines for tissues production;
ideas and suggestions of company maintenance . one converting line for kitchen paper; and
experts. .
one converting line for toilet paper.
As one can see, in terms of inventory, each
The continuous production process works seven
equipment spare is assigned to one of four possible
days a week The converting department works six
strategies:
days a week, 24 hours a day.
(1) No stock – under a MASTA program,
In terms of spares inventories, more than
unavailability of a spare is a conscious decision
10,000 articles must be potentially managed. In
reached after analysis of what facility
practice, about 7,500 different components are
function(s) would be affected by the absence
stocked in the warehouse for a value of more than
of the spare in case of failure versus the cost/
two million of dollars. The basic inventory policy
problems sustainable to have an immediate
adopted is based on the EOQ theory which makes
availability of the item.
it possible to obtain low risks of stock-out but with
high levels of inventory.
Table II Inventory management policy matrix
To reduce the dimension of the problem, a pre-
Spare part classification analysis has been conducted to recognise the most
Inventory policy A B C D “critical” spares. In this case the pre-screening has
No stock X X not be based on a FMECA approach, but on a
Single item inventory X X crossed ABC analysis concerning the historical
Just-in-time policy X X stock levels recorded for the different spare
Multi item inventory X components. In particular, the two ABC analyses
61
Multi-attribute classification method for inventory management Journal of Quality in Maintenance Engineering
Marcello Braglia, Andrea Grassi and Roberto Montanari Volume 10 · Number 1 · 2004 · 55-65

have been based on slow moving and dead stock .


85 items are now managed following JIT
concepts. philosophy;
With slow moving items we refer to the spares .
the single inventories of 52 items have been
which have not been used for a pre-defined given eliminated; and
period (five months). This concepts is shown in
.
for 61 items the average stock levels have been
Figure 5. considered too high and, as a consequence,
With dead stock items we refer to the spares reduced. For some of these spares the stock
characterised by a level of inventory which has has been reduced to a single item.
never gone under a pre-determined level (dead The monetary amount of this reduction can be
stock) in a given interval of time (generally two quantify in more than 100,000 dollars. This
years). This concept is shown in Figure 6. reduction is mainly due to the (general) electrical
The 1,600 spare components (20 per cent of the components (65-70 per cent), followed by
total) classified contemporarily in A-class in terms mechanical components of the tissue production
of slow moving and dead stock are defined as lines.
“critical” (Figure 7) and subsequently analysed Finally, only to test the goodness of the effected
with the MASTA approach. choices in terms of plant service level, a simple
Downloaded by UNIVERSIDAD DE SEVILLA At 01:14 21 November 2014 (PT)

Briefly, the results of the MASTA application simulation study has been conducted. In other
were the following: words, based on managerial data of the company,
we have simulated the failures that occurred in the
Figure 5 Slow moving concept last five years, assuming the new inventory
configuration was used instead of the old one.
Imposing the costs per hour for a downtime of the
different lines (e.g. 4,000 dollars/hour for the most
expensive continuous process), and estimated the
average lead time provision (in emergency
condition), we have analysed the expected costs for
production loss due to the protraction of downtime
when a breakdown occurs and spare part is not
immediately available. The analysis shows only
one stock-out occurrence more than in old
inventory situations for an estimated additional
cost of about 20,000 dollars.

Figure 6 Dead stock concept


Conclusions
The spare parts inventories management in
industrial plants represents a very complex
problem due to the difficulties concerning data
collection, the number of factors to be considered,
and the large amount of the items involved. A new
multi-attribute technique to define the “best”
strategies of spare inventories management is
presented. The methodology is based on two
instruments. First, similarly to RCM analysis, a
decision diagram drives the analyst towards the
best criticality classification of the items through a
set of decision nodes. To help the analyst during
Figure 7 Critical spares
the choice in each node, several (reduced) AHP
models are proposed. Second, to join an
appropriate inventory strategy to each class, an
inventory management policy matrix is developed.
In this way, a rational and complete analysis can be
quickly conducted considering up to 17 possible
attributes influencing the spares inventory
management policy.
62
Multi-attribute classification method for inventory management Journal of Quality in Maintenance Engineering
Marcello Braglia, Andrea Grassi and Roberto Montanari Volume 10 · Number 1 · 2004 · 55-65

The different aspects of the methodology Petrovic, D. et al. (1992), “SPARTA II: further development in an
framework have been defined following the expert system for advising on stocks of spare parts”,
suggestions of the maintenance experts of an International Journal of Production Economics, Vol. 24
No. 3, pp. 291-300.
international company operating in the paper Saaty, T.L. (1980), The Analytic Hierarchy Process, McGraw-Hill,
industry. However the model maintains a general New York, NY.
validity and modifications of the suggested Saaty, T.L. (1982), Decision Making for Leaders, Lifetime
procedures are possible for a different use/ Learning Publications, Belmont, CA.
industrial context, without altering the basic Saaty, T.L. (1990), “How to make a decision: the analytic
hierarchy process”, European Journal of Operational
“philosophy” of the approach proposed.
Research, Vol. 48, pp. 9-26.
Based on our experience, we can affirm that the Sharaf, M.A. and Helmy, H.A. (2001), “A classification model for
use of AHP methodology is a valid alternative inventory management of spare parts”, Proceedings of 7th
approach to solve the problems of decision making International Conference on Production, Industrial
step in RCM procedure, and it represents the first Engineering, Design and Control (PEDAC 2001),
such proposal, as far as the authors know, to Alexandria, Egypt, 13-15 February, Vol. 1, pp. 375-82.
Sherbrooke, C.C. (1968), “METRIC: a multi-echelon technique for
appear in the literature. It is evident that, as a
recoverable item control”, Operation Research, Vol. 16,
possible future study, the integration between pp. 122-41.
AHP technique and decision diagrams could be Vargas, L.G. (1990), “An overview of the analytic hierarchy
Downloaded by UNIVERSIDAD DE SEVILLA At 01:14 21 November 2014 (PT)

tested also in the original version of RCM (i.e. to process and its applications”, European Journal of
select the “best” maintenance for the different Operational Research, Vol. 48, pp. 2-8.
causes of failures).

Appendix 1. The analytic hierarchy process


References The AHP (Saaty, 1980, 1982, 1990) is a powerful
and flexible multi-criteria decision making tool for
Becker, A.J. (1987), “The scaling of repairable spares”,
complex problems where both qualitative and
Maintenance Management International, Vol. 6 No. 1,
pp. 239-47. quantitative aspects need to be considered. The
Cohen, M.A. and Ernst, R. (1988), “Multi-item classification and AHP helps the analysts to organise the critical
generic inventory stock control policies”, Production and aspects of a problem into a hierarchical structure
Inventory Management Journal, Vol. 29 No. 3, pp. 6-8. similar to a family tree. By reducing complex
de Almeida, A.T. (2001), “Multicriteria decision making on decisions to a series of simple comparisons and
maintenance: spares and contracts planning”, European rankings, then synthesising the results, the AHP
Journal of Operational Research, Vol. 129, pp. 235-41.
not only helps the analysts to arrive at the best
Dhakar, T.S., Schmidt, C.P. and Miller, D.M. (1994), “Basestock
level determination for high cost low demand critical decision, but also provides a clear rationale for the
repairable spares”, Computer and Operations Research, choices made.
Vol. 21 No. 4, pp. 411-20. Briefly, the step-by-step procedure in using
Duchessi, P. et al. (1988), “A conceptual approach for managing AHP is the following:
of spare parts”, International Journal of Physical (1) Define decision criteria in the form of a
Distribution & Materials Management, Vol. 18 No. 5, hierarchy of objectives. The hierarchy is
pp. 8-15.
structured on different levels: from the top
Eisinger, S. and Rakowsky, U.K. (2001), “Modeling of
uncertainties in reliability centered maintenance – a (i.e. the overall objective) through
probabilistic approach”, Reliability Engineering and intermediate levels (criteria and sub-criteria
System Safety, No. 71, pp. 159-64. on which subsequent levels depend) to the
Fonseca, D.J. and Knapp, G.M. (2000), “An expert system for lowest level (i.e. the alternatives).
reliability centered maintenance in the chemical industry”, (2) Weigh the criteria, sub-criteria and
Expert Systems with Applications, Vol. 19, pp. 45-57. alternatives as a function of their importance
Fuller, J.B., O’Connor, J. and Rawlinson, R. (1993), “Tailored
for the corresponding element of the higher
logistics: the next advantage”, Harvard Business Review,
Vol. 71 No. 3, pp. 87-98.
level. For this purpose, AHP uses simple
Gajpal, P.P., Ganesh, L.S. and Rajendran, C. (1994), “Criticality pairwise comparisons to determine weights
analysis of spare parts using the analytic hierarchy and ratings so that the analyst may
process”, International Journal of Production Economics, concentrate on just two factors at the same
Vol. 35 No. 1-3, pp. 293-7. time. The verbal judgements are then
Huiskonen, J. (2001), “Maintenance spare parts logistics: special translated into a score via the use of discrete
characteristics and strategic choices”, International
nine-point scales (i.e. Equally ¼ 1,
Journal of Production Economics, Vol. 71, pp. 125-33.
Kennedy, W.J., Patterson, J.W. and Fredendall, L.D. (2002), “An Moderately ¼ 3, Strongly ¼ 5,
overview of recent literature on spare parts inventories”, Very strongly ¼ 7, Extremely ¼ 9).
International Journal of Production Economics, Vol. 76 (3) After that the judgement matrix has been
No. 2, pp. 201-15.. developed, to calculate a priority vector to
63
Multi-attribute classification method for inventory management Journal of Quality in Maintenance Engineering
Marcello Braglia, Andrea Grassi and Roberto Montanari Volume 10 · Number 1 · 2004 · 55-65

weigh the elements of the matrix. This is the .


The AHP procedure is readily available in
normalised eigenvector of the matrix. decision-making software packages from
(4) Evaluate the goodness of judgements with the several commercial software sellers.
inconsistency ratio IR. This is a peculiarity of
the AHP technique. Before determining an
inconsistency measurement, it is necessary to
introduce the consistency index CI of an n£ n
Appendix 2. A numerical example of AHP
matrix (of judgements) defined by the ratio:
evaluation
lmax  n
CI ¼
n1 A brief numerical example concerning a bearing-
criticality-class determination in the usage-rate
where lmax is the maximum eigenvalue of the node of MASTA approach is here reported.
matrix. Then, IR is defined as the ratio: As one can see in Figure 4, three different
CI attributes (the number of identical parts in the
IR ¼ plant, the redundancies, the frequency of failure)
RI
are considered and three are the criticality classes
Downloaded by UNIVERSIDAD DE SEVILLA At 01:14 21 November 2014 (PT)

where RI is the corresponding average random that are taken into account (desirable, important,
value of CI for an n£ n matrix. The values of critical). In Table AI the pertinent bearing data for
RI are shown in Saaty (1980) (e.g. for n ¼ 3 the three attributes are reported.
we have RI ¼ 0:58, etc.). The judgements can In order to determine the bearing-criticality-
be considered acceptable if IR # 0.1. In cases class adopting the AHP approach, the main steps
of inconsistency, the assessment process for to be followed are:
the inconsistent matrix is immediately (1) Calculate a square, reciprocal symmetric
repeated. An inconsistency ratio of 0.1 or matrix whose elements are pairwise
more may warrant further investigations. comparisons of attribute with reference to the
usage rate. This matrix is carried out by a
After its introduction by Saaty (1980), AHP has questionnaire compilation of an experts
been widely used in many applications (Vargas, committee (Table AII).
1990). The AHP was developed more than 20 (2) Compute and normalise the principal engine
years ago and continues to be the most highly vector of the previous matrix (see Table AII).
regarded and widely used decision-making (3) Calculate the pairwise comparison matrixes
technique. for the criticality classes for to each attribute
The AHP approach is characterised by some and compute the corresponding the principal
important properties: engine vectors. As the previous matrix, these
.
Such a technique makes it possible to tackle are carried out by a questionnaire compilation
the problem in a complete and thorough way, of an experts committee (Table AIII).
taking several factors into account. This (4) Compute the composite weights (W(i) C ) for
capacity is more difficult to obtain when using each attribute “i” and utilise them for the
conventional methodologies such as ABC absolute measurement of the bearing
analysis or mathematical approaches (see, for criticality (Table AIV).
example, multi-echelon models). (5) Define the boundary conditions for the usage
.
It must also be considered that the AHP is rate classification. When the number of
able to manage a large number of possible identical parts in the plant is “only one in the
alternatives in an efficient way. plant” (WCð1Þ ¼ 0:159) and when the
. A direct quantitative judgement of the relevant redundancy is “simple redundancy”
factors is not required by the maintenance (WCð2Þ ¼ 0:138) the bearing is classified as
manager. The pairwise comparisons can be “Desirable” even if its frequency of failure is
preferred by the decision-maker when “moderate” (WCð3Þ ¼ 0:066). This
intangible criteria have to be treated as is the represents the lower boundary condition of
case with spare parts criticality classification. the desirable-usage-rate classification and its
In addition, the coherence of experts’ numerical value is computed as the total of the
judgements can be validated by using the
inconsistency ratio. However, one can note
Table AI Bearing attributes data
that, if reliable quantitative judgements are
Attribute
available for some criteria, they can easily be
Number of identical parts in the
included in AHP analysis.
plant Redundancies Frequency of failure
. The robustness of the approach can be
evaluated performing a sensitivity analysis. 30 No redundancy Moderate

64
Multi-attribute classification method for inventory management Journal of Quality in Maintenance Engineering
Marcello Braglia, Andrea Grassi and Roberto Montanari Volume 10 · Number 1 · 2004 · 55-65

Table AII AHP judgement matrix for the attributes


Number of identical Frequency Normalised
Attribute parts in the plant Redundancies of failure engine vector
Number of identical parts in
the plant 1 1 1/2 0.25
Redundancies 1 1 1/2 0.25
Frequency of failure 2 2 1 0.5

Table AIII AHP judgement matrix for the criticality classes


Class Desirable Important Critical Normalised engine vector
Number of identical parts in the plant
Desirable 1 3 4 0.634
Important 1/3 1 1 0.192
Critical 1/4 1 1 0.174
Redundancies
Downloaded by UNIVERSIDAD DE SEVILLA At 01:14 21 November 2014 (PT)

Desirable 1 2 3 0.550
Important 1/2 1 1 0.240
Critical 1/3 1 1 0.210
Frequency of failure
Desirable 1 3 2 0.547
Important 1/3 1 2 0.263
Critical 1/2 1/2 1 0.190

Table AIV Composite weights for attribute and criticality-class combinations


Criticality-class Composite weights
Attribute Attribute weights Desirable Important Critical Desirable Important Critical
Number of identical parts in
the plant 0.250 0.634 0.192 0.174 W(1)
C 0.159 0.048 0.044
Redundancies 0.250 0.550 0.240 0.210 W(2)
C 0.138 0.060 0.053
Frequency of failure 0.500 0.547 0.263 0.190 W(3)
C 0.137 0.066 0.048

above three composite weights Table AV Usage rate boundary conditions


WC(1)+WC(2)+WCð3Þ ¼ 0:363. Clearly Usage-rate classification Critical Important Desirable
when the number of identical parts in the Composite weights 0.145 0.260 0.261 0.362 0.363 0.434
plant is “only one in the plant”
(WCð1Þ ¼ 0:159), when the redundancy is ten” (WCð1Þ ¼ 0:044), when the redundancy
“simple redundancy” (WCð2Þ ¼ 0:138), and is “no redundancy” (WCð2Þ ¼ 0:053), and
when frequency of failure is “low” when the frequency of failure is “high”
(WCð3Þ ¼ 0:137) the bearing is classified as (WCð3Þ ¼ 0:048) the bearing is classified as
“Desirable” too and its desirable-usage-rate- “critical” too and its critical-usage-rate-
classification value is the desirable-usage-rate classification value is the critical-usage-rate
upper bound WC(1)+WC(2)+ lower bound WC(1)+WC(2)+
WCð3Þ ¼ 0:434. When the frequency of WCð3Þ ¼ 0:145. The boundary conditions for
failure is “high” (WCð1Þ ¼ 0:048) and when the usage rate classification can therefore be
the redundancy is “no redundancy” specified as reported in Table AV.
(WC(2)¼0.053) the bearing is classified as (6) Compute the bearing total composite weight
“critical” even if the number of identical parts and define the bearing-usage-rate-
in the plant is “only one in the plant” classification value. From the bearing data of
(WCð3Þ ¼ 0:159). This represents the upper Table AI (WCð1Þ ¼ 0:044; WCð2Þ ¼ 0:053;
boundary condition of the critical-usage-rate WCð3Þ ¼ 0:066) the bearing total composite
classification and its numerical value is weight is WC(1)+WC(2)+WCð3Þ ¼ 0:163.
computed as the total of the above three Comparing this value in Table AV the bearing-
composite weights WC(1)+WC(2)+ usage-rate classification is obtained. In
WCð3Þ ¼ 0:260. Evidently, when the number particular, for this case, the final classification
of identical parts in the plant is “more then is judged as “critical”.

65
This article has been cited by:

1. Golam Kabir. 2014. Multiple Criteria Inventory Classification Under Fuzzy Environment. International Journal of Fuzzy System
Applications 2:4, 76-92. [CrossRef]
2. Adolfo Crespo Márquez, Beno^t Iung, Marco Macchi and Khairy Kobbacy, Irene Roda, Marco Macchi, Luca Fumagalli, Pablo
Viveros. 2014. A review of multi-criteria classification of spare parts. Journal of Manufacturing Technology Management 25:4,
528-549. [Abstract] [Full Text] [PDF]
3. P. Lynch, K. Adendorff, V.S.S. Yadavalli, O. Adetunji. 2013. Optimal spares and preventive maintenance frequencies for constrained
industrial systems. Computers & Industrial Engineering 65:3, 378-387. [CrossRef]
4. A. Bacchetti, F. Plebani, N. Saccani, A.A. Syntetos. 2013. Empirically-driven hierarchical classification of stock keeping units.
International Journal of Production Economics 143:2, 263-274. [CrossRef]
5. Andrea Bacchetti, Nicola Saccani. 2012. Spare parts classification and demand forecasting for stock control: Investigating the gap
between research and practice. Omega 40:6, 722-737. [CrossRef]
6. An Molenaers, Herman Baets, Liliane Pintelon, Geert Waeyenbergh. 2012. Criticality classification of spare parts: A case study.
International Journal of Production Economics 140:2, 570-578. [CrossRef]
7. Natalie Scala, Jayant Rajgopal, Kim NeedyAn Inventory Criticality Classification Method for Nuclear Spare Parts 365-392.
Downloaded by UNIVERSIDAD DE SEVILLA At 01:14 21 November 2014 (PT)

[CrossRef]
8. Nachiappan Subramanian, Ramakrishnan Ramanathan. 2012. A review of applications of Analytic Hierarchy Process in operations
management. International Journal of Production Economics 138:2, 215-241. [CrossRef]
9. Paakki Jouni, Janne Huiskonen, Timo Pirttilä. 2011. Improving global spare parts distribution chain performance through part
categorization: A case study. International Journal of Production Economics 133:1, 164-171. [CrossRef]
10. Ruben Proano, Sheldon Jacobson, Janet Jokela. 2010. A multi-attribute approach for setting pediatric vaccine stockpile levels.
Journal of Industrial and Management Optimization 6:4, 709-727. [CrossRef]
11. Yurong Zeng, Lin Wang. 2010. A hybrid decision support system for slow moving spare parts joint replenishment: a case study
in a nuclear power plant. International Journal of Computer Applications in Technology 37:3/4, 287. [CrossRef]
12. Fredrik Persson, Nicola Saccani. 2009. Managing the after-sales logistic network–a simulation study. Production Planning &
Control 20:2, 125-134. [CrossRef]
13. N.A. Sharif. 2009. The inventory management systems of a production unit and some academic institutions: a comparative study.
International Journal of Business Performance Management 11:1/2, 134. [CrossRef]
14. M. Ali Ilgin, Semra Tunali. 2007. Joint optimization of spare parts inventory and maintenance policies using genetic algorithms.
The International Journal of Advanced Manufacturing Technology 34:5-6, 594-604. [CrossRef]
15. G. J. Bailey, M. M. Helms. 2007. MRO inventory reduction—challenges and management: a case study of the Tennessee Valley
Authority. Production Planning & Control 18:3, 261-270. [CrossRef]
16. Dr Panayiotis H. Ketikidis, Dr S.C. Lenny Koh, Professor Angappa Gunasekaran, Konstantinos Danas, Abdul Roudsari,
Panayiotis H. Ketikidis. 2006. The applicability of a multi‐attribute classification framework in the healthcare industry. Journal
of Manufacturing Technology Management 17:6, 772-785. [Abstract] [Full Text] [PDF]

You might also like