The Emerald Research Register for this journal is available at The current issue and full text archive

hive of this journal is available at

www.emeraldinsight.com/researchregister www.emeraldinsight.com/1355-2511.htm

APPLICATIONS AND CASE STUDIES Spare parts

estimation
Reliability and operating
environment-based spare parts
169
estimation approach
A case study in Kiruna Mine, Sweden
Behzad Ghodrati and Uday Kumar
Division of Operation and Maintenance Engineering, Luleå University of
Technology, Luleå, Sweden

Abstract
Purpose – With continuous technological development in the twenty-first century, the industry and
industrial systems have become complex and making their availability more critical. In this context,
the product support and its related issues such as spare parts play an important role. Lack of timely or
incomplete support, such as the lack of spare parts when required, is likely to cause unexpected
downtimes, which in turn often lead to incompensatable losses. Therefore the importance of predicting
the correct support to keep the system functionally available needs to be emphasized. Required
number of spare parts could be obtained based on technical and life parameters. This paper seeks to
examine the system reliability and operating environment, which are the two parameters to be
considered in this article.
Design/methodology/approach – A model is provided in this paper to determine the number of
required spare parts with respect to the effect of the external factors, except time, on the reliability
characteristics of components through the proportional hazard model. The model is verified with
estimation of the number of spare hydraulic jacks, used on a load-haul-dump (LHD) machine, as
non-repairable components. The reliability of this non-repairable part and its operational impact are
assessed, while considering environmental factors and ignoring them.
Findings – The results indicate that the operating environment of system/machine has considerable
influence on system performance. Forecasting the required support/spare parts based on technical
characteristics and the system-operating environment is an optimal way to prevent unplanned
stoppages.
Practical implications – The environmental conditions in which the equipment is to be operated,
such as temperature, humidity, dust, road conditions, maintenance facilities, maintenance crew
training, operators’ skill, etc., often have considerable influence directly on the system/machine or
component reliability and indirectly on the product supportability characteristics. Spare parts, are
classified as a product support item whose availability is important when planned or unplanned
maintenance is to be carried out. Forecasting the required number of spare parts, based on technical
characteristics and operating environmental conditions of a system, is one of the best ways to optimize
unplanned stoppages.
Originality/value – Previously, the state of the specific technology and other factors have
demonstrated the need for support in enhancing system effectiveness and preventing unexpected
Journal of Quality in Maintenance
downtime. This paper sets the required number of spare parts necessary to fulfil this need. Engineering
Vol. 11 No. 2, 2005
Keywords Spare parts, Operations management, Distribution and inventory management, pp. 169-184
Systems and control theory, Sweden q Emerald Group Publishing Limited
1355-2511
Paper type Case study DOI 10.1108/13552510510601366
JQME Introduction
11,2 Generally, the industrial products need support through out their lifetime. Some typical
forms of support needed to perform their expected function includes installation,
maintenance, repair services, availability of spare parts, and documentation to user
guiding and training. Such supports extended by original equipment manufacturer
(OEM) and/or suppliers, entails all activities “to ensure that a product is available for
170 trouble-free use to consumers over its useful life span” (Loomba, 1998).
Product support can briefly be defined as any form of assistance that
manufacturers/suppliers offer to users/customers to help them gain maximum value
(profit) from the manufactured products and it is important in the recent industrial
world scenario. Meanwhile, the management considers and pays more attention to
product support, because it:
(1) plays a key role for many products in achieving customer satisfaction;
(2) can be a considerable source of revenue and profit; and
(3) can provide a competitive advantage in marketing.
Maintenance and subsequently spare parts support are two basic and critical issues of
product support. Usually, due to state of the art of technology and other compelling
factors, such as economy, environmental situations, end-user technological
competence, etc., in the design phase, it is impossible to design a product that will
fulfill its total functions. So the need for support has become vital to enhance system
effectiveness and prevent unexpected downtime. In this paper we will consider and
discuss the required number of spare parts to enhance the availability of
system/machine by minimization of the equipment downtime for repair and service.
In a previous paper (Ghodrati and Kumar, 2004) we assumed constant failure rate
for non-repairable components and provided an exponential time to failure distribution
based model. But in this paper, considering the best fit distribution for time to failure of
mechanical components as Weibull distribution we have attempted to provide a new
model that coincides with the real situations and actual requirements.

Spare parts estimation – background

With the advancement of high technology equipment in industries worldwide the need
for spare parts to optimize the utilization of equipment is becoming paramount. Sound
spare parts management improves productivity by reducing idle machine time and
increasing resource utilization. It is obvious that spares provisioning is a complex
problem and requires an accurate analysis of all conditions and factors that affect the
selection of appropriate spares provisioning models.
In the literature, there are a large number of papers in the general area of spare
provisioning, especially in spare parts logistics (Chelbi and Ait-Kadi, 2001; Kennedy
et al., 2002; Langford, 1995; Orsburn, 1991). Most of these papers deal with the
repairable systems and spares inventory management (Aronis et al., 1999; Sarker and
Haque, 2000; Smith and Schaefer, 1985). They mostly provide a queuing theory
approach to determine the spare parts stock on hand to ensure a specified availability
of the system (Graves, 1985; Huiskonen, 2001).
These models have been further extended to incorporate the inventory management
aspect of maintenance (Gross et al., 1985; Hall and Clark, 1987; Ito and Nakagawa, 1995;
Sherbrooke, 1992; Kumar et al., 2000).
 The following common features have been presented in the literature: Spare parts
(1) They mostly deal with repairable system. estimation
(2) They have used queuing theory with demand rate l, and repair rate m. There is
a catch in this, since the failure rate is based on operational time to failure,
where the demand rate (used in inventory models) and repair rate (used in
availability models) are based on calendar time. This distinction was not very
clearly dealt with in the papers. 171
(3) These queuing theory based models primarily deal with constant failure rates,
and constant repair rates (exponential time to failure and time to repair),
whereas this assumption is restrictive, particularly for mechanical parts.
Mechanical parts often fail due to aging with time. The aging or wear out
mechanisms such as creep, fatigue, corrosion, oxidation, diffusion, and wear are
all time dependent processes.
In addition, quantitative techniques based on reliability theory have been used for
developing the failure rates of the required parts to be purchased and/or stocked
(Jardine, 1998; Gnedenko et al., 1969; Kales, 1998; Lewis, 1996; Lipson and Sheth, 1973;
Wååk and Alfredsson, 2001; Xie et al., 2000). This failure rate was used to determine
more accurate demand rates.
In the specific area of spare parts management of non-repairable components
(mechanical), which often fail with time dependent failure rates (ageing), there are some
renewal theory based prediction models available for forecasting the needs of spares in
a planning horizon (Gnedenko et al., 1969; Kumar et al., 2000).
Generally, most of the work and research in the spare parts domain have been done in
the inventory management. It has always been attempted to secure the availability of
systems/machines with existence of spares in stock. Estimation and calculation,
however, of required number of spare parts for storage to guarantee its availability when
required, with respect to techno-economical issues (reliability, maintainability, life cycle
cost, etc.), has rarely been considered and studied. Fewer researches have been relatively
accomplished in this area (for example the notable exceptions are Sheikh et al., 2000;
Tomasek, 1970). In the surveyed literature about the required spare parts calculations
based on reliability characteristics of product, the operating environment conditions
have not been considered (see for example Jardine, 1998; Lewis, 1996). It seems that the
estimations are not supported confidently, as we know that the reliability characteristics
of a product is a function of operation time and operating environmental factors.
Therefore, it appears reasonable to take operating environment issues into account
when studying and analyzing the systems’ reliability, according to the estimate and
forecast of the required spare parts, which has been neglected up till now.

Reliability characteristics of equipment

Generally, system reliability characteristics and factors such as mean time to failure
(MTTF) and mean time to repair (MTTR) for both the component and the whole system
are required for reliability analysis and spare parts forecasting. System operating
environmental factors such as dust, temperature, humidity, pollution, vibration,
operators’ skill, etc. (known as covariates) are required as well in this context. The
covariates influence system’s (includes components) hazard (failure) rate so that the
observed hazard rate is either greater or smaller than the baseline hazard rate.
JQME Thus, for better estimation of the reliability characteristics, the use of regression
11,2 models is suggested because of the possibility of including the covariates.
The proportional hazard model (PHM) was introduced by Cox (1972), which is a
regression type model. The PHM is a complement to the set of tools use in reliability
analysis and provides some particular advantageous features (Kumar and Klefsjö,
1994). This model is classified as multiplicative and non-parametric regression model
172 considering covariates that assumes the hazard rate of a system/component is a
product of baseline hazard rate l0(t), dependent on time only and a positive functional
term c(z ·, a), basically independent of time, incorporating the effects of a number of
covariates such as temperature, pressure and changes in design. Thus:

lðtÞ ¼ lðt; zÞ ¼ l0 ðtÞcðz;

_ aÞ;

where z is a row vector consisting of the covariates, and a is a column vector consisting
of the regression parameters.

Mathematical model for required spare parts forecasting

Spare parts forecasting and inventory management is one of the most challenging
problems in the whole integrated logistic support process. On one hand the operators
want replacement parts to be in stock when required but on the other hand they cannot
afford to have capital tied up in inventory. The cost of spares for an operator of a fleet
of trucks, or loaders over the life of the system far exceeds the cost of the original
system (truck or loader), but it depends on how the fleet is operated, maintained and
supported (Kumar et al., 2000).
For large quantities of items, even a very small error in forecasting the demand for
spares can make a huge deference in the support cost.
Usually, the manufacturer/supplier provides the information on the required
number of spares of each component of the system for a stated period of time (initial
provisioning). However, as mentioned by Pironet (1998), demand prediction for spare
parts as well as maintenance requirements is the weakest aspect of stock management
today, in all armed forces and industries alike.
Two popular mathematical models that are used in spare parts provisioning are
based on Poisson process and renewal theory. The Poisson process can be used
whenever the failure rate (equal to the demand rate and/or hazard rate for the
non-repairable equipments) is constant. This means that each failure mode and other
factors, which influence the demand, should follow the exponential distribution (this
model was discussed in Ghodrati and Kumar, 2004). Whenever the failure rate is not
constant we use renewal theory to forecast demands for spares, which is discussed in
this paper.

Renewal process models for forecasting spares

Let N(t) denote the number of renewals (in our case the number of demands) that occur
by time t. Assuming that the time between renewal random variables X i ; i $ 1, is
independent and have common distribution F(t), then the probability distribution of
number of renewals is given by:

P ½ N ðtÞ ¼ n
¼ F n ðtÞ 2 F nþ1 ðtÞ;

where F n ðtÞ is the n-fold convolution of FðtÞ and given by: Spare parts
Zt estimation
n
F ðtÞ ¼ F n21 ðt 2 xÞdFðxÞ:
0
n
F (t) denotes the probability that the nth renewal occurs by time t. The expected 173
number of renewals, M(t), during a length of t is given by:
X
1
M ðtÞ ¼ F n ðtÞ:
n¼1

The above equation is known as the Renewal Function.

The renewal rate function mðtÞ ¼ dM ðtÞ=dðtÞ gives the expected number of
renewals per unit time.
By substituting the Weibull cumulative distribution function for time to failure,
FðtÞ ¼ 1 2 RðtÞ, as:
FðtÞ ¼ 1 2 exp½2ðt=hÞb
:
These functions can be evaluated for the Weibull model.
Considering replacements of a part having an average time to failure as T
and
standard deviation of time to failures as s(T ) (coefficient of variation of time to
failures, V ¼ sðTÞ=T
). If the operation time t of the system or machine on which this
part is installed is quite long and several replacements need to be made during this
period, then the average number of failures E½N ðtÞ
¼ M ðtÞ will stabilize to the
asymptotic value of the renewal function as (Gnedenko et al., 1969):
t V221
Nt ¼ M ðtÞ ¼ E ½ N ðtÞ
¼
þ
T 2
¼ Average number of failures in time t

And the failure intensity or renewal rate function is given by:

dM ðtÞ dE ½ N ðtÞ
1
mðtÞ ¼ ¼ ¼
:
dt dt T
The standard deviation of number of failures in time t is:
rffiffiffiffi
t
s½ N ðtÞ
¼ V
:
T
If time t in above equations representing a planning horizon is large, then N(t) is
normally distributed (based on central limit theorem) with mean ¼ N ðtÞ and the
number of spares Nt needed during this period with a probability of shortage ¼ 1 2 p
is given by (Sheikh et al., 2000):
rffiffiffiffi
t V221 t
Nt ¼
þ þ V
F 21 ðpÞ;
T 2 T
JQME where F 21 ðpÞ is the inverse normal function and is available in probability textbooks.
While assuming the Weibull reliability model as a most versatile model for
11,2 characterizing the life of machine (mechanical) parts and integrating the effect of
covariates with regard to proportional hazard model, we have:
  !
b t b21 X n
lðtÞ ¼ exp a j zj
174 h h j¼1
!
X n
Assuming: c ¼ exp aj zj
j¼1
   
b t b21 b ðb21Þ
1 t b21
lðtÞ ¼ £c¼ c
h h h h
1
Assuming: k ¼ c ðb21Þ
 
b kt b21
lðtÞ ¼
h h

The reliability model with assuming the h0 ¼ baseline scale parameter, can be defined as:
0 t 1
Z  b21   !
@ b kx A 1 kt b
FðtÞ ¼ 1 2 RðtÞ ¼ 1 2 exp 2 dx ¼ 1 2 exp 2
h0 h 0 k h0
0
 b
kt
FðtÞ ¼ 1 2 exp 2
k 1=b h0
21
This equation indicates the Weibull distribution with the scale parameter h ¼ h0 k ð1=bÞ
1 1 12b
and with substitute of k ¼ c b21 , we have h ¼ h0 ðc b21 Þ b ¼ h0 c 2ð1=bÞ .Thus, it can be
concluded that the influencing covariates change the scale parameter only and the shape
parameter remains unchanged. So:
(
b ¼ b0
h ¼ h0 c 2ð1=bÞ

can be calculated based on
The coefficient of variation of time to failures ðV ¼ sðTÞ=TÞ
the shape and scale parameters as follows:
 

1
T ¼ hG 1 þ ;
b
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
   
2 1
sðTÞ ¼ h G 1 þ 2 G2 1 þ :
b b
Model verification – case study Spare parts
The dominating machine for loading rock in the Kiruna underground iron ore mine in estimation
Sweden is the load-haul-dump (LHD) machine, which is used to pick up ore or waste
rock from the mining points and for dumping it into trucks or ore passes. An
investigation of a fleet of LHD machines deployed at this mine shows that the
hydraulic systems are most critical sub-systems. The lifting cylinder (jack) (Figure 1) is
a part of hydraulic system, which has been considered and studied in this paper. 175
The operation and maintenance cards for a fleet of LHD machines were collected
and required information such as time to failure was obtained from the cards (see
appendix for list of data). For the purpose of preliminary investigations into the
statistical nature of breakdowns of lifting jack, data were classified in their
chronological order and the reordering was avoided to study the nature of trends
presents in the data sets. Three T2500 (25 tone bucket) model machines that are
working in the same condition with the same age were studied. With non-repairable
assumption for hydraulic jacks some minor parts of it (such as gasket or seal),
however, could be replaced in order to restore the failed jack in functional mode. The
time to failures (TTFs) of the hydraulic jacks are given in the Appendix.
The plot of cumulative number of failures of the hydraulic jacks against cumulative
time to successive failures explores the presence of trends in the TTFs (convexity of
the curve, Figure 2) (see for detail Kumar and Klefsjö, 1989). A test for serial correlation
is also done by plotting the ith TTF against the ði 2 1Þth TTF, i ¼ 1; 2; 3; . . . ; n, as
shown in Figure 3, which indicates that no correlation in general among the TTFs was
present.
Thus, we can conclude that the collected data are independent (serial correlation test
based) and not identically distributed (trend test based).
For the purpose of this study, the available information about the operating
conditions of the hydraulic jack was determined and codified by a numeric value
wherever required.
Selection and definition of covariates are very important in the reliability analysis
with covariates because any statistical inference is based on the way they were
formulated. This process should be based on the failure mechanism of the
system/components (hydraulic jack in our case) under study. In this case the
formulation of covariates (influencing factors except time) were carried out based on
observation and the experience of operators and maintenance crew, and are as follows:

Figure 1.
Hydraulic jack
JQME
11,2

176

Figure 2.
Trend test for TTFs of the
hydraulic jack

Figure 3.
Serial correlation test for
TTFs of the hydraulic jack

(1) Human factors. In mining industry, 25-35 percent of the machine breakdowns
are attributed to human related causes (Kumar, 1990). Therefore it will not be an
exaggeration to define and discuss the human factor as a critical covariate. This
covariate can be considered as:
.
Operator skill: this covariate refers to the operator’s experience in driving,
loading and hauling. It is denoted by OPSK and is assigned 2 1 for unskilled
and þ 1 for an expert operator.
.
Maintenance crew skill: this factor affects the quality of service, repair and
maintenance and the condition of jacks after service, and denoted by MCSK.
This covariate like operator skill is assigned 2 1 for unprofessional and þ 1
for an expert crew.
(2) Machine (LHD) factors. This item indicates the parameters which belongs to
machine and operating system and in this case for instance includes:
 .
Hydraulic oil quality: the indicator “HOILQ” which is used to denote the Spare parts
quality of hydraulic oil in the system at work is assigned 2 1 for estimation
non-standard and non-manufacturer/supplier recommended oil and þ 1 for
standard and manufacturer recommended hydraulic oil.
.
Hydraulic system temperature: this factor has influence on the viscosity of
hydraulic oil and elasticity of rubber components in jack (such as seals and
gaskets). For this covariate, the indicator “STEMP” is used to denote the 177
condition and it is assigned the value 2 1 when the temperature is higher
than 55-608C and þ 1 for better (optimum) condition (less than 508C).
(3) Environmental factors. This parameter indicates the effect of operating
environmental factors such as existence of dust, chemical materials, etc. on the
jack. This covariate is present the jack is exposed to corrosive conditions when
the pollution and dust exist in the hydraulic oil and operating environment.
Indicator “ENDUS” denotes this covariate. The value 2 1 signifies presence of
dust and pollution and vice versa for þ 1.
We assumed that a replaced jack had the same baseline failure rate [l0(t)] when the
failed jack was considered to be a non-repairable item. For modeling covariates we use
the proportional hazards regression (Cox, 1972), which is a hybrid model – partly
nonparametric, which allows for an arbitrary survivor function like the Kaplan-Meier
estimator, and partly parametric, in which covariates are assumed to induce
proportional shifts of the arbitrary hazard function. The Kaplan-Meier (product limit)
estimator is equivalent to the Cox model without covariates. When comparing the
parameter estimates of a Cox model with those of a fully parametric model such as the
Weibull, it is important to note that the coefficients are expected to have opposite signs
and will differ by a scale factor.
In the Cox regression analysis, when there is little theoretical reason to prefer one
model specification over another, stepwise methods of covariate selection can be useful.
The software SYSTAT, which is used for estimation of regression coefficients, uses a
“step down procedure” where all the covariates are first considered together in the
model. In SYSTAT, because the forward selection (step up procedure) cannot be used
with the Cox model unless at least one covariate is forced into the model. For this
reason we used backward elimination (step down procedure) with all stepwise
procedures as it is less likely to miss potentially valuable predictors.
Thus, covariates found to have no significant value were eliminated in the
subsequent calculations. The corresponding estimates of a (regression coefficient)
were obtained and were tested for their significance on the basis of t-statistics (the ratio
of the estimated a to the standard error of the estimates) and/or p-value (obtained from
the table of unit normal distribution). One minus the p-value for a covariate gave a
measure of importance when we considered whether to retain any particular covariate
in the model. The estimates of a for the five covariates are listed in step number 0 in
Table I. By following the step down procedure we found that the effects of three
covariates (ENDUS, OPSK and STEMP) were significant at the 10 percent p-value.
So, the best hazard rate model based on the PHM analysis can be defined as:

lðt; zÞ ¼ l0ðtÞ £ expð21:201OPSK 2 1:425ENDUS 2 0:748STEMPÞ:

JQME
11,2

178

Table I.
Estimation of covariates
(the estimates of a and
standard error (S.E.) were
obtained by maximizing
the likelihood function)
Goodness-of-fit tests Spare parts
The constant ratio of any two hazard rates with respect to time is the basic assumption estimation
of the proportional hazard model. This can be indicated as:
lðt; z1 Þ l0 expðz1 aÞ
¼ ¼ exphaðz1 2 z2 Þi;
lðt; z2 Þ l0 expðz2 aÞ
where z1 and z2 are any two different levels of a covariate assumed to be associated 179
with the system. To satisfy the proportionality assumption of the hazard rates
(whether the PHM fits a given data set), the plot of logarithm of the estimated
cumulative hazard rates against time should simply be shifted by an additional
constant a, the estimate of the regression parameter a of the covariate is taken as
strata (Kalbfleisch and Prentice, 1980).
For instance Figure 4 represents the result of this test for the covariate OPSK as a
strata.
As it is seen in Figure 4, the plots are approximately parallel and separated
appropriately corresponding to the different values of the regression parameter a. It
implies that the proportionality assumption is correct.
Based on the results from trend test, the time to failure cannot be exponentially
distributed and on the other hand it follows the power law process with shape parameter
b0 ¼ 3 and scale parameter h0 ¼ 4; 500 hour (manufacturer recommendation).
With this assumption the hazard rate is equal to:
  !  
b t b21 Xn
b t b21
lðtÞ ¼ exp aj zj ¼
h h j¼1
h h
£ expð21:201OPSK 2 1:425ENDUS 2 0:748STEMPÞ:
In this study, the LHD operators were not expert enough. This is because the operators
are not driving the LHD directly by sitting in it; rather the LHD is remote controlled far
away from the operational location. Therefore, the operator is not in the working place
and does not feel the realities. Some times the hydraulic pressure is more than the

Figure 4.
A graphical test for
goodness-of-fit of the PHM
JQME allowed range, which causes rise in hydraulic oil temperature. This in turn causes the
deterioration of jack’s component and finally leads to failure of jack. So, the
11,2 corresponding covariates are assigned by 2 1 for both operator’s skill and the
hydraulic system temperature (OPSK and STEMP). Dust is the most significant factor
as an environmental covariate, which the jacks have protection, in various ways,
against it. For instance there is a filter, which refines the hydraulic oil of the system
180 from dust and other physical particles. So, due to dust protection, the corresponding
covariate for ENDUS is assigned by þ 1. While considering these concepts, we have:
!
Xn
c ¼ exp aj zj ¼ expð21:201 £ ð21Þ 2 1:425 £ ðþ1Þ 2 0:748 £ ð21ÞÞ ¼ 1:69;
j¼1

(
b ¼ b0 ¼ 3
h ¼ h0 c 2ð1=bÞ ¼ 4500 £ 1:692ð1=3Þ ¼ 3; 779 ðhrÞ

   
1 1
T
¼ hG 1 þ ¼ 3; 779 £ G 1 þ ø 3; 374 ðhrÞ
b 3
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
   
2 1
sðTÞ ¼h G 1 þ 2 G2 1 þ ¼ 3; 779 £
b b

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
   ffi
2 1
G 1þ 2 G2 1 þ ø 1; 227 ðhrÞ;
3 3

sðTÞ 1; 227
V¼ ¼ ¼ 0:364:
T
3; 374
The expected number of required spare jacks in one year (two working shifts per day)
when T
¼ 3; 374 hours is considered to be the real mean time to failure of the jack with
a 95 percent confidence of availability is equal to:
rffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffi
t V221 t 21 5; 600 0:3642 2 1 5; 600
Nt ¼
þ þ V
F ðpÞ ¼ þ þ 0:364 £ 1:645
T 2 T 3; 374 2 3; 374

N t ø 2 ðpiece=yearÞ:

If we ignore the effect of covariates on the jack hazard rate, then the required number of
spare jacks will be calculated as:
   

1 1
T 0 ¼ h0 G 1 þ ¼ 4; 500 £ G 1 þ ø 4; 020
b 3
 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
   
sðT 0 Þ ¼ h0 G 1 þ
2
2G 1þ 2
1
¼ 4; 500 £
Spare parts
b b estimation
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
   ffi
2 1
G 1þ 2 G2 1 þ ø 1; 448
3 3
181
sðT 0 Þ 1; 448
V¼ ¼ ¼ 0:364
T
0 4; 020
sffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffi
t V221 t 21 5; 600 0:3642 2 1 5; 600
N0 ¼
þ þV
F ðpÞ ¼ þ þ 0:364 £ 1:645
T0 2 T0 4; 020 2 4; 020

N 0 ø 1:65 ðpiece=yearÞ:
This difference in required spare parts might not seem to be important and significant,
but in one year it is 4 jacks for a fleet of 12 LHD (existent number of LHD-T2500 in
Kiruna mine) and is considerable in the sense of spare parts forecasting and inventory
management.

Conclusion
Some times the LKAB Company faces downtime of LHDs due to shortage in
availability of required spare parts. This is because of the manufacturer/supplier’s
recommended less number of required spare parts to be kept in stock. In most cases the
manufacturer is not aware of the environmental factors or has not considered these
issues in the estimation of the number of required spare parts (like in this case). So, to
avoid downtime regarding the unavailability of spare parts, it is suggested that the
mine company should take the operating environmental factors into consideration
while estimating the spare parts need. The operating environment of system/machine
has considerable influence in system performance and its technical characteristics such
as reliability and maintainability. Forecasting required support/spare parts based on
technical characteristics and the system-operating environment is an optimal way to
prevent unplanned disruptions or stoppages.

References
Aronis, K.P., Magou, I., Dekker, R. and Tagaras, G. (1999), “Inventory control of spare parts
using a Bayesian approach: a case study”, Econometric Institute Report, No. EI-9950/A,
Erasmus University, Rotterdam.
Chelbi, A. and Ait-Kadi, D. (2001), “Spare provisioning strategy for preventively replaced
systems subjected to random failure”, International Journal of Production Economics,
Vol. 74 No. 1, pp. 183-9.
Cox, D.R. (1972), “Regression models and life-tables”, Journal of the Royal Statistical Society,
Vol. B34, pp. 187-220.
Ghodrati, B. and Kumar, U. (2004), “Operating environment-based spare parts forecasting and
logistics – a case study”, accepted for publication in International Journal of Logistics:
Research and Applications.
JQME Gnedenko, B.V., Belyayev, Y.K. and Solovyev, A.D. (1969), Mathematical Methods of Reliability,
Academic Press, New York, NY.
11,2
Graves, S.C. (1985), “A multi-echelon inventory model for a repairable item with one-for-one
replacement”, Management Science, Vol. 31 No. 10, pp. 1247-56.
Gross, D., Miller, D.R. and Soland, R.M. (1985), “On some common interests among reliability,
inventory and queuing”, IEEE Transactions on Reliability, Vol. 34 No. 3, pp. 204-8.
182 Hall, F. and Clark, A.J. (1987), “ACIM: availability-centered inventory model”, Proceeding of the
Annual Reliability and Maintainability Symposium, IEEE, New York, NY, pp. 247-52.
Huiskonen, J. (2001), “Maintenance spare parts logistics: special characteristics and strategic
choices”, International Journal of Production Economics, Vol. 71 No. 3, pp. 125-33.
Ito, K. and Nakagawa, T. (1995), “An optimal inspection policy for a storage system with high
reliability”, Micro Electron Reliability, Vol. 36 No. 6, pp. 875-82.
Jardine, A.K.S. (1998), Maintenance, Replacement and Reliability, Preney Print and Litho Inc.,
Windsor.
Kalbfleisch, J.D. and Prentice, R.L. (1980), The Statistical Analysis of Failure Time Data,
John Wiley & Sons, New York, NY.
Kales, P. (1998), Reliability: For Technology, Engineering, and Management, Prentice-Hall Inc.,
Englewood Cliffs, NJ.
Kennedy, W.J., Patterson, J.W. and Fredendall, L.D. (2002), “An overview of recent literature on spare
parts inventories”, International Journal of Production Economics, Vol. 76 No. 2, pp. 201-15.
Kumar, D. and Klefsjö, B. (1994), “Proportional hazards model: a review”, Reliability Engineering
and System Safety, Vol. 44 No. 2, pp. 177-88.
Kumar, U. (1990), “Reliability analysis of load-haul-damp machines”, PhD thesis, Luleå
University of Technology, Luleå.
Kumar, U. and Klefsjö, B. (1989), “Reliability investigation for a fleet of load-haul-dump machines
in a Swedish mine”, Reliability Engineering and System Safety, Vol. 26, pp. 341-61.
Kumar, U.D., Crocker, J., Knezevic, J. and El-Haram, M. (2000), Reliability, Maintenance and
Logistic Support – A Life Cycle Approach, Kluwer Academic Publishers, New York, NY.
Langford, J.W. (1995), Logistics: Principles and Applications, McGraw-Hill Inc., New York, NY.
Lewis, E.E. (1996), Introduction to Reliability Engineering, John Wiley & Sons Inc., New York, NY.
Lipson, C. and Sheth, N.J. (1973), Statistical Design and Analysis of Engineering Experiments,
McGraw-Hill, New York, NY.
Loomba, A.P.S. (1998), “Product distribution and service support strategy linkages – an
empirical validation”, International Journal of Physical Distribution & Logistics
Management, Vol. 28 No. 2, pp. 143-61.
Orsburn, D.K. (1991), Spares Management Handbook, McGraw-Hill Inc., New York, NY.
Pironet, F. (1998), “Multiple provisioning strategy for weapons system used by multiple
missions”, in Knezevic, J., Kumar, D. and Nicholas, C. (Eds), Proceedings of the 8th
International MIRCE Symposium, Exeter University, pp. 147-60.
Sarker, R. and Haque, A. (2000), “Optimization of maintenance and spares provisioning policy
using simulation”, Applied Mathematical Modeling, Vol. 24 No. 10, pp. 751-60.
Sheikh, A.K., Younas, M. and Raouf, A. (2000), “Reliability-based spare parts forecasting and
procurement strategies”, in Ben-Daya, M., Duffuaa, S.O. and Raouf, A. (Eds), Maintenance,
Modeling and Optimization: The State-of-the-Art, Kluwer Academic Publishers, Boston,
MA, pp. 81-108.
Sherbrooke, C.C. (1992), Optimal Inventory Modeling of Systems, John Wiley & Sons Inc., Spare parts
New York, NY.
Smith, C.H. and Schaefer, M.K. (1985), “Optimal inventory for repairable redundant systems with
estimation
aging components”, Journal of Operation Management, Vol. 5 No. 3, pp. 339-49.
Tomasek, K.F. (1970), “Technical notes – calculation of the required number of spare parts”,
Microelectronics and Reliability, Vol. 9 No. 1, pp. 77-8.
Wååk, O. and Alfredsson, P. (2001), Constant versus Non-constant Failure Rates: Some 183
Misconceptions with Respect to Practical Applications, Systecon, Stockholm, available at:
www.systecon.se/publications.html
Xie, M., Kong, H. and Goh, T.N. (2000), “Exponential approximation for maintained Weibull
distributed component”, Journal of Quality in Maintenance Engineering, Vol. 6 No. 4, pp. 260-9.

Appendix
Table AI contains the data sets used for demonstrating the concept. The column with TTFs
exhibits time to failure of a particular type of hydraulic jack.

Table AI.
Field data used for
analyzing and
demonstrating the concept
JQME Table explanation: if we consider row 1, for instance column 2 indicates that the hydraulic
jack failure takes place after working 2536 hours. Considering column 3 with MCSK, þ 1
11,2 indicates that the maintenance and service crew were skilled, column 4, þ 1 indicates that the
LHD operators were expert enough in their job, column 5, 2 1 indicates physical environment
was not good and acceptable (such as existence of dust), in column 6, þ 1 indicates the hydraulic
oil used was standard oil recommended by manufacturer, and finally column 7, 2 1 indicates
system’s temperature condition was not good (for example high temperature).
184