Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

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

net/publication/243085075

Benchmarking of maintenance performance: a case study in two


manufacturers of furniture

Article  in  IMA Journal of Management Mathematics · July 2004


DOI: 10.1093/imaman/15.3.253

CITATIONS READS

11 280

1 author:

Basim Al-Najjar
Linnaeus University
68 PUBLICATIONS   1,133 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Integration of the data gathered from maintenance, production, quality and accountancy View project

Predictive Cognitive Maintenance Decision Support System (PreCoM) View project

All content following this page was uploaded by Basim Al-Najjar on 14 June 2016.

The user has requested enhancement of the downloaded file.


IMA Journal of Management Mathematics (2004) 15, 253–270

Benchmarking of maintenance performance: a case study in


two manufacturers of furniture

BASIM A L -NAJJAR†
School for Industrial Engineering, Växjö University, 351 95 Växjö, Sweden

M ATS -O LA H ANSSON
Information Technology Adviser, WM-Data Ellips AB, 35104 Växjö, Sweden

AND

PATRIK S UNNEG ÅRDH


System Safety Consultant, AerotechTelub, 351 08 Växjö, Sweden

[Received on 10 January 2003; revised on 7 May 2004]

When profit margins of a plant are decreasing, the need for reliable and efficient
maintenance policy becomes more important. Measuring maintenance performance is
important for companies to recognize whether their planned goals are achieved or not.
Also, such measurements can be utilized for benchmarking, which is one of the tools
for never-ending improvement. But, these objectives cannot be achieved without well-
documented data of the relevant variables. Better data coverage and quality is necessary for
following maintenance performance development and it would, in many cases, clarify the
ambiguity concerning the main problem in the context, namely that neither the production
nor maintenance department can show what effect maintenance has on profitability.
A more effective maintenance policy indirectly implies improvements in product
quality and manufacturing process effectiveness. Elongation of the production time, i.e.
reducing the downtime due to failures, planned replacements and repair, in addition to
the improvement in the total maintenance activities, are also some of the results that can
be expected when an efficient maintenance policy is used. Measuring and monitoring
maintenance performance measures is required partly for detecting, and eventually treating
as soon as possible, undesirable changes and partly to make benchmarking with the best in
the branch, which saves appreciable economic losses for companies.
In this paper, a model for how to identify the measurable variables, which are needed
to develop measures for monitoring maintenance performance behaviour systematically,
is developed. Five maintenance performance measures are proposed and applied. An
additional model for systematically analysing the trend of maintenance performance
measures, for an overall assessment of the company’s situation, is presented. Two case
studies in manufacturers of furniture are conducted to verify these models.

Keywords: maintenance performance measures; benchmarking; maintenance cost; quality


rate; operating cost; overall process effectiveness (OPE).

† Email: basim.al-najjar@ips.vxu.se

IMA Journal of Management Mathematics Vol. 15 No. 3 


c Institute of Mathematics and its Applications 2004; all rights reserved.
254 B . AL - NAJJAR ET AL.

1. Introduction
In Dunn (1987), the cost of maintenance is estimated to be between 15 and 40 per cent
of the production costs. According to Al-Najjar (1997), maintenance performance can be
assessed by different measures. Each of these measures consists of two or more measurable
variables and reflects some of the real situation of maintenance. This is why it is not easy
to find one comprehensive measure for controlling maintenance performance.
Measuring maintenance performance is important for companies to know whether
the planned goals are achieved or not. Also, such measurements can be utilized for
benchmarking, which is one of the tools for never-ending improvement. But, these
objectives cannot be achieved without well-documented data of the relevant variables.
This is necessary for following maintenance performance development and it would, in
many cases, clarify the ambiguity concerning the main problem in the context that neither
the production nor maintenance department can show what effect maintenance has on
profitability (Löfsten, 1999).
Selection of the relevant variables that are required for developing maintenance
performance measures is influenced by the objectives expected when using these measures.
When monitoring the performance of a maintenance policy, both technical and economic
measures should be considered to reflect the significance of the changes done in the
maintenance policy and manufacturing process. The technical measure, which consists of
variables such as product quality rate, machine availability and performance efficiency such
as overall equipment effectiveness (OEE) reflects only machine effectiveness (Al-Najjar,
1996). Maintenance performance measures should not be used only as an information
source for management control and decision making. They should be considered as a pow-
erful tool for motivating and driving actions that are consistent with the advocated strategy.
In Australia, De Jong (1997) conducted a survey focused on maintenance practices of
small and medium sized enterprises. It was found that the main measure of maintenance
performance used in the responding companies is the ratio of the total cost of maintenance
to the estimated equipment replacement value. Eccles (1995) proposed that customer-
oriented measures such as response time, service commitments and customer satisfaction
should serve as lead indicators of business success. The economic measures are necessary
to express, for example, a company’s different cost factors, cost-effectiveness, losses,
profitability, benefits and their development during operation and according to the improve-
ments introduced within the enterprise. According to the European Benchmarking Study
on Maintenance group (EBSOM, 1993), there is a growing interest in judging maintenance
activities through the measurment of achieved maintenance results by specific variables.

2. Difficulties in evaluating maintenance performance


In general, improvements in technical and economic maintenance effectiveness are usually
motivated by actual needs. Some of these needs can be explored by asking questions
regarding whether:

1. maintenance costs can be reduced and efficiency improved; maintenance efficiency


is defined here by two measures, effectiveness and accuracy (see Al-Najjar, 1991,
1997);
BENCHMARKING OF MAINTENANCE PERFORMANCE 255

2. the applied maintenance policy is better or worse than that being used in other
companies (within the same branch or other branches);
3. there were reasons behind why special arrangements and investments in mainte-
nance were being enacted at other companies and not at ours.
The study performed by the EBSOM (1993) group showed that several difficulties exist
when trying to measure maintenance performance. These difficulties concern
1. identification of activities that should be included by maintenance: for example
repair, cleaning, greasing and adjusting;
2. distinguishing production assets from buildings and installations;
3. operator’s activities that can be considered to be a part of maintenance;
4. distinguishing spare parts and the material being consumed when performing
maintenance actions;
5. difficulties when judging and calculating availability.

Dwight (1995) identified the following shortcomings of the performance measures that
are currently used in industry:

1. the concept of accumulation of risk is not captured;


2. the focus is on the immediate rather than the overall requirement;
3. measures are not related to business requirements.

Many attempts to develop general measures to assess maintenance performance have


been made. Henriksson (1993), for example, used a number of measures to compare
maintenance performance at multinational companies. Identification and development of
relevant measures for monitoring maintenance performance is necessary before starting
benchmarking maintenance. The selected maintenance performance measures should fulfil
requirements such as ability to

1. describe the performance behaviour of the manufacturing process in question;


2. provide relevant data and less noise;
3. offer possibility to perform periodic and continuous measurements;
4. use measures that are understandable and acceptable within the company and
externally;
5. the measures should be able to assess the improvements in maintenance performance
and their impact on the company’s results.

3. Variables for measuring, monitoring and benchmarking maintenance perfor-


mance
Benchmarking means striving to equal the best. In Japanese the equivalent word for
benchmarking is dantotsu. It is the continuous process of measuring products, service and
processes against the industrial leaders or the toughest competitors (Oakland, 1995). In
general, improvements can easily be realized with respect to some established standards.
Internal benchmarking is a comparison of internal operations.
Usually, a wide range of areas may be benchmarked, with short, medium and long-
term targets. Benchmarking maintenance is a way to identify the best result gained by a
256 B . AL - NAJJAR ET AL.

competitor (or another company in a different branch) which makes use of approximately
the same machinery. This may ultimately lead to the adoption of a more cost-effective
maintenance policy or to performing the required changes within the one currently
available. The implementation of a new maintenance policy or performance of the
required changes should lead to an exceptional improvement in the effectiveness of the
manufacturing process, in particular in the production cost of quality products.
Many parameters that can be utilized in monitoring maintenance performance of a
particular equipment, such as downtime, availability, reliability, performance efficiency,
product quality rate, overall equipment effectiveness (OEE), maintenance-related costs,
income losses, total maintenance cost, delivery delay penalties, number and cost of
accidents, spare parts, tied-up capital, planned stoppages, failures, cost of manufacturing a
quality product, etc. (Al-Najjar, 1996). Before benchmarking it is necessary to collect data
on the relevant variables, which is needed for estimating maintenance measures.
The aim of maintenance benchmarking is to select and improve the most cost-effective
policy through comparison of its concept, objectives, tools, methods and benefits, etc with
the ‘best in class’ company. This is why it is always beneficial to explore the within-
branch competitors’ economic situations to be used for benchmarking. But, the information
required from the ‘best in class’ to be able to perform such a comparison is not always
easy to find. Also, it is difficult for the ‘best in class’ company to choose which company
to compare itself with if it does not implement the total quality maintenance concept
(TQMain). Implementation of TQMain opens the possibilities for further improvements
and provides opportunities to improve quality, the cost of manufacturing quality products
and the annual profit continuously ‘without’ the need to get additional information
concerning competitors (Al-Najjar, 1996, 1997). It allows a company to take the lead rather
than just stay competitive.
The economic measures reflect the economical impact of maintenance performance on
the plant activities and they are required to describe maintenance costs, effectiveness, role
and importance, and to give a reasonable platform for judging a company’s economic situa-
tion from the standpoint of maintenance. The primary feature of using maintenance perfor-
mance measures is to monitor one of the crucial factors influencing a company’s economic
development, i.e. maintenance-related costs. Many variables can be considered for measur-
ing economic development, such as solidity, liquidity, profitability, turnover per employee,
rate of return on investment, total maintenance cost, operating cost and income reduction.
Technical measures are used to describe manufacturing process effectiveness develop-
ment technically. The most interesting variables for measuring effectiveness and accuracy
of maintenance are stoppage time, machine availability, number of planned replacements,
failures, product quality rate, OEE, Overall process effectiveness (OPE) (Al-Najjar, 1997).
Some of these measures include other measures such as OEE, which expresses availability,
product quality rate and performance efficiency.
A cost-effective maintenance policy is one which results in benefits that justify all the
costs incurred by driving it. Some aspects of the economic and technical effectiveness can
sometimes be assessed simultaneously in one measure: for example the total maintenance
cost divided by the number of the manufactured and accepted items (i.e. items of good
quality).
It is very important to clearly identify the goals behind use of each measure to avoid
confusion and to systemize the analysis of maintenance performance. In order to identify
BENCHMARKING OF MAINTENANCE PERFORMANCE 257

how much one-unit maintenance-related increment in OEE costs a company during a


particular period, the total maintenance cost should be divided by the increment created
in OEE due to using better maintenance during the same period. This measure makes
it easier for the company to explore the role of maintenance improvement on machine
availability and effectiveness, and product quality rate. There are many criteria that
should be fulfilled by the selected measure, which will be used to monitor maintenance
performance. Maintenance performance measures are preferably
1. simple to understand and implement,
2. applicable in different circumstances,
3. informative enough so that it is possible to judge the situation and trace the basic
reasons behind deviations,
4. cost-effective,
5. sensitive to changes in maintenance performance and manufacturing process,
6. not easily influenced by disturbing factors,
7. representative, i.e. provide relevant information with respect to the specified goals.
If the measures are complex, they are not likely to be widely used in the long run.
In order to achieve a reliable comparison between different companies, these companies
should define their maintenance performance measures in a similar way. Different
departments at a company may use different measures for monitoring their maintenance
performances. It is important that the measures which will be used for monitoring
maintenance performance should be applicable in the plant’s different departments.

4. Maintenance performance measures


The model developed below describes, in a logical and systematic way, how to identify
the relevant variables that are needed to develop measures for monitoring maintenance
performance. Four criteria have been considered in developing this model: namely that
it should be simple, illustrative, usable and logical. It is developed on the basis of the
concept of never-ending improvement when using a plan–do–control–act (PDCA) cycle,
i.e. the model incorporates continuous improvement, see Fig. 1.
The first step is adoption of the concept of never-ending improvement, which is
usually based on internal or external demands stated by, e.g. technical analysis, customers’
complaints and new innovations, shareholders, etc. The second step is naturally initiated
when the company realizes where the improvements should be focused. Identifying and
gathering relevant data at the right frequency (third step) is necessary to describe and assess
the situation. The first three steps are very important to build a solid platform for the coming
work.
Analysis of the gathered data in step four leads to a description of the current
condition of the maintenance and manufacturing process and identifies which processes
or equipment that should be improved. In step five, relevant measures should be identified
and defined clearly to avoid any misinterpretation. The variables previously selected should
be categorized in step six to develop (according to the criteria mentioned in Section 3) the
required measures for monitoring maintenance performance.
In step seven, calculations of the observed variables give numerical results of the
measures, which will be used for monitoring changes in the maintenance performance
258 B . AL - NAJJAR ET AL.

Step

Concept of Never-ending
1
improvement

Selection of the area


2
for improvement

Identifying relevant variables


3
and gathering data

Data analysis Process improvement 4

Identification of the relevant 5


measures for monitoring
maintenance performance

Definition and development of


6
the variables and measures

Comparison between Variables and measures calculation


steps 2 and 9 7
(Measures numerical results)
10

Actions Analysis
8
9

F IG . 1. Ten-step model to identify the relevant variables needed to develop measures for monitoring maintenance
performance.

and manufacturing process. Also, it can be used to compare different departments at


the same company or in different companies. Analysis of changes in the variables’ (and
consequently measures) values and identification of the basic reasons behind changes
helps the company to make the right decisions at the right time, i.e. steps eight and nine
respectively. In the tenth step, a comparison between planned and actual improvements will
be done to control the level of success and to decide whether a revision of the improvement
policy is necessary.

5. Developing maintenance performance measures


The measure OEE suggested by Nakajima (1988) represents the effects of six major losses
on the equipment effectiveness. These losses are divided into three major areas; availability
BENCHMARKING OF MAINTENANCE PERFORMANCE 259

(A), performance efficiency (η), and quality rate (1 − q), where q is the proportion of
defectives produced, and η may be defined as the theoretical mean time for producing
one item to its actual time. The latter is a measure of the effect of small adjustments and
production speed reduction on the theoretical performance cycle time. Thus, OEE may be
expressed as OEE = Aη(1 − q).
The major reason for using OEE is to measure the technical effectiveness of a
manufacturing machine. The manufacturing process is often improved through finding
solutions to the problems causing losses. Applying these solutions is usually associated
with a new measure of OEE. Positive results lead to a new trial for reducing losses. But
reasons behind negative results cannot be traced via the OEE, a particular investigation
is usually required: i.e. failures lead to failure which causes investigation. Sometimes the
improvements in OEE become expensive, especially when it concerns highly automated
production lines, which may not be justified by the gained benefits.
A suitable scale for measuring equipment condition and accurate equipment operation
record is of great importance to achieve high equipment technical effectiveness. Better
understanding and a more accurate evaluation of the machine condition can be achieved
through collecting and analysing data concerning all the essential elements influencing
a manufacturing process, such as maintenance, operation/production, quality control,
environmental conditions (Al-Najjar, 1996, 1997).
Below, five measures are presented for detecting changes in maintenance performance:
specifically, losses in time and speed of production, product quality rate, machine
effectiveness and maintenance cost.

5.1 Overall process effectiveness


OPE is a modified version of OEE. It is defined as: a measure of process technical
effectiveness revealing the contribution of the basic process elements to the process
total effectiveness, e.g. the effect of environmental conditions on machinery availability,
performance of manufacturing procedures or on product quality (Al-Najjar, 1996, 1997).
I. Availability, A:
A = 1 − [equipment downtime/loading time]
= 1 − [(number of stoppages × average stoppage time) / loading time]
= 1 − [(Ns ta )/tl ] (1)
where Ns , ta , tl are number of stoppages, average stoppage time and loading time,
respectively. Loading time is the time the machine is scheduled for production. Equipment
downtime includes stoppage losses due to failures or replacements just before failures, set-
up procedures, etc. The reason behind the variation in A, i.e. the changes in Ns and/or
ta , can be identified and traced when using (1). If more accurate indications are required
to determine the actual time-share of each stoppage type experienced by the system, Ns
should be divided into failures, planned replacements, etc. ta should then be replaced by
the time of each stoppage or the average stoppage time for each type of stoppage. It is not
practical to perform this task manually, but using software, this can be handled rapidly.
II. Proportion of the accepted produced items, i.e. quality products, is usually influenced
by
260 B . AL - NAJJAR ET AL.

1. number of rejected items due to technical-based stoppages (n f ), because this results


in extra losses due to low quality production, which is associated with failures
(Ahlmann, 1994);
2. number of rejected items due to common causes (n c ) that arise due to the variation
within each element of the manufacturing process, e.g. raw material, machines
performance, operating conditions, operators and tools, which are in general not
easily detectable;
3. number of rejected items due to special causes (n s ). These causes arise due to
some detectable reasons such as faults in the equipment, a new batch of material,
insufficiently-trained operator, or dramatic change in the operating or environmental
conditions.
So,
quality rate(q) = 1 − [(defect items) / total processed items]
= 1 − [(n f + n c + n s )/n] (2)
where n is the total manufactured number of items. Inability of the machine to
manufacture the specified amount of quality products is considered as a machine
failure.
III. Performance efficiency (η) is often determined by the theoretical cycle time, processed
amount and the total operating time, i.e.
η = (theoretical cycle time × number produced)/ operating time (Nakajima, 1988).
But it is expressed as a function of the reasons behind the reduction in the performance
efficiency such as minor stoppages, time losses due to the reduced speed, tr s , set-up and
small adjustments which may arise due to, for example, the quality deviation arising from
common causes, i.e.
η = 1 − [(n m tam + tr s )/to ] (3)
where n m , tam , to are the number of minor stoppages, their average stoppage time and
the operating time, respectively. To monitor further details the above variables should be
analysed once more, e.g. classifying the minor stoppages, the average time of each class.
Thus, OPE can be written as a multiplication of (1), (2) and (3):
OPE = {1 − [(Ns ta )/tl ]} × {1 − [(n f + n c + n s )/n]} × {1 − [(n m tam + tr s )/to ]}. (4)

5.2 Maintenance related costs per unit time


The maintenance-related cost consists of several factors as stated by Al-Najjar (2000).
These factors can be used to assess and monitor maintenance cost and its behaviour.
Maintenance cost can be used together with production cost and lack of quality cost to
highlight the impact of more efficient maintenance on the manufacturing process economic
effectiveness. The measure that is suggested for this task is the total maintenance cost per
unit time, i.e.
total maintenance-related cost
. (5)
running time
BENCHMARKING OF MAINTENANCE PERFORMANCE 261

This is to achieve a measure of maintenance cost that can be used independent of the
machine operating time or running time (if they are different), because longer (or faster)
use of a manufacturing machine usually increases maintenance cost. Such a measure is
necessary when comparing machines of different production speed and operating/running
times.

5.3 Maintenance impact on the accepted manufactured products (MIAMP)


This variable is expressed as

total maintenance-related cost


. (6)
total accepted products

This measure is used to display the technical and economic impact of a more efficient
maintenance policy on manufacturing process effectiveness. It reveals how better
maintenance efficiency, which is measured by the cost paid for maintaining the process
capability and effectiveness, may influence the quality rate. Maintenance cost is considered
to include

1. labour, including expenses for repair, adjustment, service such as machine and
surrounding cleaning, and greasing;
2. overhead cost for maintenance department, e.g. building, instrumentation, adminis-
tration;
3. spare parts expenses (including tied up capital expenses), e.g. rolling element
bearing, oil;
4. expenses of investments in more redundancies in equipment and personnel to avoid
longer waiting times;
5. production and quality losses arise due to failures, malfunctions and unavailability;
6. extra insurance premium according to failure-based accidents;
7. expenses (penalty) due to damage done to the environment (air, earth and water);
8. delivery delay expenses, due to failures and malfunctions;
9. loss of market share.

5.4 Impact of better process effectiveness on maintenance cost (IPEMC)


This measure is defined as
OPE
. (7)
total maintenance cost
When monitoring the behaviour of this measure, it is possible to trace how improvements in
maintenance influenced OPE and how improvements in OPE (due to changes in production
and quality systems) influenced maintenance cost. Changes in this measure should be
monitored and analysed in order to detect the reasons behind its behaviour. Limits to
control the increment in OPE are necessary to justify the cost incurred by technical
improvements.
262 B . AL - NAJJAR ET AL.

5.5 Maintenance cost related to operating cost (MCROC)


In many cases, the increase in maintenance cost may occur due to, for example, increase
of production or rate of production. Then, the measure

maintenance cost
(8)
operating cost

can be used to display, in conjunction with the measure in Section 5.2, how
increase/decrease in the operating cost influences maintenance cost, and how more efficient
maintenance may reduce operating cost.

6. Monitoring maintenance performance measures


When the values of relevant variables and measures are calculated (step seven, Fig. 1),
their behaviour during running time can be monitored, analysed and judged according to
step eight to detect changes at an early stage. To ease the task of monitoring maintenance
performance measures, a decision rule tree for identifying the reasons behind changes is
constructed, see Fig. 2. It is specially developed for the measures used in this study and
according to the following criteria:

1. logical sequence between steps;


2. low level details;
3. ability to illustrate changes easily.

The decision rule tree is divided into two parts. In part one, observe that
improvements/degradation in OPE, which is achieved due to improvements/malfunction
in production and quality respectively, may influence maintenance cost per unit time.
Increments (reductions) in the maintenance cost per unit time also arise due to increased
(reduced) production rate, production time or exposing machinery to harsh (better)
operating environment conditions. But operating cost often increases due to longer
production time and more consumption of raw material. Thus, a comparison of changes
in maintenance cost with those in OPE and MCROC is necessary to identify the basic
reasons behind maintenance cost behaviour. Trends of maintenance and operating costs (or
OPE) would reveal if there are interactions between changes in these measures.
If it is not possible to find interactions between operating cost (or OPE) and
maintenance cost, measurements of OPE and MIAMP should be analysed to highlight
whether MIAMP is increased or decreased and the reason behind that: whether it was
higher/lower maintenance cost or lower/higher quality rate, respectively. In many cases
improved OPE yields higher maintenance cost especially when it concerns automations and
continuous controlling production. Following the behaviour of maintenance performance
measures in this systematic way opens reliable opportunities for identifying the basic
causes behind changes.
In part two of the tree, IPEMC displays the effect of technical related changes such as
improvements in the process effectiveness on maintenance cost per unit time. For example,
more automated production may result in higher maintenance costs. It is a matter of
economic justification and the company should decide how high the maintenance cost per
BENCHMARKING OF MAINTENANCE PERFORMANCE 263

Monitored changes in the measures

Maintenance cost IPEMC


Part 1 Part 2

MCROC OPE MCROC

MIAMP
OPE MIAMP

F IG . 2. Decision rule tree for analysing maintenance performance measures.

unit time (or per quality product) should be for the process to be still considered as cost-
effective. It is always possible to improve the manufacturing process technically, but it is
not always the case that it is also cost-effective.

7. Case studies
Two Swedish companies are considered in this study. They are AB Recta and Swedwood
AB.
AB Recta has been certified according to ISO 9000 since 1992 and has about 180
full-time employees. It manufactures bookshelves and coffee (sofa) tables and its yearly
turnover is about 230 million SEK. About 99% of its production delivers to IKEA and the
rest to another customers, i.e. the market buys all AB Recta’s production. The department
for surface treating (with ultraviolet) is selected in AB Recta for further study. This
department works in shifts, about 111 hours per week. Surface treatment is accomplished
by using several operations, which are done by rolling, polishing, staining, drying and
exposing to ultraviolet rays. In order to increase the production, the machine speed was
increased from 17 to 24 m/min. This increment also means some increment in the rejected
products because material feed and product collection is done manually. According to the
personnel, the biggest problem is related to rolling and polishing operations. Maintenance
is mostly performed by the operator, except for large repairs and purchasing of spare parts,
which are handled by a repair team. Lack of information and documentation regarding
repair, waiting and stoppage times was noticed at AB Recta. Maintenance actions were
usually found to be about 90% based on break-down maintenance (BDM) and 10% on
preventive maintenance (PM) policy. The company struggles to reduce production cost,
increase product quality and improve the share of PM.
Swedwood AB is a part of Swedwood International AB, which is a daughter company
of IKEA. It manufactures different products, e.g. frontispieces to kitchens and bathrooms,
and about 95% of the production sales are to IKEA. Its annual turnover is about 135 million
SEK. The company has 173 employees and it has been certified according to ISO 9000
since 1993. In this company, the surface treatment department, which is the bottleneck
in the production line similar to that at AB Recta, is also selected for further analysis.
The production runs in three shifts from Sunday to Friday, i.e. about 111 hours weekly,
264 B . AL - NAJJAR ET AL.

excluding overtime. The most troublesome areas are the turn-over operation and sandpaper
change in the polisher. Manufacturing speed is about 8 m min−1 . The main problem
confronting the company is the maintenance cost. This is because BDM takes a long time
and demands more resources compared with PM. BDM represents about 90% of the total
maintenance effort; the rest is outsourced. As at AB Recta, there is a lack of relevant data
and documentation regarding maintenance. The repair team is also responsible for large
repairs and for the purchasing of spare parts. Outsourced maintenance including spare
parts costs about 4·4 million SEK annually.

7.1 Data gathering


The data presented in this section are partly copied from the companies’ databases and
partly gathered by the authors. The data collected by the authors covered a period of six
weeks. The general observations when data were collected are:

• There was no register documenting the scrapping and reworking of bad quality
products. Therefore, it will not be possible to assess OPE accurately.
• Planned and unplanned stoppage times are mixed in both companies.
• No data are found concerning theoretical and practical cycle time and manufacturing
speed. Therefore, we used four assumed values to denote machine performance
efficiency (η), which are 0·4, 0·6, 0·8, and 1·0. This is in order to explore how this
variable may influence OPE at the two companies, as shown in Fig. 3.
• Availability: the companies measure availability in different ways. Reworking time of
the rejected items, i.e. those which were rejected due to lack of quality, is included into
Recta’s stoppage time, but not at Swedwood. At the latter company, the rejected items
were included in product quality rate.
• Losses at AB Recta were calculated to be about 2643 SEK per one hour stoppage
of production. This cost includes man-hours, expenses for invested capital, insurance,
delivery delay expenses, interest rate losses and bad quality cost.
• Losses at Swedwood AB, which included parts of maintenance cost, were calculated
to be about 3300 SEK per one hour production stoppage.
• There were no special investments in the maintenance in these two departments during
the period of study.
• Operating and maintenance costs are related to the surface treating departments at the
two companies.
• The cost factors included in maintenance cost were repair (man-hours), spare parts,
overhead cost, production losses and operator costs during stoppages.
• Operating cost includes man-hours, invested capital expenses, insurance, penalties
according to delivery delay and affecting the environment and the cost of bad quality
products, but material cost is excluded.
• The products from (and the manufacturing process at) the two companies’ departments
are comparable.
BENCHMARKING OF MAINTENANCE PERFORMANCE 265

TABLE 1 Variables to calculate (a) availability and (b) OPE


(a)
Variable AB Recta Swedwood AB
Stoppages (Ns )× average
stoppage time (ta ) 187·6 h 103·8 h
Loading time (tl ) 888 h 888 h
A = 1 − (Ns ts )/tl 0·7919 0·8831

(b)
Variable Recta AB Swedwood AB
Number of rejected items due to
technical-based stoppages (n f )+
number of rejected items due to
common causes (n c )+ number
of rejected items due to special
causes (n s ) 6550 11 747
Total processed items (n) 443 209 311 747
Quality rate
(q) = 1 − (n f + n c + n s )/n 0·985 0·9623
OPE = A ∗ q ∗ η 0·7919 ∗ 0, 985 ∗ ηRecta 0·8831 ∗ 0·9623 ∗ ηSwedwood
= 0·780ηRecta = 0·850ηSwedwood

7.2 Analysis

I. Past Data from AB Recta and Swedwood AB

The companies suffer from serious lack of efficient maintenance. Maintenance perfor-
mance measures are calculated, analysed and judged according to the seven steps of the
model shown in Fig. 1 and the decision rule tree shown in Fig. 2. Tables 1(a) and (b)
display the variables needed to assess availability and OPE respectively. Maintenance
performance measures’ values are assessed based on the data gathered from the surface
treating departments at the two companies.
From Table 1(b) we notice that the number of rejected items experienced at Swedwood
is approximately twice the number experienced at AB Recta although the total number
of manufactured items at the latter is much more than that at the former company. This
describes why the quality rate at AB Recta is greater than that at Swedwood AB. But, due
to the lower availability at AB Recta, its OPE is less than that at Swedwood AB given that
their performance efficiencies η are equal.
From Table 2(b), it can be seen that AB Recta pays more in maintenance for each
manufactured and accepted item (and for each unit in OPE) than Swedwood AB does.
This measure can be used to compare these two companies as long as the products from
(and the manufacturing process at) the two companies are comparable. Operation in AB
Recta costs more in maintenance than is the case in Swedwood.
266 B . AL - NAJJAR ET AL.

0,9
0,8
0,7
0,6
OPE

0,5 OPE hos Recta


0,4 OPE hos Swedwood
0,3
0,2
0,1
0
0,4 0,6 0,8 1
Machine performance efficiency

F IG . 3. OPE at AB Recta and Swedwood AB with respect to the four assumed values of performance efficiency.

TABLE 2 Comparison of maintenance performance measures at AB Recta and


Swedwood AB

(a)
Maintenance performance measure AB Recta Swedwood AB
Maintenance cost during 6 weeks, Table 3 1066 654.60 342 101.01 SEK
Accepted products 436 659 300 000 items
OPE 0·780ηRecta 0·850ηSweedwood
Operating cost 13 464 984 6156 637 SEK

which yields

(b)
Maintenance AB Recta Comparison Swedwood AB
performance
measure
MIAMP 2.4428 SEK (MIAMP)Recta > 1.1403 SEK
(MIAMP)Swedwood
IPEMC 0.731×ηRecta × 10−6 (IPEMC)Recta < 2.484×ηSwedwood × 10−6
1/SEK (IPEMC)Swedwood 1/SEK
for equal η
MCROC 0.0792 (MCROC)Recta > 0.0557
(MCROC)Swedwood
Maintenance 177 775.77 (Maintenance 57 016.84 SEK/week
cost/week SEK/week cost/w.)Recta >
(Maintenance
cost/w.)Swedwood

To benchmark companies from the same branch, consideration of measures of


maintenance performance such as those described in Sections 5.2 and 5.3 is necessary.
This is in order to normalize the effect of some factors that may confuse the importance of
BENCHMARKING OF MAINTENANCE PERFORMANCE 267

Availability

0,8

0,6 Recta
0,4 Swedwood

0,2

0
9 10 11 12 13 14 15 16
Week

F IG . 4. Availability at AB Recta and Swedwood AB.

what a maintenance cost means, such as operating and running time (if these variables
are different) and production speed (rate of production), otherwise the results may be
misleading. For example, faster production and longer production time yields faster
deterioration of the machines, which means more maintenance and costs. But it is not
necessarily unjustified economically. This is why when maintenance cost increases it is
usually interpreted negatively. On the other hand, the increase will be accepted if it means
more quality production, i.e. no increase in the maintenance cost per quality product,
and this is what is reflected by the performance measures such as maintenance cost
per quality product. This is important to take into consideration because maintenance
performance measures should be used carefully when two companies are compared due
to the differences in the problems and norms (Karlöf, 1997).
Figure 4 shows availability for both companies. Because the measurements were few,
it was not easy to display a trend describing the behaviour of availability. Note that in the
data acquired from AB Recta, the reworking was included, which was not the case in the
data collected from Swedwood AB. This creates differences in availability calculation at
these companies.

II. Additional measurements from the case companies


Due to the lack of past data, two intervals of three weeks of data gathering at AB Recta were
selected, see Table 3. The results achieved through calculating maintenance performance
measures are analysed and judged using the seven steps of the model shown in Fig. 1 and
the decision rule tree of Fig. 2. The data which were collected during these two 3-week
intervals gave an opportunity to compare the values of maintenance cost per unit time,
OPE and IPEMC during these two intervals, see Table 3.
It would be more beneficial if the data that were collected for comparing maintenance
performance measures covered longer intervals, such as a year. In that case it would be
268 B . AL - NAJJAR ET AL.

TABLE 3 Changes in maintenance performance measure’s values at AB Recta


Maintenance 3 weeks, period 1 3 weeks, period 2 Difference
performance (weeks 9–12) (weeks 13–16)
measure
Maintenance cost 631 118·30 435 536·30 +195 582 SEK
Maintenance 210 732·77 145 178·77 −65 554 SEK/week
cost /week
Operating cost 6340 092 SEK 7124 892 SEK +784 800 SEK
OPE 0·7832 × 0·985 × ηRecta 1 0·8006 × 0·985 × ηRecta 2
+0·0174 × |ηRecta 1
= A(1 − q)η (ηRecta 1 is considered (ηRecta 1 is considered
−ηRecta 2 | (the
for period 1) for period 2) average of OPE for the
whole period = 0,7919)
IPEMC 1·2223 × 10−6 × ηRecta 1 1·8106 × 10−6 × ηRecta 2 −0·2618 1/SEK

possible to capture changes in the maintenance economic and technical effectiveness more
accurately, especially after investment (if any) in maintenance.
The cause of any increase in maintenance cost should be analysed and described. Part
1 of Fig. 2 is used to judge the case before using part 2 because it is easier to handle. First,
MCROC is analysed and judged because changes in maintenance cost per unit time can be
related to changes in production rate and production time.
From Table 3, it is obvious that operating cost increased from 6340 092 to 7124 892
SEK, while maintenance cost decreased from 631 118·30 to 435 536·30 SEK. The
production is increased due to the increase in availability, from 0·7832 to 0·8006, and
production speed, from 17 to 24 m min−1 . The performance efficiency is unknown and the
number of rejected products during the period of study is considered constant and equal to
1.5%. Notice that maintenance performance measures’ values at AB Recta are improved
although no changes were introduced at the company concerning production and quality
during the period of study.
Judging MIAMP is the last step in Fig. 2, part 1, which shows a reduction from
2·69 to 2·16. Such a reduction means smaller maintenance cost per accepted item. If the
observation time had been longer than six weeks, it might have been interpreted with high
certainty that it was the better maintenance performance which caused this reduction in
the maintenance cost per accepted product; otherwise it may be considered as a natural
variation in the process output. The value of IPEMC increased and this increase is due to
the increase in OPE and reduction in maintenance cost per unit time.
To summarize the situation, operating cost increased due to more production,
maintenance cost decreased, OPE and IPEMC increased when MIAMP decreased. This
means that when implementing an effective maintenance (here the period of study was too
short to be sure of this), higher availability and longer production time at a reduced cost
can be achieved.

7.3 Results and discussion


These measures are applicable for a company’s internal monitoring and comparison, and
for comparing the performance of similar activities in different companies. The behaviour
BENCHMARKING OF MAINTENANCE PERFORMANCE 269

of OPE may be misleading because the number of rejected products at AB Recta is


assumed constant during the measurement interval (three weeks). Further, it was not
possible to measure machine performance efficiency due to the lack of data and the
incomplete data acquisition and analysis system. In general, such a problem should have
the highest priority and should be treated before trying to assess maintenance performance.
Due to the lack of data it was not possible to determine the sensitivity of maintenance
performance measures. Short intervals for data gathering made the results function
more as indicators instead of as statistical evidence. This is why no significant test has
been performed. Better data coverage and quality would give more reliable analysis of
maintenance performance measure behaviour. The first six steps of the model shown in
Fig. 1 were used fully but steps seven and eight did not work properly due to the lack of
data. A feedback loop is necessary for the model to adjust itself continuously to each use
and environment and that can be achieved at step ten.

Conclusions
The measures identified for monitoring maintenance performance are informative and
helpful to follow changes in maintenance performance and to decide which action is
necessary if any deviation from the predetermined specifications is detected. The models
developed can be used for benchmarking with other companies of comparable products
and it will be beneficial for internal monitoring, comparison and following the trend
(behaviour) of maintenance performance measures. With better data coverage and quality,
it would be possible to get statistical evidence instead of the achieved indications. Using
the decision rule tree, results can be improved continuously based on better knowledge and
experience, which is usually provided by training.
Using the five maintenance performance measures described in Section 5 means fewer
variables, measurements and less cost, and increased consistency. Monitoring maintenance
cost per unit time (and per unit quality product) in conjunction with operating cost and OPE
would lead to an identification of maintenance profits and motivate increasing maintenance
cost, which is more or less impossible when monitoring maintenance costs separately.
Other reasons behind changes in the maintenance cost can be detected at an early stage
in order to be eliminated.

Recommendations
Based on the results achieved by this study, the following recommendations are suggested
for the companies:
1. It is important to establish a relevant data acquisition and analysis system, which
makes the assessment and monitoring of the behaviour of maintenance performance
measures easier. By means of a reliable data acquisition and analysis system it
would be easier to detect and eliminate deviations in maintenance performance,
operating and quality costs and technical effectiveness, and to motivate the required
investments.
2. The variables needed for calculating maintenance performance measures should be
identified and defined clearly in the data acquisition system to make the use of these
measures reliable for internal and external benchmarking.
270 B . AL - NAJJAR ET AL.

Acknowledgements
The authors are grateful to the staff at AB Recta and Swedwood AB for the help received
during gathering data, which is partly used for this paper and partly for a previous MSc
project report.

R EFERENCES
A HLMANN , H. (1994) Industriell Driftsäkerhet, Service och Underhåll (Availability, Performance,
Servicing and Maintenance in Industry). Department of Industrial Management, Lund Institute
of Technology.
A L -NAJJAR , B. (1991) On the selection of condition based maintenance for mechanical systems.
Operational Reliability and Systematic Maintenance. London: Elsevier, pp. 153–175.
A L -NAJJAR , B. (1996) Total Quality Maintenance: An approach for continuous reduction in costs
of quality products. J. Quality Mainten. Engng, 2, 2–20.
A L -NAJJAR , B. (1997) Condition-Based Maintenance: Selection and Improvement of a Cost-
Effective Vibration-Based Policy in Rolling Element Bearings. Doctoral Thesis, Lund
University, Institute of Industrial Engineering.
A L -NAJJAR , B. (2000) Impact of Integrated Vibration-Based Maintenance on Plant LCC: A Case
Study. 3rd Int. Conf. on Quality, Reliability and Maintenance (Oxford). (G. J. McNulty. ed.).
pp. 105–110.
D E J ONG , A. (1997) Maintenance Practices in Small to Medium Sized Enterprises (SMEs). National
Key Centre for Advanced Material Technology, Monash University, Australia.
D UNN , R. (1987) Advance maintenance technologies. Plant Engng, 40, 80–82.
DWIGHT , R. (1995) Concepts for measuring maintenance performance. New Development in
Maintenance. The Netherlands: Moret Ernst and Young Management Consultants, pp. 109–125.
EBSOM G ROUP (1993) Nyckeltal inom underhållet i Svensk tillverkningsindustri Föreningen
Underhållsteknik (UTEK), Franzén Kaj, Utek-nytt nr. 1.
E CCLES , R. G. (1995) The performance measurement manifest. Performance Measurements and
Evaluation. (J. Holloway, J. Lewis & G. Mallory, eds). London: Sage, pp. 5–14.
J OHN , S. O. (1995) Total Quality Management. Oxford: Butterworth-Heinemann.
K ARL ÖF , B (1997) Benchmarking i Verkligheten (Benchmarking in Reality). Svenska förlaget liv &
ledarskap AB.
L ÖFSTEN , H. (1999) Management of industrial maintenance-economic evaluation of maintenance
policies. Int. J. Oper. Product. Manage., 19, 716–737.
NAKAJIMA , S. (1988) Introduction to Total Productive Maintenance, TPM. Cambridge: Productivity
Press.
T HOMAS , H. (1993) Underhållets ekonomiska betydelse. Lund: kortversion av UTC-projekt F9020.

View publication stats

You might also like