See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/228974772

Condition Based Maintenance Systems–An

investigation of technical constituents and
organizational aspects

Thesis · October 2004

CITATIONS READS

8 2,355

1 author:

Marcus Bengtsson
Mälardalens högskola i Eskilstuna och Västerås
41 PUBLICATIONS 215 CITATIONS

SEE PROFILE

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

Develop and implement CBM by vibration monitoring technology on machine tools' spindles at Volvo
Group Trucks Operations in Köping, Sweden View project

Reducing maintenance-related waste View project

All content following this page was uploaded by Marcus Bengtsson on 23 June 2014.

The user has requested enhancement of the downloaded file.

 Mälardalen University Licentiate Thesis
No. 36

CONDITION BASED MAINTENANCE SYSTEMS –

AN INVESTIGATION OF TECHNICAL CONSTITUENTS AND
ORGANIZATIONAL ASPECTS

Marcus Bengtsson

2004

Department of Innovation, Design, and Product Development

Mälardalen University

I
Copyright © Marcus Bengtsson, 2004
ISBN number: 91-88834-68-9
ISSN number: 1651-9256
Printed by Arkitektkopia, Västerås, Sweden
Distribution: Mälardalen University Press

II
“The major difference between a thing that might go
wrong and a thing that cannot possibly go wrong is
that when a thing that cannot possibly go wrong goes
wrong it usually turns out to be impossible to get at
or repair”

- Douglas Adams

III
IV
 Abstract

P R O D U C T I V I T Y I S A key strategy for manufacturing companies to stay competitive

in a continuously growing global market. Increased availability of production equipment is
crucial for competitiveness and, to achieve increased availability, a good maintenance
strategy is necessary. Good maintenance is achieved by having few corrective maintenance
tasks performed while the preventive maintenance is kept to a minimum. Condition based
maintenance is one possible way of achieving this.

Condition based maintenance and the use of condition based maintenance systems ensures
that the condition of an item is always monitored and alarm limits can be programmed or
specified to alert attention if the condition exceeds specified accepted levels. In-depth
analysis of the monitored data provides a diagnosis and prognosis of an incipient fault.
Everything is taken into consideration when scheduling predictive maintenance actions,
autonomously or in interaction with humans. Investigations indicate that there is a lot of
money to gain in increasing item availability, with the implementation of condition based
maintenance as one possibility. Other investigations have shown that condition based
maintenance is not widely utilized within the Swedish industry and that the monitoring that
takes place is mostly subjective monitoring performed by humans.

The objective of this thesis is to investigate technical constituents and organizational aspects
in a condition based maintenance system in order to make the condition based maintenance
strategy more accessible within Swedish industry. Within technical constituents, standards
and standardization proposals have been investigated as one possible constituent. Several
standards already published and a couple of standardization proposals have been found for
developers of condition based maintenance system technology to use, as well as for
customers to demand. These standards would increase a system’s technology
modularization and flexibility performance. Condition based maintenance system
technology, e.g. how to technically design a system, has also been highlighted. Seven
modules have been investigated in order to achieve a comprehensive system function.
Sensors, signal processing, condition monitoring, diagnosis, prognosis, decision support, and
presentation all need to exist in the form of hardware/software or as specifications for
humans to perform. Regarding the organizational aspects, investigations in implementing
aspects have been performed. Results from the organizational investigations show that it
would be feasible to design an implementation strategy for condition based maintenance, an
implementation strategy that would help companies to go from a strategy that is not
condition based to a strategy that is condition based. Several aspects that are to be
considered when deciding to implement condition based maintenance, such as support from
top management, training of employees, increased cooperation between departments, pilot
projects on single systems, etc., are also highlighted.

V
VI
 Acknowledgement

F I R S T L Y , I W O U L D like to send my greatest appreciations and thanks to my

supervisor Mats Jackson and co-supervisor Peter Funk. Without their help, support, fruitful
discussions, and guidance this work would never have been a possibility. Also, my deepest
appreciation goes to Erland Olsson, my late supervisor, for believing in me and trusted me
with this, my first job.

I would also like to acknowledge my co-workers at the Department of Innovation, Design

and Product Development at Mälardalen University. Special thanks go to my research
group, Product- and Process Development, with the senior staff of Mats Jackson, Mats
Deleryd, Rolf Lövgren, and Sabah Audo. Carina Andersson, Anette Brannemo, Anna
Andersson, Milun Milic, Rolf Olsson, and Sofi Elfving in particular, all PhD-student at the
department who at different times has treated me with interesting discussions in and around
research and life in general. Hopefully, our cooperation has just begun and will last for many
years to come. Also, thanks to the library at Mälardalen University, the staff and Juliana
Cucu in particular, you have all been an invaluable resource in my work.

Göran Wikinson, Ingemar Andreasson, together with the staff at the Swedish Maintenance
Society (UTEK), for their support and help. The engineers at Bombardier Transportation and
Tomas Robertsson in particular, thanks for inspiring discussions, your enthusiasm in setting
up and tutoring interesting final thesis, and not least your financial contributions. Anders
Nilsson, and the co-workers at ABB, for interesting discussions and enthusiasm in setting up
and tutoring final thesis. Also thanks to all other experts from the Swedish and International
industry that has taken an interest in my work during the years. EU goal2 Regional
Structural Fund, Länsstyrelsen in Västmanland, and the Knowledge foundation through
research project 2002/0232, without their financial support this project would not have been
a possibility.

Finally, I would like to send my appreciations and thanks to my family and friends, you
have all taken an interest in my work but at the same time made me think of other than just
maintenance and school work.

Eskilstuna, September 2004

Marcus Bengtsson

VII
VIII
 Publications

T H I S T H E S I S I S based on the following papers.

Paper I
Bengtsson, M. (2003). Standardization Issues in Condition Based Maintenance. The 16th
Conference of Condition Monitoring and Diagnostic Engineering Management, Sweden,
Växjö: Växjö University Press.
Paper II
Bengtsson, M. (2004b). Condition Based Maintenance Systems Technology – Where is Development
Heading? The 17th Conference of Euromaintenance, Spain, Barcelona: Puntex - Publicaciones.
Paper III
Bengtsson, M., Olsson, E., Funk, P., and Jackson, M. (2004). Technical Design of Condition Based
Maintenance Systems – A Case Study using Sound Analysis and Case-Based Reasoning. The 8th
Conference of Maintenance and Reliability, USA, Knoxville.
Paper IV
Bengtsson, M. and Jackson, M. (2004). Important Aspects to take into Consideration when
Deciding to Implement Condition Based Maintenance. The 17th Conference of Condition
Monitoring and Diagnostic Management, England, Cambridge: The Central Printing
Services, University of Birmingham, UK.

IX
X
 List of Figures

FIGURE 1 Graph of the different studies performed during the research process and roughly
the amount (in percentage) of time put into the different studies. The papers are also
marked in time with roman numerals when they where published.................................. 10
FIGURE 2 Different maintenance types as depicted in SS-EN 13306 (2001).............................. 16
FIGURE 3 The functional layer of OSA-CBM as presented by Thurston (2001). ...................... 21
FIGURE 4 Another look at the OSA-CBM with slightly different module names (Lebold et.al,
2003)............................................................................................................................................. 21
FIGURE 5 The overall architecture of the prognostic system as presented by Vachtsevanos
and Wang (2001). ....................................................................................................................... 21
FIGURE 6 The basic elements of diagnostics and prognostics for condition based
maintenance as presented by Chinnam and Baruah (2004)................................................. 22
FIGURE 7 Model presented by Utterback (1996)........................................................................... 25
FIGURE 8 The case development system structure with its input and output signals,
compare to the OSA-CBM system structure in figure 4....................................................... 34
FIGURE 9 The technical constituents and organizational aspects that have been investigated
in this thesis. ............................................................................................................................... 43

XI
XII
 List of Tables

TABLE 1 Outline of the papers........................................................................................................... 6

TABLE 2 The research methods and data collection tools used in the different papers and
studies. CBM is an abbreviation for condition based maintenance.................................... 10
TABLE 3 The papers’ contributions to answering the research questions, where x indicates
strong focus while o indicates that the research question has been discussed but is not
the main scope of the paper. .................................................................................................... 12

XIII
XIV
 Table of Contents

ABSTRACT ......................................................................................................................................................... V
ACKNOWLEDGEMENT............................................................................................................................... VII
PUBLICATIONS ............................................................................................................................................... IX
LIST OF FIGURES ............................................................................................................................................ XI
LIST OF TABLES ........................................................................................................................................... XIII
TABLE OF CONTENTS..................................................................................................................................XV

1 INTRODUCTION............................................................................................................................ 1
1.1 PROBLEM STATEMENT ......................................................................................................................... 2
1.2 OBJECTIVE ............................................................................................................................................ 3
1.3 RESEARCH QUESTIONS ........................................................................................................................ 3
1.4 DELIMITATIONS ................................................................................................................................... 4
1.5 OUTLINE OF THE THESIS ...................................................................................................................... 5
2 RESEARCH METHODOLOGY .................................................................................................... 7
2.1 SCIENTIFIC APPROACH........................................................................................................................ 7
2.2 RESEARCH METHODS .......................................................................................................................... 8
Case Studies......................................................................................................................................... 8
Comparative Studies............................................................................................................................ 8
Literature Reviews ............................................................................................................................... 9
Interviews ............................................................................................................................................ 9
Surveys ................................................................................................................................................ 9
2.3 RESEARCH PROCESS ............................................................................................................................ 9
2.4 REFLECTIONS ON THE METHODS ...................................................................................................... 12
3 THEORETICAL FRAMEWORK ................................................................................................. 15
3.1 MAINTENANCE TYPES ....................................................................................................................... 15
Corrective Maintenance..................................................................................................................... 16
Preventive Maintenance .................................................................................................................... 17
3.2 CONDITION BASED MAINTENANCE ................................................................................................. 17
Condition Based Maintenance Systems............................................................................................. 18
Condition Based Maintenance System Architectures........................................................................ 20

XV
 3.3 TECHNICAL STANDARDS ................................................................................................................... 22
Aims and Purposes of Standardization.............................................................................................. 23
Setting up Standards ......................................................................................................................... 24
3.4 IMPLEMENTATION STRATEGIES ......................................................................................................... 26
Organizational Change...................................................................................................................... 26
Implementing Maintenance............................................................................................................... 27
3.5 SUMMARY OF THE THEORETICAL FRAMEWORK ............................................................................... 29
4 THESIS CONTRIBUTION AND SUMMARY OF PAPERS.................................................. 31
4.1 PAPER I – STANDARDIZATION ISSUES IN CONDITION BASED MAINTENANCE................................ 31
4.2 PAPER II - CONDITION BASED MAINTENANCE SYSTEM TECHNOLOGY .......................................... 32
4.3 PAPER III - TECHNICAL DESIGN OF CONDITION BASED MAINTENANCE SYSTEMS......................... 33
4.4 PAPER IV – IMPORTANT ASPECTS TO TAKE INTO CONSIDERATION WHEN IMPLEMENTING
CONDITION BASED MAINTENANCE ................................................................................................. 34
4.5 SUMMARY OF THESIS CONTRIBUTIONS ............................................................................................. 36
5 CONCLUSION AND FUTURE RESEARCH............................................................................ 39
5.1 CONCLUSIONS ................................................................................................................................... 39
RQ1 – The Standardization Issue...................................................................................................... 39
RQ2 – The Condition Based Maintenance System Design Issue ...................................................... 40
RQ3 – The Implementation Issue ...................................................................................................... 41
Summary of Thesis Conclusions........................................................................................................ 41
5.2 ACADEMIC AND INDUSTRIAL RELEVANCE ....................................................................................... 42
5.3 FUTURE RESEARCH ............................................................................................................................ 44
REFERENCES .................................................................................................................................................... 45
APPENDED PAPERS ....................................................................................................................................... 51

XVI
 1 Introduction
This chapter covers the underlying
background and problem area of the
research project. It also discusses the
research objective and formulates it into
research questions with delimitations.
Finally, the thesis outline is specified.

P R O D U C T I V I T Y I S A key strategy for manufacturing companies to stay competitive

in a continuously growing global market. Increased productivity can be achieved through
increased availability. Maintenance of production assets is a prerequisite in order to uphold a
high availability. Maintenance costs have increased rapidly during the past years and the
United States alone had an estimated maintenance cost of 200 billion dollars in 1979
(Wireman, 1990) with expected increases of 10-15 % per year to follow. It was also reported
that as much as one-third of the total maintenance cost was spent unnecessarily due to
certain circumstances. These circumstances are e.g. bad planning, overtime costs, poor usage
of work order systems, bad use of preventive maintenance, etc.

Good maintenance has been defined as when very few corrective maintenance actions are
undertaken and when as little preventive maintenance as possible is performed (Cooke and
Paulsen, 1997). This, of course, put a great deal of stress on planning proper preventive
maintenance intervals and preventive maintenance tasks, i.e. scheduling. When as few
corrective maintenance events as possible should take place, it can be seen as a good thing to
perform as much preventive maintenance as possible. That would of course lead to
decreased availability and uncontrolled direct maintenance costs, for e.g. labor and spare
parts. The preventive maintenance should, for the most effective execution, be planned for
when an item’s pre-set normal condition is exceeded. In some cases a machine can actually
be run until just before failure (Al-Najjar, 1997).

Condition based maintenance is a preventive maintenance type that utilizes condition

monitoring technologies in order to determine the current condition of an item, such as a
machine, and thereby plan the maintenance schedule. The current condition and thus
condition monitoring can be decided through highly complex equipment with objective
measurements, such as vibration, temperature, shock pulse measurements, etc., or by
humans subjective senses, such as sight, smell, hearing, and touch (Johansson, 1993).

The need of condition based maintenance with the use of proper condition monitoring
technologies was revealed as early as in the 1960’s through a study performed during the
development of the preventive maintenance program for the Boeing 747. The study’s
purpose was to determine the failure characteristics of aircraft components (Overman, 2004).

1
The study was, at the request of the Department of Defense (USA), documented and
published by Nolan and Heap in 1978. It was found that a relatively small part of all
components (11%) had clear ageing characteristics which would enable a scheduled
overhaul, i.e. predetermined preventive maintenance. The rest of the components (89%) did
not show such ageing characteristics which leads to the conclusion that they were random
failures which are thus not applicable to scheduled overhauls (Nolan and Heap, 1978). Nolan
and Heap (1978) also conclude that the failure rate is not a very important characteristic
within maintenance programs as, although a good figure when planning maintenance
intervals it tells nothing of “…what tasks are appropriate or the consequences that dictate
their objective” (p. 48). Page (2002) presents similar conditional-probability curves within the
manufacturing industry, stating that only 30% of all components have clear ageing
characteristics and that this percentage decreases as complexity and technology increases.
Evidently, when planning appropriate maintenance schedules, the ageing feature of a
component is not the best approach, and in some applications not even possible, thereby
introducing condition based maintenance and condition monitoring as one solution to the
issue.

Condition based maintenance and the use of condition based maintenance systems in
industry have been reported to be one way of decreasing maintenance budgets. Less
corrective maintenance actions (decreases spare parts and labor costs) and better planned
preventive maintenance (increases availability of items and assets) will lead to a decreased
maintenance budget. More research on maintenance costs, budgets, and potential savings
has been performed by Ahlmann (2002), Alsyouf (2004), Fararooy and Allan (1995), and
Wireman (1990).

1.1 Problem Statement

A survey1 conducted by (Bengtsson, 2004a), indicates that the use of condition based
maintenance in Swedish industry is quite low. The respondents of the questionnaire mostly
came from larger corporations and manufacturers in Sweden within the process, energy,
engineering, and provisions/pharmaceutical industries. The 28 respondents gave the mean
value of the utilization grade of condition based maintenance over predetermined preventive
maintenance to 22%. The mean value of the utilization grade of preventive maintenance in
contrast to corrective maintenance was only 44%, i.e. the utilization grade of condition based
maintenance was at this time and for these companies’ assets only 10%. The mean value of
the utilization grade of condition monitoring equipment far exceeds the utilization grade of
condition based maintenance, with a mean value of 2.43 on a scale from 1-5, where 1 is equal
to “not at all” and 5 is equal to “on all assets”. The mean value of the automation level of
condition monitoring used was 1.32, where 1 is equal to “manual inspection” and 5 is equal
to “completely automated”. This alone indicates that Swedish companies are using condition
monitoring equipment more as tool for inspecting quality rather than a tool to help schedule
maintenance activities. The companies seem to be in need of more general research
concerning condition based maintenance and how to use the maintenance technology

1The survey was conducted at a maintenance fair and during a condition based maintenance seminar
in Sweden, 2004.

2
practically, organizationally, and technically. Alsyouf (2004) reports on another study within
the Swedish industry indicating that the maintenance approaches used within the Swedish
industry were ranked as: preventive maintenance (use-based or time-based) as the most
frequently used, condition based maintenance and failure-based maintenance (corrective
maintenance) were second on equal scores, total productive maintenance was third, and
reliability-centered maintenance was fourth. The question in relation to which condition
monitoring techniques that were used coincides quite well with the questionnaire performed
by (Bengtsson, 2004a), indicating that the most common technique is manual visual
inspection.

When turning to literature and, in particular, journal and conference proceedings within the
topic of condition based maintenance, not a lot of comprehensive research within the area
has been published. Most of the published literature focuses on technical aspects of condition
based maintenance and goes into detailed specifications of a small part of a condition based
maintenance system. Very few studies have been published within organizational aspects
and comprehensive system theories. Carnero (2003) goes as far as pointing out that there is
no international standard on managing a predictive maintenance program, i.e., more
research concerning how to implement and run a comprehensive condition based
maintenance system would possibly be of good help to the Swedish industry.

1.2 Objective
The problem statement indicates that there is a need for more research in both
comprehensive system technologies as well as on a more organizational basis within
condition based maintenance. Therefore, the objective of this thesis can be formulated as: to
investigate technical constituents and organizational aspects in a condition based maintenance
system2 in order to make the condition based maintenance strategy more accessible within the Swedish
industry.

1.3 Research Questions

In order to fulfill the objective, three research questions have been formulated. The research
questions were formulated through literature studies and through discussions with
professionals from the Swedish industry and academia. The research questions are
formulated in such a way that both hard and soft factors are considered.

RQ1) What standards and standardization proposals can be found within the
condition based maintenance system technology, and what effects might they have on
future research?

2 In this project a condition based maintenance system is defined as: “A system that uses condition based
maintenance to determine and schedule predictive maintenance actions autonomously or in interaction with
other systems or humans”, see section 3.2 for more information on the definition. The term will be used
in order to get a more holistic view of condition based maintenance.

3
In developing technical systems, standards and standardized means enhance the technology
in several ways, e.g. compatibility and modularization. Standards also, for most parts, mean
a better deal for customers. The question sets out to answer if there are any standards that
condition based maintenance system developers can apply when developing the technology
as well as what customers can demand of developers. The question also sets out to answer if
the standards are compatible.

RQ2) How can a condition based maintenance system be designed technically, what
methods and techniques can be used for a comprehensive system function, and where
can a human being enter such a system?

The complexity and automation level of a condition based maintenance system virtually has
no boundaries, other than financial, but for a lot of assets a high complexity and automation-
level of a condition based maintenance system is not economically justifiable today. Some of
the steps can be performed by humans instead. Is there a system structure that can work for
both complex automated condition based maintenance systems as well as for condition
based maintenance systems with human interaction? Also, is there a system structure that
can serve as a template for technical condition based maintenance system development and
implementation? This research question focuses on technical constituents of a condition
based maintenance system and human interaction.

RQ3) What aspects does a company need to consider when implementing a condition
based maintenance system?

Work methods in a maintenance organization can be deeply rooted in the ordinary day to
day work and changing those methods sometimes imposes a problem. Following an
accepted implementation strategy, the shift from one maintenance strategy to another, such
as condition based maintenance, can mean that the change is performed in a more controlled
manner. The question sets out to find if there are any major aspects a company has to
consider when deciding to implement condition based maintenance and if it would be
possible to construct a general implementation strategy for condition based maintenance.
This research question focuses on organizational aspects of a condition based maintenance
system.

1.4 Delimitations
Even though condition based maintenance could be used for virtually any process or
product, it will, in this project, only be treated in relation to physical assets such as motors,
machines, and pumps, etc., i.e., assets that can be found within the ordinary manufacturing
industry. Software, buildings, and services will be excluded.

Case studies that have been performed within the project have focused on companies in
Sweden and the Swedish manufacturing industry. Comparisons between Swedish and
international industry would have been an interesting scenario, but due to time and budget
constraints it has not been feasible at this stage in the project.

4
Condition based maintenance systems, and within the diagnosis, prognosis, and decision
support in particular, analysis techniques and methods such as mathematical modeling and
different artificial intelligence methods, etc., might possibly be needed. This research project
acknowledges this but does not have the objective of performing research within these
specified questions; the project has been formulated to approach the problem statement in a
more comprehensive manner.

1.5 Outline of the Thesis

Chapter 1 starts with the background of the project as well as the problem statement. The
research objective is then derived and the research questions are formulated. The chapter
also presents the delimitations that have been set for this thesis, and finally the outline of the
thesis is presented.

Chapter 2 presents the methodological approach to the research. It discusses some scientific
approaches and different research methods. Thereafter it explains, in terms of the theoretical
research methodology that has been presented, how this research has been conducted as well
as presenting a reflection upon the validity and reliability of the conducted research.

Chapter 3 presents the theoretical framework that the research has been based on. The
chapter presents and summarizes the relevant literature, namely for maintenance types and
strategies, condition based maintenance systems, technical standards, and implementation
strategies.

Chapter 4 presents the results and the contribution of the research. The contribution will
refer to the papers that have been published within the research project.

Chapter 5 presents the conclusions and the industrial and academic relevance of the
research, as well as giving suggestions for future research.

Section 6 is a list of all references used in the thesis.

Section 7 contains the appended papers this thesis is based on. The outline of the papers, i.e.
which papers that answer which research questions and what the main message for each
paper is, can be found in table 1.

5
TABLE 1 Outline of the papers
Research question
RQ.1 What standards and
standardization proposals can
be found within the condition
based maintenance system
technology, and what effects
might they have on future
research?
RQ.2 How can a condition
based maintenance system be
designed technically, what
methods and techniques can be
used for a comprehensive
system function, and where can
a human being enter such a
system?
RQ.3 What aspects does a
company need to consider
when implementing a condition
based maintenance system?
Paper
I. Standardization Issues in
Condition Based Maintenance.
II. Condition Based
Maintenance System
Technology – Where is
Development Heading?
III. Technical Design of
Condition Based Maintenance
Systems – A Case Study using
Sound Analysis and Case-
Based Reasoning.
IV. Important Aspects to take
into Considerations when
Deciding to Implement
Condition Based Maintenance.
6
Main message
To give an overview of the
standards and standardization
proposals that can be found
within condition based
maintenance systems.
Through standards and
standardization proposals try to
look ahead and see where
development and research
within condition based
maintenance systems might
have its focus in the future.
To describe and illustrate the
necessary technical
constituents of a condition
based maintenance system and
point out where the human can
interact.
To illustrate and highlight
important aspects a company
has to consider when deciding
to implement condition based
maintenance.
 2 Research Methodology
This chapter deals with the type of research methodology that
has been used within this research project. Section 2.1 gives
some theoretical views on scientific approaches, while section
2.2 discusses some of the most common research methods.
Section 2.3 presents the research approach and the research
methods that have been chosen for this research project, while
section 2.4 gives some reflection in the form of a discussion
regarding the research project’s validity and reliability.

T R A D I T I O N A L L Y , M A I N T E N A N C E R E S E A R C H has been viewed from an

operations research perspective, based in the mathematical theory of reliability. Renewal
theory, reliability tests, failure rate estimates, fatigue life in materials, etc., have been
calculated and researched (Jonsson, 1999); also see (Barlow, 1984). As other conclusions are
sought in this research project, a somewhat different method has been used.

2.1 Scientific Approach

At least three different approaches to how to perform research exist. These are the analytical
approach, the system approach, and the actor’s approach (Arbnor and Bjerke, 1994). The
analytical approach strives to explain reality as objectively as possible. The researcher seeks
explanations of effects by certain causes. The approach strives to find causes that are
independent; the classical laws of physics can be regarded as a model (Arbnor and Bjerke,
1994). The system approach also considers reality to be objective but somewhat differently
constructed as components are mutually dependent. The system approach strives for an
explanation (or understanding) of a situation by applying it into a comprehensive
perspective (Arbnor and Bjerke, 1994). Winroth (2004) states that an open system interacts
with the surrounding world and that the total solution is not built up mainly by
summarizing the sub-solutions, which is a more holistic approach. The actor’s approach,
regarding objectivity is quite different to the two other approaches. The actor’s approach
suggests that it is difficult not to influence the phenomenon being studied and that reality
exists as a social construction, not independent from humans (i.e. researchers), (Arbnor and
Bjerke, 1994).

The research conducted within this thesis lies within the art of engineering. Technical
constituents and organizational aspects in a condition based maintenance system are being
investigated in order to make the condition based maintenance technology more accessible
within the Swedish industry. The researcher has applied the system approach when
conducting the research; this as the system approach with its reality constructed as
components with mutual dependence is in fact how a condition based maintenance system is
treated in this project. Also, the research is performed in order to place a certain strategy

7
(condition based maintenance) into a comprehensive perspective - to make condition based
maintenance more accessible within the Swedish industry. The analytical approach would be
too lengthy for this objective. The analytical approach would possibly work well in order to
find different energies to measure when analyzing the condition of an item. The actor’s
approach would not be valid as too much focus is put on humans in a social context.

In order to perform research within the system approach and within condition based
maintenance systems, it is important that the researcher defines his or her views on what a
system is and what its purpose is. Banks et.al. (1999) has defined a system as “…a group of
objects that are joined together in some regular interaction or interdependence toward the
accomplishment of some purpose” (p. 8). Hubka (1982) states that the purpose of a technical
system is to “…transform certain well-defined input quantities, particularly materials (e.g.
auxiliary materials), energy, and information (e.g. commands), into desired effects (output
quantities) in space and time (e.g. position, movement, velocity, force).” (p. 12).

2.2 Research Methods

The choice of research methods has included both theoretical and empirical studies.
Theoretical studies have been performed in order to find out where the state of the art lies
within condition based maintenance system technology and as input to what empirics that
should be collected.

Case Studies
According to Yin (1994) “A case study is an empirical inquiry that investigates a
contemporary phenomenon within its real-life context, especially when the boundaries
between phenomenon and context are not clearly evident.” (p. 13). Yin (1994) also means that
case study research is preferred when ‘how’ and ‘why’ questions are being posed. One
prejudice to case study research is that it is impossible to get generalizable results. Yin does
state that the result can not be generalized to populations or the universe but that it can be
generalized to theoretical propositions. In this research project case studies have been used
to gather empirics when ‘how’ questions have been asked.

Comparative Studies
Comparative studies or just simply comparison is used as a research method within several
scientific fields (literature, anthropology, etc.), (Ejvegård, 1996). The method is not entirely
easy, even though the researcher is basically just comparing different events, products, or
such. Researchers might believe that they are comparing two objects that are of the same
kind, but in real life they are completely different (Ejvegård, 1996). In comparative studies
the researchers’ objects are cases which are similar in some respects but different in others,
and the goal is to find out why the cases are different (Routio, 2004). Two different styles are
available when performing a comparative study; the descriptive and the normative. The first
style aims at explaining the differences between the cases, and to describe and possibly also
explain them. The latter style has the same base as descriptive studies but also focuses on
improving the present state of the case being compared (Routio, 2004). In this research
project a comparative study has been performed in order to test the present theory on a
certain topic.

8
Literature Reviews
All research should be based on, or take into account, previous research that has been
undertaken within the same subject (Merriam, 1994). This means that it is an advantage for
research projects to start with a broad literature review that is narrowed down and more
specific at later stages. In this research project, literature studies have been performed
throughout the whole project. At first, on a comprehensive level within maintenance and
condition based maintenance in particular and, thereafter, narrowed down into the different
studies performed in order to answer the research questions.

Interviews
One of the most important data-collecting techniques within case study research is
interviews (Yin, 1994). Interviews are a good way of doing research when the researcher is
looking for answers about views, feelings, opinions, knowledge, etc., within a population
(Ejvegård, 1996). Interviews can be performed in many different ways, depending on what
knowledge or information that is being sought. In an open-ended interview, where the
researcher and respondent more or less are have an ordinary conversation, questions
regarding ‘what something is’ and ‘what meaning it has’ are being sought. The respondents’
perception and experience are being captured. The structured interview gives the
opportunity to answer questions regarding ‘how much’ of different phenomena. Thus, the
open-ended interview regards qualities and the structured interview regards quantities
(Lantz, 1993). In this research project, interviews have been performed to gain knowledge of
phenomena within organizational theories, and a mixture of questions, both open-ended and
structured, have been used.

Surveys
Besides interviews, also surveys and questionnaires are suitable when answers to questions
about views, feelings, opinions, knowledge, etc., are being sought (Ejvegård, 1996). The
result of a structured survey can be quantitative. In this research project, surveys have been
used in conjunction to interviews within phenomena of organizational theories and solely for
seeking a comprehensive view of the maintenance types used today in Swedish industry.

2.3 Research Process

According to Yin (1994), each type of empirical study has some sort of research design. Yin
also defines a research design as “an action plan for getting from here to there” (p. 19), with
“here” meaning from the initial sets of questions and “there” meaning some sets of
conclusions. This section will explain the research process, i.e. how the research has been
performed from the start to the final result of this licentiate thesis. Below, in figure 1, the
different studies performed during the project are illustrated in time as well as in size. Table
2 depicts the research methods, the unit of analysis, and the objective of the different papers.

9
FIGURE 1 Graph of the different studies performed during the research process and roughly the amount
(in percentage) of time put into the different studies. The papers are also marked in time with roman
numerals when they where published.

As can be seen in figure 1, literature studies have been performed throughout the project. In
the beginning, more general literature studies were performed as the research questions
were being formulated. As papers and reports were being written, more directed literature
studies were performed. The literature was collected through books (including licentiate and
doctoral dissertations), conference proceedings, journals, and in some cases the Internet.

TABLE 2 The research methods and data collection tools used in the different papers and studies. CBM is
an abbreviation for condition based maintenance.
Study Paper Research Unit of Objective
No. Method Analysis
CBM Paper I Review Literature, To give an overview of
Standardization Standard Documents existing condition based
Issue maintenance system
standards
CBM Paper II Review Literature, To look ahead and see
Standardization Standard Documents where the state of the art
Issue within condition based
maintenance system might
have its focus
CBM System Paper III Comparative An architectural To illustrate a condition
Design Issue Study standardization based maintenance
proposal and a case system’s constituents and
development of a to verify a standardization
condition monitoring proposal’s
system comprehensiveness
CBM Paper IV Case Study Respondents at an To illustrate aspects to
Implementation Interview/ International consider when
Issue Questionnaire/ Manufacturing implementing condition
literature Company based maintenance and as
a feasibility study to see if
it is possible to develop an
implementation model for
CBM

10
The first study regarded condition based maintenance system technology standardization
issues, but also technical standards and standardization in general. The study was performed
theoretically and the research method was comprised of literature reviews and standard
documents. To answer the second part of the research question, a literature search in the
different modules within condition based maintenance systems was performed. The
literature search was conducted in order to see if any trends in published papers, regarding
different modules, during the last ten years were present. The technical database IEEE
Xplore® (http://ieeexplore.ieee.org/Xplore/DynWel.jsp) was used in the literature search.
The standardization study led, as indicated in figure 1, to two published papers, Paper I and
Paper II. The papers were published in order to answer RQ1, but they also, indirectly,
discussed the condition based maintenance system design issue, as indicated in table 3.

The second study on the condition based maintenance system design issue was based on the
literature overview performed for the standardization study. In that, an open system
architectural standardization proposal for condition based maintenance systems was found.
The study started out theoretically by explaining and giving definitions or clear purposes of
the different modules within a condition based maintenance system. The architectural
standardization proposal was later compared in a comparative study with a development of
a system constructed within a near project. The comparative study was performed as a
descriptive study, where the similarities were discussed as the architectural standard
proposal’s comprehensiveness was tested. The second study was reported on in Paper III
and the paper also discusses the issue of standardization within condition based
maintenance systems, as indicated by table 3.

The third study, on condition based maintenance implementation issues, was based on a
literature study, in which it became clear that quite limited results had been published in
relation to organizational aspects within condition based maintenance. Empirics were
collected in a case study, conducted at an international manufacturing company operating in
Sweden, in the form of a combined interview/questionnaire study. The questionnaire was
built on theory from organizational change literature and literature from maintenance
implementation strategies. The sample of respondents was recommended by a manager at
the company, i.e., the sample was strategically but non-probabilistically chosen. The sample
size was set to be 20 respondents but due to different declines, lack of time, and in one case a
respondent not feeling well-acquainted with the topic, the sample size came to be 16
respondents. The questionnaires were sent to the respondents well in time before an
interview was booked. During all interviews the questionnaire was answered and the
respondent was given the opportunity to motivate their responses with qualitative
comments. At the end of the interview the respondent was also asked an open-ended
question regarding their beliefs on the biggest threats to why condition based maintenance
would not be implemented for their processes and products. The analysis was performed in
order to find important aspects that companies have to consider when implementing
condition based maintenance. The first five steps of the implementation model on total
productive maintenance, as explained in Nakajima (1988), were used as a frame of reference
for the analysis.

11
TABLE 3 The papers’ contributions to answering the research questions, where x indicates strong focus
while o indicates that the research question has been discussed but is not the main scope of the paper.
Paper I Paper II Paper III Paper IV
RQ1: What standards and standardization
proposals can be found within the condition
based maintenance system technology, x x o
and what effects might they have on future
research?

RQ2: How can a condition based maintenance

system be designed technically, what methods
and techniques can be used for a o o x
comprehensive system function, and where
can a human being enter such a system?

RQ3: What aspects do a company have to

consider when implementing a condition x
based maintenance system?

2.4 Reflections on the Methods

There are different ways of judging the quality of a research project. The most common way
is to discuss the validity and reliability of the research. Yin (1994) defines four tests of
validity and reliability: construct validity, internal validity, external validity, and reliability.

Construct validity is achieved by establishing correct operational measures for the concept
being studied. This test of validity is more problematic within case studies than in other
research, although equally important all round (Yin, 1994). There are three different tactics to
increase construct validity: to use multiple sources of evidence, to establish a chain of events,
and to have the draft of the reports studied by key informants (Yin, 1994). In this project
multiple sources of evidence have been used in one case study; interviews, questionnaires,
and literature was analyzed. In another project a comparative study was performed where
literature and a standardization proposal were compared to a real development in order to
see differences, but particularly similarities. Papers and other technical reports have been
presented at both international conferences and to the Swedish industry at fairs and
seminars.

Internal validity is achieved by establishing causal relationships in order to follow that

certain conditions lead to other conditions (Yin, 1994). It also deals with the question of if the
results agree with reality (Merriam, 1994). The test of internal validity has been given most
attention in experimental and quasi-experimental research (Yin, 1994). In this research
project, internal validity has not been treated to a great extent; causal relationships have thus
not been in focus.

External validity is achieved by establishing the domain from which the findings of a study
can be generalized to other populations (Yin, 1994). A theory springing from a case study

12
must be replicated with roughly the same results several times before a researcher can state
that the results are generalizable for other populations than the one being studied. In this
research project, external validity has, so far, been difficult to prove. The study has only been
of the duration of approximately two years; some of the studies even shorter. The
generalizations that possibly can be made are, so far, limited to the case study companies
and possibly also to companies within the same industry, i.e. the manufacturing industry.

Reliability is achieved by demonstrating that a study can be performed at a later occasion

with the same results (Yin, 1994), (Merriam, 1994). Yin (1994) states that the objective of
reliability is “…if a later investigator followed exactly the same procedures as described by
an earlier investigator and conducted the same case study all over again, the later
investigator should arrive at the same findings and conclusions.” (p.36). The most important
thing in achieving reliability is to document the procedures a case study has followed. In this
research, project reliability has been treated in that the data collection routines are
documented; interview questions, questionnaires, reviewed literature are all documented.
This makes replications of the studies possible. However, this does not mean that the results
will be the same in every single way. This is due to the constantly changing world we are
living in as well as the market situations for the companies that have been part of the studies.
It is also very difficult to stay completely objective, as a researcher and some personal views
are bound to affect the results in even the slightest way.

It is hard to believe that a research project can be completely perfect; there is always room for
improvements. For this research project it can be said that many of the questionnaires and
the interview studies could have been used at additional companies with more respondents.
However, for the objective of this project, it was believed that the number of companies and
respondents would be enough to be able to decide which the most important technical
constituents and organizational aspects were, and to decide which would be the most
interesting ones for the continuation of the project.

13
14
 3 Theoretical Framework
The theoretical framework gives a background to the issues
concerning different maintenance types and strategies in section
3.1. Section 3.2 discusses condition based maintenance and its
system technology in theory. Section 3.3 covers technical
standards and the purpose and aim of standardization and section
3.4 discusses different organizational change and implementation
strategies, while section 3.5 gives a brief summary of the theory.

T H I S C H A P T E R A I M S to describe and analyze the existing theories that can be found

within the research topic in order to place the research into a context as well as give the
reader an overview of related research. When performing research it is also necessary to
build one’s results on both theories and empirics in order to reach reliable conclusions.

3.1 Maintenance Types

Maintenance, as a process, is merely a name for all activities and resources necessary to
uphold items’ specified performance and condition within a given time frame. Various
definitions of maintenance have been suggested through the years, the common point being
that they have moved away from the traditional perception of maintenance, as explained by
Tsang et.al. (1999), to repair broken items. In this research project, the maintenance
terminology standard SS-EN 13306:2001, a European terminology standard approved by
CEN (Comité Européen de Normalisation, European Committee for Standardization), has
been used. The SS-EN 13306:2001 is partly based on the standard IEC 60050 (191). The
terminology standard SS-EN 13306 (2001) defines maintenance as:

“Combination of all technical, administrative, and managerial actions during the life
cycle of an item intended to retain it in, or restore it to, a state in which it can perform
the required function.”

Focusing on “…retain it in, or restore it to…” of the definition, it becomes clear that there are
two main types (or strategies) to perform maintenance, see figure 2. The first is a preventive
approach; to perform maintenance in order to prevent breakdowns or failures. The second is
a corrective approach; to perform maintenance after breakdowns or obvious
malfunction/failure. These can in turn be divided further, and the way of planning and
controlling the maintenance function can be performed in even more approaches. Below, a
brief discussion of these follows.

15
FIGURE 2 Different maintenance types as depicted in SS-EN 13306 (2001).

Corrective Maintenance
The corrective maintenance is thus an approach undertaken when a breakdown or an
obvious fault has been located. The terminology standard SS-EN 13306 (2001) defines
corrective maintenance as:

“Maintenance carried out after fault recognition and intended to put an item into a state
in which it can perform a required function.”

Corrective maintenance can for some items be deferred in time if the consequences of a fault
do not affect the comprehensive function. This deferred corrective maintenance can instead
be planned to be executed at a time more appropriate for the production capacity. Faults or
breakdowns of an item that have direct affect on the comprehensive system function or that
are hazardous to safety, the environment, or other specified maintenance rules, must without
delay be maintained. If applied, corrective maintenance is best suited to non-critical item
where: capital costs are small, consequences of failure are not critical, there is no safety risk,
the failure will be identified quickly, and where the repair work will be fast and easy (Starr,
1997).

Corrective maintenance can thus be looked upon as normal repair work or exchange of
components. For repair work, some modeling approaches are available. With a minimal
repair, the failed item is only restored back to the functioning state and the item continues as
if nothing has happened. The likelihood of a failure (failure rate) stays the same as it was
immediately before the failure. Using the minimal repair approach means that the item is
only restored to an “as bad as old” condition (Høyland and Rausand, 1994). Minimal repair
can be executed for various reasons (lack of time, spare parts, competence, etc.). If the item
instead is replaced by a new component of the same type, or if it is restored to an “as good as
new” condition, the failure rate will decrease to the level when the item was just put into use.
This is called a renewal process or sometimes a maximal repair. These types of repairs are,
according to Høyland and Rausand (1994), the extremes of repair work and, accordingly,
most repair actions are somewhere in between, often called imperfect repair.

16
Preventive Maintenance
The preventive maintenance type has been defined as (SS-EN 13306, 2001):

“Maintenance carried out at predetermined intervals or according to prescribed criteria

and intended to reduce the probability of failure or the degradation of the functioning of
an item.”

Focusing on “…predetermined intervals or according to prescribed criteria…” of the

definition, it becomes visible that preventive maintenance can be performed in two different
ways. The first, Predetermined maintenance has been defined as (SS-EN 13306, 2001):

“Preventive maintenance carried out in accordance with established intervals of time or

number of units of use but without previous condition investigation.”

The predetermined maintenance is thus scheduled and planned without any occurrence of
condition monitoring. The scheduled maintenance can be planned after number of hours in
use, number of times an item has been used, number of kilometers the item has been used,
according to certain dates, etc. Predetermined preventive maintenance is best suited to
failures that have a visible age or wear-out characteristic and where maintenance tasks can
be made at a time which for sure will prevent a failure from occurring (Starr, 1997).

The second category of preventive maintenance is condition based maintenance; this

maintenance type will be further discussed in chapter 3.2.

3.2 Condition Based Maintenance

Various researchers have defined condition based maintenance. Mitchell (1998) defines it as
“Maintenance actions based on actual condition (objective evidence of need) obtained from
in-situ, non-invasive tests, operating and condition measurement.”. Butcher (2000) defines
the maintenance technology as “CBM is a set of maintenance actions based on real-time or
near real-time assessment of equipment condition, which is obtained from embedded
sensors and/or external tests & measurements taken by portable equipment.” (p. 6). Moya
and Vera (2003) defines the purpose of a condition based maintenance program as:
“…improve system reliability and availability, product quality, security, best programming
of maintenance actions, reduction of direct maintenance costs, reduction of energy
consumption, facilitates certification, and ensures the verification of the requisites of the
standard ISO 9000” (p. 54). For this research project, the definition in SS-EN 13306 has been
applied. Condition based maintenance has been defined as (SS-EN 13306, 2001):

“Preventive maintenance based on performance and/or parameter monitoring and the

subsequent actions.”

The performance and parameter monitoring may be scheduled, on request, or continuous.

Condition based maintenance is thus a maintenance technology that utilizes condition
monitoring tools to analyze the current condition of the item and, through that knowledge,
set up proper preventive maintenance schedules. Since the maintenance intervals and tasks

17
are based on the condition of the item, they are predictive. Predictive maintenance has been
defined as (SS-EN 13306, 2001):

“Condition based maintenance carried out following a forecast derived from the analysis
and evaluation of significant parameters of the degradation of the item.”

Since both set-up costs and operation costs are available when introducing condition based
maintenance (Starr, 1997), it is important to structurally decide what components that can
benefit from a condition based strategy. Financially, some components benefit from a
corrective approach, others from a predetermined preventive approach. For items that are
capital-intense, safety-critical, production-critical, etc., a condition based approach would fit
in the best, if the failure modes that are diagnosed are detectable. Methods and techniques on
how to decide on the most appropriate maintenance approach and whether or not to utilize
condition based maintenance is discussed by (Al-Najjar and Alsyouf, 2003); (Starr, 1997).

Condition Based Maintenance Systems

In order to get a more holistic view of condition based maintenance as a maintenance type,
‘condition based maintenance system’ will be used as a term in this research project and it is
therefore included in the objective. This section will define condition based maintenance
systems with theory based on the SS-EN 13306 (2001) terminology standard and compared to
the results from other researchers.

In order to define condition based maintenance systems, other key terms need to be
understood and defined.

Condition based maintenance

If turning to the maintenance terminology SS-EN 13306 (2001), condition based
maintenance is defined as:

“Preventive maintenance based on performance and/or parameter monitoring and the

subsequent actions.”

Additionally, the definition has one notation saying that: Performance and parameter
monitoring may be scheduled, on request, or continuous. A keyword to put focus on and
explore in the definition is “monitoring”. Monitoring is according to (SS-EN 13306, 2001)
defined as:

“Activity, performed either manually or automatically, intended to observe the actual

state of an item.”

Three notations are attached to the definition: (1) Monitoring is distinguished from
inspection in that it is used to evaluate any changes in the parameters of the item with time;
(2) Monitoring may be continuous, over time intervals, or after a given numbers of
operations; (3) Monitoring is usually carried out in the operating state. A keyword to put
more focus on and explore in the definition is “item”. An item is according to (SS-EN 13306,
2001) defined as:

18
 “Any part, component, device, subsystem, functional unit, equipment or system that can
be individually considered.”

The definition has one notation: A number of items, e.g. a population of items, or a sample,
may itself be considered as an item. Since the maintenance intervals and tasks are based on
the condition of the item, they are predictive. Predictive maintenance is according to (SS-EN
13306, 2001) defined as:

“Condition based maintenance carried out following a forecast derived from the analysis
and evaluation of significant parameters of the degradation of the item.”

A keyword to put more focus on and explore in the definition is “degradation”. Degradation
is, according to (SS-EN 13306, 2001) defined as:

“Irreversible process in one or more characteristics of an item with either time, use or
external cause.”

The following two notations are attached to the definition; (1) Degradation may lead to
failure, (2) Degradation is often referred to as wear-out.

Here, the word “degradation”, as it is defined in the terminology standard, might actually be
too strong to use. Condition based and predictive maintenance should be performed in order
to find disturbances before they become irreversible, e.g. if an unbalance in a rotor is found
by vibration technology, it does not decidedly mean that the problem is irreversible and lead
to failure, but can be corrected by applying a counter balance. Therefore, in this research
project, the word “degradation” is replaced with the word “condition”. Thus redefining
predictive maintenance as:

“Condition based maintenance carried out following a forecast derived from the analysis
and evaluation of significant parameters of the condition of the item.”

Condition based maintenance system

In this research project a condition based maintenance system is defined as:

“A system that uses condition based maintenance to determine and schedule predictive
maintenance actions autonomously or in interaction with other systems or humans.”

Using Hubka’s (1982) purpose of a technical system, one can say that the purpose of a
condition based maintenance system is to transform certain well-defined input quantities in
the form of energy (e.g. vibration, temperature etc.) into desired effects (e.g. information of
the condition of the item, a prognosis on the future condition, etc.) in space and time. A
condition based maintenance system is thus a system that uses performance and/or
parameter monitoring techniques (e.g. vibration, thermal, visual, etc.) in order to find
disturbances in the performance of or changes in the characteristic parameters of an item. As
seen in the definition and notations of “monitoring”, the activity can be performed manually
or automatically, i.e. the definition includes every technique from [human] subjective
monitoring, to [handheld] condition monitoring equipment, and to [continuous]

19
autonomous condition monitoring equipment. It also points out that the monitoring activity
is usually performed while the item is in its operating state, which in many cases enhances
the ability to get a reliable condition assessment, while not decreasing the availability of the
item. The monitoring may also, according to the notations, be scheduled, on request, or
continuous.

Following the performance and/or parameter monitoring, the system should schedule the
subsequent actions. In this case, the subsequent actions would be predictive maintenance
actions, as the definition says: maintenance carried out following a forecast derived from
analysis and evaluation of significant parameters of the condition of the item.

The scheduling of the predictive maintenance actions can, as pointed out in the definition of
condition based maintenance system, be autonomous or in interaction with other systems or
humans. This means that advanced condition based maintenance systems should be able to
establish maintenance schedules without interference of humans. An advanced condition
based maintenance system might actually also be able to perform the maintenance actions
autonomously (e.g. for software). For most parts though, a condition based maintenance
system will give input to what current condition the item is in and give a forecast as to where
the condition will be at a future time. A human being will then, in interaction with
maintenance scheduling software (or such), decide the maintenance schedule. Thus, putting
the human as a part of the system, actually being able to perform all tasks within the
definition of a condition based maintenance system.

Condition Based Maintenance System Architectures

Condition based maintenance systems have been discussed by various researchers. Thurston
(2001) presents the work of the organization Open System Architecture for Condition Based
Maintenance (OSA-CBM) in which a proposal for open system architecture solutions for
condition based maintenance is discussed. The architecture was developed intended to
address the need for a standard to handle the flow of information between software
components in such a system. No definition of the term ‘condition based maintenance
systems’ is presented but the functional capabilities of such a system are discussed and
penetrated. Seven different layers, all representative of the functional capabilities are
specified: (1) Sensor Module, (2) Signal Processing, (3) Condition Monitor, (4) Health
Assessment (diagnosis), (5) Prognostics, (6) Decision Support, and (7) Presentation. For more
information on each layer turn to (Thurston, 2001); (Bengtsson, 2003); (Bengtsson, 2004b);
(Bengtsson et.al., 2004). The sensor module feeds the system with signals which in the signal
processor is processed into features, the condition monitor detects abnormal features which
are classified in the health assessment module, the prognostic module then predicts the
remaining useful life; all is taken into consideration when deciding and scheduling
maintenance actions in the decision support module, see figures 3 and 4.

20
FIGURE 3 The functional layer of OSA-CBM as presented by Thurston (2001).

FIGURE 4 Another look at the OSA-CBM with slightly different module names (Lebold et.al, 2003).

Another, yet similar, architecture for condition based maintenance, in which focus has been
put on diagnostics and prognostics, is presented by Vachtsevanos and Wang (2001). The
system is called a prognostic system, with the output of time-to-failure in order to perform
condition based maintenance actions, see figure 5. The diagnostician assesses the real
condition of an item through on-line sensor measurement. Its purpose is to decide about the
existence of impending or incipient failure conditions. The prognosticator, whose input
comes from the diagnostician, decides the need to maintain the item on the basis of historical
failure rate data, appropriate fault models, and a maintenance scheduler.

FIGURE 5 The overall architecture of the prognostic system as presented by Vachtsevanos and Wang
(2001).

21
A third architecture for condition based maintenance systems is presented by Chinnam and
Baruah (2004). The focus in this system is also on diagnostics and prognostics but, for their
overall presentation of what condition based maintenance typically involves, they explain it
accordingly. Non-intrusive sensors are mounted on the monitored components in order to
capture degradation signals that by subsequent interpretations can lead to a development of
a customized maintenance policy. This, in conjunction with advances in sensor technology,
data acquisition hardware and signal processing algorithms, reductions in cost for
computing and networking, and the increased ease for information technology products,
makes diagnostics and prognostics more effective as well as cheaper. See figure 6 for the
basic elements in a condition based maintenance system as explained by Chinnam and
Baruah (2004).

FIGURE 6 The basic elements of diagnostics and prognostics for condition based maintenance as
presented by Chinnam and Baruah (2004).

3.3 Technical Standards

Standards and standardizing can be looked upon as a tool for a rational, environmental, and
energy saving production, distribution and utilization of commodities, processes, systems,
and services. It can also be looked upon as a means to enhance technical cooperation and
technology transference. Standardization work demands involvement from affected parties
in society, such as the users, manufacturers, public authorities, and other interest groups
(Engström, 1995). Standards have been defined as (Granö, 1998):

“A document that has been established by consensus and approved by a recognized body
and states rules, guidelines and characteristics for activities or their results for general
and recurring use, with a view to achieving an optimum degree of order in a given
context.”

22
Aims and Purposes of Standardization
According to (Engström, 1995) the purpose of standardization is defined as: to reach the
optimal technical and economical solution to recurrent problems. Standardization aims to:
• facilitate communication by creating distinct conceptions with definitions and
terms,
• secure compatibility and inter-operability through restrictions of size and
weight, dimensions and interfaces,
• accomplish variety reduction through selection of size and weight, dimensions,
and designs,
• facilitate flexibility through modularization,
• standardize characteristics, functions, qualities and safety for products,
processes, systems, and services, and
• specify distinct testability methods.

de Vries (2001) states that standardization can function as a lubricant in the modern
industrial society and that the main aim of standardization can be found in the contribution
to the business results and the effectiveness and efficiency of the organization. Further,
standardization can reduce costs of products and services and the issue of following
standards or not can be the difference between success and failure in the market. de Vries
also presents some additional aims of standardization:
• variety reduction of products and procedures in human life,
• enabling communication,
• its contribution to the functioning of the overall economy,
• its contributions to safety, health, and protection of life,
• its protection of consumers and community interests, and
• eliminating trade barriers.

Standardization should be looked upon as a tool for a rational, environmentally sound, and
energy saving development of production, distribution and utilization of commodities,
processes, systems, and services. Standardization is also a means to facilitate technology
collaboration and technological transference. According to Brunsson and Jacobsson (1998),
there are three different actors both issuing and following standards: individuals,
organizations, and states.

The arguments and motivations for following standards are presented by Brunsson and
Jacobsson (1998) in four topics:
• Standards are an effective instrument for transportation of information. Just
knowing that a potential producer or customer is following an accepted
standard means that no detailed questions need to be asked. In some cases
knowing a lot about the standard is not even important, especially if a trusted
third party has attested that the standard is enforced, i.e. some sort of
certification.
• Standards constitute a method for coordination. Standards are often developed
so that different products will be compatible, for example the electric plug and
electric outlet.

23
 • Standards also mean simplifications; there is less to choose from (variety
reduction). This facilitates the overview and understanding of a problem.
• Standards are usually also motivated that they constitute the best solution (in
some cases merely a good solution, in others the only solution). This argument
can actually in itself be independent - there exists no reasons not to follow the
standard, simply because it is the best alternative.

The unifying effects that often follow from standards are usually not promoted by the
standards organizations. They like to turn it around and state that standards help variation,
arguing that by standardizing everything that already is invented, innovators can focus on
developing on what has not yet been developed and standardized. However, there is one
important aspect to the issue of unifying by standardization, and that is large-scale
production. This was a central argument when standards first began to emerge -
standardization can almost be looked upon as a prerequisite to industrialism (e.g.
dimensions of bolts, sizes of clothing, etc.), (Brunsson and Jacobsson, 1998).

Setting up Standards
According to Hawkins et.al. (1995), historians, economists, and sociologists have shown that
standards have played a key role at the core of technological processes, innovation, and
industrial development. They also say that standards will play an even more significant role
in the future.

One should question the positive and negative effects of a standard before embracing it
completely. Will the standard’s goals really be desirable for your organization? Will the
standards lead you to your goals? Is there any better, or perhaps an even easier way to reach
the goals? There are several examples of how standardization can create poor solutions. For
example, the so called QWERTY-standard, which specifies how the letters on the keyboard
are supposed to be located was originally developed to reduce the typing speed to prevent
the type arms (in old type writers) to get hooked together. Today the problem no longer
exists, but the standard does. The standard bumper height of cars is about 440 mm, which
happens to be almost the exact height of the kneecaps of a normal-sized human being
(Brunsson and Jacobsson, 1998).

Tassey (1992) stresses that from a corporate strategy and public policy point of view, no
standards should be accepted before the technology being standardized has matured. This is
especially important in complicated system technologies like communications or factory
automation (e.g. condition based maintenance systems), where standardization should be
decided upon in evolutionary steps according to the technology’s progress. The progress is
set by three different factors: (1) the pace of technology change in each component category;
(2) the disembodied technology development (sets the limits for the overall system
architecture and organization); and (3) the market structure. Complete standardization too
early in a technology life cycle can put a restraint on innovation.

Utterback (1996) states that in the early phase of the life cycle of a product, neither the
customer nor the producer have developed a sense of either the ideal product design or the
optimal manufacturing process. Everyone is learning and experimenting as the product

24
matures. Industry standards at this stage are usually imperfect or undeveloped, if they even
exist. Out of what can be seen as an experimenting chaos of development and competition, a
dominant design usually emerges. The dominant design, being the product architecture that
wins the trust in the marketplace, is the one that competitors must turn to if they hope to
keep their market share. The dominant design usually has the effect of enforcing
standardization on a product. With the appearance of a dominant design, the innovative
effort shifts focus from the product’s functionality and performance features to improving
the manufacturing process, especially with regard to production efficiency and costs.

Utterback (1996) also presents a model, see figure 7, illustrating that the changes in product
and process innovation occur in patterns with three visible phases: the fluid, the transitional,
and the specific phase. In the fluid phase, the rate of major innovations are high for the
product, while being low for processing. The products often have diverse designs and are
usually customized while the production is flexible but inefficient. In the transitional phase,
the rate of major innovations subsides for products but increases for processes. The product
design becomes stable enough for at least a few to have significant production volume and
the production becomes more controlled as the changes occur in larger steps. In the specific
phase, the rate of major innovation stays low for products and decreases for processes.
Standards are followed in new product development and innovation is incremental with
improvements focused on productivity and quality; the production is inflexible and the cost
of change is high (Utterback, 1996).

FIGURE 7 Model presented by Utterback (1996).

Sometime in the transitional phase, dominant designs tend to emerge and industry
standards are set. Interface standards become important to facilitate modularization (which
can help variation through compositions of different components). Poorly designed or
delayed interface standards could have negative effects on the market structure. Inadequate
standards (no standards or poorly designed standards) can increase the economic costs. The
costs can range from interoperability problems in different products to the need of having to
buy several vendors’ versions of similar software (Tassey, 2000).

25
3.4 Implementation Strategies
Work methods in an organization are often deeply rooted in the ordinary day to day work.
Changing work methods can sometimes be looked upon with scepticism by both employees
and management, sometimes more by one than the other. When changing work methods, an
implementation or introduction phase must be followed if success is to be achieved.

Organizational Change
Kotter (1996) states that successful implementation of change in organizations has to follow
two important patterns; firstly, the change has to follow a multi-step process that creates
power and motivation to overwhelm the reactionaries and, secondly, the process has to be
driven by high-quality leadership. Kotter (1996) derives this process in an eight-stage
process:
1. establishing a sense of urgency,
2. creating the guiding coalition,
3. developing a vision and strategy,
4. communicating the change vision,
5. empowering broad-based action,
6. generating short-term wins,
7. consolidating gains and producing more change, and
8. anchoring the approaches in the culture.

To be successful, the organizational change needs a well-organized implementation strategy.

Vrakking (1995) states that the success of an implementation of an innovation is in relation to
the time between the generation of the innovation to the implementation of it, and that
success is achieved if this time is kept to a minimum. Vrakking argues that these
implementations are only successful in companies that follow a very strict implementation
strategy. Vrakking also presents the four phases that the innovation process follows:
Phase 1, Generation of ideas
o Research phase
o Development phase
Phase 2, Initiation
o Dissemination phase
o Acceptance phase
Phase 3, Implementation
o Implementation phase
Phase 4, Incorporation
o Evaluation phase
o Adjustment phase

Vrakking (1995) also presents eleven practical ideas on implementation:

1. good communication and information,
2. training,
3. learning process,
4. top-down and bottom-up communication,
5. project approach,
6. powerful leaders, support from opinion leaders,

26
 7. prevent “group think”,
8. create support,
9. implementation is not separate from the design process,
10. prevent resistance (if possible), and
11. line management must support the change.

Kotter and Cohen (2002) point out that empowerment to remove barriers is one very
important aspect when it comes to implementing change. They also mean that it is often an
immediate manager or someone higher in the hierarchy that is the biggest obstacle for
change. Kotter and Cohen present four steps for empowerment and four steps that do not
lead to empowerment. The four steps that work are:
• finding individuals with change experience who can bolster people’s self-
confidence with we-won-you-can-too anecdotes,
• recognition and reward systems that inspire, promote optimism, and build self-
confidence,
• feedback that can help people make better vision-related decisions, and
• “retooling” disempowering managers by giving them new jobs that clearly
show the need for change.

The four steps that do not work are:

• ignoring bosses who seriously disempower their subordinates,
• solving the boss problem by taking away their power (making them mad and
scared) and giving it to their subordinates,
• trying to remove all the barriers at once, and
• giving in to your own pessimism and fears.

Implementing Maintenance
Within maintenance-related issues there exist some references to implementation strategies,
though within condition based maintenance they are scarce. Spare (2001) states that
condition based maintenance programs should be designed and implemented through:
“Well-defined goals and a cost-effective investment strategy...” (p. 954). Reichard et.al. (2000)
give a more technically oriented aspect by stating: “The implementation of such systems
requires a combination of sensor data fusion, feature extraction, classification, and prediction
algorithms.” (p.329). Jiang et.al. (2002) point out that the human aspect cannot be forgotten in
condition based maintenance technology by stating: “Correct analysis and diagnosis based
on the collected information is essential for right maintenance decisions: when, where, and
what maintenance actions should be carried out for a specific piece of equipment. Obviously,
participation and intervention of the human experts are necessary for all these activities.”
(p.1957). However, Carnero (2003) points out that there is no international standard on
managing a predictive maintenance program, little less to say no standard for implementing
it either.

Mobley (2002) presents how a Predictive Maintenance (PdM) program should be established
focusing on: (1) goals, objectives, and benefits; (2) functional requirements; (3) selling
predictive maintenance programs; (4) selecting a predictive maintenance system; (5)
database development; and (6) getting started. No model with steps in time is presented

27
though. In the first section on goals, objectives, and benefits, focus is on the importance of
creating a reference or baseline dataset of the existing maintenance costs and other
parameters that will be affected through the introduction of PdM. The second section,
functional requirements, focuses on the importance of management support and dedicated
personnel, efficient data collection and analysis procedures, and the initial creating of a
database. The third section, selling predictive maintenance programs, focuses on keys to
success, such as: formulating a program plan, knowing the audience, creating an
implementation plan, doing your homework, taking a holistic view, and getting absolute
buy-in. The section on creating an implementation plan states that the most important part of
the establishment is to construct a concise, detailed program plan. The plan must include
well-defined goals and objectives that will be achievable within the expected time limit. The
plan should also be of a phased approach so that the capital investment can be spread out
over a longer period of time. Also mentioned is the importance of assigning responsibility to
specific individuals and that there must be a start date and an end date for all projects. The
fourth section, selecting a predictive maintenance system, focuses on system requirements,
such as: software and hardware, automated data acquisition, reliability, cost, training and
support, etc. The fifth section, database development, focuses on data acquisition frequency,
analysis parameters, defining alarm limits and alerts, and selection of transducers. The sixth
section, getting started, focuses mostly on training and technical support.

Mitchell and Murry (1995) present how a Predictive Maintenance (PdM) program was
implemented at the United States Department of Energy (DOE) Strategic Petroleum Reserve
(SPR). The SPR was reported to, at the time, have the country’s largest and most diverse
collection of pumps and motors, with the main task of moving large quantities of oil that
were stored within the reserve. The SPR’s equipment was seldom or infrequently operated
but was required to have high reliability and a high degree of readiness and, at the same
time, it suffered from ageing problems so it was decided to implement the PdM program.
The program was composed of five major elements: tests and exercises, condition
monitoring, data storage and retrieval, training, and program goals and reports. The PdM
program consisted of eight logical steps:

1. Develop PdM Policy and Program Plan

2. Assign responsibilities for
a. Development of program requirements
b. Implementing the program
c. Assessing the program
3. Define and specify required equipment
4. Purchase equipment
5. Develop implementing procedures
6. Train employees
7. Implement program
8. Report results and assess program effectiveness

The PdM program was specified and developed by an Engineering department,

implemented by a Maintenance department, and maintained by the Operations and
Maintenance departments. During the first year after implementation, the program showed
some success stories and payback but the majority of time was spent on building a

28
satisfactory data base, tailoring of the program, accomplishing the required training, and
lobbying for a positive attitude towards the program. Mitchell and Murry (1995) also present
a few suggestions and actions on what they would do differently if they had to go through
the implementation process again (in short);
• to do a complete analysis and definition of the project needs, tailor the program,
• ensure that support from management, clients, and budget are available before
the implementation process starts,
• be clear about that the process takes time, a satisfactory database can take 18-24
months to build, additional manpower is required,
• involve effected groups early; maintenance, engineering, and operations; and
develop a project team,
• be early in the procurement of equipment and services, be sure potential
products have been developed, been debugged, and have been proven to work
in applications in environments similar to your own,
• implement in a phased approach, stay small, do not incorporate the whole
program in the first phase, and
• keep management and clients informed; publicize any success no matter how
small in the early phases of the implementation.

According to Nakajima (1988) it takes at least three years to implement Total Productive
Maintenance (TPM) before really good results can be presented and, according to them,
trying to do it in less time will most likely lead to failure. Before implementing TPM, three
requirements need to be taken into consideration in order to achieve fundamental
improvements. The first two are to increase the motivation and competence of the personnel
which will increase the effectiveness and operations of the equipment; the third is the work
environment that must be developed to support the establishment of a systematic program
for implementing TPM. Nakajima has divided the implementation process into three stages
with 12 steps. In the preparation stage, the work environment is created by the establishment
of a plan for the introduction of TPM. This stage can be compared to the initial stages of
product development. In the implementation stage, the program is put into work and it is up
and running. This stage can be compared to the production of a product. In the last stage,
called the stabilization stage, evaluation and higher goal setting is performed. The 12 steps
are divided into five steps in the preparation stage, five steps in the implementation stage,
and one step in the stabilization stage. Another step is located between the preparation and
the implementation stage, in the so called preliminary implementation stage.

3.5 Summary of the Theoretical Framework

Maintenance, in general, can be executed in two different ways - either in order to retain or
restore an item to a state in which it can perform the required function. These approaches are
normally referred to as preventive and corrective maintenance. Many different strategies
exist to schedule and perform maintenance actions for both preventive and corrective
maintenance. A way to achieve ‘good maintenance’, as defined by Cooke and Paulsen (1997),
when very little corrective maintenance takes place and when as little preventive
maintenance as possible is performed, can be to utilize condition based maintenance.
Condition based maintenance, by utilization of condition monitoring to reveal incipient

29
failure, is one approach to schedule preventive maintenance according to the definition SS-
EN 13306 (2001). In this project, the phrase ‘condition based maintenance system’ has been
used in order to visualize condition based maintenance in a more holistic view. A condition
based maintenance system has, within this project, through theory, been defined as:

“A system that uses condition based maintenance to determine and schedule predictive
maintenance actions autonomously or in interaction with other systems or humans.”

Different designs of condition based maintenance systems were discussed in section 3.2.2.
They are all roughly similar with small differences.

The section devoted to standardization gives a discussion of the purpose and aims of
standards, as well as a small discussion on setting up standards. Several positive purposes,
such as variety reduction, communication, modularization and flexibility, coordination,
elimination of trade barriers, etc., for using standards and standardization are brought up
and discussed. Some possible drawbacks in the use of standards are also discussed.

In the implementation strategy section, some general organizational change processes have
been discussed where several general tips on implementing new strategies, etc., are pointed
out. Some general implementation strategies for condition based maintenance were also
discussed. The strategies are mostly aspects to think of and are not divided into models with
successive aspects to think of in terms of ‘do this first’ and ‘follow up like this’, as could be
found within the general organizational change processes. Some of the most outspoken
aspects in the literature are to let the implementation process take time and not stress the
change process just in order to chase good results. Also, to set the implementation process
and the goals for it in a phased approach and not try to incorporate condition based
maintenance in one step. Set aside time and a budget for training and education.
Communicate and motivate the decision to implement condition based maintenance in the
whole company.

30
 4 Thesis Contribution and
Summary of Papers

This chapter deals with the thesis contribution in the form of

summaries of the appended papers. The conclusions from each paper
are presented at the end of the chapter.

T H E C O N T R I B U T I O N O F this thesis derives from different studies presented in four

appended papers. This chapter gives a summary of these papers.

4.1 Paper I – Standardization Issues in Condition Based Maintenance

This paper (Bengtsson, 2003) aims to find out what standards and standardization proposals
that can be found within the condition based maintenance system technology. The objective
of the paper was thus to answer a part of the first research question. The paper was
developed based upon a literature review of technical standards in general as well as
standards and standardization proposals within condition based maintenance system
technology. The literature has been gathered through a review of journal and conference
proceedings databases, hard copies of journals and conference proceedings, books, and the
Internet.

Today, and for the past few years, the development within condition based maintenance has
been strong. Prototypes and products have been developed in a rapid pace. Companies that
are selling condition monitoring products and condition based maintenance services can
easily be found on the Internet. However, the development is in many cases directed to
specific applications and the results of a prototype are not always generally reproducible.
One way of directing the development might be to follow standards and standardization
proposals.

Standards made large-scale production in the industrial area possible (Brunsson and
Jacobsson, 1998) and it has thereafter been regarded differently by different actors. The
unifying effect that sometimes comes from standardization might be regarded as mind-
numbing and innovation hostile. Standard organizations and standard bodies defend the
unifying effect by stating that it has the advantage of pushing innovators into developing
something new, something that is not yet standardized (Brunsson and Jacobsson, 1998). The

31
unifying effect also brings forward the advantage of compatibility and modularization which
brings better deals for customers and users of the standardized products.

Within condition based maintenance system technology, two major standard organizations,
MIMOSA (the Machinery Information Management Open System Alliance) and OSA-CBM
(Open System Architecture for Condition Based Maintenance), were found. MIMOSA
(www.mimosa.org) has since the middle of the 1990’s hosted open conventions for
information exchange between plant and machinery information systems, i.e. a way to
enhance, amongst other things, the compatibility issue between different vendor products.
OSA-CBM (www.osacbm.org) has developed an open system architecture for condition
based maintenance, i.e. a way to enhance the modularization of different vendor systems,
while not locking customers into a single-source solution. The organization claims that an
accepted non-proprietary open system architectural standard would bring an improved ease
of upgrading for system components, a broader supplier community, more rapid technology
development, and reduced prices (Thurston, 2001).

Other standards found and discussed within the literature review were the IEEE Std 1451
and the IEEE Std 1232, see Paper I (Bengtsson, 2003) for more information and references.
The first, a standard at the basic level of a condition based maintenance system, i.e. the
sensor network. A standard that helps customers to integrate different vendor products
(transducers, actuators, and sensors) when networking. The latter is a standard to address
the problems and difficulties of diagnosing complex and costly technical systems.

4.2 Paper II - Condition Based Maintenance System Technology

The objective with this paper (Bengtsson, 2004b) is, together with Paper I, to answer the first
research question, “What standards and standardization proposals can be found within the
condition based maintenance system technology, and what effects might they have on future
research?”. The paper is a combined literature review and literature search. For the review,
the literature has been gathered through searches in journal and conference proceedings
databases, hard copies of journals and conference proceedings, books, and the Internet. For
the literature search, the technical database IEEE Xplore® was used
(http://ieeexplore.ieee.org/Xplore/DynWel.jsp).

The contribution of the literature review in this paper is similar to that of the first paper,
although this review also brings up the standard ISO 13373-1 – Condition monitoring and
diagnostics of machines, as another possible standard to use within condition based
maintenance system technology. The standard provides general guidelines for measuring
and data collection functions for machine condition monitoring, focusing on machine
vibrations. The literature review sets out to link the module standards that exist to the OSA-
CBM’s open system architecture standard, to see what has been standardized and what has
not, in order to draw conclusions to what modules that possibly need more immediate
research. The OSA-CBM divides a condition based maintenance system into seven modules:
the sensor module, the signal processor, the condition monitor, health assessment
(diagnostics), prognostics, decision support, and presentation. The sensor module could be
linked to the published standard IEEE Std 1451. The signal processing could be executed in

32
several standard ways, such as Fast Fourier Transform (FFT), spectral analysis, filtering, etc.
Parts of the condition monitor could be linked to the published standard ISO 13373-1. The
health assessment module could partially be linked to the ISO 13373-1 and the IEEE Std 1232,
thus leaving out the prognostic, decision support, and presentation module from standards
and standardization proposals.

The literature search was conducted in order to see what modules of the OSA-CBM’s open
system architecture standard research has been performed for and if any possible trends in
the research were visible. The modules of the OSA-CBM were searched together with the
word maintenance, one module at a time (with exception to the presentation module) from
the year 1993 to 2002. The word maintenance was also search solely in order to get a
reference point, later to be calculated as a percentage of hits in publications (see paper II for
more information). With exceptions to the condition monitor and the prognosis module, no
trends in percentage of hits were visible, and the most discernible trend was seen in the
prognosis module. The figures show a steady increase in the number of publications from
the year 1993 to 2001, with a small dip in the year 2002 compared to 2001.

4.3 Paper III - Technical Design of Condition Based Maintenance

Systems
The objective of this paper (Bengtsson et.al, 2004) is to describe and illustrate the necessary
technical constituents of a condition based maintenance system, thus answering research
question 2, “How can a condition based maintenance system be designed, technically, what
methods and techniques can be used for comprehensive system function, and where can a
human enter such a system?”. The paper is conducted of a theoretical discussion of condition
based maintenance system technology and its necessary modules, based on literature on the
topic. The theoretical discussion is then compared to a case study development of a
condition based maintenance system, conducted on industrial robots.

Condition based maintenance systems can be constructed of highly complex technical

systems with sensitive data acquisition systems and automatic diagnostic and prognostic
abilities all the way to humans manually monitoring the conditions by either handheld
devices or by their subjective senses. Normally, a mixture of all the above is used within a
condition based maintenance strategy. The on-line semi or fully automatic system has not yet
been widely accepted in Swedish industry and the reason for this may come from several
different sources. One reason might be the fact that the maturity level within complex
technical condition based maintenance systems might be too low. This paper, once again,
discusses the open system architecture proposed by OSA-CBM, but here in more depth and
illustrating the modules by a case study development of a monitoring system on industrial
robots.

The modules are theoretically discussed one by one and either defined and/or given a clear
purpose of their function, this in order to be able to see the comprehensive function of a
condition based maintenance system. The theory is later compared to the case study
development consisting of a microphone, a signal processing module, and a combined
condition monitor and diagnosis module, see figure 8.

33
FIGURE 8 The case development system structure with its input and output signals, compare to the OSA-
CBM system structure in figure 4.

Sound is recorded to a computer and the recording is taken as input to the pre-processing
step. The pre-processing process is responsible for filtering and removing unwanted noise. It
also extracts period information from the sound. In the feature identification process, the
system uses a two-pass model, first identifying features and then creating a vector with
features. Once the features are identified, the system classifies the feature vector. The
classification is based on previously classified measurements (case library), i.e. a Case-Based
Reasoning approach was chosen. Case-Based Reasoning is a method from Artificial
Intelligence, based on a cognitive model of learning from experience. This method was
chosen as it is was found to have some advantages in this category of applications where the
end-testing personnel’s (of the industrial robots) tacit knowledge of the sound profile was
used as input to the monitoring system. When a new sound has been classified, the new case
is added to the case library.

4.4 Paper IV – Important Aspects to take into Consideration when

Implementing Condition Based Maintenance
The objective of this paper (Bengtsson and Jackson, 2004) is to illustrate and highlight
important aspects companies have to consider when deciding to implement condition based
maintenance. The paper also tries to find out if there is a possibility to construct an
implementation model for condition based maintenance, models such as the ones that can be
found within the maintenance strategies Total Productive Maintenance and Reliability
Centered Maintenance. The papers’ objective is thus answering the third research question,
“What aspects do a company need to consider when implementing a condition based
maintenance system?”. The paper is based upon a case study, performed at an international
manufacturing company, conducted through a questionnaire/interview study. The
questionnaire was constructed with input from literature on the topic of implementation
strategies within maintenance as well as general implementation and change processes. The
results of the questionnaire/interview study were later compared to the literature.

As more companies have started seeing the maintenance department as a profit generating
function, more focus has been directed on different maintenance strategies. Total Productive
Maintenance (TPM) and Reliability-Centered Maintenance (RCM) are two of the most well-

34
known maintenance strategies that, in the last 20 years, have been used frequently in the
manufacturing and aircraft industries. However, to perform maintenance actions in the
corrective or the preventive approach is still a present issue. The preventive maintenance
technology, condition based maintenance, has in the past years received good reviews
although it is not utilized to a great extent within Swedish industry. One reason for this
might be that there is a lack of organizational research within condition based maintenance
in terms of implementation strategies and applied condition based maintenance, i.e. how to
implement and run a successful condition based maintenance program.

The questionnaire consisted of 38 statements that the respondent was asked to either agree
or disagree with, with the help of a six-level scale. All respondents were also given the
chance to leave qualitative comments to all statements. At the end of the questionnaire the
respondents were also asked to state the biggest threats to why condition based maintenance
would not be implemented on their products and processes. The questionnaire with the
qualitative comments made it possible to single-out important aspects a company has to
consider within the first phase of implementing condition based maintenance. The open
question regarding the biggest threats to why condition based maintenance would not be
implemented resulted in that most threats were purely organizational with questions such as
whom should pay the initial costs and that the wrong people were left to decide. Only two
(out of twelve) that chose to leave a comment mentioned something about the technology at
hand.

The first phase of implementing condition based maintenance that this case study set out to
answer came to be presented in five brackets. The brackets are partially following
Nakajima’s (1988) first five steps in his implementation strategy for Total Productive
Maintenance, the ‘starting up phase’.

The Managerial Issues:

- Determine the existing maintenance philosophy and costs, direct maintenance costs
(manpower, direct materials, overheads, etc.), and indirect maintenance costs (cost of
lost production, cost of quality deficiencies, accidents, etc.), in order to set a reference
point.
- Determine the possible incentives in the form of higher productivity, Key
Performance Index (KPI), increased safety for humans and the environment, etc,
against the calculated reference point.
- Promotion of CBM must come from top management.

The Employee, Information, and Education Issues:

- Communicate the decision to develop and implement CBM.
- Develop an information strategy to communicate the coming results, in particular the
positive ones.
- Develop a plan for the education of employees and managers.
- Use the employees’ tacit knowledge when applicable.

The Organizational Issues:

- Increase cooperation with contractors and involve customers in specification issues.

35
- Create temporary organizations (project groups) and create empowerment to lead the
development and implementation of CBM.
- Try to involve people from the contractors and customers into the temporary
organizations.

The Goal Setting Issues:

- The temporary organizations should develop the visions and the goals.
- Set the goals in a phased approach, one or two long-term visions (~5 years) and
several achievable short-term goals (~1 year).
- Finalize goals at top management.
- Communicate goals to the company, the contractors, and the customers.

Getting started:
- Create a master plan.
- Decide on the use of standards.
- Decide how to select which systems to monitor and select them.
- Pilot on single systems:
o Select pilot systems that have high downtime cost, high failure rates, etc., i.e.
capital intensive systems.
o Select pilot systems that the company has knowledge of, e.g. failure rates, fault
patterns, etc.
- Use simpler monitoring techniques at first (e.g. trend analysis) to learn.
- Develop a comprehensive database, including measuring data, maintenance history,
failure history, mission history, model information, spare parts assets, etc.

4.5 Summary of Thesis Contributions

The contribution from paper I and paper II is a review of standards and standardization
proposals within condition based maintenance systems. The conclusion that can be drawn is
that several modular standards (IEEE Std 1451, IEEE Std 1232, and ISO 13373-1) exist within
the condition based maintenance system technology and that two standardization proposals,
from the organizations of MIMOSA and OSA-CBM, also exist as guidance on a more
comprehensive level when developing condition based maintenance systems. The two
comprehensive standardization proposals are complimentary and the possibility of
integrating them is considered good. The modular standards can also, to some extent, be
integrated into the comprehensive standards; on some a slight modification will be necessary
though.

Paper II also suggests that the standards that do exist, and the research that has been
performed on condition based maintenance system technology modules, point towards a
possible need for an increase in research on prognostic modules, but it also points towards
that there has been a possible increase in research on prognostic modules over the past ten
years, thus finding the state of the art within condition based maintenance system
technology.

36
Paper III suggests that the OSA-CBM system approach of seven different modules as a base
was one appropriate way of developing condition based maintenance systems in order to
achieve a comprehensive system function. The modules can also be seen as a template of
technical aspects of condition based maintenance implementation. All seven are equally
important to uphold a condition based maintenance strategy. The level of automation can
actually be seen to be of secondary importance within the implementation process. For some
processes, the level of automation does not have to be taken to the extreme, for other more
complex, large-sized, critical, and processes that are hazardous to monitor, more automation
is a necessity.

Paper IV suggests that more research on the topic of organizational issues and
implementation strategies within the condition based maintenance technology needs to be
performed. Answers to the interview question on the issue of what the biggest threat against
implementation of condition based maintenance on the companies’ products were
constituted of mostly organizational factors that needed to be resolved before technical
implementation could commence. This, single-handed, points in the direction that more
research needs to be conducted on the issue. If the relatively small amount of research that
was found in the initial literature search is also taken into account, then the fact that
additional research needs to be conducted increases even more.

Paper IV also suggests that it is feasible to construct an implementation strategy for

condition based maintenance. The case findings were built on the first five steps of
Nakajima’s implementation strategy for Total Productive Maintenance, which is relatively
general, and it clearly indicates that it is feasible to enhance the ‘starting-up phase’ (called
the preparatory stage in Nakajima’s implementation strategy) with an ‘introductory stage’, a
‘CBM introduction stage’, and finally a ‘stabilizing stage’.

37
38
 5 Conclusion and Future Research

This chapter summarizes and discusses the conclusions of the research presented in
Chapter 3 Theoretical Framework and Chapter 4 Thesis Contribution and Summary of
Papers. The chapter also discusses the academic and industrial relevance as well as
future research.

THE O B J E C T I V E O F this thesis was to investigate technical constituents and

organizational aspects in a condition based maintenance system in order to make the condition based
maintenance strategy more accessible within Swedish industry. The objective was broken down
into three research questions that have been answered through different research projects
and reported through submissions of papers to different conferences on the topic. The
conclusions of these research questions can be summed up as conclusions for the overall
objective of this thesis.

5.1 Conclusions
Following, there will be a discussion of the findings and results that have been obtained from
the three research questions.

RQ1 – The Standardization Issue

The first Research Question, “What standards and standardization proposals can be found within
condition based maintenance system technology, and what effects might they have on future
research?” was answered by Paper I (Bengtsson, 2003) and Paper II (Bengtsson, 2004b).

Paper I gives a review of which standards and standardization proposals that exist within
condition based maintenance and implicitly within automated condition based maintenance
system technology. The overview shows that there are several standards and standardization
proposals within condition based maintenance system technology and that they, to a great
extent, are compatible. The paper also discussed the drawback that can come if new
innovative technologies become standardized too quickly.

Paper II once again gives a review of possible standards to use when developing condition
based maintenance system technologies. The paper also, through the architectural
standardization proposal of the organization OSA-CBM, presents which modules in a
condition based maintenance system that can be developed today using standards or
standardized means. Through this disposition, and through the literature search that was
conducted in the paper, it was possible to draw some conclusions as to where the future

39
automated condition based maintenance system technology research will have its focus, in
terms of standardized modules.

Conclusions for RQ1 can be summed up as:

a) Several of the standards and standardization proposals found can be
integrated and possibly simplify the development of automated condition
based maintenance system technology, e.g. by modularization and flexibility.
b) Standards or standard means exist for at least the four first modules of the
OSA-CBM architecture proposal. With this in mind, together with the
literature search that was carried out, the next step in commercializing and
possibly standardizing for researchers and vendors is the automated
prognostic ability of condition based maintenance systems.
c) The state of the art within condition based maintenance system lies within the
prognostication and the decision support modules.

RQ2 – The Condition Based Maintenance System Design Issue

The second Research Question, “How can a condition based maintenance system be designed
technically, what methods and techniques can be used for a comprehensive system function, and where
can a human being enter such a system?” was answered by Paper III (Bengtsson et.al., 2004).

Paper III had the objective to describe and illustrate the necessary technical components of a
condition based maintenance system. Following a theoretical discussion of the OSA-CBM
architectural standardization proposal and a more in-depth theoretical discussion of the
modules proposed by OSA-CBM, it was, through a case study, illustrated that condition
based maintenance systems could be developed using the architecture proposal as a
template. Although the case study system did not include all layers (prognostic ability has
not yet been implemented), it was shown to be a good way of thinking to not miss any
crucial performances of a comprehensive condition based maintenance system. The modules
of the standardization proposal could also be used as an organizational way of thinking
when using the subjective condition monitoring technique, visual inspection. All system
functions need to be accounted for, the level of automation is of subordinate significance. As
a secondary objective, the paper set out to prove if it would be feasible to use tacit
knowledge of the end-testing personnel when developing a diagnostic function and if such a
system would be accepted as a tool by potential users.

Conclusions for RQ2 can be summed up as:

a) A definition of condition based maintenance systems as: “A system that uses
condition based maintenance to determine and schedule predictive maintenance
actions autonomously or in interaction with other systems or humans”.
b) The seven modules proposed by the OSA-CBM were found to be a good
template for a comprehensive condition based maintenance system
development.
c) Human interaction with such a condition based maintenance system could be
one possible set up if not deciding to run a completely automated system.
d) Case-Based Reasoning was shown to be one good Artificial Intelligence
method to analyze the sound signals measured and to diagnose abnormalities.

40
 e) For condition monitoring, when using sound signals, qualified end-testing
personnel’s tacit knowledge proved to be an effective parameter for analyzing
and diagnosing the abnormalities, thus also meaning that machine operators’
tacit knowledge possibly would be effective to use.

RQ3 – The Implementation Issue

The third Research Question, “What aspects do a company need to consider when implementing a
condition based maintenance system?” was answered by Paper IV (Bengtsson and Jackson,
2004).

Paper IV had the objective to illustrate and highlight important aspects a company has to
consider when deciding to implement condition based maintenance. This objective was
solved by a literature review and through a case study at an international manufacturing
company. A 38-statement questionnaire was answered by 16 co-workers at the company that
all had been in or around maintenance and/or condition based maintenance, although the
respondents were spread out over several divisions within the company. The questionnaire,
with qualitative comments that the respondents gave during the interview and the literature
review gave results that put forward important aspects a company has to consider when
deciding to implement condition based maintenance. An open question that was given to the
respondents at the end of each interview, of what threats they could see against condition
based maintenance at their company, gave input that more research on organizational
implementation strategies of condition based maintenance is necessary.

Conclusions for RQ3 can be summed up as:

a) It would be feasible to design a general implementation strategy for condition
based maintenance that also takes organizational issues into consideration.
b) A summary, as depicted in chapter 4, of important aspects to take into
consideration when deciding to implement condition based maintenance.
c) More research on the organizational issues within implementation strategy of
condition based maintenance would help companies make the transition from
one maintenance strategy to a condition based maintenance strategy easier.

Summary of Thesis Conclusions

The research questions were formulated and answered in that both technical and
organizational factors of a condition based maintenance system would be highlighted as a
way to make the condition based maintenance type more accessible in Swedish industry. The
purpose of this thesis was also to be a starting point for deciding what topics that need to be
investigated further in future research towards a PhD degree.

The state of the art regarding the technical aspects of a condition based maintenance system
can now be determined to be the prognosis and the decision support modules of the system.
Even though ISO has now started to work for a standardization for prognosis in the
approved project ISO/DIS 13381-1 - Condition monitoring and diagnostics of machines –
Prognostics – Part 1: General guidelines (Culverson, 2004), it can probably be safe to say that
the state of the art will stay on the prognosis module for a while ahead, contemplating the
complexity of the art of prognosticating.

41
The definition deduced in chapter 3 was also concerning the technical constituents of
condition based maintenance systems. Before deciding on possible methods and tools for a
comprehensive system function in condition based maintenance, it was defined as: “A system
that uses condition based maintenance to determine and schedule predictive maintenance actions
autonomously or in interaction with other systems or humans”. The modules in a condition based
maintenance system were also defined or given a clear purpose (Bengtsson et.al., 2004), as
well as tested in a real system development. A contemplation regarding the technical aspects
of a condition based maintenance system is that all parts of the system do not have to be
managed by technical components; humans with expertise in certain areas might function
just as well in many of the processes the technical systems handle. This might come in handy
for companies that do not need an automated system technology, which might be the case
for smaller companies with less money invested in the maintenance organization. This might
also be the case for processes that do not need around the clock monitoring, processes that
are not safety critical, safety or environmental hazardous, or processes with clear ageing
features, etc.

Aspects regarding organizational issues within condition based maintenance were shown
not to be extensively published. There was not a lot of research found within the topic. The
performed case study gave results of aspects a company needs to consider when deciding to
implement condition based maintenance. The open-ended questions that were asked during
the interview resulted in that several organizational issues need to be resolved before
technical implementation can commence. The organizational aspects documented in this
thesis might be one way of structuring the start of an implementation.

5.2 Academic and Industrial Relevance

The conclusions in relation to the first research question can be put into use by vendors of
condition monitoring equipment. The findings can also possibly give customers of condition
monitoring equipment some knowledge of demands that can be put on vendors. The
vendors, if they are not already using standards, will most likely not turn to standardized
solutions immediately, but with more customers asking for standardized products, it might
result in more standardized products in the future, which will give customers positive
experiences. The fact that most modules in a complete condition based maintenance system
can be developed using standard means, except for the prognostic module and the decision
support, put them in focus for future development, both within the condition monitoring
equipment industry as well as in academic research.

The conclusions that arose from the second research question can be put into use within
companies deciding to implement condition based maintenance. All seven modules
explained in Paper III need to be specified in order to uphold a comprehensive condition
based maintenance strategy, not necessarily saying what level of automation the system
should have. This was also discussed in the theoretical framework where a ‘condition based
maintenance system’ was defined as: “A system that uses condition based maintenance to
determine and schedule predictive maintenance actions autonomously or in interaction with other
systems or humans.”, thus stating that a condition based maintenance strategy can be upheld

42
by humans keeping track of the condition of items subjectively. The seven modules, as
explained in Paper III, together with the definitions of ‘condition based maintenance’ and
‘condition based maintenance systems’ is also a good way of explaining what this particular
maintenance strategy really is and how to execute it.

The conclusions from the third research question can be put to use by vendors of condition
monitoring equipment as well as companies within the industry. The five brackets with
important aspects to consider when deciding to implement condition based maintenance can
be seen as a first level of implementation and to see whether the condition based
maintenance strategy fits or not.

It is now in order to discuss if these constituents and aspects are the most important parts.
This cannot be verified within this research project and not within this thesis. These
constituents and aspects were chosen to be researched upon in order to fulfill the thesis’
objective and to serve as an investigation for the greater goal of a doctoral thesis. For
example, as in research question 2, where sound signals were analyzed using Case-Based
Reasoning, one can quickly state that it might work equally well or perhaps even better with
e.g. vibration technology analysis using Neural Networks. In this research project, sound
signals and Case-Based Reasoning were chosen to serve as an example and to show new
methods while, at the same time, answering research question 2. In research question 3
several aspects concerning organizational implementation of condition based maintenance
were highlighted. These have not been validated in other projects than this one, and it is the
researchers’ intention that, at a later stage, to both validate and expand this study. The
objective of the thesis was stated as an investigation of technical constituents and
organizational aspects. Technical standards within condition based maintenance systems,
technical components for a comprehensive system function, human interaction with the
system, and implementation aspects when deciding to implement condition based
maintenance, as illustrated in figure 9, has been investigated in order to fulfill the objective.

Technical Constituents Organization Aspects

Standards Human Integration

System Components Implementation Aspects

Sensors Managerial Issues

Signal Processor The Employee, Information, and Training Issues

Condition Monitor The Organizational Issue

Diagnosis The Goal Setting Issue

Prognosis Getting Started

Decision Support

Presentation

FIGURE 9 The technical constituents and organizational aspects that have been investigated in this thesis.

43
5.3 Future Research
Maintenance departments and the maintenance function have, historically, been seen as a
cost centre. Alsyouf (2004) reports the results of a recent study within the Swedish industry
stating that 70% of the respondents considered maintenance as a cost centre. The fact that
this figure still is high is a testimony to that more research needs to be performed in order to
change the perception of the maintenance function.

Interesting future research objectives would be to investigate if and to what extent standards
are being used in the developing process of condition based maintenance systems. It would
also be interesting to see if the companies or researchers that do use the standards,
experience any positive or negative effects as a result of their choice.

Other interesting research objectives would be to enhance the development of the sound
analyzing system and to investigate if the system would be feasible to use when robots are
used in actual production. It would also be interesting to investigate if any other diagnostic
methods (other than Case-Based Reasoning) would function equally well or perhaps even
better, to enhance the systems prognostic ability, and to see if it would be possible to develop
a more modularized system structure as proposed in the OSA-CBM.

The important aspects to consider when deciding to implement condition based maintenance
needs to be validated, possibly by performing more case studies at additional companies.
The aspects illustrated in this thesis could also be enhanced with more aspects concerning
staying with a condition based maintenance strategy, how to work with the technology, and
how to lobby for a positive environment. It would also be interesting to see a general
implementation strategy of condition based maintenance designed. An implementation
strategy that companies could use to make the transition from one maintenance strategy to
the condition based maintenance strategy easier.

44
 References

Ahlmann, H. (2002). From Traditional Practice to the New Understanding: The Significance of the
Lice Cycle Profit Concept in Management of Industrial Enterprises. In proceedings of
IFRIMmmm Conference, Sweden, Växjö: Växjö University Press.

Al-Najjar, B. (1997). Condition-Based Maintenance: Selection and Improvement of a Cost-effective

Vibration-Based Policy for Rolling Element Bearings. Doctoral Thesis. Sweden, Lund: KFS AB.

Al-Najjar, B. and Alsyouf, I. (2003). Selecting the most efficient maintenance approach using
fuzzy multiple criteria decision making. International Journal of Production Economics. Vol.
84, Issue 1, pp. 85-100.

Alsyouf, I. (2004). Cost Effective Maintenance for Competitive Advantages. Doctoral Thesis.
Sweden, Växjö: Växjö University Press.

Arbnor, I. and Bjerke, B. (1994). Företagsekonomisk metodlära. Sweden, Lund: Studentlitteratur.

Banks, J., Carson, J., and Nelson, B. (1999). Discrete-Event System Simulation. USA, New
Jersey: Prentice-Hall Inc.

Barlow, R. E. (1984). Mathematical Theory of Reliability: A Historical Perspective. IEEE

Transactions on Reliability. Vol. 33, No.1, pp. 16-20.

Bengtsson, M. (2004a). Användandet av tillståndsbaserat underhåll i svensk industri – en

enkätundersökning genomförd under Underhållsmässan 2004. Technical Report,
IDPOPTR:04:01. Department of Innovation, Design and Product Development.
Mälardalen University: Sweden.

Bengtsson, M. (2004b). Condition Based Maintenance System Technology – Where is Development

Heading? In proceedings from the 17th Conference of Euromaintenance, Spain, Barcelona:
Puntex - Publicaciones.

45
Bengtsson, M. and Jackson, M. (2004). Important Aspect to take into Considerations when
Deciding to Implement Condition Based Maintenance. In proceeding from the 17th Conference
of Condition Monitoring and Diagnostic Management, England, Cambridge: The Central
Printing Services, University of Birmingham, UK.

Bengtsson, M., Olsson, E., Funk, P., and Jackson, M. (2004). Technical Design of Condition Based
Maintenance System – A Case Study using Sound Analysis and Case-Based Reasoning. In
proceedings from the 8th Conference of Maintenance and Reliability, USA, Knoxville.

Bengtsson, M. (2003). Standardization Issues in Condition Based Maintenance. In proceedings

from the 16th Conference of Condition Monitoring and Diagnostic Engineering
Management, Sweden, Växjö: Växjö University Press.

Butcher, S. W. (2000). Assessment of Condition-Based Maintenance in the Department of Defense.

LOGISTICS MANAGEMENT INSTITUTE, USA, McLean, VA.
http://www.acq.osd.mil/log/logistics_materiel_readiness/organizations/mppr/assetts/
senior_steering/condition/LMI%20CBM%20Report.pdf (2003-10-08).

Brunsson, N. and Jacobsson, B. (1998). Standardisering. Sweden, Stockholm: Nerenius &

Santérus Förlag AB.

Carnero, C. M. (2003). How To Set Up a Predictive Maintenance Program. Quality Press. Vol.
36, Issue 3, pp. 56-61.

Chinnam, R. B. and Baruah, P. (2004). A neuro-fuzzy approach for estimating mean residual
life in condition-based maintenance systems. International Journal of Materials and Product
Technology. Vol. 20, Nos. 1-3, pp. 166-179.

Cooke, R. and Paulsen, J. (1997). Concepts for measuring maintenance performance and
methods for analysing competing failure modes. Reliability Engineering and System Safety.
Vol. 5, Issue 2, pp. 135-141.

Culverson, K. J. (2004). Working Within the ISO Structure for Condition Monitoring and
Diagnostics. In proceedings from the 17th Conference of Condition Monitoring and
Diagnostic Management, England, Cambridge: The Central Printing Services, University
of Birmingham, UK.

de Vries, H. J. (2001). Standardization – A New Discipline? In the proceedings from the 2nd
Conference on Standardization and Innovation in Information Technology, 2001.

Ejvegård, R. (1996). Vetenskaplig metod. Sweden, Lund: Studentlitteratur.

Engström, C. (Editor). (1995). Nationalencyklopedin, 17th volume. Sweden, Höganäs:

Bokförlaget Bra Böcker AB.

46
Fararooy, S. and Allan, J. (1995). Condition-Based Maintenance of Railway Signalling Equipment.
In International Conference Electric Railways in a United Europe.

Granö, N. (Project Leader). (1998). ABB Industrial Manual. Sweden, Lund: Wallin & Dalholm
Tryckeri AB.

Hawkins, R., Mansell, R., and Skea, J. (Eds.). (1995). Standards, Innovation and Competitiveness.
UK, Cornwall: Edward Elgar Publishing Limited.

Høyland, A. and Rausand, M. (1994). System Reliability Theory – Models and Statistical Methods.
USA: John Wiley & Sons, Inc.

http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-12-17)

Hubka, V. (1982). Principals of Engineering Design. England: Butterworh Scientific.

Jiang, G., Zhaohui, L., Yitao, C., Yuewu, W., and Shijie, C. (2002) Virtual environment
conceptions for CBM of hydro-electric generating units. In proceedings from Power System
Technology, PowerCon.

Johansson, K-E. (1993). Driftsäkerhet och underhåll. Sweden, Lund: Studentlitteratur.

Jonsson, P. (1999). ”The Impact of Maintenance on the Production Process – Achieving High
Performance. Doctoral Thesis. Sweden, Lund: KFS in Lund AB.

Kotter, J. P. (1996). Leading Change. USA, Boston: Harvard Business School Press.

Kotter, J. P. and Cohen, D. S. (2002). Creative Ways to Empower Action to Change the
Organization: Cases in Point. Journal of Organizational Excellence. Vol. 22, Issue 1, pp. 73-82.

Lantz, A. (1993). Intervjumetodik. Sweden, Lund: Studentlitteratur.

Lebold, M., Reichard, K., and Boylan, D. (2003). Utilizing DCOM in an Open System
Architecture framework for Machinery Monitoring and Diagnostics. In proceedings from
Aerospace Conference, USA, Irvine.

Merriam, S. B. (1994). Fallstudien som forskningsmetod. Sweden, Lund: Studentlitteratur.

Mitchell, J. S. (1998). Five to ten year vision for CBM, ATP Fall Meeting – Condition Based
Maintenance Workshop. USA, Atlanta, GA, 1998 (PowerPoint Presentation),
http://www.atp.nist.gov/files/3 (2003-10-08).

Mitchell, B. F. and Murry, R. J. (1995). Predictive Maintenance Program Evaluation – Lessons

Learned. In proceedings from the 1995 Annual Reliability and Maintainability Symposium,
USA, Washington.

47
Mobley, R. (2002). An Introduction to Predictive Maintenance. USA: Butterworth-
Heinemann/Elsevier Science.

Moya, C. C. and Vera, J. C. H. (2003). Evaluation of Condition Based Maintenance through

Activity Based Cost. Maintenance Journal. Vol. 16, No. 3, pp. 54-61.

Nakajima, S. (1988). Introduction to Total Productive Maintenance. USA, Cambridge:

Productivity Press.

Nolan, F. S. and Heap H. F. (1978). Reliability-Centered Maintenance. USA, Springfield, VA:

Department of Commerce.

Overman, R. (2004). RCM Past, Present and Future. In proceedings from the 8th Conference of
Maintenance and Reliability. USA, Knoxville.

Page, R. (2002). Maintenance Management and Delay Reduction. Maintenance and Asset
Management. Vol. 17, No. 1, pp. 5-13.

Reichard, K. M., Van Dyke, M., and Maynard, K. (2000). Application of sensor fusion and signal
classification techniques in a distributed machinery condition monitoring system. In proceedings
from the SPIE – The International Society for Optical Engineering.

Routio, P. (2004). Methods of arteology. www2.uiah.fi/projects/metodi/e00.htm.

Spare, J. H. (2001). Building the Business Case for Condition-Based Maintenance. In proceedings
from the Transmission and Distribution Conference and Exposition, USA, Atlanta, GA.

Starr, A. G. (1997). A Structured Approach to the Selection of Condition Based Maintenance. In

proceedings from the 5th International Conference on FACTORY 2000, UK, Cambridge.

Swedish Standards Institute, “Maintenance terminology”, SS-EN 13305, 2001.

Tassey, G. (1992). Technology Infrastructure and Competitive Position. USA, Norwell: Kluwer
Academic Publishers.

Tassey, G. (2000). Standardization in Technology-Based Markets. Research Policy. Vol. 29,

Issue 4-5, pp. 587-602.

Thurston, M. G. (2001). An Open Standard for Web-Based Condition-Based Maintenance Systems.

In proceedings from the IEEE System Readiness Technology Conference, Autotestcon
Proceedings 2001. USA, Valley Forge, PA.

Tsang, A. C., Jardine, A. K. S., and Kolodny, H. (1999) Measuring maintenance performance:
a holistic approach. International Journal of Operations & Production Management. Vol. 19,
No. 7, pp. 691-715.

48
Utterback, J. M. (1996). Mastering the Dynamics of Innovation. USA, Boston: Harvard Business
School Press.

Vachtsevanos, G. and Wang, P. (2001). Fault Prognosis using Dynamic Wavelet Neural Networks.
In proceedings from the IEEE System Readiness Technology Conference, Autotestcon
Proceedings, 2001. USA, Valley Forge, PA.

Vrakking, W. J. (1995). The Implementation Game. Journal of Organizational Change

Management. Vol. 8, No. 3, pp. 31-46.

Winroth, M. (2004). On Manufacturing Strategies – Competing Through Inter-Organizational

Collaboration. Doctoral Thesis. Sweden, Linköping: JustNu-Tryck.

Wireman, T. (1990). World Class Maintenance Management. USA: Industrial Press Inc.

Yin, R. K. (1994). Case Study Research, Design and Methods – Second Edition. USA:Sage
Publications, Inc.

49
50
 Appended Papers

Paper I
Standardization Issues in Condition Based Maintenance
Paper II
Condition Based Maintenance System Technology – Where is Development Heading?
Paper III
Technical Design of Condition Based Maintenance System – A Case Study using Sound Analysis and
Case-Based Reasoning
Paper IV
Important Aspects to take into Consideration when Deciding to Implement Condition Based
Maintenance

51
52
 Paper I

Standardization Issues in Condition Based Maintenance

Marcus Bengtsson

PROCEEDINGS OF THE 16TH INTERNATIONAL CONFERENCE OF

CONDITION MONITORING AND DIAGNOSTIC ENGINEERING MANAGEMENT
COMADEM 2003

53
54
 STANDARDIZATION ISSUES
IN CONDITION BASED MAINTENANCE

Marcus Bengtsson

Department of Innovation, Design and Product Development,

Mälardalen University, P.O. Box 325, SE-63105, Eskilstuna, Sweden,
marcus.bengtsson@mdh.se, phone +46 16 15 34 86, fax +46 16 15 36 10

ABSTRACT

Swedish industries maintenance-related costs run into billions on a yearly basis. Research has shown that
up to 20% of the costs go to waste in non-realized revenue due to poor maintenance decisions. Both
excessive and lacking maintenance can be harmful to an asset. A technology that strives to make the right
judgements in maintenance intervals is the Condition Based Maintenance (CBM) technology. During the
last decade the development in CBM technology has increased rapidly. A variety of different products
can be found on the market. However developments are, in many cases, directed to specific applications
and the results of a prototype are not always generally reproducible. One way to direct development in
CBM technology to more general solutions can be to follow standards and standardization proposals.
Within the technology, two standards and two standardization proposals can be found. IEEE 1451 and
IEEE 1232, MIMOSA and OSA-CBM have been found to be the most interesting standards that lie
within the scope of the CBM technology.

KEYWORDS

Condition Based Maintenance, standardization, MIMOSA, OSA-CBM, IEEE 1451, IEEE 1232

INTRODUCTION

In the Condition Based Maintenance (CBM) technology, one takes condition monitoring results in
account and plans the maintenance action by it. The purpose of CBM is to eliminate breakdowns and
prolong the preventive maintenance intervals. With CBM, an increase in availability of an asset will
follow. With CBM technology one analyzes condition monitoring data in depth to determine whether the
asset is running at a normal operation condition or not. If the preset limits for normal condition are
exceeded, one also wants to know the reasons behind it and how long before a critical breakdown will
occur. With this information it will be easier to plan the maintenance actions more effectively. In
maintenance related costs, Swedish industries spend 190-200 billion SEK (~US$22-24 billion) per year.
With a more effective maintenance execution and an increase in availability, the heavy industry in
Sweden could save up to as much as 20% in non-realized revenue, due to loss of productivity when non-
effective maintenance is carried out [1].
For a completely automated CBM system, with the ability to diagnose an abnormality and calculate the
remaining useful life (RUL), new analysis techniques with Artificial Intelligence (AI), will be a necessity.
AI, such as Neural Networks, Case Based Reasoning, and Fuzzy Logic are tools for quick analysis. AI
techniques are also good for handling large amounts of data, pattern recognition, complex non-linear data
etc. An overall view of different AI techniques is presented by [2], [3] gives a tutorial in Neural
Networks.

There are numerous papers written on system development in CBM technology. A lot of them are built
with analysis tools such as Neural Network, Fuzzy Logic, Expert System and those between hybrid
solutions, see [4]-[7] for examples. The progress of prototypes of CBM systems and components are
strong but the developments are diverging in all kinds of directions. With strong standards or an
emergence of a dominant design, the developments can be directed for the benefit of both the customers
and developers/producers.

Literature overviews of technical standards in general and the standards and standardization proposals
that exist today within the CBM technology have been performed. The paper will briefly present a
standard for smart transducer interface for sensors and actuators (IEEE 1451), a standardization proposal
in CBM architecture (OSA-CBM), and a standardization proposal in communication between different
CBM modules (MIMOSA). AI-ESTATE (Artificial Intelligence and Expert System Tie to Automatic
Test Equipment), the IEEE 1232 standard will also be presented in short. The paper will discuss some of
the effects one might expect when following, or not following, standards.

STANDARDIZATION

According to [8] the purpose of standardization is defined as: to reach the optimal technical and
economical solution to recurrent problems. Standardization aims to:

o Facilitate communication by creating distinct conceptions with definitions and terms.

o Secure compatibility and interoperability through restrictions of size and weight, dimensions and
interfaces.
o Accomplish variety reduction through selection of size and weight, dimensions and designs.
o Facilitate flexibility through modularization.
o Standardize characteristics, functions, qualities and safety for products, processes, systems, and
services.
o Specify distinct testability methods.

Standardization should be looked upon as a tool for a rational, environmentally sound and energy saving
development of production, distribution and utilization of commodity, processes, systems and services.
Standardization is also a means to facilitate technology collaboration and technological transference.
According to [9], there are three different actors both issuing and following standards: individuals,
organizations, and states.

The arguments and motivations for following standards are presented by [9] in four topics.

I. Standards are an effective instrument for transportation of information. Just knowing that a potential
producer or customer is following an accepted standard means that no detailed questions need to be
asked. In some cases knowing a lot about the standard is not even important, especially if a (trusted)
third party has attested that the standard is enforced, i.e. some sort of certification.
 II. Standards constitute a method for coordination. Standards are often developed so that different
products will have compatibility, for example the electric plug and electric outlet.

III. Standards also mean simplifications, there are less to choose from (variety reduction). This
facilitates the overview and understanding of a problem.

IV. Standards are usually also motivated that they constitute the best solution (in some cases merely a
good solution, in others the only solution). This argument can actually in itself be independent -
there exists no reasons not to follow the standard, simply because it is the best alternative.

The unifying effects that often follow from standards are usually not promoted by the standard
organizations. They like to turn it around and mean that standards help variation, arguing that by
standardizing everything that already is invented, innovators can focus development on what has not yet
been developed and standardized. However, there is one important aspect to the issue of unifying by
standardization, and that is large-scale production. This was a central argument when standards first
began to emerge - standardization can almost be looked upon as a prerequisite to industrialism (e.g.
dimensions of bolts, sizes of clothing etc.) [9].

According to [10], historians, economists, and sociologists have shown that standards have played a key
role at the core of technological processes, innovation and industrial development. They also mean that
standards will play an even more significant role in the future.

One should question the positive and negative effect of a standard before embracing it completely. Will
the standards’ goals really be desirable for your organization? Will the standards lead you to your goals?
Is there any better, or perhaps an even easier way to reach the goals? There are several examples of how
standardization can create poor solutions. For example, the so called QWERTY-standard, which specifies
how the letters on the keyboard are supposed to be located was originally developed to reduce the typing
speed to prevent the type arms (in old type writers) to get hooked together. Today the problem does no
longer exist but the standard does. The standard bumper height of cars is about 440 mm (17 inches),
which happens to be almost the exact same height as the kneecaps of a normal sized human being [9].

Tassey [11] stresses that from both corporate strategy and public policy point of view, no standard should
be accepted before the technology being standardized has matured. This is especially important in
complicated systems technologies like communications or factory automation (e.g. Condition Based
Maintenance systems), where standardization should be decided upon in evolutionary steps of the
technology’s progress. The progress being set by three different factors: (1) the pace of technology
change in each component category, (2) the disembodied technology development (sets the limits for the
overall system architecture and organization), and (3) the market structure. Complete standardization too
early in a technology life cycle can put a restraint on innovation.

Utterback [12] means that in the initial cycle of a new product the customer and the producer have not
developed a sense of neither the ideal product design nor the optimal manufacturing process. Everyone is
learning (experimenting) as development moves along. Industry standards at this stage are usually
imperfect or undeveloped, if they even exist. Out of what can be seen as an experimenting chaos of
development and competition a dominant design usually emerges. The dominant design, being the
product architecture that wins the trust in the marketplace, is the one that competitors must turn to if they
hope not to lose their market share (compare the battle of Sony’s Betamax VCR versus JVC’s VHS
VCR). The dominant design usually has the effect of enforcing standardization on a product. With the
appearance of a dominant design the innovative effort shifts focus from the product’s functionality and
performance features to improving the manufacturing process, especially with regards to production
efficiency and costs.
Utterback also presents a model (Figure 1), illustrating that the changes in product and process innovation
occur in patterns with three visible phases: the fluid, the transitional, and the specific phase. In the fluid
phase the rate of major innovations are high in product, while being low in process. The products often
have diverse designs and are usually customized while the production is flexible but inefficient. In the
transitional phase the rate of major innovations subsides in products but increases in processes. The
product design becomes stable enough for at least a few to have significant production volume and the
production becomes more controlled as the changes occur in larger steps. In the specific phase the rate of
major innovation stays low on the products as well as decreases in processes. Standards are followed in
new product development and innovation is incremental with improvements focused on productivity and
quality, the production is inflexible and cost of change is high [12].

Figure 1. Model presented by Utterback [12, p. 91].

Sometime in the transitional phase, dominant designs tend to emerge and industry standards are set.
Interface standards become important to facilitate modularization (which can help variation through
compositions of different components). Poorly designed or delayed interface standards could have
negative effects on the market structure. Inadequate standards (no standards or poorly designed standards)
can increase the economic costs. The costs can range from interoperability problems in different products
to the need of having to buy several vendors versions of similar software [13].

There are different kinds of standards; one way to separate them is through the standards producers and
whether they are open or proprietary standards. Standards are usually referred to as either de facto or de
jure standards. According to [14] the words “de facto” and “de jure” are Latin and mean “according to the
circumstances” and respectively “according to the law”. Within standards the de jure standards are
recommended and produced by standard organizations while the de facto standards are proposed by the
marketplace and/or industrial consortia. To be considered as an open standard it must be published and
easily available at a low cost [15].

STANDARDIZATION IN CBM SYSTEMS

Below, a standard for smart transducer interface for sensors and actuators (IEEE 1451), a standard in AI
techniques in test equipment (IEEE 1232), a standardization proposal in CBM system architecture (OSA-
CBM) and a standardization proposal in communication between different CBM modules (MIMOSA)
will be presented. Following standards like these the CBM community would achieve interchangeable
hardware and software components, more technological choices for users, more rapid developments of
technology, reduced prices and improved ease of upgrading of system components. Interface standard that
enhance the ability of integration between different vendor-products in a system, have several effects.
Overall system costs will be reduced, performance will be optimized, and competition between different
vendors will lower prices as well as the ability to replace obsolete components in a system (at a lower
cost) will be enhanced [11]. When developers of CBM systems start to follow standards and
standardization proposals it will be easier to direct the development towards new innovative ways of
predicting remaining useful life.

IEEE 1451

At the basic level of CBM systems there are sensors or other devices to pick up the data needed for
analysing the health of an asset. This is often referred to as a distributed measurement and control system
(DMC). Due to the customers problem of integrating different vendor products (transducer, sensors and
actuators) when networking, a standard for the hardware interconnection level is needed. But there is also
need for standards in the software module of the transducers to achieve network interoperability at the
network-node level [16].

Looking to develop a standardized interface to network smart sensors the National Institute of Standards
and Technology (NIST) started to work together with the Institute of Electrical and Electronics Engineers
(IEEE) in the middle of the 1990’s. To achieve easy installation and upgrading of sensors, one should link
them together like personal computers via a local area network (LAN). Through this connection one will
be able to connect many sensors via a single cable or bus. This will mean that sensors can be detached
without affecting other sensor nodes [17].

The entire IEEE 1451 family consists of four sub-standards, IEEE 1451.1, IEEE 1451.2 and the proposed
sub-standards IEEE P1451.3 and IEEE P1451.4. According to [18] all sub-standards are complementary,
made to be used either as a family or by themselves. The benefits of the entire IEEE 1451 standard are
presented by [19]:

o Self-identification of transducers
o Self-configuration
o Easier to maintain long term self-documentation
o Easier to upgrade and maintain transducers
o An increase in data and system reliability
o Allows for transducers to be calibrated remotely or even to calibrate themselves

The IEEE 1451.2 standard specifies the transducer to microprocessor communication protocols and
transducer electronic data sheet (TEDS) formats. The IEEE 1451.2 interface defines the Smart Transducer
Interface Module (STIM). Up to 255 sensors and actuators of various digital and analog mixes can be
connected to a STIM. The STIM in its turn is connected to a network capable application processor
(NCAP) [20]. The IEEE 1451.2 also defines a Transducer Electronic Data Sheet, TEDS. According to
[16] TEDS will achieve self-identification of IEEE 1451-based sensors or actuators. This will be realized
through a memory chip physically attached to the sensor. The chip will be able to store information of:

o Manufacturer’s name
o Identification number
o Type of device
o Serial number
o Calibration data

According to [20] the risk of losing manufacturing and calibration data (transducer paper data sheet) will
decrease due to the fact that the information will be stored within the sensor or actuator.
The IEEE 1451.1 standard specifies the network capable application processor (NCAP) information
model. One key reason to standardize the interface at the hardware interconnection level is the current
compatibility problems transducer manufacturers’ face when integrating their devices into multi-vendor
networks [16]. The NCAP in 1451.1 could be looked at as a small computer that resides to a specific
network node. NCAP’s are defined as sensor network nodes. NCAP nodes allow multiple sensors to be
attached to the network using one common point of access [17].

The proposed standard IEEE P1451.3 attempts to define a means to connect the TEDS to the transducer
via a bus. Due to harsh environment this will be necessary in some applications. The P1451.3 document
proposes this should be done with a “mini-bus”, small and cheap enough to fit into a transducer. The
proposed standard IEEE P1451.4 defines a specification that will add self-describing and configuration
capabilities to analog sensors [19].

IEEE 1232

According to [21-22] the technical systems of today are more complex, costly, and difficult to diagnose
and repair. To address these problems the Diagnostic and Maintenance Control (D&MC) subcommittee
of IEEE SCC20 developed the IEEE 1232 standard family, AI-ESTATE, The Artificial Intelligence
Exchange and Service Tie to All Test Environments.

According to [22] the goals with the IEEE 1232 standard are to:

o Incorporate domain specific terminology

o Facilitate portability of diagnostic knowledge
o Permit extensibility of diagnostic knowledge
o Enable the consistent exchange and integration of diagnostic capabilities

Even before the vision of AI-ESTATE was fully developed the IEEE Std 1232-1995 was published. The
IEEE Std 1232-1995 defines the architecture of an AI-ESTATE conformant system. IEEE Std 1232.1-
1997 defines a standard for how knowledge and data exchange should be accomplished. The IEEE Std
1232.2-1998 addresses the issues in system-level diagnosis. After the IEEE 1232.2-1998 standard was
published, the standards all together were published as a “trial-use” standard. This means the standard
was not finalized and could be revised after comments from organizations trying to implement or use the
standard [21]. In 2002 the three standards merged into the today published IEEE Std 1232-2002.

OSA-CBM

OSA-CBM is an abbreviation for Open System Architecture for Condition Based Maintenance and is a
proposal for a de facto non-proprietary standard. In the mission statement from the OSA-CBM
organization it is declared that the standard proposal shall cover the whole range of functions of a CBM
system, for both hardware and software components [23]. Due to the difficulty of integrating different
vendor products most CBM system users limit the flexibility and performance of a system. The many
proprietary standards that exist today have a tendency to lock customers into a single source solution. An
accepted non-proprietary open system architecture standard would, according to the organization [15],
provide:

o Improved ease of upgrading for system components

o A broader supplier community
o More rapid technology development
o Reduced prices

The OSA-CBM proposed standard divides a CBM system into seven different layers/components [15].
Layer 1 Sensor Module: The sensor module provides the CBM system with digitized sensor or transducer
data.

Layer 2 Signal Processing: The signal processing module receives signals and data from the sensor
module or other signal processing modules. The output from the signal processing module includes
digitally filtered sensor data, frequency spectra, virtual sensor signals and other CBM features.

Layer 3 Condition Monitor: The condition monitor receives data from the sensor modules, the signal
processing modules and other condition monitors. Its primary focus is to compare data with expected
values. The condition monitor should also be able to generate alerts based on preset operational limits.

Layer 4 Health Assessment: The health assessment module receives data from different condition
monitors or from other health assessment modules. The primary focus of the health assessment module is
to prescribe if the health in the monitored component, sub-system or system has degraded. The health
assessment module should be able to generate diagnostic records and propose fault possibilities. The
diagnosing should be based upon trends in the health history, operational status and loading and
maintenance history.

Layer 5 Prognostics: The prognostic module should have the possibility to take account data from all the
prior layers. The primary focus of the prognostic module is to calculate the future health of an asset, with
account taken to the future usage profiles. The module should report the future health status of a specified
time or the remaining useful life (RUL).

Layer 6 Decision Support: The decision support module receives data from the health assessment module
and the prognostic module. Its primary focus is to generate recommended actions and alternatives. The
actions can be related to maintenance or how to run the asset until the current mission is completed
without occurrence of breakdown.

Layer 7 Presentation: The presentation module should present data from all previous modules. The most
important layers to present would be the data from the health assessment, prognostic and decision support
modules as well as alerts generated from the condition monitors. The ability to look even further down in
the layer should be a possibility. The presentation module could be built into a regular machine interface.

There exists several different means to run the communication between networks of components. The
three most popular are multicast, broadcast, and client-server. For more information, see [15]. For the
OSA-CBM the client-server approach would work best as a core of the architecture. The current
Information Technology (IT) already provides a solid backbone for CBM systems. PC’s, Networking and
the Internet are already available, cost-effective, easily implemented communications technologies for
CBM systems. If using a Web-based solution for a CBM system, HTTP servers can function as document
servers. As transportation medium either the HTML (Hypertext Markup Language) or the XML
(eXtensible Markup Language) would work. The HTML describes the content and appearance of a
document while the XML describes the information content and information relationship.

MIMOSA

The Machinery Information Management Open System Alliance, MIMOSA, was founded in 1994 and
introduced in the September issue 1995 of Sound and Vibration. In December 1996 the not-for-profit
organization was incorporated. The purpose and goal of MIMOSA is to develop open conventions for
information exchange between plant and machinery maintenance information systems. The development
of MIMOSA CRIS (Common Relational Information Schema) has been openly published at their website
[24]. The CRIS provides coverage of the information (data) that will be managed within a CBM system.
This is done by a relational database schema for machinery maintenance information. The typical
information that will need to be handled is presented by [25]:

o A description of the configuration of the system being monitored

o A list of specific assets being tracked
o A description of system functions, failure modes, and failure mode effects
o A record of logged operational events
o A description of the monitoring system and characteristics of the monitoring components
o A record of sensor data
o Resources of describing signal processing algorithms and resulting output data
o A record of alarm limits and triggered alarms
o Resources describing degradation in a system as well as prognostics of system health trends
o A record of recommended actions
o A complete record of work request

INTEGRATION OF THE ABOVE STANDARD AND STANDARDIZATION PROPOSALS

In order to achieve benefits from the IEEE 1451, OSA-CBM and MIMOSA, the standard and
standardization proposals were carefully examined by [26] for entry points or hooks that could provide
links. When OSA-CBM was developed the MIMOSA/CRIS was used as a core architectural
infrastructure for information communication. The OSA-CBM also recommended interfaces between the
standardization proposal and other already published standards. The signal module in the OSA-CBM
architecture could be built on the IEEE 1451 standard and the signal processing module could consist of
AI-ESTATE (Artificial Intelligence and Expert System Tie to Automatic Test Equipment), the IEEE
1232 standard. According to [26] it would be feasible to provide a link from the OSA-CBM to the IEEE
1451.1 with only slight modification.

CONCLUSIONS

This paper is meant to be an informative paper regarding the standards and standardization proposals that
exist within the Condition Based Maintenance technology. The author does not mean to imply that it is a
necessity to follow standards when developing new products within the technology. However, if a
standard constitutes a good solution (for new product development) it will most likely mean a stronger
and faster development. Standardized products also mean a better deal for customers. As different
vendor-products of e.g. DVD players and television sets can be interfaced, why should CBM components
be any different? Though, one has to stop and ponder how mature the technologies of CBM really are, as
Tassey remarks, standardization too early in a technology life cycle can put restrains on innovations [11].
The dominant design may not yet have emerged within the CBM technology, so according to the model
(Figure 1) presented by Utterback [12] it might yet be too soon to decide over standards. More research
and prototypes following the standards and standardization proposals need to be completed for a
resolution to the issue.

ACKNOWLEDGEMENT

The author would like to acknowledge Professor Erland Olsson of Mälardalen University, a great person,
teacher, and tutor. The author also gives thanks to EU goal2 Regional Structural Fund, Bombardier
Transportation Sweden, ABB Automation Technology Products AB, UTEK, and Länsstyrelsen in
Västmanland, who all donated sponsorship for this work.
REFERENCES

1 Ahlmann, H., R., From Traditional Practice to the New Understanding: The Significance of the
Life Cycle Profit Concept in the Management of Industrial Enterprises, “IFRIMmmm
Conference”, Växjö University, Sweden, Växjö, 2002.
2 Fenton, B., McGinnity, T.M., Maguire, L.P., Whither AI in Test and Diagnosis,
“AUTOTESTCON Proceedings, 2001”, USA, Valley Forge, PA, 2001, pp. 333-351,
http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).
3 Jain, A.K., Mao, J., Mohiuddin, K.M., Artificial Neural Networks: A Tutorial, “Computer”, 1996,
Vol. 29, Issue 3, pp. 31-44, http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).
4 Rao, M., Sun, X., Feng, J., Intelligent System Architecture for Process Operation Support, “Expert
Systems with Applications”, 2000, Vol. 19, Issue 4, pp. 279-288, www.sciencedirect.com/ (2003-
03-10).
5 Hu, W., Starr, A., Zhou, Z., Leung, A., An Intelligent Integrated System Scheme for Machine Tool
Diagnostics, “International Journal of Advanced Manufacturing Technology”, 2001, Vol. 18, Issue
11, pp. 836-841, http://link.springer.de/search.htm (2003-03-10).
6 Hadden, G.D., Bergstrom, P., Samad, T., Bennett, B.H., Vachtsevanos, G.J., Van Dyke, J.,
Application Challenges: System Health Management for Complex Systems, “IPDPS 2000
Workshops”, 2000, pp. 784-791, http://link.springer.de/search.htm (2003-03-10).
7 Yam, R. C. M., Tse, P. W., Li, L., Tu, P. Intelligent Predictive Decision Support System for
Condition-Based Maintenance, “International Journal of Advanced Manufacturing Technology”,
2001, Vol. 17, Issue 5, pp. 383-391, http://link.springer.de/search.htm (2003-03-10).
8 Engström, C. (Chief Editor) “Nationalencyklopedin”, 17th volume, Bokförlaget Bra Böcker AB,
Höganäs, 1995.
9 Brunsson, N. and Jacobsson, B., “Standardisering” Nerenius & Santérus Förlag AB, Stockholm,
1998.
10 Hawkins, R., Mansell, R., Skea, J. (Editors) “Standards, Innovation and Competitiveness” Edward
Elgar Publishing Limited, Cornwall, 1995.
11 Tassey, G., “Technology Infrastructure and Competitive Position” Kluwer Academic Publishers,
Norwell, USA, 1992.
12 Utterback, J. M., “Mastering the Dynamics of Innovation”, Harvard Business School Press
(Pocket version), Boston, 1996.
13 Tassey, G., Standardization in technology-based markets, “Research Policy”, 2000, Vol. 29, Issue
4-5, pp.587-602, www.sciencedirect.com/ (2003-03-10).
14 Marklund, K. (Chief Editor) “Nationalencyklopedin”, 4th volume, Bokförlaget Bra Böcker AB,
Höganäs, 1990.
15 Thurston, M. G., An Open Standard for Web-Based Condition-Based Maintenance Systems,
“AUTOTESTCON Proceedings, 2001”, USA, Valley Forge, PA, 2001, pp. 401-415,
http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).
16 Lee, K. and Schneeman, R., Distributed Measurement and Control Based on the IEEE 1451 Smart
Transducer Interface Standards, “IEEE Transactions on Instrumentation and Measurement”, 2000,
Vol. 49, Issue 3, pp. 621-627, http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).
17 Gilsinn, J. and Lee, K., Wireless Interfaces for IEEE 1451 Sensor Networks, “SIcon’01 Sensor
for Industry Conference”, USA, Rosemont, IL, 2001, pp. 45-50,
http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).
18 Potter, D., Implementation of a Plug and Play Sensor System Using IEEE P1451.4, “SIcon’01
Sensor for Industry Conference”, USA, Rosemont, IL, 2001, pp. 162-166,
http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).
19 Conway, P., Heffeman, D., O’Mara, B., Burton, P., Miao, T., IEEE 1451.2: An Interpretation and
Example Implementation, “Instrumentation and Measurement Technology Conference”, USA,
Baltimore, MD, 2000, pp. 535-540, http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).
20 Lee, K., IEEE 1451: A Standard in Support of Smart Transducer Networking, “Instrumentation
and Measurement Technology Conference”, USA, Baltimore, MD, 2000, pp. 525-528,
http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).
21 Sheppard, J. W. and Kaufman, M., Formal Specification of Testability Metrics in IEEE P1522,
“AUTOTESTCON Proceedings, 2001”, USA, Valley Forge, PA, 2001, pp. 71-82,
http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).
22 IEEE Std 1232-2002, IEEE Standard for Artificial Intelligence Exchange and
Service Tie To All Test Environments (AI-ESTATE), 2002, pp. 01-113,
http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).
23 www.osacbm.org (2003-03-10)
24 www.mimosa.org (2003-03-10)
25 Thurston, M. and Lebold, M., Open Standards for Condition-Based Maintenance and Prognostic
Systems, “MARCON”, 2001, www.osacbm.org (2003-03-10).
26 Lee, K., Gao, R. X., Schneeman, R., Sensor Network and Information Interoperability –
Integrating IEEE 1451 with MIMOSA and OSA-CBM, “IEEE Instrumentation and Measurement
Technology Conference”, USA, Anchorage, AK, 2002, pp. 1301-1305,
http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).

BIBLIOGRAPHY

IEEE Std 1451.1-1999, IEEE Standard for a Smart Transducer Interface for Sensors and Actuators -
Network Capable Application Processor (NCAP) Information Model, 2000, pp. 01-349,
http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).

IEEE Std 1451.2-1997, IEEE standard for a smart transducer interface for sensors and actuators -
transducer to microprocessor communication protocols and Transducer Electronic Data Sheet (TEDS)
formats, 1998, pp. 01-120, http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).
 Paper II

Condition Based Maintenance System Technology –

Where is Development Heading

Marcus Bengtsson

PROCEEDINGS OF THE 17TH EUROPEAN MAINTENANCE CONFERENCE

EUROMAINTENANCE 2004

55
56
Condition Based Maintenance System Technology –
Where is Development Heading?
Marcus Bengtsson, Mälardalen University, Sweden

Keywords: Condition Based Maintenance, standards, future development

Abstract: Examining the standards and standardization proposals within the Condition
Based Maintenance (CBM) technology and specific within CBM systems, conclusions might
be drawn as to where the technical development are or perhaps should be heading.
Reviewing the standardization proposals promoted by the organizations of MIMOSA and
OSA-CBM and the published standards IEEE Std 1451, IEEE Std 1232, and ISO 13373-1 it
can be concluded that the future technical development might or should be heading towards
prognostication and calculations of remaining useful life.

The OSA-CBM divides a CBM system into seven different modules: sensor; signal
processing; condition monitoring; health assessment (diagnostics); prognostics; decision
support; and the presentation module. If compared to the standards above it can be
concluded that the sensor module could be developed using the IEEE Std 1451, for signal
processing several standard means are available, the condition monitor could be developed
using the ISO 13373-1, and the diagnosis module could be developed partly using the ISO
13373-1 and IEEE 1232. It might thus be time to start directing the future technical
development and research towards what has not yet been standardized, finding effective
ways to perform condition prognosis and calculations of remaining useful lifetime. Through
literature and standard reviews this paper will give a discussion of where the research within
CBM systems might be heading.

1. Introduction
Maintenance is traditionally performed in either time based (or distance based) fixed
intervals, so called preventive maintenance, or by corrective maintenance. With the
preventive approach, maintenance is performed in order to prevent equipment breakdown by
performing repair, service, or components exchange. With the corrective approach,
maintenance is performed after a breakdown or when an obvious fault has occurred, for
some equipment and faults the maintenance action must be performed immediately, for
others the maintenance action can be deferred in time, all depending on the equipments
function. In the Swedish standard SS-EN 13306 [2001] it can be seen that also the
preventive maintenance has been divided into two categories, Condition Based Maintenance
(CBM) and predetermined maintenance. The predetermined is scheduled in time, while the
condition based can have dynamic or on request intervals (see figure 1).
Figure 1. Overview of the Swedish maintenance terminology standard SS-EN 13306.

CBM has been defined as “Maintenance actions based on actual condition (objective
evidence of need) obtained from in-situ, non-invasive tests, operating and condition
measurement.” [Mitchell, 1998]. The main point being that the assets condition is assessed
under operation with the intention of making conclusions to whether it is in need of
maintenance or not and if so at what time does the maintenance actions needs to be
executed not to suffer a breakdown or malfunction. The degree of automation in assessing
the condition can vary from human visual inspection to fully automated systems with sensors,
data manipulation, condition monitoring, diagnosis, and prognosis.

The development within CBM system technology is strong but diverging. Following standards
can be one way to direct the developments towards more general solutions [Bengtsson,
2003]. Several general standards and standardization proposals within CBM system
technology can be found as well as several standards on more modular levels. Another way
to direct development could be to review previous research. If combining these two
approaches it might be possible to draw some conclusions as to where development and
research within the CBM system technology might or/and should be heading. The objective
with this paper is thus to try to look ahead and see where the future development within CBM
system technology will have its focus. The methods used to solve the objective have been to
study secondary data in form of a literature review and a literature search. The secondary
data has been collected from reviewed scientific journals and conference proceedings (in
both printed and database versions), the internet, and standards. The literature has been
reviewed from its scientific value and sources with unprofessional look have been rejected.
The method does not give a certain result; merely tendencies can be drawn as conclusions.

2. Technical Standards
Generally speaking, technical standards are to be looked upon with the purpose of reaching
the optimal technical and economical solution to recurrent problems, the aims of
standardizing can be presented as to [Engström, 1995]:
• Facilitate communication by creating distinct conceptions with definitions and terms.
• Secure compatibility and interoperability through restrictions of size and weight,
dimensions and interfaces.
• Accomplish variety reduction through selection of size and weight, dimensions and
designs.
• Facilitate flexibility through modularization.
• Standardize characteristics, functions, qualities and safety for products, processes,
systems, and services.
• Specify distinct testability methods.
According to Brunsson [1998] the motivations and arguments for following standards can be
presented in four topics:
• Standards are an effective instrument for transformation of information.
• Standards constitute a method for coordination.
• Standards mean simplifications.
• Standards usually constitute the best solution.

2.1 CBM System Technology Standard Review

The pace of development in CBM system technology has increased rapidly the past years.
Prototypes and products can easily be found in scientific journals, conference proceedings,
on the internet, and in other media. To direct development and to make the results of the
development more general, several standards and standardization proposals have been
developed within the system technology.

2.1.1 OSA-CBM
The Open System Architecture for Condition Based Maintenance organization (OSA-CBM)
have developed a de facto standard that encompasses all the rage of functions a CBM
system need to consist of, from sensing hardware through diagnosis and prognosis all the
way to presentation of the asset condition with recommended maintenance actions. If
accepted as a non-proprietary standard, it will result in a free market for CBM components; it
will be easier to upgrade system components, there will be a broader supplier community,
more rapid technology developments, and reduced prices [www.osacbm.org].

The OSA-CBM divides a CBM system in seven different layers (components), with a modular
solution. The OSA-CBM standard includes more than the system architecture of a CBM
system, e.g. means of communication, explaining the whole standard is out of the scope of
this paper though, for more information see [Thurston, 2001]; [www.osacbm.org]. Here will
just a brief explanation of the architecture proposal follow.

1. Sensor Module: The sensor module provides the system with digitized sensor or
transducer data. The sensor module could be developed using the published standard
IEEE Std 1451 (see 2.1.2 for more information and references).

2. Signal Processing Module: The signal processing module receives data from the sensors
or transducers or other signal processors and performs signal transformations and
feature extractions. Its output data includes digitally filtered sensor data, frequency
spectra, virtual sensor signals, and other CBM features. There exists many standard
means of performing signal processing, such as filtering, spectrum analysis, multi-
resolution decomposition etc. [Wen et.al, 2000].

3. Condition Monitoring Module: The condition monitoring module receives data from the
sensor modules, the signal processing modules and other condition monitoring modules.
The condition monitoring module compares on-line data with expected values; if
necessary it should also be able to generate alerts based on preset operational limits.
The condition monitor could be developed using the published standard ISO 13373-1
(see 2.1.3 for more information and references).

4. Health Assessment Module: The health assessment module receives data from condition
monitors and other health assessment modules. The health assessment determines if the
condition of the monitored system, sub-system, or component has degraded. The module
also generates a diagnostic record and suggests fault possibilities. The health
 assessment module could be developed using the published standard IEEE 1232 (see
2.1.4 for more information and references) and ISO 13373-1.

5. Prognostic Module: The prognostic module predicts the future condition of the monitored
system, sub-system or component. The module should be able to require data from all
the previous modules all depending on the modelling approach.

6. Decision Support Module: The decision support module receives data from the health
assessment module and the prognostic module. Its primary function is to provide
recommended maintenance actions or alternatives on how to run the system, sub-
system, or component until completed mission.

7. Presentation Module: The presentation module receives data from all previous modules;
the most important are the data from the health assessment, the prognostic, and the
decision support module. The presentation module could be built into a regular machine
interface.

When analyzing the OSA-CBM standard it clearly shows that some of the different modules
can be standardized. Several module standards within the CBM system technology are
available on the market and if used, development might become more directed and even
somewhat simplified.

2.1.2 IEEE Std 1451

The IEEE Std 1451, a standard for smart transducer interface for sensors and actuators, has
been developed since the middle of the 1990’s. NIST (National Institute of Standards and
Technology) started to work together with the IEEE (Institute of Electrical and Electronics
Engineers) to develop a standard to achieve, among other, easy installation and upgrading of
sensors [Gilsinn & Lee, 2001]. The IEEE Std 1451 are an entire family of standards,
consisting of four sub-standards, according to Potter [2001] all complementary, made to be
used as a family or by themselves. Benefits of the complete IEEE Std 1451 are presented by
Conway et.al [2000]:
• Self-identification of transducers
• Self-configuration
• Easier to maintain long term self-documentation
• Easier to upgrade and maintain transducers
• An increase in data and system reliability
• Allows for transducers to be calibrated remotely or even to calibrate themselves

For more information on the IEEE Std 1451 see [IEEE Std 1451.1-1999, 2000] and [IEEE Std
1451.2-1997, 1998].

2.1.3 ISO 13373-1

The ISO 13373-1 - Condition monitoring and diagnostics of machines, provides general
guidelines for measuring and data collection functions for machine condition monitoring,
focusing on machine vibrations. The standard was developed to ensure consistency in
measurement procedures and practices and contains recommendation of following topics [Ali
et.al, 2003]:
• Measurement methods
• Measurement parameters
• Transducer selection
• Transducer location
• Transducer attachment
• Data collection
• Machine operating condition
• Vibration monitoring systems
• Interfaces with data-processing systems
• Continuous monitoring
• Periodic monitoring

For more information on ISO 13373-1:2002 see [ISO 13373-1:2002, 2002]

2.1.4 IEEE Std 1232

The IEEE 1232, The Artificial Intelligence Exchange and Service Tie to All Test Environment,
AI-ESTATE, were developed by the Diagnostic and Maintenance Control (D&MC)
subcommittee of IEEE SCC20. The purpose of the standards is to “…provide formal models
of diagnostic information to ensure unambiguous access to an understanding of the
information supporting system test and diagnosis” [IEEE Std 1232-2002, 2002]. The goals of
the IEEE Std 1232 are to:
• Incorporate domain specific terminology
• Facilitate portability of diagnostic knowledge
• Permit extensibility of diagnostic knowledge
• Enable the consistent exchange and integration of diagnostic capabilities

For more information on the IEEE Std 1232 see [IEEE Std 1232-2002, 2002]

2.1.5 MIMOSA
MIMOSA, the Machine Information Management Open System Alliances, is a non-profit
organization that has developed open conventions for information exchange between plant
and machinery maintenance information systems. The relationship-based platform is called
the Open System Architecture for Enterprise Application Integration (OSA-EAI) with the core
of the Common Relation Information Schema, CRIS. The CRIS enables communication of
equipment diagnostic and prognostics. The specification, CRIS Version 2.2, is openly
published at the MIMOSA website (www.mimosa.org) [Kahn, 2003]. The typical information
that will need to be handled, presented by Thurston & Lebold [2001]:
• A description of the configuration of the system being monitored
• A list of specific assets being tracked
• A description of system functions, failure modes, and failure mode effects
• A record of logged operational events
• A description of the monitoring system and characteristics of the monitoring components
• A record of sensor data
• Resources of describing signal processing algorithms and resulting output data
• A record of alarm limits and triggered alarms
• Resources describing degradation in a system as well as prognostics of system health
trends
• A record of recommended actions
• A complete record of work request

2.1.6 Integration of CBM system technology standards

The OSA-CBM was developed using the MIMOSA CRIS as a core architectural infrastructure
for information communication. Lee et.al [2002] shows that with slight modification, the OSA-
CBM and the IEEE1451.1 can be integrated. Lebold et.al [2002] states that the OSA-CBM
was developed around input from existing and emerging standards within CBM system
technology, such as the MIMOSA CRIS, IEEE 1451, and the AI-ESTATE.
2.2 Result of Standard Review
To be able to develop a standard or a standardization proposal of something as complex and
technical as a CBM system module, a great deal of research must have been performed. If
looking at the standards and standardization proposals that exist today within the CBM
system technology, it can be concluded that the sensor module, the signal processing
module, the condition monitoring module, and the diagnostic module can all be partially
developed using standards or standard means. Conclusions that can be drawn from the
CBM system technology standard review is that it is time for researchers and developers of
CBM system technology to start focus research on the next steps and modules of CBM
system, namely prognosis and decision support.

Figure 2. Simplified schematic figure of the modules of OSA-CBM, highlighting the module
standards that can be used when developing CBM systems.

3. Literature search
In order to see what has been published within maintenance, related to the modules
discussed in section 2.1.1, a literature search was conducted. The all technical IEEE Xplore®
database was used as a reference [http://ieeexplore.ieee.org/Xplore/DynWel.jsp]. The
database contains IEEE journals, transactions, and magazines, IEEE conference
proceedings, IEE journals, IEE conference proceedings, and current IEEE standards, all
included over 976’857 documents available (2003-12-17).

The purpose of the literature search was to visualize possible trends in existing number of
publications in respect to the different modules within a CBM system. The search was
conducted yearly from 1993 to 2002. The solely keyword “maintenance” was searched for all
years in order to get a reference point, then the keywords “maintenance” and “module
keywords” was searched together, one layer of the OSA-CBM at a time (see appendix, table
1, for results and all used keywords). A percentage of the number of hits with the keywords
“maintenance” and “module keywords” was calculated in respect to the hits of the solely
keyword “maintenance”, the results was graphed for visualization (see section 3.1).

3.1 Results of the literature search

The literature search shows a tendency of a trend in percentage of publications in the area of
prognosis and condition monitoring. Other modules percentages of publications are more
spread per year and it is hard to see any tendencies of possible trends.
Figure 3. Graph of the percentage of the keyword “maintenance” and “module keywords” in
respect to the solely keyword “maintenance”, a tendency of trend is discernible on the
condition monitoring module as well as on the prognostic module (see graph below).

Figure 4. Graph of the percentage of the keyword “maintenance” and “module keywords” of
the prognostic module in respect to the solely keyword “maintenance”, a tendency of trend is
discernible.

4. Conclusions
By comparing the architecture standardization proposal of Open System Architecture for
Condition Based Maintenance with different module standards, some conclusions for future
development within the CBM system technology can be drawn. To write a standard proposal
for a CBM system module, years of research must have been conducted, consequently,
research and development on the modules that have been standardized must have been
comprehensive. To standardize the prognostic module is perhaps more difficult than other
modules, due to the great differences in the processes one are trying to prognosticate. The
analyses of the module standards can not tell us if it would be smart to try to standardize the
prognostic module, but it can tell us that a lot of research has been made on the foregoing
modules. Though, not saying further research on those modules are unimportant, merely
hinting that future development in CBM system technology might perhaps focus more on the
prognostication issue. To develop an automated CBM system all modules accounted for in
the OSA-CBM standardization proposal must be included. With the literature search in mind,
this might lead to the conclusion that research on the prognostic module might not have
increased in respect to the other modules but increased research in automated CBM
systems has made the prognostication issue more visible, this leading to increased number
of hits of publications on the prognostic module.

Conclusions that can be drawn of the secondary data that has been collected and analyzed
in this study can thus tell us that the complexity and the grade of automation in CBM system
has for the past ten years increased and that it in the future most likely will increase even
more. New methods and techniques, such as Artificial Intelligence (AI), has made this a
possibility and with the new hybrids of different AI methods and techniques that are now
emerging even more automated systems are to be expected. An interesting issue are now
coming into light – If the systems are becoming improved, more complex, more automated,
and thus easier to use, why is not the CBM technology implemented at more processes,
factories, vehicles etc.? One hypothesis of why this might be so, might be that the necessary
change in the organization (to implement CBM technology) has not yet happened at many
companies or the change is occurring over a long time. Accepted implementation strategies,
as can be found in Total Productive Maintenance [Nakajima, 1998] and Reliability Centred
Maintenance [Backlund, 2003], seems to be missing within the CBM technology and makes
an interesting future research objective. In addition to the implementation strategy, an
interesting research issue to make the CBM technology more accepted and used within
industry might be to highlight and visualize the technical and organizational aspects and
advantages of CBM system technology.

Acknowledgement
This work was supported by The Knowledge Foundation (KK-Stiftelsen) through research
project 2002/0232, EU goal2, Bombardier Transportation, ABB Automation Technology
Products AB, and the Swedish Maintenance Society (UTEK).

References:

1. Ali, A-H., Starr, A., Doyle, A., “International Standards in Maintenance Data: A Review”,
COMADEM 2003 Proceedings of the 16th International Congress, August 27-29, 2003,
Växjö University, Sweden, Edited by Shrivastav, O. and Al-Najjar, B., Växjö University
Press, ISBN 91-7636-376-7, pp. 671-681.
2. Backlund, F., “Reliability-centred maintenance: identification of management and
organisational aspects when introducing RCM”, Doctoral thesis, Luleå Univeristy of
Technology, ISSN 1402-1544, 2003:16.
3. Bengtsson, M., “Standardization Issues in Condition Based Maintenance”, COMADEM
2003 Proceedings of the 16th International Congress, August 27-29, 2003, Växjö
University, Sweden, Edited by Shrivastav, O. and Al-Najjar, B., Växjö University Press,
ISBN 91-7636-376-7, pp. 651-660.
4. Brunsson, N. and Jacobsson, B., “Standardisering” Nerenius & Santérus Förlag AB,
Stockholm, 1998.
5. Engström, C. (Chief Editor), “Nationalencyklopedin”, 17th volume, Bokförlaget Bra Böcker
AB, Höganäs, 1995.
6. Gilsinn, J. and Lee, K., “Wireless Interfaces for IEEE 1451 Sensor Networks”, SIcon’01
Sensor for Industry Conference, USA, Rosemont, IL, 2001, pp. 45-50,
http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).
7. http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-12-17)
8. IEEE Std 1232-2002, IEEE Standard for Artificial Intelligence Exchange and
Service Tie To All Test Environments (AI-ESTATE), 2002, pp. 01-113,
http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).
9. IEEE Std 1451.1-1999, IEEE Standard for a Smart Transducer Interface for Sensors and
Actuators - Network Capable Application Processor (NCAP) Information Model, 2000, pp.
01-349, http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).
10. IEEE Std 1451.2-1997, IEEE standard for a smart transducer interface for sensors and
actuators - transducer to microprocessor communication protocols and Transducer
Electronic Data Sheet (TEDS) formats, 1998, pp. 01-120,
http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).
11. ISO 13373-1:2002, Condition Monitoring and Diagnostics of Machines – Vibration
Condition Monitoring – Part 1: General Procedures, 2002.
12. Kahn, J., D., “Overview of MIMOSA and the Open System Architecture for Enterprise
Application Integration”, COMADEM 2003 Proceedings of the 16th International
Congress, August 27-29, 2003, Växjö University, Sweden, Edited by Shrivastav, O. and
Al-Najjar, B., Växjö University Press, ISBN 91-7636-376-7, pp. 661-670.
13. Lebold, M., Reichard, K., Byrington, C. S., Orsagh, R., “OSA-CBM Architecture
Development with Emphasis on XML Implementations”, MARCON 2002,
www.osacbm.org (2003-12-18).
14. Lee, K., Gao, R. X., Schneeman, R., “Sensor Network and Information Interoperability –
Integrating IEEE 1451 with MIMOSA and OSA-CBM”, IEEE Instrumentation and
Measurement Technology Conference, USA, Anchorage, AK, 2002, pp. 1301-1305,
http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).
15. Mitchell, J. S., “Five to ten year vision for CBM, ATP Fall Meeting – Condition Based
Maintenance Workshop”, USA, Atlanta, GA, 1998 (PowerPoint Presentation),
http://www.atp.nist.gov/files/3 (2003-10-08).
16. Nakajima, S., “Introduction to TPM”, Productivity Press, Cambridge, Massachusetts,
USA, ISBN 0-975299-23-2, 1988.
17. Swedish Standards Institute, “Maintenance terminology”, SS-EN 13305, 2001.
18. Thurston, M. and Lebold, M., “Open Standards for Condition-Based Maintenance and
Prognostic Systems”, MARCON, 2001, www.osacbm.org (2003-03-10).
19. Thurston, M. G., “An Open Standard for Web-Based Condition-Based Maintenance
Systems”, AUTOTESTCON Proceedings, 2001, USA, Valley Forge, PA, 2001, pp. 401-
415, http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-03-10).
20. Wen, F., Willett, P., Deb, S., “Signal Processing and Fault Detection with Application to
CH-46 Helicopter Data”, Aerospace Conference Proceedings, USA, Big Sky, MT, 2000,
pp. 15-26, http://ieeexplore.ieee.org/Xplore/DynWel.jsp (2003-12-13).
21. www.mimosa.org (2003-12-13)
22. www.osacbm.org (2003-12-13)

Marcus Bengtsson, M.Sc. (PhD student)

Department of Innovation, Design and Product Development
Mälardalen University
P.O. Box 325, SE-63105, Eskilstuna, Sweden
phone +46 16 15 34 86
fax +46 16 15 36 10
e-mail: marcus.bengtsson@mdh.se
Appendix

Table 1. The number of hits (yearly) of the keyword “maintenance” with “module keywords” of
the different modules, number of hits (yearly) of the solely keyword “maintenance”, and a
calculated percentage between the two.

*The search was conducted with more keywords than the ones in the table;
maintenance and (sensors or actuators or transducers)
maintenance and (signal processor or signal processing)
maintenance and (condition monitor or condition monitoring)
maintenance and (diagnostic or diagnostics or diagnosis or health assessment)
maintenance and (prognostic or prognostics or prognosis or prediction or remaining useful
life)
 Paper III

Design of Condition Based Maintenance System –

A Case Study using Sound Analysis and Case-Based Reasoning

Marcus Bengtsson, Erik Olsson, Peter Funk, and Mats Jackson

PROCEEDINGS OF THE 8TH CONFERENCE OF

MAINTENANCE AND RELIABILITY
MARCON 2004

57
58
 Technical Design of Condition Based Maintenance System
-A Case Study using Sound Analysis and Case-Based Reasoning

Marcus Bengtsson1, Erik Olsson2, Peter Funk2, Mats Jackson1

1
Department of Innovation, Design and Product Development
2
Department of Computer Science and Engineering
Mälardalen University
Box 325, SE-631 05 Eskilstuna, Sweden
{marcus.bengtsson; erik.olsson; peter.funk; mats.jackson}@mdh.se
1
phone +46 16 153 486, fax +46 16 153 610

Abstract
Productivity is a key weapon for manufacturing companies to stay competitive in a continuous
growing global market. Increased productivity can be achieved through increased availability. This has
directed focus on different maintenance types and maintenance strategies. Increased availability
through efficient maintenance can be achieved through less corrective maintenance actions and more
accurate preventive maintenance intervals. Condition Based Maintenance (CBM) is a technology that
strives to identify incipient faults before they become critical which enables more accurate planning of
the preventive maintenance. CBM can be achieved by utilizing complex technical systems or by
humans manually monitoring the condition by using their experience, normally a mixture of both is
used. Although CBM holds a lot of benefits compared to other maintenance types it is not yet
commonly utilized in industry. One reason for this might be that the maturity level in complex
technical CBM system is too low. This paper will acknowledge this possible reason, although not
trying to resolve it, but focusing on system technology with component strategy and an open approach
to condition parameters as the objective is fulfilled. This paper will theoretically discuss the technical
components of a complete CBM system approach and by a case study illustrate how a CBM system
for industrial robot fault detection/diagnosis can be designed using the Artificial Intelligence method
Case-Based Reasoning and sound analysis.

Introduction
Industrial competition today is truly global with fragmented markets and customers expecting to get
the best product at the best price with immediate availability. Success in manufacturing, and indeed
survival, is increasingly more difficult to ensure and it requires continuous development and
improvement of the way products are produced. Meeting customer demands require a high degree of
flexibility, low-cost/low-volume manufacturing skills, and short delivery times. These demands make
manufacturing performance a strategic weapon for competition and future success. This view is
supported by Rolstadås who state that many managers believe that the greatest potential for
improvement of competitiveness lies in better production management (Rolstadås, 1995).

One important weapon in securing the productivity is to have a well functioning maintenance
organization. The maintenance organization in a company probably has one of the most important
functions, looking after assets and keeping track of equipment in order to secure productivity. With no
or a poor maintenance organization a company will loose a lot of money due to lost production
capacity, cost of keeping spare parts, quality deficiencies, damages for absent or late deliveries etc.

Today, most maintenance actions are carried out by either the predetermined preventive- or the
corrective approach. The predetermined preventive approach has fixed maintenance intervals in order
to prevent components, sub-systems or systems to degrade. Corrective maintenance is performed after
an obvious fault or breakdown has occurred. Both approaches have shown to be costly in many
applications due to e.g. lost production, cost of keeping spare parts, quality deficiencies etc. Since a
few decades some industries have started to perform maintenance action in a predictive manner, where
the assets condition is the key parameter to set the maintenance intervals and appropriate maintenance
tasks. The condition can be assessed through different levels of automation, from human visual
inspection, to condition monitoring of e.g. vibration-levels (with human diagnosis and prognosis) all
the way to completely automated Condition Based Maintenance (CBM) systems.

On-line, semi- or fully automated CBM systems has not been widely accepted within Swedish
industry. The reason for this can come from many different sources. The maturity level within
complex technical systems might be too low. The fear of investing a lot of money without knowing
exactly what will come out of it might be yet another reason. The methods and techniques to diagnose
faults might also be on a too abstract level. This paper will not try to resolve the issue of what reason
might be the biggest, although taking the aspect of condition parameters and system architecture into
context as the objective is fulfilled. The objective of this paper is to describe and (by a case study)
illustrate the necessary technical components of a CBM system. The paper contains a short theoretical
frame of reference (covering maintenance and Condition Based Maintenance in particular); a
theoretical discussion of technical components within a CBM system; a case study illustrating the
technical design and components of a CBM system for industrial robot fault detection, using Case-
Based Reasoning and sound analysis; and conclusions.

Theoretical frame of reference

Maintenance is traditionally performed in either time based (or distance based) fixed intervals, so
called preventive maintenance, or by corrective maintenance. With the preventive approach,
maintenance is performed in order to prevent equipment breakdown and do this by performing repair,
service or components exchange. With the corrective approach, maintenance is performed after a
breakdown or when an obvious fault has occurred, for some equipment the maintenance action must
be performed immediately, for others the maintenance action can be deferred in time, all depending on
the equipments function. In the Swedish standard SS-EN 13306 (2001) one can see that also the
preventive maintenance have been divided into two categories (see Fig. 1), Condition Based
Maintenance and predetermined maintenance. The predetermined is scheduled in time were as the
condition based can have dynamic or on request intervals. The Condition Based Maintenance is
sometimes referred to as predictive maintenance, see figure 2 for strengths and weaknesses of the
different maintenance types.

Figure 1. Overview of the different maintenance types (SS-EN 13306, 2001).

Figure 2. Strengths and weaknesses of different maintenance types.

As early as 1978, Nowlan and Heap (1978) presented a study of conditional-probability curves of
United Airlines aircraft components. The study showed that the conditional-probability curves fell into
six different patterns (see Fig. 3) were only 4% of the components fell into the commonly known
bathtub curve. Further, it showed that only a total of 6% of the components showed a well-defined
wear out region, another 5% had no well-defined wear out region but it was visible that the probability
of failure was higher as age increased. This means that 89% of the tested components had no wear out
region; therefore the performance of the components can not be improved by introduction of an age
limit. Nowlan and Heap also concludes that the failure rate of a component is not a very important
characteristic within maintenance programs; although a good figure for setting up maintenance
intervals it tells nothing of “…what tasks are appropriate or the consequences that dictate their
objective.” (p 48). Corresponding conditional-probability curves for the manufacturing industry is
presented by (www.wmeng.co.uk) and it is estimated that 30% of all components have well-defined
wear out regions, consequently 70% does not. Evidently, the ageing feature of a component is not the
best approach of deciding appropriate maintenance tasks, introducing Condition Based Maintenance is
one solution to the issue.

Figure 3. The six different conditional-probability curves generated by the United Airlines aircraft
study.
Condition Based Maintenance (CBM) has been defined as “Maintenance actions based on actual
condition (objective evidence of need) obtained from in-situ, non-invasive tests, operating and
condition measurement.” (Mitchell, 1998). Butcher (2000), defines the maintenance technology as
“CBM is a set of maintenance actions based on real-time or near-real time assessment of equipment
condition which is obtained from embedded sensors and/or external tests & measurements taken by
portable equipment.”. Moya and Vera (2003) defines that the purpose of a CBM Program is to
“…improve system reliability and availability, product quality, security, best programming of
maintenance actions, reduction of direct maintenance costs, reduction of energy consumption,
facilitates certification and ensures the verification of the requisites of the standard ISO 9000.”. The
Swedish maintenance terminology standard SS-EN 13306 (2001) defines CBM as “Preventive
maintenance based on performance and/or parameter monitoring and the subsequent actions.”. More
similar quotes can be found in literature and on the internet, the common point being that maintenance
actions are not considered until there is an obvious need, which will increase the availability of an
asset as well as lower maintenance cost (labour and spare parts). CBM systems (programs) will also
increase quality and improve environmental aspects.

Condition Based Maintenance System Architecture

In order for a system to achieve full potential as a Condition Based Maintenance system, it needs to be
constructed of a number of different functional capabilities. The Open System Architecture for
Condition Based Maintenance organization (OSA-CBM) has specified an open standard proposal how
a CBM system should be designed technically. The OSA-CBM is an industry consortium that includes
industrial, commercial, and military participants, the Applied Research Laboratory at Penn State, and
the MIMOSA (Machinery Information Management Open System Alliances) are two of the team
participants. The open, non-proprietary, standard proposal was developed in order to create a free
market for CBM components, where users of CBM technology will be able to choose CBM
components from different manufactures. The organization has divided a CBM system into seven
different technical modules (Thurston, 2001) (see Fig. 4). The standard proposal covers more than the
technical design of CBM systems, e.g. means of communication within the system etc., this paper
though, will solely focus on the architecture design.

Layer 1 Sensor Module: The sensor module provides the CBM system with digitized sensor or
transducer data.

Layer 2 Signal Processing: The signal processing module receives signals and data from the sensor
module or other signal processing modules. The output from the signal processing module includes
digitally filtered sensor data, frequency spectra, virtual sensor signals and other CBM features.

Layer 3 Condition Monitor: The condition monitor receives data from the sensor modules, the signal
processing modules and other condition monitors. Its primary focus is to compare data with expected
values. The condition monitor should also be able to generate alerts based on preset operational limits.

Layer 4 Health Assessment: The health assessment module receives data from different condition
monitors or from other health assessment modules. The primary focus of the health assessment
module is to prescribe if the health of the monitored component, sub-system or system has degraded.
The health assessment module should be able to generate diagnostic records and propose fault
possibilities. The diagnosing should be based upon trends in the health history, operational status and
loading and maintenance history.

Layer 5 Prognostics: The prognostic module should have the possibility to take account data from all
the prior layers. The primary focus of the prognostic module is to calculate the future health of an
asset, with account taken to the future usage profiles. The module should report the future health status
of a specified time or the remaining useful life (RUL).
Layer 6 Decision Support: The decision support module receives data from the health assessment
module and the prognostic module. Its primary focus is to generate recommended actions and
alternatives. The actions can be related to maintenance or how to run the asset until the current mission
is completed without occurrence of breakdown.

Layer 7 Presentation: The presentation module should present data from all previous modules. The
most important layers to present would be the data from the health assessment, prognostic and
decision support modules as well as alerts generated from the condition monitors. The ability to look
even further down in the layer should be a possibility. The presentation module could be built into a
regular machine interface.

When studying scientific reports and papers one can see that several developments within CBM
systems more or less have followed the OSA-CBM architecture, giving the proposal positive
feedback. Garga et.al (2001) has used a CBM system approach with sensors, data processing, fault
classification techniques, and for some features prognostic models. Discenzo et.al (1999) presents a
nine step hierarchy of intelligent machines with data acquisition, monitor, detect, diagnose, prognosis,
prognostics & control, system-level prognosis & control, dynamic optimization/multi-objective
control, and adaptive/reconfigurable.

Figure 4. The seven modules in the OSA-CBM architecture standard proposal as presented in Lebold
et.al (2003).

Sensors
Sensors have been defined as a “device that receives and responds to a signal or stimulus.” (Fraden,
1996, p. 2). The definition is according to Fraden (1996) broad, and that it could contain everything
from the human eye to the trigger of a gun. He, instead, would like to use the definition “a sensor is a
device that receives a signal and responds with an electrical signal.”(p. 3). A sensor or other technical
measurement devices have some advantages to human inspection: they are reliable and precise, they
can measure in unhealthy and hazardous conditions, they work fast, they work continuously, and they
can perform measurements to a relatively low cost.

A sensor never functions on its own but is a part of a larger system with other tools, such as signal
processors. Those tools are in its turn a part of an even bigger system, such as a condition monitor.
When an engineer chose an appropriate sensor for monitoring she/he needs to ask what the simplest
way of discovering the stimuli is without constitute any degradation of the comprehensive function of
the system (Fraden, 1996).

The rapid technology developments within the sensor industry have pressured both the prices and sizes
of sensors. MEMS (micro electromechanical systems) and smart sensors have made it into the market
of CBM as the lowest level of system health management (Lewis and Edwards, 1997). Sensors that
have decreased in size can perform more tasks than conventional sensors. Takeda (2001) classifies
industrial inspections in four categories:
• Inspection of infrastructure or facility,
• Inspection of equipment,
• Monitoring of products, and
• Monitoring of the environment

Lewis and Edwards (1997) means that MEMS devices are more reliable and produce more accurate
sensor data than conventional sensors due to:
• Sensor redundancy,
• Low drift and increased temperature stability,
• Sensor self-test,
• Multi-parameter output,
• Operation in harsh environment, and
• All-optical versions for operation in hazardous environments

Signal Processing
The purpose of signal processing in diagnostic applications and CBM is: (1) remove distortions and
restore the signal to its original shape, (2) remove sensor data that is not relevant for diagnostics or
predictions, and (3) transform the signal to make relevant features more explicit (may be hidden in the
signal, FFT analysis is an example of such a transformation). Distortions in sensor data may be caused
by an imperfect:
• Sensor,
• Media (metal, water, air etc.) in which the signal travels before reaching the sensor, and
• Media from the sensor to an analogue/digital converter

Signal processing may also manipulate the signal that some characteristics enabling prognosis are
more visible (for an analysis program or a human). Creating a feature vector from a signal is an
abstraction of the signal, preserving the features used in diagnosis and prognosis.

Condition Monitoring
Condition monitoring has been defined as “A means to prevent catastrophic failure of critical rotating
machinery.” and as “A maintenance scheduling tool that uses vibration, infrared or lubricating oil
analysis data to determine the need for corrective maintenance actions.” (Davies, 1998). The
parameters to monitor should be characteristics that will indicate an assets condition. The parameters
to monitor should be selected by the ones that in normal mode remain stable but in abnormal or
unhealthy mode will indicate some sort of a trend, e.g. increased vibration levels, increased noise, or
decreased pressure etc. (Yam et.al, 2001).

According to Johansson (1993), condition monitoring has been divided into two separate techniques:
a) Subjective condition monitoring:
- Sight,
- Hearing,
- Sense, and
- Smell
b) Objective condition monitoring:
- Sensors and other measuring equipment give data for either immediate condition
assessment or as basis for trend analysis.
The subjective parameters can of course be used objectively if collected through sensors or other
measuring equipment. Tsang (1995) divides condition monitoring techniques into six categories:
• Dynamic effects, such as vibrations and noise levels,
• Particles released into the environment,
• Chemicals released into the environment,
• Physical effects, such as cracks, fractures, wear and deformation,
• Temperature rise in the equipment, and
• Electrical effects, such as resistance, conductivity, dielectric strength etc.

Tsang (1995) also shortly presents a few different common condition monitoring techniques such as
vibration monitoring, process-parameter monitoring, thermography, tribology, and the subjective
technique visual inspection.

Diagnosis
Diagnosis has been defined as “…fault recognition and identification” (Lewis and Edwards, 1997, p.
8.5-5), i.e. a means to find out where something will go wrong and possibly even why. According to
Yam et.al (2001) condition based fault diagnosis can be divided into three categories:
• Rule-based diagnostic systems,
• Case-based diagnostic systems, and
• Model-based diagnostic systems

Rule-based diagnostic systems comprise of a knowledge-base and a set of rules the system use to
diagnose or predict a fault. These rules may be derived from experts in their field, and are then
compiled into a set of rules. Extracting, validating, and verifying the rule base is essential in such
systems since one faulty rule may wreck the complete result and make the system unreliable. This
problem is often referred to as the “brittleness” of rule-based systems. The expert becomes the so
called “knowledge acquisition bottleneck” and the rule-base needs maintenance, updates and
extensions once circumstances change or new knowledge is developed. A set of rules in a rule-based
diagnostic system may be translated to a decision tree traversed to determine the fault. This is only
possible if the rules meet a number of criteria’s (e.g. being deterministic). In some applications the
rules may be induced automatically. For many applications rule-based diagnostics systems is the most
appropriate solution. If statistics and fuzzy logic are used, these systems become powerful diagnostic
tools for industry.

Case-based diagnostic systems are based on Case-Based Reasoning (CBR), a method from artificial
intelligence, based in a cognitive model of learning from experience. Cases capture both a specific
situation/problem and the solution to the problem. When a new problem occurs, it is compared with
the case library and similar cases are retrieved. These cases are adapted, using domain knowledge, to
fit the current problem. The solution in the case is reused after validation/verification and if necessary
revised (performed by a human or by the system). The problem and the new solution are added to the
case library as a new case. Case-based diagnostic is used in situations where the task to create a large
and consistent rule base is too difficult or where model based diagnosis is inappropriate (example of
these is given in the following section). If statistics and feedback (automatic, semi automatic or
manually) is included in the cases, the system will not only improve performance with the addition of
new cases, but also with experience derived from feedback.

Model-based diagnostic system is a powerful solution if a complete model of the equipment to

monitor can be created. The model is used to detect any deviations and if a deviation is detected the
model is used to identify what the problem is. The abstraction level of the model is the limiting factor
for what faults are detectable. If it is possible to build a model based diagnostic systems, this is the
most desirable diagnostic system. Unfortunately it is a manual process to build a model and it is
difficult to build a model detailed enough for a majority of industrial applications where diagnostic
systems are desirable. If a model can be built, real-time simulation is to computationally costly or
impossible with available computers.

To build intelligent diagnostic systems, combination of the above mentioned methods is often
necessary. Also other techniques and methods from artificial intelligence are needed, e.g.
reinforcement learning to learn and adapt to normal conditions, reducing diagnostic mistakes. Genetic
algorithms may be used to find the cause of multiple faults, avoiding the “state space explosion” when
using traditional search strategies.

Prognosis
Prognosis has been defined as “…prediction of when a failure may occur” (Lewis and Edwards, 1997,
p. 8.5-5), i.e. a means to calculate remaining useful life of an asset. In order to make a good (reliable)
prognosis it must be followed after a good (reliable) diagnosis has been made.

Some diagnostic systems are able to make predictions when a fault may occur and with what
probability. This information may be used as input by a prognostic system to predict a future health
profile and calculate remaining useful life of some asset, given a required reliability level and safety
limits. Further on this calculations may be used to produce a prognosis of the overall reliability of a
large system. The weakest links in a system may be identified and counter measurements taken to stay
within some specified reliability and safety limits. These measurements may be different maintenance
tasks, but also measurements to ensure that certain replacement parts are available or redundant
production capacity is within access within a certain time frame.

Thurston and Lebold (2001) present a proposal to a generic prognostic module where they present a
standard set of input and output requirements for an OSA-CBM prognostic module. Input
requirements cover historic data in form of e.g. prognostics, health, failure, mission, and maintenance
history, as well as model information and spare assets capacity. Output requirements cover
information about the current health along with remaining useful life with confidence levels of the
prediction. The prognostic algorithms can be generic and can range from simple historical failure rate
models to physical models.

Decision support
Decision support systems are computer systems aiding in the decision making process. A human
expert is needed to make the final decision and the system provides the necessary information for
making the decision. It may also be legal reasons for using decision support systems instead of fully
automated systems, e.g. if decision making is not time critical, but the consequences of a faulty
decision is large. Computer systems can be reliable up to certain level, but are often so complex that it
is known that there are faults in them. The price of finding and removing all faults in a system may be
too expensive, and even software sold in very large quantities contain large number of faults, e.g.
Windows based systems.

A decision support system may have a number of diagnostic and prognostic tools, human experience
and statistical data, all accessible by the human to aid in the decision making process. In intelligent
human computer collaboration both humans and computers takes initiative and action. The computer
system may notice, based on previous experience that a human operator tries to do something that may
damage the equipment and intertwine. A dialogue between the system and the operator may result in a
modified procedure, acceptable for both parts. Hence the step beyond decision support systems is
human computer collaboration system.
Case Study
This chapter presents a case study of a Case-based fault diagnosis system implemented as a part of a
master thesis in computer science at Mälardalen University (Olsson, 2003). The system was
implemented to show how sound comparison and Case-Based Reasoning (CBR) can be used to detect
faults in the gearboxes of industrial robots. In this paper it will serve as an illustration of how a
technical CBM system can be designed.

Mechanical faults in industrial robots (and other machines) often show their presence as audible
deviations compared to a normal sound profile. As a part of the end-test of industrial robots, a
subjective condition monitoring based on hearing is used in order to detect audible deviations. Correct
classification of those deviations is a critical part of the end-test. An incorrect classification of the
sound can result in the delivery of a faulty robot to the customer. An operator needs long experience in
order to make a correct classification. Artificial Intelligence (AI) methods, such as Case-Based
Reasoning, have some advantages in this category of applications. The fundamental idea of CBR –
applying old knowledge of problem solving to solve new problems is very feasible for this type of
industrial applications. The method preserves experience that is often lost if personnel leave their
employment.

The proposed fault diagnosis system uses a hybrid Case-Based Reasoning method using a nearest
neighbour approach for a light weight solution of recognizing and diagnosing audible faults on
industrial robots. Sound is recorded with a microphone and compared with previous recordings.
Similar cases are then shown to the user corresponding diagnosis and actions based on previous
experience. The system aids engineers in making a correct objective diagnosis of the industrial robot
based on earlier classifications of similar sounds. The system is able to successfully diagnose faults in
an industrial robot based on sound recordings (6 recordings from faulty robots and 30 recordings from
normal robots are used in the evaluation).

The Case-based fault diagnosis system uses three different steps in its classification process; pre-
processing, feature identification and classification. Sound is obtained from the robot to be diagnosed
via a microphone as shown at the left in Fig. 5. The sound is recorded to a computer and the recording
is taken as input to the pre-processing step. The pre-processing process is responsible for filtering and
removal of unwanted noise. It also extracts period information from the sound. In the feature
identification process, the system uses a two-pass model, first identifying features and then creating a
vector with features. Once the features are identified, the system classifies the feature vector. The
classification is based on previously classified measurements (case library). When a new sound has
been classified, the new case is added to the case library.

Figure 5. Schematic picture of the Case-based fault diagnosis system with its three steps to condition
diagnosis.
Condition Monitor and Diagnostics
Sound from 24 healthy robots and 6 faulty robots were recorded during the case study. All recordings
where made during the end-test of the robots. Among other tests, the end-test includes a separate axis
test. In the separate axis test, all axes of the robot are individually tested. A microphone is mounted
close to the axis of the industrial robot being measured (in this case axis 4). The robot is set to separate
axis test and axis 4 is chosen. Two types of faults were recorded; fault 1 is caused by a notch on the
big gear wheel in the gearbox of axis 4. The fault is hearable and it is characterized by a low frequency
impulse sound in the middle of the rotation of the axis. Fault 2 is caused due to a slack between the
gear wheels in the gearbox. The fault can be heard as bumps at the end of each rotation of the robot
arm.

In the time/frequency plot in figure 4 the sound of the notch is seen as two repeating prominent peaks
(see Fig. 6). The frequency of the plot is sound intensity for the frequency 180-220 Hz during 12
seconds. The normalised sound intensity level is a value indicating the peak intensity at a specific
time. The plot shows four successive rotations of the robot arm. The peak is only visible in one
direction of the rotation of the arm.

Figure 6. Illustration of X+1 recording sample with time/frequency plot from axis 4 on a faulty robot.

A feature vector is assembled from the sound and matched with the vectors in the case library. Table 1
displays a list with the five best matching cases in the case library ordered according to similarity in
present. The similarity measurement is based on a straight forward nearest neighbour algorithm. The
matching could be significantly improved using domain knowledge (Zhao et.al, 1991), but is already
giving good results and classifying sound recordings correctly in 91% of all tests. As can be seen in
table 1; a previously diagnosed notch faults is ranked close to a current recording verified to be a notch
fault. The case ranked as second candidate (case #12) and third candidate (case #4) comes from
normal recordings in the case library.

Case name Similarity Case ranking

Notch fault #2 98% 1.
Normal case #12 84% 2.
Normal case #4 83% 3.
Table 1. The three best-ranked classifications of cases in the case library

Case Study Results

CBR was found to be a feasible method to use to identify faults based on sound recordings in
industrial robot fault diagnosis. Sound recordings where made under realistic industrial conditions.
The CBR system has a number of benefits as an industrial diagnostic tool:
• New cases are easy to add to the library, one sound recording is sufficient,
• The method is easily accepted by engineers and is seen as a tool enabling them to perform
better,
• It transfers experience; engineers are able to listen to different sounds and make manual
comparisons,
 • The system does not need to be “complete” from the start, a list of similar sounds and their
classification are shown to the engineer, and
• Performance increases continuously, if a new “not normal” sound is recorded that cannot be
classified, the engineer contributes to the systems experience by classifying the sound after the
fault has been identified and corrected.

It has been shown in the validation that one recording is sufficient for identification of a similar sound
in the case library. Also producing a straight forward feature vector from the original sound recording
is sufficient for good results in the matching based on nearest neighbour. The feature vector and
matching has potential for improvement. Potential users have been interviewed and their reaction to
the research prototype tool is positive and they all judge it would improve their performance and
productivity.

After analyzing the architecture of the developed Case-based fault diagnosis system, some similarities
to the OSA-CBM standard proposal can be visualized. The microphone can be regarded as the sensor
module. The pre-processing and the feature extraction process can be deduced to the signal processing
module. The classification (with the case-library) performs both the condition monitoring and the
diagnosis as it both detects deviations in the sound profiles and can classify different sound profiles to
different fault modes.

Conclusions
This paper has presented a theoretical discussion of the components necessary for a complete CBM
system approach and illustrated the first four of them through a case study. The OSA-CBM approach
of seven modules has proven to be an appropriate way of developing CBM system technology.
Although the development in the case study of this paper does not have complete modular design, the
modules as they have been described has been a good template for development. The modules can also
be seen as a template as technical aspects of CBM implementation. All of them are necessary to
uphold a CBM strategy, level of automation can be seen as secondary importance. Although, for
logistic reasons, the more complex, critical, and big sized processes to monitor the more automation
will be necessary, but for smaller companies with less critical machines it might be enough just to
have specified the way to think within the different modules. It might also be enough to take the level
of automation to e.g. the condition monitor and leave the diagnosis, prognosis, and decision to the
human intellect, thus making CBM a strategy not to complex for small companies to handle, but as a
tool used by the maintenance personnel to plan more accurate maintenance intervals.

The case study did not only show that the modules specified by OSA-CBM were a good template for
development, it also showed how a CBM system can be developed using maintenance personnel’s
tacit knowledge as condition parameters. End-testing of the industrial robots had previously been done
manually by listening and recognizing audible deviations in the sound profile as the test program was
launched, an experience taken long to acquire. With the development of the system the tacit
knowledge of the testing personnel was recorded, secured to be used as a condition based quality
assurance tool and possibly as a training tool for new employees. The system can also be further
developed to involve condition diagnosis while the robots are used in actual production, as a tool to
assess maintenance need, thus making it more necessary to include prognosis ability and possibly a
more enhanced decision support function.

Acknowledgements
This work was supported by The Knowledge Foundation (KK-Stiftelsen) through research project
2002/0232, EU goal2 Regional Structural Fund, Länsstyrelsen in Västmanland, Bombardier
Transportation, ABB Automation Technology Products AB, and the Swedish Maintenance Society
(UTEK).
References
Author Unknown, Maintenance Improvement Guide, Section 2 Failure Theory, Wolfson Maintenance,
www.wmeng.co.uk/maintenance-improvement-guide/failure_theory.html (2004-03-22).
Butcher, S. W. Assessment of Condition-Based Maintenance in the Department of Defense,
LOGISTICS MANAGEMENT INSTITUTE, USA, McLean, VA, 2000, pp. 1-70,
http://www.acq.osd.mil/log/logistics_materiel_readiness/organizations/mppr/assetts/senior_steerin
g/condition/LMI%20CBM%20Report.pdf (2003-10-08).
Davies, A. (Ed.). (1998). “Handbook of Condition Monitoring”. Chapman & Hall, Great Britain,
Cornwall. ISBN 0 412 61320 4.
Discenzo, F. M. et.al, Self-diagnosing intelligent motors: a key enabler for next generation
manufacturing systems, “IEE Colloquium on Intelligent and Self-Validating Sensors”, UK,
Oxford, 1999, pp. 3/1-3/4.
Fraden, J. (1996). “Handbook of Modern Sensors.” Springer-Verlag, USA, Ann Arbor, MI. ISBN 1-
56396-538-0.
Garga, A. K, McClintic, K.T., Campbell, R.L., Chih-Chung Yang, Lebold, M.S., Hay, T.A., Byington,
C.S., Hybrid Reasoning for Prognostic Learning in CBM Systems, “Aerospace Conference
Proceedings”, USA, Big Sky, MT, 2001, pp. 6-2957-6-2969.
Johansson, K-E. (1993). “Driftsäkerhet och underhåll”. Studentlitteratur, Sweden, Lund. ISBN 91-44-
39111-0.
Lebold, M., Reichard, K., Boylan, D., Utilizing DCOM in an Open System Architecture framework
for Machinery Monitoring and Diagnostics. “Aerospace Conference Proceedings”, 2003, pp.
3_1227-3_1236.
Lewis, S.A. and Edwards, T.G., Smart Sensors and System Health Management Tools for Avionics
and Mechanical Systems, “Digital Avionics Systems Conference”, USA, Irvine, CA, 1997, pp.
8.5-1-8.5-7.
Mitchell, J. S., Five to ten year vision for CBM, ATP Fall Meeting – Condition Based Maintenance
Workshop, USA, Atlanta, GA, 1998 (PowerPoint Presentation), http://www.atp.nist.gov/files/3
(2003-10-08).
Moya, C. C. and Vera, J. C. H., Evaluation of Condition Based Maintenance through Activity Based
Cost, “Maintenance Journal”, Vol. 16, No. 3, 2003.
Nowlan, F.S. and Heap, H. F. (1978). “Reliability-Centered Maintenance”. U.S. Department of
Commerce, USA, Springfield, VA.
Olsson, E., Diagnosis of Industrial Robots Using Case-Based Reasoning, “Master Thesis at Dept. of
Computer Science”, Mälardalen University, Sweden, Report number: TR0307.
Rolstadås, A., Enterprise Modelling for Competitive Manufacturing, “Control Engineering Practice”,
1995, Vol. 3, pp. 43-50.
Swedish Standards Institute, “Maintenance terminology”, SS-EN 13305, 2001.
Takeda, M., Application of MEMS to Industrial Inspection, “The 14th IEEE International Conference
on Micro Electro Mechanical Systems”, Switzerland, Interlaken, 2001, pp. 182-191.
Thurston, M. and Lebold, M., Open Standards for Condition-Based Maintenance and Prognostic
Systems, “MARCON, 2001”, www.osacbm.org (2003-03-10).
Thurston, M. G., An Open Standard for Web-Based Condition-Based Maintenance Systems,
“AUTOTESTCON Proceedings, 2001. IEEE Systems Readiness Technology Conference.”, 2001,
USA, Valley Forge, PA, 2001, pp. 401-415.
Tsang, A. H. C., Condition-Based Maintenance: Tools and Decision Making, “Journal of Quality in
Maintenance Engineering”, Vol. 1, No. 3, 1995, pp. 3-17.
Yam, R. C. M., Tse, P. W., Li, L., Tu, P., Intelligent Predictive Decision Support System for
Condition-Based Maintenance, “International Journal of Advanced Manufacturing Technology”,
2001, Vol. 17, Issue 5, pp. 383-391.
Zhao, Z., Cha, J., Tang, H., Zhu, M., Application of principal component analysis in gear fault
diagnosis, “1991 ASME Design Technical Conference”, USA, Miami, FL, 1991, pp. 159-163.
 Paper IV

Important Aspects to take into Consideration when Deciding to Implement

Condition Based Maintenance

Marcus Bengtsson and Mats Jackson

PROCEEDINGS OF THE 17TH INTERNATIONAL CONFERENCE OF

CONDITION MONITORING AND DIAGNOSTIC ENGINEERING MANAGEMENT
COMADEM 2004

59
60
 IMPORTANT ASPECTS TO TAKE INTO
CONSIDERATION WHEN DECIDING TO IMPLEMENT
CONDITION BASED MAINTENANCE

Marcus Bengtsson1 and Mats Jackson

Department of Innovation, Design and Product Development

Mälardalen University, P.O Box 325, SE-631 05, Eskilstuna, Sweden
{marcus.bengtsson; mats.jackson}@mdh.se
1
phone no. + 46 16 153 486, fax no. +46 16 153 610

ABSTRACT

As more companies have started seeing the maintenance department as a profit generating function
more focus has been directed on different maintenance strategies. Total Productive Maintenance
(TPM) and Reliability-Centered Maintenance (RCM) are two of the most well known maintenance
strategies that in the last 20 years have been used frequently in the manufacturing and aircraft industry.
To perform maintenance actions in the corrective or the preventive approach is though still a present
issue. The predictive maintenance technology Condition Based Maintenance (CBM) has in the past
years been receiving good reviews although it is not to a great extent utilized within Swedish industry.
One reason for this might be that there is a lack of organizational research within CBM in terms of
implementation strategies and applied CBM, i.e. how to implement and run a successful CBM
program. The objective with this paper is to, through a case study, illustrate and highlight important
aspects companies have to consider when a decision to implement CBM has arisen. The results of the
case are not very controversially and many similarities can be drawn to general literature on change
and implementation processes.

KEYWORDS

Condition Based Maintenance, Implementation Strategies, Organization, Organizational Aspects

INTRODUCTION

As industrial competition is growing harder more interest has been taken to maintenance strategies.
More companies have begun to realize that maintenance is something one can make money on and
stopped seeing it as just a cost centre. This has led to the implementation of different maintenance
strategies such as Total Productive Maintenance (TPM) and Reliability Centred Maintenance (RCM),
which better takes care of assets in a production environment. The question of using the corrective or
the preventive maintenance approach is though still present. Cooke and Paulsen define “good
maintenance” as when very little corrective maintenance tasks takes place and when as little preventive
maintenance as possible is performed [1]. The predictive maintenance technology Condition Based
Maintenance (CBM) utilizes condition monitoring techniques to assess the equipments condition in
order to decide of maintenance intervals as well as appropriate maintenance tasks. Thus decreasing the
corrective maintenance (impending faults are found before they become critical) and the
predetermined preventive maintenance (maintenance intervals are decided dynamically when need for
such has arisen and when best appropriate according to planned production).

The technical development within CBM has, for years, been strong and by looking at the Internet one
can see that a lot of companies are selling CBM services. Further, when scanning through journal and
conference proceeding papers within the topic of CBM one can also draw the conclusion that research
within the technical aspect of CBM is strong. Several prototypes for all kinds of different applications
are being developed in a rapid pace.

However, looking at the organizational aspects of CBM, or in depth, implementation strategies for
CBM one can see a lack in published research. Comparing to other, perhaps more accepted,
maintenance strategies such as TPM and RCM, one can see that the research within the organizational
issue of CBM might be lacking, thus making it difficult for companies to introduce. Carnero points out
that there is no international standard on managing a Predictive Maintenance program [2], little less to
say no standard for implementing it either. One view on how CBM should be implemented, focusing
on goals and financial issues is [3] “Well defined goals and a cost-effective investment strategy are
essential to designing and implementing a successful condition-based maintenance program.” (p. 954).

The implementation strategies that have been published (scientifically and on the Internet) are mostly
directed to special applications and hard to generically reproduce. Mitchell & Murry presents an eight
step implementation plan of a Predictive Maintenance program; they also presents the aspects that they
would focus on if they were to go through the implementation process again [4]. Mobley presents how
a Predictive Maintenance program can be established [5]. Trodd presents some practical
implementation tips of Predictive Maintenance, focusing a great deal on the key areas focused on in
order to achieve their objectives [6]. Nolan & Cesarone presents the implementation of prognostic
framework on a navy battle group [7]. Kotter [8] presents a process for organizational change and
Nakajima [9] presents a model for TPM implementation.

The objective of this study is to illustrate and highlight important aspects companies has to consider
when deciding to implement CBM. Could a model be developed that companies can follow when
implementing CBM? Models of such kind can be found within TPM and RCM, and have helped
companies make the transition from one strategy to another easier.

THEORETICAL FRAME OF REFERENCE

In the literature overview two general implementation strategies of Predictive maintenance and
Nakajima’s implementation strategy of Total Productive Maintenance (TPM) was have been included.
This section will briefly cover general theory on organizational change and theory behind those
strategies.
Organizational Change

Work methods in an organization are often deeply rooted in the ordinary day to day work. Changing
work methods can sometimes be looked upon with scepticism by both employees and management,
sometimes more by one than the other. When changing work methods an implementation or
introduction phase must be followed if success is to be achieved. Kotter [8] mean that successful
implementation of change in organizations has to follow two important patterns; first, the change has
to follow a multi-step process that create power and motivation to overwhelm the reactionaries and
second, the process has to be driven by high-quality leadership. Kotter and Cohen [10] points out that
empowerment, to remove barriers, is one very important aspect when it comes to implement change.
They also mean that it is often an immediate manager or someone higher in the hierarchy that is the
biggest obstacle for change. Kotter derives a multi-step process in eight steps [8]:
1. Establishing a sense of urgency
2. Creating the guiding coalition
3. Developing a vision and strategy
4. Communicating the change vision
5. Empowering broad-based action
6. Generating short-term wins
7. Consolidating gains and producing more change
8. Anchoring the approaches in the culture

Implementation of Predictive Maintenance

Mobley [5] presents how a Predictive maintenance program should be established focusing on: (1)
goals, objectives, and benefits; (2) functional requirements; (3) selling predictive maintenance
programs; (4) selecting a predictive maintenance system; (5) database development; and (6) getting
started. No model with steps in time is presented.

Mitchell and Murry [4] presents how a Predictive Maintenance (PdM) program was implemented at
the United States Department of Energy (DOE) Strategic Petroleum Reserve (SPR). The PdM program
was specified and developed by an Engineering department, implemented by a Maintenance
department, and maintained by the Operations- and Maintenance departments. During the first year
after implementation the program showed some success stories and payback but the majority of time
was spend on building a satisfactory database, tailoring the program, accomplish required training, and
lobbying for a positive attitude towards the program. Mitchell and Murry also presents a few
suggestions and actions of what they would do differently if they had to go through the implementation
process again, so called ‘lessons learned’.

Implementation of TPM

Before implementing Total Productive Maintenance (TPM), three requirements need to be taken into
considerations in order to achieve fundamental improvements. The first two are increasing the
motivation and competency of the personnel that will increase effectiveness and operations of the
equipment; the third is that the work environment must be created to support the establishment of a
systematic program for implementing TPM.

Nakajima has divided the implementation process into three stages with 12 steps. In the preparation
stage the work environment is created by the establishment of a plan for the introduction of TPM. This
stage can be compared to the initial stages of product development. In the implementation stage the
program is put into work and up and running. This stage can be compared to the production of a
product. In the last stage, called the stabilization stage, evaluation and higher goal setting is performed
[9].

METHOD
To gain more knowledge of the organizational aspects and implementation strategies of CBM a case
study, in form of a questionnaire/interview study, was performed at an international manufacturing
company. The company has been interested in, and had to some extent, introduced themselves to the
maintenance market of the products they manufacture. Therefore an implementation of a CBM
strategy would have to include both the contractors as well as customers of the product.

Case Study Approach

According to Yin [11] “A case study is an empirical inquiry that investigates a contemporary
phenomenon within its real-life context, especially when the boundaries between phenomenon and
context are not clearly evident.” (p. 13). Yin also means that case study research is preferred when
‘how’ and ‘why’ questions are being posed. One prejudice to case study research is that it is
impossible to receive generalizable results. Yin does state that the result can not be generalized to
populations or universe but that it can be generalized to theoretical propositions. In this case the
company that constituted the case study object had not implemented CBM in full scale, although some
projects had been done and some were on the way, it was decided to investigate the company’s look on
the first phase of implementation. The case was solved by a literature study and by an interview study
with a questionnaire approach, although leaving it up to the respondents to give qualitative comments
on all statements.

Data Collection and Respondents

As a basis for the case study a literature study within the CBM technology was performed. The
literature was sought after on the internet as well as in journal and conference proceeding data bases,
such as IEEE, Emerald full text, and Science Direct. The keywords that were used included: condition
based maintenance, condition monitoring, predictive maintenance, implementation, implementation
strategy, organizational aspects etc. The purpose of the literature overview was to find different
implementation strategies of CBM. General models were hard to be found, the ones that were found
were mostly directed and developed for special applications only. In order to gain more information in
the area of organizational issues within maintenance technology, the literature study expanded to
include Total Productive Maintenance and Reliability Centred Maintenance. These strategies were
therefore also used as keywords.

The questionnaire/interview study was constructed and performed in order to gather more data within
the issue. The questionnaire was constructed as an alternative questionnaire [12] with a number of
statements that the respondents were asked to either agree or disagree on. They did this with the help
of a six level scale, where 1 was equal to strong disagreement and 6 was equal to strong agreement. A
total of 38 statements were constructed with input from the literature study.

The sample (respondents at the company) was recommended to be a part of the study by a manager at
the company, i.e. the sample was nonprobabilistic but strategically chosen. When a sample is taken
from a population several errors can occur. In this case the sample was very small and it was chosen
only to reflect this particular company’s view on implementation aspects at that point in time. The
study may be hard to repeat with the same result for several reasons. One of the most important to
remember is that the company is in an expanding mode within the CBM technology so within a year or
two the same company’s maturity level will probably have increased, in that process the answers might
probably alter some. Other nonsampling errors can also exist; the population that is sampled might not
be relevant, the questionnaire subjects might give inaccurate or dishonest answers, and the possibility
of nonresponse which can come from choosing the wrong population which in its turn can result that a
bias in the results is present [13].

Analysis

The analysis of the interview study was performed through calculations of mean, maximum, and
minimum values of the statements respectively. All the data was entered in an excel data file were the
values was calculated. Differences between the whole group of respondents, the technicians group, and
the managers group was calculated and entered into the analysis. The comments on the statements
made during the interview occasion were also entered into the analysis. In order to analyze the
statements in a controlled way; they were sorted into five brackets. The preparation phase of
Nakajima’s [9] TPM implementation process, including five steps, was used as a frame as it was
recognized as a well tested implementation strategy of maintenance technology.

RESULTS OF CASE STUDY

The results come from an analysis based on both the quantitative and qualitative data obtained from
the interview study that included a questionnaire. The results of the collected data were then also
compared to the literature that was found on the topic. During the interview occasion the respondents
were also asked, openly, of what they though was the largest threat that CBM would not be developed
and used at their company, the results of that open question will also be discussed. The results of the
questionnaire will be displayed in tables and in a summary in the next section.

The Managerial Issues

This step sets out to identify if CBM really is what the company should focus on in terms of
maintenance related issues. The statements deal with the importance of the managements promoting
spirit when decision is on the way. It also deals with why the management should promote CBM in
terms of a few incentives (see table 1 for case results). Mitchell and Murry [4] states in their ‘lessons
learned’ that support from management, clients, and budget must be present through the whole
implementation process. They mean that the need for a new maintenance strategy must be visualized
and that implementation of such will lead to payback. Mobley [5] states that support from management
and from the personnel that will use the new maintenance strategy (i.e. operators and maintenance
technicians) must be outspoken if success is to be achieved. Nakajima [9] also concludes that the
support and total commitment to the new maintenance strategy (in this case TPM) is essential.
 TABLE 1. Statements included in the managerial step with mean values for all respondents
together (16), the technicians group (6), and the management group (10). (The statements are
shortened to better fit the table.)
The Managerial Issues All Tech Mgt
1) Company's interest in CBM of today 3.9 4.2 3.8
3) Management support of implementation of CBM is looked upon as important 4.9 5.0 4.8
17) Incentives for implementing CBM 5.4 5.5 5.4
18) Positive effect on Key Performance Index when introducing CBM 5.1 5.6 4.8
19) Increased productivity with CBM 5.4 5.5 5.3
28) Quality assurance of today 4.4 4.7 4.3
29) Quality assurance will be more effective with CBM 3.6 3.7 3.5

The Employee, Information, and Education Issues

This step covers the human aspects of CBM implementation, the importance of the human knowledge
of the products to monitor, and the importance of education and information (see table 2 for case
results). Training and education of the employees in order to run the new maintenance strategy
successful is a prerequisite and many state that the failure of many predictive maintenance programs
are a result from poorly or no training [4]; [5]; [9]. Mitchell and Murry [4] mean that any success,
especially in the early phase of implementation, should be publicized, no matter how small. This is of
course an effective way of selling in the maintenance strategy to top management and all personnel.

TABLE 2. Statements included in the employee, information, and education step with mean
values for groups respectively.
The Employee, Information, and Education Issues All Tech Mgt
7) It is important to inform of decision to introduce CBM 5.4 5.5 5.4
8) Continuous information on implementation process is important 5.3 5.3 5.2
20) The human is looked upon as important when introducing CBM 4.8 4.3 5.0
21) It is better to educate employees than to hire new when developing CBM 4.1 4.0 4.1
22) It is better to educate employees than to hire new when implementing CBM 5.0 5.0 5.0
25) To use the tacit (silent) knowledge of employees is looked upon as important 5.0 4.6 5.2

The Organizational Issues

This step deals with the organization as a whole, how to work with sub contractors and maintenance
contractors, experts within the field, and customers (see table 3 for case results). Mitchell and Murry
[4] state in their ‘lessons learned’, the importance of developing cross-functional project teams early in
the implementation process. They mean that if these teams are involved as early as the specification
phase, difficulties regarding the change in culture and creating a spirit of cooperation will be smaller.
They also mean that, if needed, it can be a good thing to hire an expert to help define the program.
Nakajima [9] focuses on creating organizations to promote the new maintenance strategy in order to
create an enthusiastic environment toward the change in maintenance approach; this view is also
supported by Kotter [8].
 TABLE 3. Statements included in the organizational step with mean values for groups
respectively.
The Organizational Issues All Tech Mgt
2) Cooperation with contratctor/maint. contractor/customers as today 3.9 3.2 4.2
4) ”Champions” is looked upon positively when implementing CBM 4.3 4.8 4.0
9) Increased cooperation with contratctor/maint. contractor/customers is necessary when
implementing CBM 5.4 5.8 5.2
10) Temporary organizations is looked upon as positive when developing and implementing CBM 4.1 3.8 4.3
11) To use consistent terminology when developing and implementing CBM is necessary 5.4 5.3 5.4
23) Consultants when developing CBM is looked upon positively 4.1 4.7 3.8
24) Consultants when implementing is looked upon positively 3.6 4.0 3.4
32) Increased cooperation with contratctor/maint. contractor/customers is necessary when developing
CBM 5.1 5.8 4.7

The Goal Setting Issue

This step deals with the issue of how to set the goals for the new maintenance strategy; this might be
one of the most decisive points for a smooth implementation (see table 4 for case results). Mobley [5]
state that it is important that the goals and objectives are adopted by both maintenance technicians and
management. Mobley mean that before any goals can be developed the existing maintenance costs and
other parameters must be determined, the goals can then be developed against the reference point.
Nakajima [9] are like Mobley presenting the idea that the goals of a new maintenance strategy should
be in reference to the current situation. These goals must then be broken down to the different
departments responsible for the current situation.

TABLE 4. Statements included in the goal step with mean values for groups respectively.
The Goal Setting Issue All Tech Mgt
5) It is important that the implementation of CBM fit other goals and strategies 4.8 5.3 4.4
12) It is important to set realistic and clear goals when introducing CBM 5.6 6.0 5.4
13) The goals should be long-term 5.0 5.2 4.9
14) The goals should be decided by cross-functional group 5.1 4.8 5.2
15) It is important to strive to raise the goals after they have been reached 4.5 5.2 4.1
16) It is important to work with continuous improvements within CBM 5.1 5.5 4.8
30) It is important that the company formulate possibilities of CBM before introducing it 5.5 5.3 5.6

Getting Started

This step deals with the planning and the start towards a full implementation of CBM, as well as
showing to the doubters that the technology will really work (see table 5 for case results). Mitchell and
Murry [4], Mobley [5], and Kotter [8] all state that the implementation process should be in a phased
approach. Mitchell and Murry [4] mean that one should not try to incorporate the entire world in the
first phase and that the golden rule is to stay small. “Select one location, one type of equipment, and
one process to initiate the program.” (p. 10). Nakajima [9] state that a master plan for the entire
implementation process should be developed at this stage.
 TABLE 5. Statements included in the getting started step with mean values for groups respectively.
Getting Started All Tech Mgt
6) A master plan, with decisive points in time, is necessary when introducing CBM 5.5 5.5 5.5
26) The company's use of standards as today 3.9 4.2 3.7
27) The use of standards when developing CBM is a necessity 4.9 5.2 4.7
31) To perform RCM analysis is possible to decide of measuring points 4.8 5.3 4.6
33) Pilot projects on subsystems is looked upon as important 5.5 5.5 5.5
34) To start monitoring critical systems is looked upon as an advantage 5.1 4.7 5.4
35) ”Champions” is looked upon positively when developing CBM 4.8 4.3 5.0
36) To start monitoring systems the company has high knowledge of is looked upon as an advantage 4.8 4.5 5.0
37) To start to use simpler monitoring techniques is looked upon as an advantage 4.8 5.0 4.6
38) To develop a comprehensive database at an early stage is looked upon as an advantage 4.8 5.0 4.7

Threats against CBM within the Company

During the interview occasion the respondents were asked to formulate the largest threat to why CBM
would not be developed and implemented in a full scale within the company. The results showed that
the threats mentioned were almost solely focusing on the organization and on whom that should pay
the initial cost. Only two respondents mentioned the technology of CBM. Some of the respondents are
mentioning that it would be good to have some data that point towards savings and by that be able to
show the top management and the customers that CBM will be something profitable if implemented on
the products. This can be derived to Kotter’s [8] first point in the change process; establish a sense of
urgency. The important thing that are sometimes forgotten though is to gather data and costs before the
pilot commence as a reference point, which will give data in hard cash to show for potential investors.

SUMMARY OF CASE FINDINGS

The Managerial Issues:

- Determine the existing maintenance philosophy and cost, direct maintenance costs (manpower,
direct materials, overheads etc.) and indirect maintenance costs (cost of lost production, cost of
quality deficiencies, accidents etc.), in order to set a reference point.
- Determine the possible incentives in form of higher productivity, KPI, increased safety for humans
and the environment, etc against the calculated reference point.
- Promotion of CBM must come from top management.

The Employee, Information, and Education Issues:

- Communicate the decision to develop and implement CBM.
- Develop an information strategy to communicate the coming results, in particular the positive ones.
- Develop a plan for education of employees and managers.
- Use the employees’ tacit knowledge were fit.

The Organizational Issues:

- Increase cooperation with contractors and involve customers in specification issues.
- Create temporary organizations (project groups) and empower to lead the development and
implementation of CBM.
- Try to involve people from the contractors and customers into the temporary organizations.
The Goal Setting Issues:
- The temporary organizations should develop the vision and the goals.
- Set the goals in a phased approach, one or two long-term visions (~5 years) and several achievable
short-term goals (~1 year).
- Finalize goals at top management.
- Communicate goals to the company, the contractors, and the customers.

Getting started:
- Create a master plan.
- Decide about the use of standards.
- Decide how to select what systems to monitor and select them.
- Pilot on single systems:
o Select pilot systems that have high downtime cost, high failure rates etc., i.e. capital intensive
systems.
o Select pilot systems that the company has knowledge about, e.g. failure rates, fault patterns etc.
- Use simpler monitoring techniques at first (e.g. trend analysis) to learn.
- Develop a comprehensive database, including measuring data, maintenance history, failure history,
mission history, model information, spare part assets etc.

CONCLUSIONS

The objective of this paper was to illustrate and highlight important aspects a company has to consider
when deciding to implement CBM. This has been achieved by the findings from the case study
together with the literature that was found on the topic. The findings have been presented in a case
summary and have throughout the paper been compared to other researchers’ results. The case
company had not at the time of the study implemented CBM on a full scale, but was in the starting up
phase. The aspects that have been presented are therefore not applicable to a full scale implementation,
but rather the first phase of implementation.

The fact that most threats mentioned in the qualitative section of the interview occasion dealt with
organizational issues and not technical, again, illustrates that there might be a lack of knowledge on
how to deal with the change and introduction of the maintenance strategy. This points out that there is
a need of more research on the topic, at least within this particular company and within the CBM
strategy. But taking in consideration the small amount of research that was found on the topic the same
fact might apply in other companies as well.

At this time it is important to mention that the aspects have not been tested empirically on companies
that have implemented CBM in a full scale. The aspects are supposed to be seen as a first attempt to be
validated at a later occasion. The findings of this case point out that it would be feasible to develop a
multi-step process for CBM implementation. How to decide of system technology, system automation,
how to change cultural back package, how to run the program successfully, how to evaluate the
program etc. will all constitute interesting and important further research objectives.
ACKNOWLEDGEMENTS

This work was supported by The Knowledge Foundation (KK-Stiftelsen) through research project
2002/0232, EU goal2 Regional Structural Fund, Länsstyrelsen in Västmanland, Bombardier
Transportation, ABB Automation Technology Products AB, and the Swedish Maintenance Society
(UTEK). I would also like to give thanks to all respondents who were willing to be apart of this study.

REFERENCES

1. Cooke, R. and Paulsen, J., Concepts for measuring maintenance performance and methods for
analysing competing failure modes, “Reliability Engineering and System Safety”, 1997, Vol. 5,
Issue 2, pp. 135-141.
2. Carnero, C. M., How To Set Up a Predictive Maintenance Program, “Quality Progress”, 2003, Vol.
36, Issue 3, pp. 56-61.
3. Spare, J. H., Building the Business Case for Condition-Based Maintenance, “Transmission and
Distribution Conference and Exposition”, USA, Atlanta, GA, 2001, pp. 954-956.
4. Mitchell, B. F. and Murry, R. J., Predictive Maintenance Program Evolution – Lessons Learned,
“1995 PROCEEDINGS Annual RELIABILITY and MAINTAINABILITY Symposium”, USA,
Washington D.C., 1995, pp. 7-10.
5. Mobley, R. (2002). “An Introduction to Predictive Maintenance”. Butterworth-Heinemann/Elsevier
Science, USA. ISBN 0-7506-7531-4.
6. Trodd, G., Practical Implementation of Predictive Maintenance, “Pulp and Paper Industry
Technical Conference, 1998”, USA, Portland, ME, 1998, pp. 29-37.
7. Nolan, M. & Cesarone, T., Implementation of Prognostics Framework on Navy Battle Group –
Automated Maintenance Environment (BG-AME) and Total Ship Monitoring (TSM) Programs,
“AUTOTESTCON Proceedings 2002”, USA, Sparta, NJ, 2002, pp. 788-800.
8. Kotter, J., P. (1996). “Leading Change”. Harvard Business School Press, USA, Boston. ISBN 0-
87584-747-1.
9. Nakajima, S. (1988). “Introduction to Total Productive Maintenance”. Productivity Press, USA,
Cambridge. ISBN 0-915299-23-2.
10. Kotter, J., P. and Cohen, D., S., Creative Ways to Empower Action to Change the Organization:
Cases in Point, “Journal of Organizational Excellence”, 2002, Vol. 22, Issue 1, pp.73-82.
11. Yin, R. (1994). “Case Study Research: Design and Methods, Second Edition”. SAGE Publications,
USA. ISBN 0-8039-5663-0.
12. Bell, J. (1993). “Introduktion till Forskningsmetodik”. Studentlitteratur, Sweden, Lund. ISBN 91-
44-37022-9 (Swedish).
13. Newbold, P. (1995). “Statistics for Business & Economics”. Prentice-Hall Inc., USA, New Jersey.
ISBN 0-13-181595-4.

View publication stats