Available online at www.sciencedirect.

com

ScienceDirect
Procedia CIRP 22 (2014) 87 – 91

3rd International Conference on Through-life Engineering Services

Can equipment failure modes support the use of a Condition Based
Maintenance Strategy?
David Baglee, Erkki Jantunen
University of Sunderland. UK. David.baglee@sunderland.ac.uk
VTT Technical Research Centre of Finland.Erkki.Jantunen@vtt.fi

Over recent years, the importance of maintenance, and therefore maintenance management within manufacturing organizations has grown. This
is a result of increasing pressure upon manufacturing organizations to meet customer and corporate demands, and equipment availability and
performance is central to achieving these. Condition Based Maintenance (CBM) is widely accepted and used as a financially effective
maintenance strategy. The economic benefit of CBM is achieved if the tools and techniques associated with CBM are applied to the right
equipment. In particular the degradation behavior of the equipment needs to be understood. Understanding of degradation is strongly related
with failure models However, very little is known or published about the importance and the role of various failure models. Thus, if failure
models are not analyses and understood the use of CBM could be directed to the wrong equipment and therefore achieve incorrect and
expensive results. The paper examines the relationship between the failure patterns observed in industrial maintenance practice and the
corresponding impact on adoption and potential benefits of Condition-Based Maintenance (CBM). The paper will explain the need for accurate
and up to date equipment information to support the correct maintenance approach. The paper suggests the importance of further supporting
such investments by appropriately addressing the need to collect relevant data as a basis upon which to make the right decisions.

© 2014
© 2014 The
The Authors.
Authors. Published
Published by
by Elsevier
Elsevier B.V.
B.V.This is an open access article under the CC BY-NC-ND license
Peer-review under responsibility of the Programme Chair of EPSRC Centre for Innovative Manufacturing in Through-life Engineering
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Services. under responsibility of the Programme Chair of the 3rd InternationalThrough-life Engineering Conference
Peer-review

Keywords: Condition Based Maintenane, Maintenance Efficiency

1. Introduction comprises a number of functional capabilities: sensing and
data acquisition, signal processing, condition and health
Maintenance engineering represents an area of great assessment, prognostics and decision aiding. Companies are
opportunity to reduce cost, improve productivity and increase moving from traditional corrective and preventive
profitability for manufacturing companies throughout the maintenance program to CBM to reduce the maintenance cost
world. There are examples of best practice what we may call and unnecessary maintenance schedules. A CBM program
World Class Maintenance which delivers great benefits. In consists of three key steps [1]:
addition, the maintenance of the infrastructure of modern
industry has become an increasingly important, and complex, x Data acquisition, to obtain data relevant to the system
activity – particularly as automation increases and the global health
marketplace in manufacturing squeezes profit margins. The x Signal processing, to handle the data or signals collected in
opportunity exists for many companies in Europe to see step 1 for better understanding and interpretation of the
substantial improvements to their competitiveness and data
profitability by improving their maintenance performance. x Maintenance decision making, to recommend efficient
Condition monitoring of plant and equipment has now been maintenance policies based on diagnosis and prognosis
identified as a major technique in establishing the optimum extracted from the data.
repair and maintenance periods to ensure in service reliability
and maximum utilization of assets. A complete CBM system

2212-8271 © 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Peer-review under responsibility of the Programme Chair of the 3rd InternationalThrough-life Engineering Conference
doi:10.1016/j.procir.2014.07.003

Modeling to identify deterioration is often overlooked. and most importantly cost have opened up an there is no or little change that could be used to justify the entirely new area of diagnostics to the companies who until diagnosis of maintenance need (e. The functions that with regard to the degradation behavior of the equipment. (2010) and more recently in Arnaiz et al. in order Fig. CBM Modelling 2) Rapid wear out after long useful life (Fig. In recent years there has been an increase in the use of 4) No infant mortality followed by indefinite useful life CBM as companies need to reduce maintenance and logistics (constant failure rate) (Fig. 2. 1) Bathtub curve. This research activity aims at x management and control of data flows or test sequences encouraging the research community to examine and discuss x Modelling to identify deterioration. sudden/random failure of recently were either unaware of the benefits or unsure on how electronic components).e. size impractical to monitor the remaining three (4. useful life. This requires detailed cost analyses of the current cost of maintenance and the necessary investment required to increase planned maintenance activities. 2013): industrial sectors where the production forms a chain i. Condition Monitoring tools have proven successful in reducing unplanned downtime by preventing equipment or process failure. In order to identify the financial and productivity benefits from a CBM strategy it is Fig. x signal processing and feature extraction This paper will present the results of a survey carried out x failure or fault diagnosis and health assessment with different maintenance experts to obtain information x identification of remaining useful life about failure models. the importance and understanding of various failure models before embarking on a condition based maintenance strategy. capabilities of predictive maintenance technologies have increased in recent years with advances made in sensor The first three (1. costs. Naturally. 5 and 6) as reductions. It As the gathered data clearly shows differences between is important to use an appropriate method for modelling various industrial sectors do exist. infant mortality. The diagnostic 6) Infant mortality followed by indefinite useful life (Fig. 5). improve equipment availability and ensure that mission 5) Indefinite useful life (constant failure rate) (Fig. Nevertheless. necessary to start with a detailed range of functions from . A complete CBM system comprises a number of functional capabilities including a range of sensors and data acquisition techniques. 2 and 3) can be monitored. 1. This is often due to lack of understanding implement specific maintenance actions. Gradual wear out. demonstrating the magnitude of the savings that can be generated using CBM is difficult.88 David Baglee and Erkki Jantunen / Procedia CIRP 22 (2014) 87 – 91 Unfortunately without a plan or path CBM strategies can which to collect and analyse data in order to develop and be unsuccessful. 6). to be successful in terms of cost to implement or equipment availability it is important to: a) Determine the cost of failures b) Determine the cost-benefit of avoiding failure. 2). to best make use of the equipment. Bathtub curve: Infant mortality – useful life – rapid wear out. If facilitate CBM include but not limited to: failure modes are not identified and understood the use of CBM to support a maintenance strategy will be based upon x sensing and data acquisition false or inaccurate information. rapid wear out (Fig. This is achieved by providing asset managers with the information they need to implement real-time and need-based maintenance for deteriorating equipment. However. and the optimal selection and scheduling of the light of this study it is clear that all investments to support inspections and preventive maintenance actions. (2010) and Fumagalli et al. findings deterioration to identify the different conditions and their outline a need for accurate data to support similar studies. 3). direction have been provided by Jantunen et al. the highest benefits can be gained in grouped in six categories (Tutorial Part 14. These advances in component sensitivities. 1). 4). (2013). Rapid wear out after long useful life. whereas it is technologies. critical equipment is available when required. 3) Gradual wear out after long useful life (Fig. 3. In addition. In effects. one breakdown can affect the entire production process. First attempts in this Fig. Theoretically modern maintenance technologies have relatively short there are different types of failure characteristics often payback time. 2. This is due to internal accounting systems but mostly due to the inherent difficulties in estimating the often indirect positive impact that CBM has on savings.g.

Interviews were carried out with senior managers while questionnaires were distributed to shop floor personal. No infant mortality followed by indefinite useful life. Fig. in % followed by CBM useful life. Indefinite useful life. “who. what. and in many situations dispute the views of senior managers. out % in % in % Process France 30 % 30 % 30 % 3% 3% 3% 90 % industry Aerospace UK 10 % 10 % 70 % 0% 0% 0% 90 % Chemical Finland 10 % 10 % 70 % 0% 0% 10 % 90 % industry Mechanical Spain 10 % 30 % 50 % 0% 5% 5% 90 % components Tyre industry Russia 5% 10 % 70 % 5% 10 % 10 % 85 % Process UK 60 % 15 % 10 % 10 % 5% 5% 85 % industry Rail Spain 15 % 60 % 5% 10 % 10 % 0% 80 % Process Russia 10 % 20 % 50 % 0% 0% 20 % 80 % industry Mining Canada 30 % 20 % 30 % 0% 10 % 10 % 80 % industry Home UK 30 % 37 % 13 % 2% 0% 2% 80 % electronics Process Sweden 10 % 50 % 10 % 10 % 15 % 5% 70 % industry . infant wear out wear out mortality useful mortality to use mortality. table 2 shows the mortality and how CBM could be sued to support comparison of the process industry data. Data collection Data were collected using interviews. The interviews allowed a range of experiences. x The strategies employed to collect and analysed data It is immediately apparent that differences do exist to inform future maintenance between what the experts think regarding similar industry x Their understanding of useful and remaining life. 4. where and how”. maintenance professionals from companies who provided maintenance consultancy were contacted. varied and ‘uncensored’ view of maintenance strategy development. after long after long followed by life. in % life. 6. from 5 different countries. The aim was to collect. when. in certain situations. David Baglee and Erkki Jantunen / Procedia CIRP 22 (2014) 87 – 91 89 3. scientific company profiles from approximately 60 companies from 12 countries. For example. Data analyses Industrial Country Bathtub Rapid Gradual No infant Indefinite Infant Logical sector curve. in % useful life. analyse and present a Fig. The results are shown in table 1 below. to be discussed and analysed. In addition. 5. Although the interviews were open they did provide a systematic description on: x Their current maintenance practices. useful useful indefinite indefinite rapid wear life. sectors in different countries. x How management decisions are taken when Fig. The questions were similar to each organisation i. situations and knowledge that would otherwise be hidden. useful life. examining their maintenance practices. to support the views of senior management.e. questionnaires. (presented in table 1) maintenance decisions. Infant mortality followed by indefinite useful life. Table 1. The questionnaires were used. x Their justification for using this maintenance method.

rapid useful life. the practices. USA aerospace industry claim CBM is needed. in % % useful life. figures which fall within 90% . This is contrasted by the use of CBM Automotive production in the UK. in % % France 30 % 30 30 % 3% 3% 3% 90 % UK 60 % 15% 10 % 10 % 5% 5% 85 % Russia 10 % 20% 50 % 0% 0% 20 % 80 % Sweden 10 % 50% 10 % 10 % 15 % 5% 70 % Belgium 10 % 10% 15 % 20 % 5% 10 % 35 % It is evident from the data that there are certain similarities tolerances it would follow that industries within the aerospace i. gear boxes. Honda. Indefinite out after out after long followed by followed by Logical to Country useful life.90 David Baglee and Erkki Jantunen / Procedia CIRP 22 (2014) 87 – 91 Electric Spain 5% 35 % 30 % 0% 30 % 0% 70 % motors/batteries Manufacturing Italy 5% 20 % 40 % 20 % 14 % 1% 65 % Mining Sweden 10 % 30 % 25 % 5% 20 % 10 % 65 % industry Lifts Spain 0% 35 % 30 % 0% 35 % 0% 65 % Robotic Spain 0% 30 % 30 % 0% 35 % 5% 60 % systems Manufacturing Spain 10 % 25 % 25 % 0% 30 % 10 % 60 % industry Machine tools Spain 10 % 40 % 5% 0% 40 % 5% 55 % Cars UK 10 % 21 % 22 % 10 % 13 % 14 % 53 % Paper industry Turkey 10 % 20 % 20 % 10 % 20 % 20 % 50 % Process Belgium 10 % 10 % 15 % 20 % 5% 10 % 35 % industry Mechanical Portugal 5% 10 % 15 % 20 % 25 % 25 % 30 % components Paper industry Sweden 4% 6% 15 % 18 % 20 % 37 % 25 % Ships USA 0% 17 % 0% 0% 42 % 29 % 17 % Aircraft USA 4% 2% 5% 7% 14 % 68 % 11 % Table 2. The data presented in the table respondent countries. it is interesting to note that 11% of the mechanical components. This raises the Little or no monitoring of robots takes place. This is true on assembly lines operated by robots where from infant mortality with useful life and rapid wear out. No infant Rapid wear Gradual wear mortality Infant mortality Infant mortality. This the majority of robots weld. categories. in life. This is an interesting claim inefficient maintenance practices across the range of and one which should be examined by the UK auto-industry. This is a year by year increase of ap- is unclear if the same or similar equipment is used with the proximately 12% since 2009. which includes Nissan. there might be some consensus. 65% of robotic systems have an infant mortality The figures for Belgium indicate that they suffer from followed by indefinite useful life. high figure seems to be unique to the UK. In these components which is highly regulated and components are made to exact . are within 1%-5%. long useful useful life in indefinite indefinite useful use CBM wear out % in % life. or the data collected is according to the manufacturing companies surveyed.35%. about inaccurate and not be validated. 60% of the respondents stated that they suffer CBM. This is an area question of the type and efficiency of current maintenance of great interest to the UK auto-manufacturers. of ineffective maintenance practices. In an industry hydraulic pitches or bearing systems. This firstly questions Toyota. table reports. This. is increasing with approximately 1.e. in addition it cars produced in 2013.5 million validity of the data supplied by the respondents. as with the UK suggests they employ a range On the other hand. form and assemble small fixtures. such as spindles. Comparison of the process industry data from 5 different countries. the figures which represent gradual wear out after long life supply chain would benefit from CBM. the importance of the wear failure models regarding If we return table 1. In Spain. suggest that the industry could benefit with a wider uptake of In the UK.

introducing a CBM strategy can be accessed on the face of [7] Jantunen E. organizations. It is evident that the cost-efficiency of of e-Maintenance. issue 2. Volume 7 pp. and Alarcón J. this was References outside the scope of this initial investigation and requires in- depth analyses of the data and data sources to be able to [1] Lee J. Continuous Improvement on this study. Industrial This paper presents a discussion of how different sectors maintenance management: a survey carried out on Italian companies. Roma. However. Helsinki. Still such views would ideally as machine tools and lifts. Arles. While this is so. A- MEST'10. in these two the need to be validated by actual observation in industrial increased product complexities and the process characteristics practice. Advanced Maintenance Engineering. Konde E. 38. 11. 15-17 valid. Darning L. Baglee D. On the Need for Research on Holistic Maintenance. 2014). 34 – Although the maintenance community has for long been 35. David Baglee and Erkki Jantunen / Procedia CIRP 22 (2014) 87 – 91 91 electronics are still kept to a minimum and therefore little data is available to enable a truly data-driven decision. 5-8 May 2014. p. To be published in Euromaintenance 2014. Arnaiz A. Macchi M. 2013. A detailed data platform for diagnostics.87. why significant potential to bring substantial savings in different would the failure happen.28. 193-198. Portugal. and Macchi M. Italy. Proceedings of the 22th international congress on Condition 4. One Technologies in an eMaintenance platform. The aim is to evaluate process and manufacturing industries. . prognostics and maintenance optimization. failure types seem to follow quite a different pattern in [6] Jantunen E. In e- mining process would need to be developed in order to extract Proceedings of Intelligent Maintenance System. and Di Leone F. Risk. [3] Bengtsson M. Conclusion Monitoring and Diagnostic Engineering Management (COMADEM). France. with random failure events difficult to recording of the failure events. Services and observation made from the survey results is that the apparent Technology. p. This would include expert perception of failures occurrences is that CBM has simple analyses to determine what could go wrong. [8] Tutorial Part 14 (2013). An integrated confidently develop a set of failure modes. processes for possible failures and to prevent them by correcting the processes proactively rather than reacting to adverse events after failures have occurred. so as to facilitate more prevent. In: can benefit from the introduction of Condition-Based Proceedings of the 4th MM 2009 (Maintenance & Facility Management) Maintenance strategies. Lisboa.. 2010. 81. Procedia CIRP. 2009. Financial analysis different sectors. Economic Value of different types of industries (Jantunen et al. In mechanical failures are predominant. rail) and in consequences of each failure.43. In: 1st IFAC Work-shop. and Dragan B. Risk across the Life Cycle (continued). most typically in transport (aerospace. Andrews KSJ. comprehensible information from July. The discussion is based on an analysis Conference. Information and On-line Maintenance Technologies for Increased Cost- effectiveness. The Asset Journal.. Ramji A. and Salonen A. However. Issue 2. The conclusion is that there is an increasing and involving incorrect product usage increase the importance of rather urgent need for organizations to establish accurate indefinite useful life. of expert views regarding the type of failures occurring in [5] Fumagalli L. previously unknown. Jantunen E. Reliability and Life Cycle Cost. Finland. the current evidence from the the data into quantifiable problems. Ierace S. MAINT-WORLD. Di Leone F. and what would be the sectors. Fumagalli L. 23. Ei-bar: Fundacion Tekniker. p. such circumstances the decision can be out of necessity taken Wear mechanisms are also important in other sectors. and Fumagalli L. Vladimiro C. 2004 the organisations and individuals who supplied the data for [2] Arnaiz A. informed choices regarding the introduction of CBM The next stage would be to use failure modes to categorise strategies. p. aware of the importance of studying such failure statistics. such on the basis of expert views. [4] Fumagalli L. Identification of Wear evidence from occurring failure types across a range of Statistics to Determine the Need for a New Approach to Maintenance. and Macchi M.