Condition Based Maintenance

Byron A. Ellis

Consultant and Executive Director of The Jethro Project, (TJP), email: ellis@jethroproject.com.

Abstract

A well-implemented condition based maintenance (CBM) regime requires management support, dataanalysis to determine cost-effective monitoring points, and a knowledgeable and disciplined maintenanceorganization. The main objective of CBM is to ensure that assets fulfill their mission in the most cost-effectively manner.

Keywords: Condition based maintenance, reliability centered maintained, preventive maintenance, failuremode, computerized maintenance management system.

Introduction

Condition based maintenance (CBM) is amanagement philosophy that posits repairor replacement decisions on the current orfuture condition of assets (Raheja, Llinas,Nagi, & Romanowski, 2006); itrecognizes that change in conditionand/or performance of an asset is themain reason for executing maintenance(Horner, El-Haram, & Munns, 1997).Thus, Horner et al. noted that the optimaltime to perform maintenance isdetermined from actual monitoring of theasset, its subcomponent, or part. Theyalso noted that condition assessmentvaries from simple visual inspections toelaborate automated inspections using avariety of condition monitoring tools andtechniques.

The objective of CBM is to minimize thetotal cost of inspection and repairs bycollecting and interpreting intermittent ora continuous data related to the operatingcondition of critical components of anasset (Knapp and Wang, 1992). Knapp,Javadpour, and Wang (2000) argued thatmonitoring, where cost-effective, couldprovide adequate notice on pendingfailures, which would allow for plannedrepairs based on asset degradation, as

opposed to costly time-based (fixedintervals) repairs or emergencybreakdowns. In some instances, however,CBM might not be cost-effective or datamight not exist to justify its introduction.

Raheja, Llinas, & Nagi (2000), on theother hand, argued that currentapproaches to CBM system design areextremely specific and that a genericarchitecture for CBM is missing. Thus,they claimed that each domain area hasits own interpretation of CBM that maynot be compatible with the requirementsof other applications. However, CBM isnot applicable to all maintenance assetsand should only be applied to instanceswhere condition-monitoring techniquesare available and cost-effective (Horner etal., 1997).

CBM requires robust analysis ofreliability and financial maintenance data(Crespo Márquez & Sánchez Herguedas,2004). Furthermore, it requires anunambiguous understanding of failuremodes and rates, asset criticality, andpotential payoffs associated with differentmaintenance strategies. Moubray (2001)cited studies done on civil aircraftdescribing patterns of failure thatcontradicted the belief of a relationship

www.jethroproject.com, TJP, November 10, 2008, pp. 1-5 Condition Based MaintenanceCopyright 2008 TJP. All rights reserved. 1 Byron A. Ellis

between reliability and operating age. Henoted that this belief led to the idea thatfrequent overhaul of an asset wouldincrease its reliability.

According to Moubray (2001), unlessthere is a dominant age-related failuremode, time-based overhauls do notimprove the reliability of complex items.He further noted that it is highly possiblethat time-based overhauls wouldintroduce infant mortality failures inotherwise stable systems.

The Benefit of Monitoring AssetPerformance

Wood (1999), as well as Moubray (2001),argued that time-based preventivemaintenance (PM) often failed tomaximize the service life of eachcomponent; in many instances thecomponents were being replaced withmany hours of useful life remaining. PMtasks, based on hard time intervals, wastea lot of resources (Tsang, 1995); that is,often the plant or building is over-maintained (Mann, Saxena, & Knapp,1995).

Bloch and Geitner (1983), noted thatcertain signs, conditions, or indicationsprecede 99% of all machine failures.Therefore, by monitoring an asset, actioncan be taken prior to it having a seriouseffect on the performance of theorganization. Thus, condition-basedmonitoring offers an alternative to thePM assumption of age-related failuremode. Moreover, with condition-basedmonitoring, managers can focus on just-in-time (JIT) replacement. According toWood, JIT replacement maximizes thelife of each component and is facilitatedby building automation systems (BAS) orintelligent buildings.

Maintenance Strategies

Coetzee (1999) observed thatmaintenance management oftenimplement some highly publicizedstrategy towards increasing the efficiencyof the maintenance function, such asreliability centered maintenance (RCM),total productive maintenance (TPM),condition based maintenance (CBM),computerized maintenance managementsystems (CMMS), auditing systems andso on. However, most managers do notknow what the right regime foroptimizing maintenance is. As a result,most efforts do not materialize.

Intervention strategies are not onlytechnical in nature; they are also a matterof art (Heckscher, Eisenstat, & Rice,1994). According to Hecksher et al. mostbest practice strategies fail within a yearor so. They labeled this phenomenon assuccessful failure, which results fromlimited know-how within theorganization: few internal membersunderstanding what they are movingtowards when they start animplementation process (Heckscher, etal.).

As a result, Coetzee (1999) argued for aholistic approach within the maintenancefunction to implement CBM. A holisticapproach would include the evaluation ofthe maintenance assets, includingbuilding structures. Shonet (2003) arguedthat effective implementation of CBMrequires the development of performanceindicators for building components andsystems. He noted that the evaluation ofbuilding components should assess (1)the physical performance of buildingsystems, (2) frequency of failures ofbuilding systems, and (3) actualpreventive maintenance (PM) carried outon the systems.

www.jethroproject.com, TJP, November 10, 2008, pp. 1-5 Condition Based MaintenanceCopyright 2008 TJP. All rights reserved. 2 Byron A. Ellis

Likewise, a holistic approach should alsoinclude performance measurementsystems as a tool for delivering strategicobjectives, such as the implementation ofa balanced scorecard process(Letza,1996); the evaluation of breakdown (BD)maintenance, where the run to failurephilosophy prevails; the evaluation ofplanned schedule maintenance, whereplanning and scheduling methodologiesare rigorously applied; and the evaluationof CBM, where critical and costly assetsare monitored for degradation (Amari andMcLaughlin, 2004). Degradation is aprocess where system parametersgradually worsen.

The application of CBM requires the useof certain analytical tools, such as failuremode, effect, and criticality analysis(FMECA) and RCM to determine thelikelihood of failure and how failurewould occur; as well as a reliableinformation tool (i.e., CMMS) to captureand track repairs and associated costs ofthe assets under consideration.

FMECA is a systematic process foridentifying all possible ways in whichfailures of an element can occur andRCM is an approach for identifying themost applicable and cost effectivemaintenance task for asset elements (El-Haram & Horner, 2002). Thus, El-Haramand Horner noted that the RCM processevaluates each failure to determine itsconsequences, if any, as well as eachconsequence to determine applicable andcost effective maintenance tasks.

Failure is when an asset, component orelement cannot fulfill its mission. Formany assets, however, certain symptoms(soft faults) often occur prior to totalfailure, such as excessive vibration,abnormal heat (Knapp et al., 2000), solid

contents in oil, and so on. Sensorscapable of detecting pre-failure signalscan provide alerts and alarms notifyingthe asset operator of the potential of animpending failure. For other assets,failure (hard faults) may occurinstantaneously and cannot be predicted;generally these failures are exogenouslygenerated.

CBM is a non-intrusive technique and theactual preventive action (repair) is takenat the incipient failure level (Tsang,1995). Tsang noted that hidden failuresare often present in standby units,protective devices, or infrequently usedassets. Hidden failures are not evidentuntil the time when the proper function ofthe item is needed.

Thus, a CBM regime on standby units,such as protective devices, orinfrequently used assets would not beeffective, since upon startup hiddenfailures could occur. However, tocounteract, hidden failures Tsang (1995)recommended faultfinding (FF) tasks. FFtasks are performed at scheduled intervalsto check the state of assets or items withdormant functions. A typical example isan emergency generator that is generallyidle until needed.

According to Sherwin and Al-Najjar(1999) if downtime is very expensive,inspection and condition monitoring arenot the best overall maintenance policiesunless the combined age to failuredistribution of the various failure modesof an asset, or component, has a largevariance such that the standard deviationis greater than or (nearly) equal to themean.

Determining CBM Failure Point

www.jethroproject.com, TJP, November 10, 2008, pp. 1-5 Condition Based MaintenanceCopyright 2008 TJP. All rights reserved. 3 Byron A. Ellis

Determining when a failure mode hasbeen initiated and predicting the timeremaining for complete failure requiresreliable data and effective algorithms.Moreover, making decisions on theappropriate maintenance response toCBM alerts or alarms requires adisciplined maintenance organization.

Many maintenance organizationsequipped with CBM do not effectivelyrespond to alerts and alarms. However,some CBM systems are integrated intothe CMMS, so that a work order isautomatically generated when signalsfrom a monitored point trigger alerts oralarms. This is particularly suited forcontinuous process plants where failureand downtime can be extremely costly.

False Alarms

A key requirement for successfulimplementation of CBM is a lowincidence of false alarms and quick andreliable detection of faults (Buswell &Wright, 2004). Buswell and Wright notedthat an International Energy Agency(IEA) project indicated that abrupt faultscould be easily detected; degradationfaults, on the other hand, had to be quitelarge before detection was unambiguous.The study also noted that it is possible tobalance robustness and sensitivity.Increasing sensitivity results in earlywarning of faults, but may increase theincident of false alarms (Knapp,Javadpour, & Wang, 2000). However,condition-based monitoring industryleaders have applied quality control typetechniques in the detection of machinefaults; thus, providing increasedprotection from false alarms, as well asfrom missed faults (Mann et al., 1995).

Conclusion

The introduction of CBM withoutmanagerial support at all levels,appropriate data and analysis, adisciplined and knowledgeablemaintenance staff is unlikely to achievedesired results.

Copyright of TJP is the property ofThe Jethro Project and its contentsmay not be copied or emailed tomultiple sites or posted to a listserverwithout the copyright holder's expresswritten permission. User, however,may print, download, or email articlesfor individuals use.

www.jethroproject.com, TJP, November 10, 2008, pp. 1-5 Condition Based MaintenanceCopyright 2008 TJP. All rights reserved. 4 Byron A. Ellis

References

Amari, S.V & McLaughlin, L. (2004).Optimal design of condition basedmaintenance Model. IEEE.

Brian W. (1999). Intelligent buildingcare, Facilities, Volume 17 ·Number 5/6 , pp. 189–194.

Buswell, R. A., & Wright, J. A. (2004).Uncertainty in model-basedconditioning monitoring. BuildingServices Engineering Res.Technol, 25 (1), 65-75.

Coetzee, J.L. (1999). A holistic approachto the maintenance ``problem.''Journal of Quality inMaintenance Engineering, Vol. 5No. 3, pp. 276-280.

El-Haram, M.A., and Horner, R.M.W.(2002). Journal of Quality inMaintenance Engineering, Vol.85 No.2, pp. 135-143.

Heckscher, C., Eisenstat, R. A., & Rice,T. J. (1994). Transformationalprocesses. In C. Heckscher and A.Donnellon (Eds.), The post-bureaucratic organization: Newperspectives on organizationalchange (pp. 129-177). ThousandOaks, CA: Sage Publications.

Horner, R.M.W., El-Haram, M.A., andA.K. Munns (1997). Buildingmaintenance strategy: a newmanagement approach, Journal ofQuality in MaintenanceEngineering, Vol. 3 No. 4, pp.273-280.

Knapp, G.M. and Wang, H.P. (1992),``Machine fault classification: aneural network approach,''International Journal ofProduction Research, Vol. 30 No.4, pp. 811-23.

Knapp, G.M., Javadpour, R., and Wang,H.P. (2000), An ARTMAP neuralnetwork-based machine conditionmonitoring system. Journal ofQuality in MaintenanceEngineering, Vol. 6 No. 2, 2000,pp. 86-105.

Stephen R. Letza (1996). The design andimplementation of the balancedbusiness scorecard. BusinessProcess Re-engineering &Management Journal Vol. 2 No.3, 54-76.

Moubray, J (2001). Reliability-CenteredMaintenance. NY: NY IndustrialPress Inc.

Raheja, D., Llinas, J., Nagi, R. &Romanowski, C. (2000). Aholistic view of condition basedmaintenance. Working Paper,December 2000 (Center forMultisource Information Fusion(CMIF), SUNY.

Raheja, D., Llinas, J., Nagi, R. &Romanowski, C. (2006). Datafusion/data mining based onarchitecture for condition basedmaintenance. InternationalJournal of Production Research,

Shonet, I.M. (2003). Building evaluationmethodology for settingmaintenance priorities in hospitalbuildings. ConstructionManagement and Economics

Tsang, A. C. (1995). Condition basedmaintenance: Tools and decisionmaking. Journal of Quality inMaintenance Engineering, 1(3),1-17.

www.jethroproject.com, TJP, November 10, 2008, pp. 1-5 Condition Based MaintenanceCopyright 2008 TJP. All rights reserved. 5 Byron A. Ellis