Journal of Business and Economics, ISSN 2155-7950, USA July 2017, Volume 8, No. 7, pp. 562-574 DOI: 10.15341/be(2155-7950)/07.08.2017/005 Academic Star Publising Company, 2017 ttp://www.academicstar.us Proposal of Maintenance-types Classification to Clarify Maintenance Concepts in Production and Operations Management Flavio Troan 1, Rui F. M. Marçal 2 (1. Department of Electronics, Federal University of Tecnology-Parana (UFPR), Brazil; 2. Department of Production Engineering, Pontifical Catolic University of Parana (PUC-PR), Brazil) Abstract: Due te fact tat tere are a wide terminologies to define types of maintenance, tis paper discuss te some traditional classifications igligted in te literature and presents a proposal for sorting types of maintenance on te most appropriate terminology, taking into account criteria as: criticality, intervention modes, planning actions, costs and available resources. Te proposed metodology evaluated maintenance concepts by multi-criteria analysis and realized a sorting using te ELECTRE TRI outranking metod. Initially te main concepts about terminologies for types of maintenance were explored in order to organize classes for terminologies sorting. In sequence were defined tresolds for te classes, identifying te borders in wat eac type could be allocated. Finally an experiment wit a numerical application was built, considering te terminologies adopted by some autors of te maintenance management area. Te results obtained indicate a pattern scientifically organized for tese concepts, supported by multi-criteria analysis. Tis approac contributes significantly to clarify maintenance concepts linked to Production and Operations Management. Key words: multi-criteria analysis; maintenance types; maintenance concepts JEL codes: L20 1. Introduction Tere is still great confusion in te maintenance management regarding terminologies used for types of maintenance, mainly in te industrial sector, not only Production and Operations Management but also correlated literature. Tis can be a barrier for te definition of a standard terminology and it occurs due te fact tat tere is: Incorrect conceptualization or dissemination of te adopted names for types of maintenance, not always in a good way or satisfactorily explained or understood, but wic are local or particular abits, Neologism, often derived from translations of foreign languages, Definition of particular names by different autors and from specific scenarios. Te terminology may even vary, but it is necessary tat te concept is well understood. Careful standardization is appropriate in order to enable a clear concept to assist te maintenance decision maker to coose te most appropriate type for a piece, equipment, installation or system. Consequently, in te industrial Flavio Troan, Doctor in Production Engineering, Economist and Professor, Department of Electronics, Federal University of Tecnology - Parana (UFPR); researc areas/interests: operational Researc, maintenance management. E-mail: troan@utfpr.edu.br. Rui F. M. Marçal, Doctor in Engineering, Professor, Department of Production Engineering, Pontifical Catolic University of Parana (PUC-PR); researc areas/interests: operational researc, maintenance. E-mail: rui.marcal@pucpr.br. 562
organizations, te economic aspects will be affected by tese definitions. Te evolution of maintenance concepts is according to te productive sector expectations and related to te tecniques used to meet emerging maintenance needs for tis moment. Currently, tere are pilosopies suc as: maintaining reliable and available systems, sutting down for maintenance, sensing and monitoring parameters tat indicate te best time to perform maintenance in order to anticipate failures. Oter concepts suc as proects to ensure te reliability/maintainability and Terotecnology come as reinforcements to anticipate maintenance actions, still in te design pase. For decision making assertiveness, it is necessary to apply a maintenance tecnique effectively. Terefore, knowing te concepts of te most suitable types to be used is essential to do it. Even te combination of two or more types of maintenance sould be well conceptually clarified; because knowing were an application ends and anoter begins is essential to plan and manage efficiently te industrial maintenance. In fact, bot literature and te applications publised in scientific articles, are constantly presenting new terminologies and assignments for maintenance types, wit little variation wen compared to traditional concepts, but wic receive different names and end up causing confusion and saded concepts. Tis work troug te multi-criteria analysis tecnique looks for common caracteristics in known maintenance types from te literature, in order to classify tese types into main classes, taking into account criteria wic are common among tese traditional concepts. Te multi-criteria decision-making metodology derives from te fact tat in most situations were we ave to decide, tere is not only one goal but also different points of view are considered, wic very often conflict wit eac oter. Te ELECTRE TRI metod was used in a numerical application in tis work. Te main obective was to recognize a standard for industrial maintenance types, distributed in classes, covering te common caracteristics of te various types presented in te literature, after applying te metod of multi-criteria analysis. We attempted wit tis application to find clusters of designations tat ave common caracteristics, allocated in teir classes. Wit tis, we can also know te variability of tese concepts in a future and more detailed study on tese classes and te intrinsic caracteristics of te elements of eac set. 2. Historical Evolution of te Maintenance In accordance wit Moubray (2000) te evolution of maintenance can be divided into generations and te timeline of tese generations was adapted as presented in Figure 1. Figure 1 Timeline Maintenance Generations 563
Tis evolution is demonstrated troug te industrial needs in eac generation, in wic te central concepts about te maintenance types and ow tey could ave been classified were igligted. Wit te division of maintenance into generations, also concepts about more efficient maintenance types to be applied in te industrial context were developed. Te first generation gave support to te diversification of corrective maintenance concept, suc as: sceduled, unsceduled, curative, palliative or repair. Te corrective maintenance actually is still repair after damage and appeared as te main concept of te first generation. Te second generation in turn, brougt conceptual elements for te development of preventive maintenance, wic is based on sceduled overauls, planning and control of work systems and te evolution of applied computer science. In te tird generation te evolution of condition monitoring tecniques, failure analysis and studies of risks brougt base to conceptualize predictive maintenance. Te evolution of tese concepts was fueled by growing maintenance expectations. Tese expectations forced te development of new tecnologies tat could elp te productive sector to maintain safety, quality, availability and reliability of teir equipment and consequently of te production processes. In te current generation, inerited from aerospace industry te RCM (Reliability Centered Maintenance) brougt advanced tecniques linked wit reliability and availability and frequently adapted to industrial systems. 2.1 Types of Maintenance Traditionally in Europe, maintenance is sorted and performed by time-based fixed intervals, and it is called preventive maintenance, or by corrective maintenance after breakdown. In te preventive type, maintenance is performed in order to prevent equipment breakdown by repair, or components excange. In te corrective type, maintenance is performed after a breakdown. For some equipment te maintenance must be performed immediately, and for oters types te maintenance can be delayed, depending on te equipment s criticality (Bengtsson et al., 2004). In te European Standard of EN 13306, as sown in Figure 2 te preventive maintenance as been sorted into two categories, condition-based and pre-determined maintenance (CBM) (CEN, 2001). Figure 2 Maintenance Types by CEN (2001) In Germany tere is a specific standard, called DIN 31051, in wic all measures for maintaining and restoring te target condition as well as determining and assessing te actual condition of te tecnical equipment in a system are performed by a sector called Plant Maintenance. Te types are categorized into preventive maintenance, inspection and repairs (DIN, 2003). Tis categorization is sown in Figure 3. 564
Figure 3 Types of Maintenance by DIN (2003) According Williams et al. (2002) it is not required to be an engineer or a tecnician to value te role of maintenance in today s industrialized life tat as been largely urbanized around te application of macines in different aspects. Tsang (2002) complements tis statement spotligting maintenance as a necessary expense tat belongs to operating budget, and a common item on te it list of cost-reduction programs in industries. Around te world, maintenance as been interpreted and classified in too many different ways. In te USA for te US Department of Energy (US DOE) past and current maintenance practices must be performed after it is broken. It underlines te literal meaning of maintenance as te work of keeping someting in proper condition and declares tat maintenance sould be actions taken to prevent a device or component from failing or to repair normal equipment degradation experienced wit te operation of te device to keep it in proper working order. Suc an interpretation catalogs four types of maintenance: reactive, preventive, predictive and reliability-centered maintenance (RCM), as presented in Figure 4 (US DOE, 2004). Figure 4 Maintenance Types by US DOE (2004) Figure 5 Maintenance Taxonomy by Kazraei and Deuse (2011) Kazraei and Deuse (2011) proposed a classification based on te tactics regarded wit traditional maintenance aspects. Based on tis classification, tey present two main strategies in macinery maintenance, namely reactive and preventive maintenance and te various tactics linked wit te maintenance concepts, as 565
sown in Figure 5 and complemented in Tables 1, 2 and 3. Te autors conclude tat obviously, eac of tese tactics can be used by utilizing different metods, tecniques and tecnologies. IRM SRM DRM FBM OTF AGM BBM CIM FTM IBM LBM PPM TBM UBM ACM BCM DOM RBM RCM TPM Abbreviation Abbreviation Abbreviation Table 1 Reactive Tactics in Maintenance Brief description Immediate reactive maintenance Sceduled reactive maintenance Deferred reactive maintenance Failure-based maintenance Operate to failure Table 2 Preventive Tactics in Maintenance Brief description Age-based maintenance Block-based maintenance Constant interval maintenance Fixed time maintenance Inspection-based maintenance Life-based maintenance Planned preventive maintenance Time-based maintenance Use-based maintenance Table 3 Proactive Tactics in Maintenance Brief description Availability centered maintenance Business centered maintenance Design-out maintenance Risk-based maintenance Reliability-centered maintenance Total productive maintenance In Latin America, more specifically in Brazil, were te experimentation was applied, some autors present maintenance classification at te same way tat European and Nort American classification, compounding a miscellaneous of tese classifications. According to Viana (2002), te main types of maintenance can be subdivided as follow: Unplanned corrective maintenance: tis type of maintenance occurs after failure or loss of device performance, witout time for te maintenance preparation services. Tis maintenance-type, despite all te trouble, is often practiced still today. Planned corrective maintenance: it is te correction of lower performance tan expected or failure, for managerial decision, i.e., for lower performance from predictive tracking or te decision to operate until breaking (Pinto & Xavier, 2007). Preventive maintenance: unlike te corrective aims, it seeks to prevent equipment failure. Tis type of maintenance is carried out in equipment wic is not in failure yet, i.e., is still operating wit minimum conditions. Tis type may ave two quite different situations: te first, wen te equipment is switced off before te need for service and te second situation is wen te equipment failed, from incorrect estimate of te repairing period. 566
Predictive maintenance: tis is noting more tan preventive maintenance based on te equipment condition. It is interesting because it allows te monitoring of equipment by measurements carried out wen it is fully operational, enabling ig availability. Te predictive maintenance can be said to predict equipment failure and ten to elp decide wen to make te intervention to repair tis equipment; it is actually a sceduled corrective maintenance. Te Total Productive Maintenance or TPM was originated in Japan in te post-world War II period, wit a pilosopy of work tat must be followed by all sectors of te company. Regarding equipment, Tavares (1999) points out tat it means promoting te revolution along te production line, troug te incorporation of Zero Break, Zero Defect and Zero Accidents. It is te realization of an Equipment Management, i.e., administration of te macines troug organization. Tis pilosopy also incorporates te concept of Autonomous Maintenance, were operators are motivated and trained to carry out cecks, quick cecks and maintenance actions on te operating equipment. Taasi and Osada (1993) reinforce te meaning of te TPM as a broader preventive maintenance based on lifetime and economic applicability of equipment, dies and igs tat play te most important roles in te production. Te Reliability Centered Maintenance or RCM was related to tecnological and social processes arising from te Second World War. Te benefits of RCM were perceived and quickly te metodology was applied in various sectors: nuclear submarines, electrical industry, construction, cemical, and steel, among oters. Te generality of te RCM concepts and tecniques are applicable today to any system, regardless of tecnology. In RCM, te obectives of maintenance are defined by te functions and performance standards required for any item in teir operating environment and its application is an ongoing process and sould be reevaluated frequently, wile experience is accumulated (Lima & Castilo, 2006). According to te Frenc standards AFNOR NF X60-010 and NF X60-011 cited by Moncy (1989) te various types of maintenance can be classified as Corrective Maintenance wen Maintenance is performed after failure; Preventive Maintenance wic aims to reduce te probability of failure. Preventive maintenance can also be explored as Preventive Systematic and Preventive Conditional. Moncy (1989), complementing definitions of te AFNOR standards, divided te Corrective Maintenance in two modes wic are: Curative Corrective Maintenance and Palliative Corrective Maintenance. Te Brazilian standard ABNT, NBR 5462 (1994), classifies te various types of maintenance: Preventive maintenance; Corrective maintenance; Controlled maintenance or Predictive; Sceduled maintenance; Non-Sceduled maintenance; Maintenance in field; Maintenance off-site; Remote Maintenance; Automatic Maintenance; Deferred Maintenance. Te classification adopted by te UN (United Nations) cited by Tavares (1999), covers an interesting feature on te preventive and corrective types of maintenance. Tey were allocated togeter in a tree called planned maintenance. Tus, te UN classification considers te corrective maintenance wit some level of planning and maintenance actions are considered wile operating, not operating or until it breaks (repair by fatigue). Te concept of operation after failure, in tis classification is addressed by maintaining a break down or unplanned maintenance. Te autors Mirsawka and Olmedo (1993) adopted a classification tat separates te preventive maintenance of te corrective one, but it includes predictive maintenance witin te preventive classification, considering it as a prevention based on equipment condition. In tis context te corrective maintenance also brougt te concept of repair after damage and enanced te concept to te improvement in te implementation 567
of corrective maintenance. Te planning criterion is implicit in te concepts adopted by te autors. Summarizing, te basic conditions to establis maintenance types are: te equipment, system or installation sould allow some kind of monitoring; te equipment, system or installation must ave te coice for tis maintenance type ustified by te costs involved; failures must originate from causes tat can be monitored and ave teir progress also monitored. Te Brazilian Association of Maintenance Abraman (2005) sows a broader concept, seeking to meet te expectations of reliability and maintainability of te tird generation of maintenance. Maintenance is classified into tree main brances (or types), represented by maintenance: Corrective, Preventive and Predictive. Witin tese brances, oter nomenclatures conceptualize some of te canges adopted in eac of tese subdivisions. Altoug considered irrelevant, as stated by Viana (2002), te canges tat occur in te maintenance types create a new terminology and a new potential concept or type of maintenance, emerging from a specific application, based on pre-defined criteria. Te definition of criteria for classification is not an easy task because discussions covering maintenance concepts are constantly arising due to te increasing development of advanced maintenance tecniques in various areas of knowledge and industry. Viana (2002) provides elements for at least one criterion tat tis work addresses, wic are te intervention modes on te equipment/instruments. Tese modes can be initially divided into: Intrusive, Non-intrusive. 2.2 Proposal to Sorting Maintenance Types Te metodology in tis paper addresses te sorting problematic in maintenance and proposes to sort te maintenance types in classes, based on collected parameters in te literature and representing te relevant caracteristics to be taken into account in te definitions of te criteria and desired limits. Te proposed metodology is constructed as a sequence of steps tat represent a way to organize and make a most efficient model, placing important information in a sequence tat provides a clear view of it. Te information tat will give relevance for tis classification sould represent caracteristics of maintenance types. Tese caracteristics are defined by te decision maker in te maintenance management area; because e is te one wo can to know te problem in its specificity. Ten, te data collected can be estimated from real processes or troug literature surveys. Te definitions sall provide te kind of classes tat you want to explain to te decision maker. For example, classes are concentrated were te degrees of criticality involving equipment or even classes tat are designed to promote more efficient alternatives for system maintenance, as is te case in tis study. Te criteria for te evaluation of alternatives sould explain weter te decision maker s preference to meet certain minimum parameters for a given alternative fits te expected class and predetermined efficiency. To calculate te performance of alternatives, some tecniques are used, suc as te definition of subective scales and standardization of tese scales in a numerical order wic is a representation of te criteria true importance. Based on tese commonalities, te calculation is performed by te ELECTRE TRI metod, so tat classification is made, for eac alternative presenting numerical caracteristics tat will be taken into account in te framework of te alternatives in teir respective classes. At tis stage, te boundaries between te classes to wic te alternatives will be subect to te frame sould be listed witout any doubt. Te relevance to create a metodology to support decision-making wit multi-criteria analysis to classify maintenance types also derives from te fact tat in tis situation we ave to define based on only one goal but considering different points of view, conflicting wit eac oter, suc as developed in te traditional teory of multi-criteria decision making. Terefore, te decision process sould be guided by multi-criteria analysis 568
metods to support te decision maker in te optimization of te alternatives. According to Vincke (1992), tere are several metods and multi-criteria tecniques, among wic one can identify: te additive models, wic generate a single criterion syntesis (Multiple Attribute Utility Teory MAUT; Analytic Hierarcy Process AHP) and te outranking metods (ELECTRE and PROMETHEE). Tose models and metods follow two main lines, te Multiple Criteria Decision Making MCDM and Multicriteria Decision Aid MCDA. Te adoption of tese models is usually ustified by arguments dictated by te nature of te problem. One of te known families of outranking metods is te family ELECTRE (Elimination Et Coice Traidusaint la REalitè) (Yu, 1992). Te metod used in tis work, ELECTRE TRI, is a multi-criteria metod tat allocates alternatives in predefined categories. Tis allocation of an alternative a results from te comparison of a profiles wit defined boundaries of categories (Mousseau & Slowinski, 1998; Yu, 1992). Given a set of criteria indices {g 1,..., g i,..., g m } and a set of index profiles {b 1,..., b,...b p }, is defined (p+1) categories, in wic b represents te upper limit for te category C and te lower category limit C +1, = 1, 2,..., p. Two conditions must be verified to validate te claim asb (a outranks b ): Concordance: for an outranking asb to be accepted, most of te criteria sould be for te claim asb. Non-discordance: wen te agreement condition is not met, none of te criteria sould oppose te claim asb. In building S, a set of veto tresolds (v 1 (b ), v 2 (b ),..., v m (b )) is used in te discordance test. v (b ), wic represents te minor difference g (b ) - g (a), inconsistent wit te statement asb. Rates of partial agreement c (a,b), concordance c(a,b) and partial discordance d (a,b) are calculated by te expressions (1), (2) and (3). 0if c ( a, b) = 1if k c c( a, b) = F F k g ( b ) g ( a) p ( b ) g ) g ( a) q ) p ( b ) + g ( a) g ( b ) p ( b ) q ( b ) (1) in oter cases (2) 0 if g ) g ( a) p ) d ( a, b) = 1 if g ) g ( a) > v ) (3) g ) + g ( a) p ) in oter cases v ) p ) 3. Proposal Application For te application of te multi-criteria metod ELECTRE TRI 2.0a software was used, te additional Lamsade (Paris-Daupine University, Paris, France). Te problem was structured wit te aim to apply te ELECTRE TRI metod in a setting for te number of alternatives and weigted criteria to evaluate te classification problem for te types of maintenance efficiency. Some subectivity inerent feature of te decision-making process was also considered to establis te relative weigts for te criteria. Table 4 presents a summary of te study tat is organized on te most commonly used maintenance types, based on intervention time. ( a, b ) 569
Table 4 Summary of Researc on te Maintenance Types Autors Classification related wit intervention time Immediate Sort-term Long-term Bycondition - Unplanned - Planned - - UN, (1975) - Emergency - Preventive - - - Corrective - - AFNOR, (1985) Moncy, (1989) Mirsawkaand Olmedo (1993) ABNT, (1994) Viana, (2002) - Corrective - Preventive - Preventive Systematic bycondition - Corrective - Corrective - Preventive - Preventive Curative Palliative Systematic bycondition - Corrective - Improvement - Preventive Systematic - Preventive Conditional - Predictive - Corrective - Preventive - Predictive - At teplace - Unsceduled - Sceduled - - Out ofplace - - - - Remote - - - - Automatic - Corrective - Preventive - Predictive - - Planned - Autonomous - TPM - - Unplanned - - RCM - - Corrective - Preventive - Predictive - - Sceduled - Sistematic - Followup - Abraman, (2005) Pinto & Xavier, (2007) - Unsceduled - Reconstructions - Monitoring - - - Oportunity - Measurement - - - Repairs - Inspections - - - Lubrication - - - Corrective - Preventive - Predictive - Maintenance - Planned - - Detective Engineering - Unplanned - - - Table 5 Criteria, Classes, Boundaries and Weigts Definition Categories / Classes CL 1 CL 2 CL 3 CL 4 Type of action Advanced Predictive Proactive Reactive Boundaries Criticality g 1 Intervention g 2 Criteria Planning g 3 Costs g 4 b 1 90 90 90 25 90 b 2 60 75 50 55 50 b 3 40 25 25 90 25 Weigts 25 15 25 20 15 Resources g 5 Direction of Preferences Max Max Max Min Max Preference and Indifference tresolds 10 10 10 10 10 5 5 5 5 5 570
After te survey, te criteria for evaluation sould be set so tat te alternatives could be framed properly in classes tat represent te action levels of te various maintenance types studied. Table 5 presents te definition of tese criteria and teir classes, relative weigts, directions of preferences and tresolds related wit te maintenance classification process, based on some features found in te literature on te subect. Parameter values try to translate te scale subective languages into numerical variables, making comparisons, as in tis case. Table 6 sets out te numerical scales for te criteria in order to assign values for evaluation. Tese standard values will support te comparison made by te ELECTRE TRI metod and allocations of alternatives in teir classes. Tese scales were considered as subective values (Hig, Medium, Low, Planned, Unplanned, etc.) and teir respective numerical values (0 to 100) in order to serve as input to te ELECTRE TRI 2.0a software, wose values need to perform calculations for comparisons. Table 6 Numerical Scales for Criteria Cr W (%) Criteria Scale Value g 1 25 Criticality g 2 15 Interven-tionMode g 3 25 Planning Level g 4 20 Costs g 5 15 Resources LowLow Low Medium Regular Hig HigHig Intrusive NP IntrPlanned On operation NIntrUnPlanned NIntrPlanned None Low Medium Hig Low Medium Hig Unvailable Unplanned Planned Available 0 20 40 60 80 100 0 25 50 75 100 0 25 50 100 100 50 0 0 25 75 100 In Table 7 te evaluation of 29 alternatives is carried out (maintenance types found in tis study), specifically in relation to te listed criteria, and using te numerical scale from Table 3. 4. Results After applying te ELECTRE TRI 2.0a software, te following results were obtained. Figure 2 sows te results obtained wit te classification performed by te software. Tese results sow te ability of te metod to present an optimized version for te maintenance types exploited by tis work. From te classification results, te optimistic procedure was adopted, due to te fact tat it better represents te purpose intended to reac. It is noticed tat in te Advanced class te types of maintenance tat require iger planning and investments were allocated: Monitoring (MON); Maintenance Engineering (ME), RCM and TPM. On te oter side, i.e., te Reactive class alternatives were linked wit te reaction failure concept. 571
Table 7 Numerical Evaluation of te Types of Maintenance a n MaintenanceTypes Abbreviation Criticality Interv. mode Planning level Cost Resources g 1 g 2 g 3 g 4 g 5 a 1 At te place AP 80 50 25 50 0 a 2 Automatic AUTO 100 50 50 50 75 a 3 Autonomous AUT 60 75 25 50 25 a 4 Byopportunity OP 20 25 50 50 50 a 5 Conditional CON 60 75 50 50 75 a 6 Corrective CORR 0 0 0 100 0 a 7 Corrective Curative CC 20 0 0 100 0 a 8 Corrective Palliative CP 40 25 25 50 25 a 9 Detective DT 60 75 25 50 25 a 10 Improvements IMP 20 50 50 50 75 a 11 Inspections INS 60 25 50 50 25 a 12 Lubrication LUB 40 50 25 50 25 a 13 Maintenance engineering ME 80 50 100 100 100 a 14 Measurement MEA 20 75 50 50 25 a 15 Monitoring MON 100 100 100 50 100 a 16 Out of place OUTP 60 25 25 50 0 a 17 Planned PL 40 50 25 50 50 a 18 Predictive PRED 60 75 100 50 75 a 19 Preventive PREV 40 50 25 50 25 a 20 Preventive by condition PCOND 60 75 50 50 50 a 21 Preventive Systematic PS 40 50 25 50 25 a 22 RCM RCM 100 100 100 25 100 a 23 Reconstructions REC 60 25 25 50 100 a 24 Remote REM 20 50 50 0 25 a 25 Repairs REP 0 25 25 100 25 a 26 Sceduled SCH 40 25 25 50 0 a 27 TPM TPM 100 100 100 0 100 a 28 Unplanned UPL 0 0 0 100 0 a 29 Unsceduled USCH 20 0 0 100 0 Figure 6 Proposed Classification 572
In te intermediate classes (Proactive and Predictive) te allocation of preventive and detective maintenance was revealed, among oters wic ave common caracteristics wit respect to criticality, planning, costs and availability of resources. Te results acieved demonstrated consistency in te classification and followed compatible logic wit te concepts studied. 5. Final Remarks Te maintenance management over te years as become an important ally for te optimization and efficiency of production processes. Understanding te concepts and teir specificities is vitally important to te decision maker at te moment e decides to adopt tecniques to elp increase efficiency and industrial productivity by managing maintenance. Tis paper attempts to make a contribution to traditional concepts of maintenance by consolidating similar terminology in performance classes of te various tecniques known about te types of maintenance currently used. Te use of multi-criteria analysis associated wit te use of te ELECTRE TRI metod, enables te classification of types for maintenance to be seen from an innovative point of view, in wic te maintenance area manager may decide on te basis of an expert study on tese types caracteristics. After te development of tis study it was possible to see te similarities tat so far caused doubts in te understanding of te concepts on te classification of maintenance types. Te classes defined as: Reactive, Proactive, Predictive and Advanced now group types of maintenance tat once brougt confusion in understanding. Certainly te caracteristic variations of eac context will still exist, but now tey are listed under concepts tat give compreensive and clear guidelines on eac type developed by various autors in te field. One can also envision some future work to recognize tese variations witin eac class presenting classification proposals. Tese variations could also be individually treated in teir specificity and furter clarified troug te development of maintenance over te years Acknowledgements Te autors gratefully acknowledge te Brazilian Researc Council (CNPq) for supporting te development of tis study. References ABNT (1994). Brazilian Standards Association, NBR 5462: Reliability and maintainability, Rio de Janeiro. Abraman (2005). Brazilian Maintenance Association, Official Journal of Abraman, ABRAMAN, No. 54. Bengtsson M., Olsson E., Funk P. and Jackson M. (2004). Tecnical design of condition based maintenance system A case study using sound analysis and cased-based reasoning, in: 8t Maintenance Reliability Conference Proceedings, Knoxville, TN. CEN (European Committee for Standardization) (2001). Maintenance terminology, European Committee for Standardization, Brussels. DIN (2003). Fundamentals of maintenance, original document in German: Grundlangen der Instandaltung, Deutsces Institutfur Normung, Beut, Berlin. Kazraei K. and Deuse J. (2011). A strategic standpoint on maintenance taxonomy, Journal of Facilities Management, Vol. 9, No. 2, pp. 96-113. LAMSADE Laboratory, ELECTRE TRI 2.0a software: Metodological guide and user s manual, available online at: ttp://www.lamsade.daupine.fr/mcda/biblio/pdf/mous3docl99.pdf. Lima F. A. and Castilo J. C. N. (2006). Aspects of scientific equipment maintenance of te University of Brasilia, dissertation of economics, Business Administration Scool, (FACE), Brasília, DF. 573
Mirsawka V. and Olmedo N. L. (1993). Maintenance Combating Non-Effectiveness Costs, Makron Books, São Paulo, Brazil. Moncy F. (1989). Te Maintenance Function, Edit. Ebras/Durban, São Paulo. Moubray J. (2000). Reliability Centered Maintenance (RCM), Aladon, United Kingdom. Mousseau V. and Slowinski R. (1998). Inferring an ELECTRE TRI Model from assignment examples, Journal of Global Optimization, Vol. 12, pp. 157-174. Mousseau V., Figueira J. and Naux J. P. (2001). Using assignment examples to infer weigts for ELECTRE TRI metod: Some experimental results, European Journal of Operational Researc, Vol. 130, No. 2, pp. 263-275. Pinto A. K. and Xavier J. N. (2007). Maintenance: Strategic function, Qualitymark, Rio de Janeiro. Taasi Y. and Osada T. (1993). TPM: Total productive maintenance, Institute IMAM, São Paulo. Tavares L. (1999). Modern maintenance management, Novo Pólo Publicações, Rio de Janeiro. Tsang A. H. C. (2002). Strategic dimensions of maintenance management, Journal of Quality in Maintenance Engineering, Vol. 8 No. 1, pp. 7-39. US DOE (2004). Q&M Best Practice Guide, Release 2.0, US Department of Energy, Wasington, DC. Viana H. R. G. (2002). PCM, planning and control of maintenance, Qualitymark, Rio de Janeiro. Vincke P. (1992). Multicriteria Decision-Aid, Jon Wiley & Sons Ltd. Williams J. H., Davies A. and Drake P. R. (Eds.) (2002). Condition Based Maintenance and Macine Diagnostics, Capman & Hall, Boca Raton, FL. Yu W. (1992). ELECTRE TRI. Aspects Métodologiques et Guide d Utilisation, Document du LAMSADE, 74, Université de Paris - Daupine. 574