Advances in system reliability engineering davim - Quickly download the ebook to start your content

Page 1


https://ebookmass.com/product/advances-in-system-

Instant digital products (PDF, ePub, MOBI) ready for you

Download now and discover formats that fit your needs...

Safety and Reliability Modeling and Its Applications (Advances in Reliability Science) 1st Edition Mangey Ram (Editor)

https://ebookmass.com/product/safety-and-reliability-modeling-and-itsapplications-advances-in-reliability-science-1st-edition-mangey-rameditor/

ebookmass.com

Reliability Engineering, Third Edition (Wiley Series in Systems Engineering and Management) Elsayed A. Elsayed

https://ebookmass.com/product/reliability-engineering-third-editionwiley-series-in-systems-engineering-and-management-elsayed-a-elsayed/

ebookmass.com

Engineering Reliability and Risk Assessment Harish Garg

https://ebookmass.com/product/engineering-reliability-and-riskassessment-harish-garg/

ebookmass.com

(eTextbook PDF) for Introduction to One Health: An Interdisciplinary Approach to Planetary Health

https://ebookmass.com/product/etextbook-pdf-for-introduction-to-onehealth-an-interdisciplinary-approach-to-planetary-health/

ebookmass.com

Gateways to Democracy 4th Edition E-book PDF Version –Ebook PDF Version

https://ebookmass.com/product/gateways-to-democracy-4th-edition-ebook-pdf-version-ebook-pdf-version/

ebookmass.com

Law and Ethics for Health Practitioners 1st Edition Sonia Allan

https://ebookmass.com/product/law-and-ethics-for-healthpractitioners-1st-edition-sonia-allan/

ebookmass.com

Hazardous and Trace Materials in Soil and Plants M. Naeem

https://ebookmass.com/product/hazardous-and-trace-materials-in-soiland-plants-m-naeem/

ebookmass.com

When Franny Stands Up Eden Robins

https://ebookmass.com/product/when-franny-stands-up-eden-robins-3/

ebookmass.com

Fundamentals of Organic Chemistry for the JEE - Vol I (Main and Advanced) Ananya Ganguly

https://ebookmass.com/product/fundamentals-of-organic-chemistry-forthe-jee-vol-i-main-and-advanced-ananya-ganguly/

ebookmass.com

Organic Chemistry Janice Gorzynski Smith Dr

https://ebookmass.com/product/organic-chemistry-janice-gorzynskismith-dr/

ebookmass.com

ADVANCES INSYSTEM RELIABILITY

ENGINEERING

ADVANCES INSYSTEM RELIABILITY

ENGINEERING

J.PAULODAVIM

AcademicPressisanimprintofElsevier 125LondonWall,LondonEC2Y5AS,UnitedKingdom 525BStreet,Suite1650,SanDiego,CA92101,UnitedStates 50HampshireStreet,5thFloor,Cambridge,MA02139,UnitedStates TheBoulevard,LangfordLane,Kidlington,OxfordOX51GB,UnitedKingdom © 2019ElsevierInc.Allrightsreserved.

Nopartofthispublicationmaybereproducedortransmittedinanyformorbyanymeans, electronicormechanical,includingphotocopying,recording,oranyinformationstorageand retrievalsystem,withoutpermissioninwritingfromthepublisher.Detailsonhowtoseek permission,furtherinformationaboutthePublisher’spermissionspoliciesandourarrangements withorganizationssuchastheCopyrightClearanceCenterandtheCopyrightLicensingAgency, canbefoundatourwebsite: www.elsevier.com/permissions

Thisbookandtheindividualcontributionscontainedinitareprotectedundercopyrightbythe Publisher(otherthanasmaybenotedherein).

Notices

Knowledgeandbestpracticeinthisfieldareconstantlychanging.Asnewresearchand experiencebroadenourunderstanding,changesinresearchmethods,professionalpractices, ormedicaltreatmentmaybecomenecessary.

Practitionersandresearchersmustalwaysrelyontheirownexperienceandknowledgein evaluatingandusinganyinformation,methods,compounds,orexperimentsdescribedherein. Inusingsuchinformationormethodstheyshouldbemindfuloftheirownsafetyandthesafety ofothers,includingpartiesforwhomtheyhaveaprofessionalresponsibility.

Tothefullestextentofthelaw,neitherthePublishernortheauthors,contributors,oreditors, assumeanyliabilityforanyinjuryand/ordamagetopersonsorpropertyasamatterofproducts liability,negligenceorotherwise,orfromanyuseoroperationofanymethods,products, instructions,orideascontainedinthematerialherein.

LibraryofCongressCataloging-in-PublicationData

AcatalogrecordforthisbookisavailablefromtheLibraryofCongress BritishLibraryCataloguing-in-PublicationData

AcataloguerecordforthisbookisavailablefromtheBritishLibrary ISBN:978-0-12-815906-4

ForinformationonallAcademicPresspublicationsvisitour websiteat https://www.elsevier.com/books-and-journals

AcquisitionEditor: BrianGuerin

EditorialProjectManager: ThomasVanDerPloeg

ProductionProjectManager: SruthiSatheesh

CoverDesigner: HarrisGreg

TypesetbySPiGlobal,India

CONTRIBUTORS

MohiniAgrawal

SchoolofBusiness,GalgotiasUniversity,GreaterNoida,India

AdarshAnand

DepartmentofOperationalResearch,UniversityofDelhi,NewDelhi,India

AlessandroBarbiero

DepartmentofEconomics,ManagementandQuantitativeMethods,Università degliStudi diMilano,Milan,Italy

NavneetBhatt

DepartmentofOperationalResearch,UniversityofDelhi,NewDelhi,India

F.DeCaro

DepartmentofEngineering,UniversityofSannio,Benevento,Italy

LirongCui

SchoolofManagement&Economics,BeijingInstituteofTechnology,Beijing,China

Ant ^ onioCarlosLopesdaCosta

CentrodeDesenvolvimentodaTecnologiaNuclear—CDTN,BeloHorizonte,Brazil

VanderleydeVasconcelos

CentrodeDesenvolvimentodaTecnologiaNuclear—CDTN,BeloHorizonte,Brazil

ChenFang

SchoolofManagement&Economics,BeijingInstituteofTechnology,Beijing,China

SoufianeGasmi

Optimization,ModelingandDecisionSupport(OMAD),UniversityofTunis—Tunis NationalHigherSchoolofEngineering,Tunis,Tunisia

MiroslavKvassay

DepartmentofInformatics,UniversityofZilina,Zilina,Slovakia

DongjinLee

SchoolofComputing,Informatics,andDecisionSystemsEngineering,ArizonaState University,Tempe,AZ,UnitedStatesofAmerica

PreetiMalik

DepartmentofComputerScienceandEngineering,GraphicEraDeemedtobeUniversity, Dehradun,India

MonikaManglik

DepartmentofMathematics,UniversityofPetroleum&EnergyStudies,Dehradun,India

LataNautiyal

DepartmentofComputerScienceandEngineering,GraphicEraDeemedtobeUniversity, Dehradun,India

RongPan

SchoolofComputing,Informatics,andDecisionSystemsEngineering,ArizonaState University,Tempe,AZ,UnitedStatesofAmerica

BrianA.Polin

JerusalemCollegeofTechnology,Jerusalem,Israel

LudmilaPustylnik

DepartmentofMechanicalEngineering,AfekaTel-AvivAcademicCollegeofEngineering, TelAviv,Israel

JanRabcan

DepartmentofInformatics,UniversityofZilina,Zilina,Slovakia

MangeyRam

DepartmentofMathematics,ComputerScience&Engineering,GraphicEraUniversity, Dehradun,India

AmandaLaureanoRaso

CentrodeDesenvolvimentodaTecnologiaNuclear—CDTN,BeloHorizonte,Brazil

AlexanderRotshtein

DepartmentofIndustrialEngineering,JerusalemCollegeofTechnology,Jerusalem,Israel

PatrikRusnak

DepartmentofInformatics,UniversityofZilina,Zilina,Slovakia

LuisMejiaSanchez

CumminsInc.,Columbus,IN,UnitedStatesofAmerica

WellingtonAntonioSoares

CentrodeDesenvolvimentodaTecnologiaNuclear—CDTN,BeloHorizonte,Brazil

A.Vaccaro

DepartmentofEngineering,UniversityofSannio,Benevento,Italy

D.Villacci

DepartmentofEngineering,UniversityofSannio,Benevento,Italy

SylwiaWerbinska-Wojciechowska

DepartmentofOperationandMaintenanceofLogisticSystems,TransportationSystemsand HydraulicSystems,WroclawUniversityofScienceandTechnology,Wroclaw,Poland

PetekYontay

Intel,Inc.,Chandler,AZ,UnitedStatesofAmerica

EDITORSBIOGRAPHY

Dr.MangeyRam receivedaPh.D.degree,majoringinMathematicsand minoringinComputerScience,fromG.B.PantUniversityofAgriculture andTechnology,Pantnagar,India.Hehasbeenafacultymemberforaround 10yearsandhastaughtseveralcorecoursesinpureandappliedmathematics atundergraduate,postgraduate,anddoctoratelevels.HeiscurrentlyaprofessoratGraphicEraDeemedtobeUniversity,Dehradun,India.Before joiningGraphicEra,hewasadeputymanager(probationaryofficer)with SyndicateBankforashortperiod.Heiseditor-in-chiefof InternationalJournalofMathematical,EngineeringandManagementSciences andtheguesteditor andmemberoftheeditorialboardofvariousjournals.Heisaregular reviewerforinternationaljournals,includingIEEE,Elsevier,Springer, Emerald,JohnWiley,Taylor&Francis,andmanyotherpublishers.He haspublished131researchpublicationsinIEEE,Taylor&Francis,Springer, Elsevier,Emerald,WorldScientific,andmanyothernationalandinternationaljournalsofrepute,andalsopresentedhisworksatnationalandinternationalconferences.Hisfieldsofresearcharereliabilitytheoryandapplied mathematics.HeisaseniormemberoftheIEEE,lifememberofOperationalResearchSocietyofIndia,SocietyforReliabilityEngineering,QualityandOperationsManagementinIndia,IndianSocietyofIndustrialand AppliedMathematics,memberofInternationalAssociationofEngineers inHongKong,andEmeraldLiteratiNetworkintheUK.Hehasbeena memberoftheorganizingcommitteeofanumberofinternationaland nationalconferences,seminars,andworkshops.Hehasbeenconferredwith “YoungScientistAward”bytheUttarakhandStateCouncilforScienceand Technology,Dehradun,in2009.Hehasbeenawardedthe“BestFaculty Award”in2011andrecentlyResearchExcellenceAwardin2015forhis significantcontributioninacademicsandresearchatGraphicEra.

Dr.J.PauloDavim receivedaPh.D.degreeinMechanicalEngineeringin 1997,anM.Sc.degreeinMechanicalEngineering(materialsand manufacturingprocesses)in1991,aMechanicalEngineeringdegree(5years) in1986fromtheUniversityofPorto(FEUP),theAggregatetitle(Full Habilitation)fromtheUniversityofCoimbrain2005,andaD.Sc.from LondonMetropolitanUniversityin2013.HeisEurIngbyFEANI-Brussels andseniorcharteredengineerbythePortugueseInstitutionofEngineers withanMBAandSpecialisttitleinEngineeringandIndustrialManagement. Currently,heisProfessorattheDepartmentofMechanicalEngineeringof theUniversityofAveiro,Portugal.Hehasmorethan30yearsofteaching andresearchexperienceinManufacturing,MaterialsandMechanical& IndustrialEngineeringwithspecialemphasisinMachining&Tribology. HealsohasaninterestinManagement,EngineeringEducation,andHigher EducationforSustainability.Hehasguidedlargenumbersofpostdoctorate, Ph.D.,andMastersstudentsaswellascoordinatedandparticipatedinseveral researchprojects.Hehasreceivedseveralscientificawards.Hehasworkedas evaluatorofprojectsforinternationalresearchagenciesaswellasexaminer ofPh.D.thesesformanyuniversities.Heistheeditor-in-chiefofseveral internationaljournals,guesteditorofjournals,bookeditor,bookserieseditor,andscientificadvisoryformanyinternationaljournalsandconferences. Presently,heisaneditorialboardmemberof25internationaljournalsand actsasreviewerformorethan80prestigiousWebofSciencejournals.In addition,hehasalsopublishedaseditor(andcoeditor)morethan100books andasauthor(andcoauthor)morethan10books,80bookchapters,and400 articlesinjournalsandconferences(morethan200articlesinjournals indexedinWebofSciencecorecollection/h-index45+/6000+citations andSCOPUS/h-index52+/8000+citations).

PREFACE

Advancesinmodelingandsimulationisarguablyoneofthemostmultidimensionaltopicsthatonecanfaceinsystemreliabilityengineeringtoday. Thisrapiddevelopmentalsocreatesmanyopportunitiesandchallenges forbothindustrialistsandacademics,andhascompletelychangedtheglobal designandsystemengineeringenvironment.Moremodelingtaskscannow beundertakenwithinacomputerenvironmentusingsimulationandvirtual realitytechnologies.

Throughthisbook, RecentAdvancesinSystemReliabilityEngineering, engineersandacademicianswillgaintheknowledgetohelptheminsystem reliabilityengineering.Thisbookismeantforthosewhoarepursuing reliabilityengineeringasasubjectofstudy.Thematerialisintendedfor anaudienceatthelevelofpostgraduateorseniorundergraduatestudents. That’swhysystemreliabilityengineeringisnowawell-recognizedand rapidlydevelopingbranchofengineering.

Topicsoffocusinclude:

•Delay-time-basedmaintenancemodelingfortechnicalsystems

•Probabilisticanddeterministicsafetyanalysis

•Reliability-gametheory

•Integratingreliabilitymodelsandadaptivealgorithmsforwindpower forecasting

•Time-dependentanalysisofseries-parallelmultistatesystemsusingstructurefunctionandMarkovprocesses

•Modeling-correlatedcountsinreliabilityengineering

•Statisticalinferenceofanimperfectrepairmodelwithuniformdistributedrepairdegrees

•Methodoffuzzyperfectnessinhumanreliabilityanalysis;selectionof performanceconditions

•SystemreliabilityassessmentthroughBayesiannetworkmodeling

•Multistatemultifailuressystemanalysiswithreworkingstrategyand imperfectfaultcoverage

•Softwarepatchschedulingpolicyincorporatingfunctionalsafety standards

•Toolsandtechniquesinsoftwarereliabilitymodeling

Throughthisbook,theundergraduateandpostgraduatestudentsofengineering,engineers,researchscientists,andacademicianswillgaintheknowledge

tohelpthemintheirreliabilitycourses.Thisbookismeantforthosewhotake reliabilityandsafetyasasubjectofstudy.

TheeditorsacknowledgeElsevierforthisopportunityandfortheir professionalsupport.Finally,wewouldliketothankallthechapterauthors fortheiravailabilityforthiswork.

MangeyRam

ACKNOWLEDGMENTS

TheeditorsacknowledgeElsevierandtheeditorialteamfortheiradequate andprofessionalsupportduringthepreparationofthisbook.Also,wewould liketoacknowledgeallofthechapterauthorsandreviewersfortheiravailabilityforworkonthisbookproject.

GraphicEraDeemedtoUniversity,Dehradun,India

J.PauloDavim UniversityofAveiro,Aveiro,Portugal

MangeyRam

CHAPTER1

Delay-Time-BasedMaintenance ModelingforTechnicalSystems

TheoryandPractice

SylwiaWerbinska-Wojciechowska

DepartmentofOperationandMaintenanceofLogisticSystems,TransportationSystemsandHydraulic Systems,WroclawUniversityofScienceandTechnology,Wroclaw,Poland

Abstract

Thischapterwillbeavaluableresourceforunderstandingthelatestdevelopmentsin maintenancemodelingissuesinthefieldofdelay-timeapproachimplementation.This chapterpresentstheliteraturereviewondelay-timemodelingforsingle-andmultiunit systems.Themaintenancemodelsforsingle-unitsystemsassumetwo-stageorthreestagefailureprocesses.Inthecaseofcomplexsystems,thediscussedproblemsarewith regardtomodels’ parameters,estimationissues,casestudiesanalysis,orhybridmodelingapproachimplementation.Themaintenancemodelsformultiunitsystemsexamine theknownmodelsfornonseriessystems.Acasestudyformaintenancemodelingfor multiunitsystemsbasedonadelay-timeapproachisprovided.Thedirectionsforfurther researchworkaredefined.

Keywords: Delay-timeconcept,Multiunitsystems,Literaturereview,Maintenance modeling,Optimization

Abbreviations

AIC

Akaikeinformationcriterion

BI BlockInspection

CM correctivemaintenance

DT delaytime

DTA delay-timeapproach

DTM delay-timemodels

ELECTRE eliminationandchoiceexpressingthereality

HPP homogeneousPoissonprocess

MAUT multiattributeutilitytheory

MC MonteCarlo

MLE maximumlikelihoodestimation

MRT meanrepairtime

MTBF meantimebetweenfailures

NHPP nonhomogeneousPoissonprocess

PAR proportionalagereduction

PM preventivemaintenance

PROMETHEE preferencerankingorganizationmethodforenrichmentevaluations

AdvancesinSystemReliabilityEngineering © 2019ElsevierInc. https://doi.org/10.1016/B978-0-12-815906-4.00001-4 Allrightsreserved.

ImportantNotations

c(Tin) expectedcostovereachinspectioncycle

c(Tini) expectedcostover ithinspectioncycle

C(Tin) totalcostsresultingfromchosenmaintenancepolicy

ccs clean-upcostconnectedwithcleaningupanyspillagethatmayresultina failure

cin costofinspectionactionperformance

cp costofpreventivereplacementofaunit

cr costoffailedunitreplacement

din timeofsingleinspectionactionperformance

dp thetimeofpreventivereplacementofasystem

dr timeofcorrectivereplacementofasystem(afterafailure)

E[x] expectedvalueofrandomvariable x

Ed(Tin) expecteddowntimeinaninspectioncycleoflength Tin

F(t) probabilitydistributionfunctionofsystem/unitlifetime; F(t)¼1 F(t)

Fh(h) probabilitydistributionfunctionofsystemdelaytime

fh(h) probabilitydensityfunctionofsystemdelaytime

Fhi(h) probabilitydistributionfunctionof ithelementdelaytime

fhi(h) probabilitydensityfunctionof ithelementdelaytime

Gh(t) probabilitydistributionfunctionoftheinitialtime u,whichelapsesfrom thebeginningofoperationby“asgoodasnew”elementsofasystemuntil themomentoffirstsymptomsoffailureoccurrence

gh(t) probabilitydensityfunctionoftheinitialtime u,whichelapsesfromthe beginningofoperationby“asgoodasnew”elementsofasystemuntil themomentoffirstsymptomsoffailureoccurrence

Ghi(t) probabilitydistributionfunctionofinitialtime u of ithelementinasystem

ghi(t) probabilitydensityfunctionofinitialtime u of ithelementinasystem

h delaytimeofadefect,denotingtheperiodbetweenthemomentofappearanceofthefirstsymptomsofpotentialfailureandthemomentofan object’sfailureoccurrence

ku constantrateforfault’sarrivalinatechnicalsystemforany inspectionperiod

Mcs measureofpossibleimpactofafailureofapieceofequipmentonan environment

Menv(Tin) environmentalmodelofatechnicalobject

m numberoffailuresuntilelementreplacement

n numberofelementsinasystem

nk minimumnumberofoperatingcomponentstomakeasystemfunction (performancein nk-out-of-n reliabilitystructure)

Nrin(ti–1 in, tiin) expectednumberoffailuresoverinspectioninterval(ti–1 in, tiin)

pcs probabilityofafailureresultinginaspillagerequiringcleanup pin probabilityofimperfectinspectionoccurrence

PuF(Tin) probabilityofafaultarisingasabreakdowninasystem

pw probabilitythat,duringsysteminspection,performancesymptomsof forthcomingfailures(iftheyoccurinasystem)areidentified

R(t) reliabilityfunctionofasystem

RTin(t) componentreliabilityattime t

RTinic acubicapproximationofaveragereliabilityoveran ithinspection

RTiniL alinearapproximationofaveragereliabilityoveran ithinspection

rmr repaircostrate

rTin (i ) (t) reliabilityfunctionforacomponentattime t

r˙(k)

Tin(kTin) left-handderivativeofreliabilityfunctionat t ¼ kTin

tf randommomentoffailureoccurrenceinasystem

Tin timebetweeninspectionactions’performance

Tini timeof ithinspectioncycleperformance(betweentwoconsecutiveinspectionactions’performance)

TM(Tin) totalexpectedlengthofarenewalcycle

TM(Tini) expectedlengthof ithinspectioncycle

TOP randomvariabledenotingsystem’slifetime

TOP(Tin) expectedlengthofelement/system’slifetime

tu momentofthefirstsymptomsofpotentialfailureoccurrence u initialtimeofadefect

δh standarddeviationfornormaldistributionofadelaytime h

δr standarddeviationfornormaldistributionofatimetofailure

λh(t) intensityfunctionofasystemdelaytime

λu(u) randomrateforfault’sarrivalinatechnicalsystemforanyinspectionperiod

1INTRODUCTION

Manyreal-lifesystemsmaydisplaysomesymptomsofforthcomingfailure. Oneexamplemaybeaproductionprocess,whichmaystartproducing defectiveitemsaftersomerandomamountoftime.Ifthesituationisnot corrected,productqualitygraduallydeterioratestoalevelwhereitisselfevidenttoanoperatorthatthesystemhasfailed.Byinspectingtheproduct qualityatsomeintervals,theoperatormaybeabletoreducethecost incurredwiththesystemlifetime [1].Thesetypesofcomponentsmaybenefitfromaninspectionpolicywherebyacomponentisinspectedforadefect andconsequentlyreplacedatinspectiontopreventfailure [2].Recent reviewsoninspectionmaintenancemodelingissuesarepresentedintheliterature [3–9].

Thebasicinspectionmodelsareextendedbyusingatechniquecalled delay-timeanalysis(DTA),whichwasdevelopedformodelingtheconsequencesofaninspectionpolicyforanysystem [10].Thisapproachwasfirst describedandinvestigatedbyChristeretal. [10–16]

Thismaintenanceconceptisbasedontheassumptionthatbeforeasystembreaksdown,therewillbesomesignsofitsreducedperformance.The timebetweenthefirstidentificationofabnormalities(calledinitialpoint)

Initial time — during this time interval the defect cannot be visible

Delay time — during this time interval the defect can be identified by an inspection

Time point when a fault could be first identified

Time point when a component fails if no maintenance intervention during time h is conducted

andtheactualfailuretime(calledfailurepoint)iscalledadelaytimeand determinesthebestopportunitytocarryoutmaintenanceoraninspection (Fig.1).Asaresult,thedelay-timeconceptdefinesatwo-stageprocessfor three-statedsystems(beingeitherinupstate,partiallyupstate,orindownstate).Formoreinformation,theauthorrecommendsreadingWang etal. [17]

Theinspectionschemesindelay-timemodelsmaybeperiodicorbased oncondition-basedmaintenanceimplementation [18].Inthischapter,the authorfocusesonperiodicinspectionmaintenancemodelingissues.More informationaboutcondition-baseddelay-timemodelsmaybefoundin theliterature [19–25].

TheinterestinDTmodelingissueshasincreasedsignificantlyinrecent years.Literaturereviews,inwhichdelay-timemodelsareinvestigatedalong withotherpreventivemaintenancemodels,aregiveninseveralstudies [26–32].Thestatesofartworks,dedicatedtoDTmodeling,weremostly developedinthe1990s.Oneofthefirstliteraturereviewswasgivenby Christeretal. [19],wheretheauthorsdiscussedabasicDTmodelfora single-unitcasetakingintoaccountmodelparametersestimationproblems. Later,thestateofartwasupdatedinChristeretal. [33].Inthiswork,mathematicalmethodsforupdatingdelay-timemodelsofindustrialinspection practiceareproposed.Theypresentedalineardelay-timeupdatemethod andmodelparametervariation.Moreover,theauthorsalsoprovideacombinationofthesemethodsinthemaintenanceareaanddefinesthecriteriafor choosinganupdatingmethod.

Christerlaterpresentedabasicdelay-timemodel [34] anddiscussedthe maindevelopmentdirections,includingperfect/nonperfectinspection

Fig.1 Time-delay-modelingconcept.

cases.TherehavebeensimpleDTmodelsinvestigatedforarepairablecomponentandforacomplexplant.Moreover,theauthorsfocusedontheproblemsofaparameterestimationprocessandDTmodels’implementation possibilities.Theworkendswithadiscussionoffurtherdevelopmentsin delay-timemodeling.Thepresentedstateofartwasfurtherextendedby Christer [35].Inthiswork,theauthorfocusesoninvestigationofindustrial maintenanceproblemsandpresentsbasicmaintenancemodelsforarepairablecomponentandacomplexplant.Themainextensionsofthebasic modelsinclude,forexample,nonperfectinspectioncaseandnonsteady-state conditions.

Oneofworksthatprovidesrecentadvancesindelay-time-basedmaintenancemodelingisgivenbyWang [36].Inthiswork,theauthorfocuseson themaindevelopmentdirectionsindelay-time-basedmodels.Asaresult,he investigatesthedelay-timemodelsforoptimizedinspectionintervals(distinguishing,forexample,imperfectinspectionandmaintenanceissuesormultipleinspectionintervalsproblem)andDTmodelingapplicationareas (includinge.g.,theproblemofsparepartinventoryandmaintenance modelingormaintenanceoutsourcingissues).Theauthoralsodefinesthe maindirectionsforfurtherresearchintheanalyzedmaintenancearea.

Formoreinformation,theauthorrecommendsfurtherreadingofthe literature [2,6,37,38].Somerecentdevelopmentsinthisareaarealsoavailableforfurtherreading [17,21,39].

Followingtheseconsiderations,in Fig.2,themainclassificationofexistingdelay-timemodelsispresented.Theproposedclassificationdividesthe knownmodelsintotwomaingroupsofinspectionstrategies:single-and multiunitsystems.

Tosumup,theauthorsummarizesandreviewstheexistingliteratureon periodicdelay-timemaintenancemodelingandclusterstheexistingworks intoseveralfields,mostlybasedontheclassificationonsingle-andmultiunit models.

Classificationschemeofdelay-timemodels(periodicinspection).

DT models for multiunit system
Models for system with two-stage failure process
Models for system with three-stage failure process
Models for complex system
Delay time (DT) models for technical system
DT models for single-unit system
Models for multiunit system in nonseries reliability structure
Fig.2

ThearticlesreferredtointhisstudywerefoundusingGoogleScholarasa searchengineandScienceDirect,JStor,SpringerLink,SAGEJournals,and Taylor&Francisasonlinedatabases.Theauthorprimarilysearchedtherelevantliteraturebasedonkeywords,abstracts,andtitles.Thefollowingmain termsand/oracombinationofthemwereusedforsearchingtheliterature: delaytimemaintenance or inspectionmaintenance.Asaresult,130papersfrom 1982to2018arereviewedinthisstudy.

Theapplicabilityofthegivenmaintenanceapproachispresentedina casestudy.Thecasestudyisdevelopedbasedonhistoricaldataofmaintenanceandoperationprocessesoffourproductionplants(injectionmolding machines)operatedbyaninternationalcompanylocatedinPoland.

Theseproductionplantshavedefinedinspectionpoliciesbasedonserviceregulationsdefinedbyaproducer.However,acompanyfindsthatsome developmentsinthisareamaybeusefultomakefurtherdecisions.

Followingthis,thischapterisorganizedasfollows. Sections2and3 examinevarioustypesofdelay-timemodelingapproachesforsingle-and multiunitsystems,whicharethemostwidelyknownintheliterature.In Section4,acasestudyisproposedtoinvestigatetheoptimalinspection intervalforafour-unitsystemperforminginaseriesreliabilitystructure. Section5 concludesthechapterandhighlightsfutureresearchinDT modeling.

2DELAY-TIMEMODELSFORSINGLE-UNITSYSTEMS

First,therearedelay-timemaintenancemodelsforone-unitstochastically failingordeterioratingsystemthatwereinvestigatedinwhichthesystem’s defectivestatecanbedetectedbyactualinspection.Inthisarea,theknown modelshavebeenexploredintwomaindimensions:optimizationofinspectionpolicyforsystemswithatwo-stagefailureprocessandforsystemswitha three-stagefailureprocess.Moreover,someresearchworkscanbedefined thatareaimedatinvestigationanddevelopmentofspecialproblems(e.g., casestudies,comparativestudies).

2.1Delay-TimeModelsforSingle-UnitSystemsWith aTwo-StageFailureProcess

Thebasicdelay-timeconceptconsidersthatthedegenerativeprocessofasystemcanbedividedintotwostages:anormalworkingstageandadelayeddefect stage.Forthismodelingcase,themaindevelopmentsaregivenin Fig.3.

Systems with a single failure mode

Systems with a multiple failure mode

Models for systems with a two-stage failure process

Preparedness systems maintenance case

Systems with postponed maintenance

Approximation modeling case

Fig.3 Classificationschemeofdelay-timemodelsforsystemswithatwo-stagefailure process.

2.1.1DTMforaSingle-DefectCase

OneofthefirstdevelopedDTmodelsforsingle-unitsystemsispresentedby Christer [11].Inhiswork,theauthordevelopsthereliabilitymodelfora singlecomponentsubjecttoonetypeofinspectabledefect,whichwillsubsequentlyleadtoafailure.Inthepresentedmodel,underthepolicyof inspectingevery Tin timeunit,theauthordevelopsafunctionofcomponent reliabilityattime tRTin(t).

Takingintoaccountthefollowingmainassumptions:

•afaultmayariserandomlyattime u sincenewwith pdfgh(u), •aperfectmaintenancepolicy,

•thetimerequiredforinspectionisnegligible, •variables u and h areassumedtobeindependent, •ainfinitehorizoncase, therequiredreliabilityfunction RTin(t)isgivenby [11]:

RTin t ðÞ¼ r i ðÞ Tin t ðÞ,where i 1 ðÞTin t < iTin (1)

where i isapositiveintegerand rTin (i ) (t)isareliabilityfunctionforthecomponentattime t,givenbytheformula [11]: r i ðÞ Tin t ðÞ¼ X i 1 ðÞ j ¼1 ð jTin u¼ j 1 ðÞTin gh u ðÞMh jTin u ðÞdu!r i j ðÞ Tin t jTin ðÞ "# ( + ð ∞

u ðÞdu + ð

i 1 ðÞTin gh u ðÞMt u ðÞdu), where i 1 ðÞTin t < iTin (2)

where Mh(x)isdefinedas [11]: Mh x ðÞ¼ ð ∞ x

Intheabsenceofaninspectionprocedure(Tin ¼ ∞),thereliabilityfunctionisgivenbythefollowingformula [11]:

Forsuchamodel,anumericalexampleisprovided.

DTmodelsforsingle-unitsystemsarelaterinvestigatedinworksby RedmondandWang [2,17].Accordingtotheseworks,thereisapossibility todefinethe C.d.f. oftimetofailure, F(x),astheconvolutionof u and h such that u + h x [2,17]:

andthereliabilityfunction, R(x) ¼ 1 F(x).

Takingintoaccountthefollowingadditionalassumptionsandnotation [2,17]:

•thesystemisrenewedateitherafailurerepairoratarepairdoneatan inspectionifadefectisidentified,

•aftereitherafailurerenewalorinspectionrenewal,theinspectionprocessrestarts,

•maintenanceactionsrestoremaintainedcomponentstogoodasnew condition,

•failuresofthesystemareidentifiedimmediately,andrepairsorreplacementsaremadeassoonaspossible,

•perfectinspectionscase,thatis,everydefectisassumedtobeidentified duringinspectionactionperformance,

theexpectedcostovereachcycle, c(Tin),maybedefinedbythefollowing equation [2,17]:

cTinðÞ¼ cr FTinðÞ + cp + cin ð Tin u¼0 gh u ðÞ 1 Fh Tin u ðÞ ðÞdu + cin 1 Gh Tin ðÞ ðÞ (6)

Theexpecteddowntime Ed(Tin)ofaninspectioncycleoflength Tin is givenas [34]:

Tin ðÞ¼ dr dp din FTinðÞ + dp Gh Tin ðÞ + din (7)

Assuminginstantaneousinspectionandreplacementtimes,theexpected cyclelength, TM(Tin),isgivenby [2,34]:

Examplesofsuchmodelimplementationsaregivenintheliterature [2, 17],aswellasananalysisoftheproblemofmodelparameterestimation [17]. ThismodelislaterextendedbyJodejko-Pietruczuketal. [40,41].Inone study,theauthorsfocusonthedevelopmentofamathematicalmodelfora technicalobjectmaintenance,whichextendstheapproachusedthusfarand allowsanalyzingthelong-termoperationtimeperiodofasingle-unitsystem [40].Themodelgivesthepossibilitytofindaconstanttimeperiodbetween theinspectionactions’performancethatisoptimalduetominimalmaintenancecostsormaximalavailabilitycriterionsatisfaction.Thesolutionis basedonBlockInspectionpolicyimplementation.TheauthorspresentanalyticalandMonteCarlosimulationmodelswithcomprehensivesensitivity analysisperformance.Thismodelisanextensionofapreliminaryanalytical maintenancemodel [42]

AnotherextensionispresentedinworkbyJodejko-Pietruczuketal. [41],wheretheassumptionofimperfectinspectionsisintroduced.Inthis work,itisassumedthatthesysteminspectionactionsareimperfect,which meansthatthedefectoccurrenceinthesystemisidentifiedwiththegiven probability pw.Insuchasituationintheanalyzedsystem,thefollowing maintenanceactionsmayoccur:

•preventivereplacementduringinspectionactionperformance,provided thatthedefecthasoccurredandbeenidentifiedwiththeprobability pw, •inspectionactionperformancewithoutpreventivereplacement,providedthatthedefecthasnotoccurrednorbeenidentifiedwithprobability(1 pw).

Basedonthemainassumptionsgivenearlier [40],thelong-termexpected maintenancecostsperunittimefunctionisdefined.

2.1.2DTMforaMultidefectCase

Oneoftheextensionsindelay-timemodelingregardsconsiderationofmore thanonetypeofinspectableandindependentdefectoccurrenceinasystem. Thefirstmodelforsingle-unitsystemswith n differenttypesofinspectable defectsisdevelopedbyChrister [11] andlaterextendedbyWang [43]. Wang [43] investigatedtheproductionprocesssubjectedtotwotypesof deterioration.Thedelay-time-basedinspectionmodelwasdevelopedfor

acomponentwithaminordefectandamajordefect.Itisassumedthatthe minordefectisfixedbyminorperfectinspectionsandrepairs,whereasthe majordefectisdealtwithbymajorimperfectinspectionsandrepairs.

Themajordefects,iftheyoccur,maynotbeidentifiedbyminorinspectionsandrepairs.Theyhavetobedetectedandrectifiedbymajorinspectionsandrepairs,otherwisetheycanleadtoabreakdownoftheprocess. Themodelalsotakesintoaccountthepossiblerelationshipbetweenthe twodefineddefectsduetothepossibleinfluenceofmajordefectsonthe minordefectsoccurringinthesystem.Theauthorsaimedatfinding theoptimalvaluesforbothtypesofinspectionintervalsbyminimizing thelong-runexpectedcostperunittime.

Thecaseofimperfectmaintenancemodelforsinglecomponentswith multiplefailuremodeswaslateranalyzedbyLietal. [44].Theauthorsin theirworkdescribeanaccumulativeageconceptandimperfectinspection maintenanceduetoinsufficientallocationofmaintenanceresources.The modelisbasedontheoptimizationoftheaveragecostperunittimeover aninfiniteperiodandaveragereliabilityfunction.

2.1.3ApproximationModelingCase

Anotherextensionofthebasicdelay-time-basedmaintenancemodelfor single-unitsystemsregardsapproximationmodeling.Theinspectionmaintenancemodelgivenearlier [11] wassimplifiedbyCerone [45].The authorinhisworkdevelopedanaveragereliabilityoveran ithinspection period RTini c asacubicapproximation:

andalinearapproximation:

Theauthorshowsthatthelinearapproximationgivesarelativeerrorof theorderof10%andbyacubicresultinginarelativeerroroflessthan1%.

Inanotherstudy [46],theauthorinvestigatedaproblemnamed“the Converseproblem”.ThemodelgivenbyChrister [11] ishereextended bythedevelopmentofacostmodel,whichinvolvesdeterminingboth thenumberofinspectionsandtheinspectionintervalthatwillproduce themaximumreliabilityatsomefuturepointintimeatminimumcost. Thereisalsoanumericalexampleprovided.

Inanotherwork [47],anewsimplifieddelay-timemodelofthereliabilityfunctionwithinspectionswasobtained.Takingintoaccountthesame assumptionsasCerone [46],theauthorformulatedthereliabilityfunctions whenthedistributionsoftimeofdefective u anddelaytime h arerespectivelynegativeexponentialdistributionswithparameters λu and λh.ThereliabilityfunctionforthecomponentisgivenbyAttia [47]:

Theauthoralsoprovidedanumericalexample,wherefordifferent valuesof λu and λh,thereliabilityfunctionisgiven.

Anotherapproximationmodelwaspresented [48],wheretheauthor presentsthemodelandmethodtofindthetimeliness-basedoptimalinspectioninterval.TheproposedsolutionisbasedonMonteCarlosimulationuse, andtheoptimizationprocessusesthecostratefunction.

2.1.4PreparednessSystemsMaintenanceCase

TheissuesofapreparednesssystemmaintenancewithDTconceptusewere investigatedbyJiaandChrister [49].Theauthorsdevelopedthelong-run availabilityforapreparednesssystembasedontherenewaltheoryuse. Moreover,theauthorsintheirworkinvestigatedvariousvariantsofthebasic model,includingthesituations,when(1)adelay-timeperiodexists,butthe technologytodetectadefectisnotavailable;(2)thedelaytimeiszero,so thatonlyfailuresaredetected;or(3)thesystemisreplacedonaregularbasis withoutanystatetesting.

Theproblemofpreparednesssystemmaintenanceoptimizationwasfurtherinvestigated [50].Theauthorsintheirworkassumethatacomponent maybeeitherweakorstrong,sothatthetimeinthegoodstatehasadistributionthatisamixture.Followingthis,themainextensionsofanother study [49] regardstheimplementationofatwo-phaseinspectionpolicywith ananticipatedhighinspectionfrequencyinearlylifeandlowinspection

frequencyinlaterlifeofthecomponent.Thecostandreliabilitymodelsare developedforfiniteandinfinitehorizoncases.

Anotherextensionofthepreviouslydiscussedmodel [49] ispresentedby Wangetal. [51].TheauthorsintheirworkintroducedtheavailabilityDT modelwithimperfectmaintenanceatinspection.Theproportionalage reduction(PAR)modelwasusedtopresentthattheaccumulativeage (wear)causedbypreviousimperfectmaintenancewillaffecttheinstantaneousratesofdefectandfailureatthenextmaintenancestage.Thesolution isbasedonrenewaltheoryuse.Moreover,theauthorspresentedthemodel parameters’estimationmethodbasedonmaximumlikelihoodestimation (MLE)use.

Aninterestingblock-basedinspectionmodelforasinglecomponentwas presentedbyZhangetal. [52].Theauthorsintheirworkfocusedonthe effectofthepossibleoverlappingofinspectionspanwithfailurerenewal ondeterminingtheinspectioninterval.Basedontherenewaltheory,the s-expectedintervalavailabilityfunctionisdevelopedfortheinfinitehorizon case.ThesolutionisalsoprovidedwithMCsimulationimplementation. Moreover,thepresentedmodelwasalsocomparedwiththeage-based inspectionmodeltoshowitspracticalsense.

2.1.5PostponedMaintenanceCase

Thelastproblemregardspostponedreplacementintroduction.VanOosterometal. [53] developedaDTmodeltodeterminetheoptimalmaintenancepolicyrelaxingassumptionofinstantaneousreplacementenforced assoonasadefectisdetectedduringinspection.Theauthors’keymotivationistoachievebetterutilizationofthesystem’susefullifeandto reducereplacementcostsbyprovidingasufficienttimewindowtoprepare necessarymaintenanceresources.Asaresult,thePMreplacementcostis modeledasanonincreasingfunctionofthepostponementinterval.Moreover,thereareanalyzedtwocases:whendelaytimeiseitheradeterministic orarandomvariable.ThismodelwaslaterextendedbyYangetal. [54], wheretheauthorsinvestigatedasystemthatsuccessivelyexecutesmissions withrandomdurations.Followingthis,theinspectionsarecarriedoutperiodicallyandimmediatelyafterthecompletionofeachmission(random inspection).Asaresult,thereplacementisimmediateifthesystemfails orisdefectiveataperiodicinspection.However,inthesituationwhen thesystemisdefectiveatarandominspection,thenreplacementwillbe postponedifthetimetothesubsequentperiodicinspectionisshorterthan apredeterminedthreshold,andimmediateotherwise.Basedonthemain

assumptionsgivenbyWang [55],thecostmodelisderivedtoobtainan optimalperiodicinspectionintervalandpostponementthreshold.Moreover,thecontinuationoftheinvestigationsonpostponedreplacementis presentedinfurtherstudy [56].Theauthorsintheirworkinvestigated imperfectinspectionperformanceandopportunityreplacementsthatarise afterapositiveinspection.

2.2Delay-TimeModelsforSingle-UnitSystemsWith aThree-StageFailureProcess

Intheknownliterature,therearesomeresearchthatintroducesatwo-level inspectionpolicymodelforsingle-componentsystemsbasedonathreestagefailureprocess.Suchafailureprocessdividesthesystem0 slifeintothree stages:good,minordefective,andseveredefectivestages.

OneofthefirstworkthatinvestigatesDTAimplementationforsystems withathree-stagefailureprocessisgivenbyWang [57].Inthedeveloped maintenancemodel,theinspectionactionsmaynotbeperfect,inthesense thattheminordefectivestagemaybemissedwithagivenprobability,but theseveredefectivestageoftheitemisalwaysidentifiedperfectly.Asameasuretooptimizetheinspectioninterval,theauthorusesthelong-run expectedcostfunctionfortwomainmodelswithrespecttothetwooptions connectedwithmaintenanceactionstaken.Thesolutionisalsopresentedon theexampleofmaintenanceoptimizationforthecoldwaterpumpsusedina softdrinkcompany.

Yangetal. [58] extendedthemodeldiscussedearlier [57] byassuming theproportionofshorteningtheinspectioninterval,whentheminordefectivestageisidentifiedasadecisionvariabletobeoptimized.Theimperfect maintenancewasintroducedbyYangetal. [59].Thedevelopedmodeluses anagereductionconceptandisbasedonMCsimulationmethod implementation.

Zhaoetal. [60] introducedthethree-stagefailureprocessmodelwith inspectionandage-basedreplacement.Accordingtotheirassumptions, thesystemneedstoberepairedimmediatelyincaseofaseveredefective stageidentifiedduringaninspection,atafailure,orwhenitreachesacertain age.Theauthorsproposedtwocostmodelsandanumericalexampleto showtheprocedureofthemodeltosolvetheoptimalinspectionand age-basedreplacementintervals.

ThismodelwaslaterextendedbyWangetal. [61],wheretheauthors introducedatwo-phaseinspectionscheduleandanage-basedreplacement

policyforasingle-plantitemwithathree-stagedegradationprocess.The solutionisbasedintheimplementationofthehybridbeecolonyalgorithm.

2.3OtherMaintenanceModelsforSingle-UnitSystems WhereDTAHasBeenUsed

Otherdelay-timemaintenancemodelingproblemsaremostlyaimedatcase studiesintroduction,comparativestudiesinvestigation,orhybridmaintenanceimplementation.Followingthis,mostoftheworksareaimedat, amongothers,optimizationofinspectionandmaintenancedecisionsfor infrastructurefacilities(seeYamashinaandOtani [62] formaintenanceoptimizationforsingleelevatorperformance)ortransportationmeans(see Christeretal. [63] formodelingshipreliabilityoptimization).Moreover, Christeretal. [64] comparedsemi-Markovanddelay-timesinglecomponentinspectionmodels.Theauthors’maininterestwastoinvestigate towhatextenttheresultsofasemi-Markovdecisionmodelremainvalid, whentheMarkovpropertyisassumedbutisnotvalid.Inanotherstudy [65],theauthorspresentedtwoalternativepoliciesforpreventivereplacementofacomponent—agereplacementduringdelay-timepolicyand opportunisticagereplacementpolicy—andcomparedthemtodefinethe possibilitiesoftheirimplementationforreal-lifesystems.Thelastinteresting problemregardstheintegrationofmulticriteriadecision-makingtechniques withadelay-timemodelforoptimalinspectionmaintenancepolicydefinition [66,67].Inonestudy,Emovonetal. [67] focusedonmarinemachinery systemmaintenance,andthesolutionbasedonELECTREandMAUT methodsimplementation.Inaseparatestudy [66],thePROMETHEE decision-makingtechniquewasused.Theproblemofsupplierchoicein preventivemaintenance,includinginspectionandreplacement,wasinvestigatedinanotherstudy [68].

Delay-timemodeling,inthecontextofriskanalysisofmaintenance problems,wasinvestigatedbyWang [55].Thepresentedmodelregards single-andmultiunitcases.Itisworthmentioningthatthedevelopedprototypesoftwarepackagesgiveapossibilityofautomatingthedelay-time modelingprocess.Thesoftwareisbasedonthemultiplecomponentsystem DTMwithperfectinspection,andthemodelparametersareestimatedusing themomentmatchingmethod.

ThesafetyconstraintsareconsideredbyAvenandCastrointheirwork [69].Intheirpaper,theauthorsconsideredthebasicdelay-timemodelfor thesystem,whosefailuresaresafetycritical,thustheriskshouldbe

controlled.Themodelisaimedatdeterminingoptimalinspectionintervals, Tin,minimizingtheexpecteddiscountedcostsunderthesafetyconstraints.

Lastly,aninterestingproblemispresentedbyMahfoudetal. [70],where theauthorspresenttheproportionaldelay-timemodel,whichincludes parametersrelatedtomaintenanceeffectivenessandworkingconditions (utilizationrate)ofthemaintainedmedicalequipment.Thecontinuation ofmaintenancemodelingformedicalequipmentislatercontinuedina follow-upstudy [71]

3DELAY-TIMEMODELSFORMULTIUNITSYSTEMS

Inthischapter,theauthorpresentsthemultiunitsystemcase.Usually,inthe literature,therearedevelopedDTmodelsforcomplexsystems,wheregenerallyitisassumedthatasystemiscomprisedofmanyindependentcomponentparts,andabreakdowncanbecausedbyanyonecomponent(aseries structure).Thearrivalpatternofdefectswithinthesystemismodeledbyan instantaneousarrivalrateparameter λh(u)attime u.If λh(u)isconstant,the modelisahomogeneousPoissonprocesstype(HPP),otherwiseitisofa nonhomogeneousPoissonprocesstype(NHPP) [6].

Intheknownliterature,therearealsosomedevelopmentsfordelaytime-basedmaintenanceofsystemswithnonseriesreliabilitystructures. Theknownanalyticalsolutionsmostlyregardsystemswithparallelreliability structuresandperfectinspectioncase.Inthenextsections,themainachievementsinthismaintenancemodelingresearchareaarepresented.

3.1Delay-TimeModelsforComplexSystems

First,thecomplexsystemmaintenancecaseisanalyzed.Thegeneralclassificationofthemaindelay-time-basedmaintenancemodelsforcomplexsystemsispresentedin Fig.4.TheauthordefinesfourmaingroupsofDT modelsaccordingtothemainissuesconsideredintheinvestigatedresearch

models with perfect/imperfect inspections, HPP/NHPP defect arrival process, multiple nested inspections

Subjective, objective or mixed estimation modeling

Case problems investigations for e.g. vehicle fleets, production plants, gearboxes, work lifts

Fig.4 Thegeneralclassificationofdelay-time-basedmaintenancemodelsforcomplex system.

Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.