Ch12_Reliability testing & growth

11©1999byCRCPressLLCReliabilityTestingandGrowth11.1INTRODUCTIONJustlikeinthecaseofanyotherreliabilityactivity,reliabilitytestingisanimportantreliabilitytask.Infact,itmaybecalledoneofthemostimportantreliabilityactivitiesofareliabilityprogram.Themainpurposeofreliabilitytestingistoobtaininfor-mationregardingfailures,inparticular,theproduct/equipmenttendencytofailaswellasthefailureconsequences.Thistypeofinformationisextremelyusefulincontrollingfailuretendenciesalongwiththeirconsequences.Agoodreliabilitytestprogrammaybeclassifiedastheoneprovidingthemaximumamountofinformationconcerningfailuresfromaminimalamountoftesting[1].Overtheyears,manyimportantpublicationsonreliabilitytestinghaveappeared;inparticular,twosuchpublicationsarelistedinReferences2and3.Inthedesignanddevelopmentofnewcomplexandsophisticatedsystems,thefirstprototypesusuallycontainvariousdesignandengineeringrelateddeficiencies.Infact,accordingtoReferences4and5,thereliabilityofrevolutionarydesignproduct/systemscouldbeverylow,i.e.,15to50%oftheirmaturedesigncapability.Itmeanswithouttheinitiationofvariouscorrectivemeasuresduringthedevelopmentstagetoimprovereliability,suchproducts’/systems’reliabilitycouldverywellremainatthelowinitialvalue.Nonetheless,correctingtheweaknessesorerrorsindesign,manufacturingmethods,eliminationofbadcomponents,etc.leadstoaproduct’sreliabilitygrowth[6].Theterm“reliabilitygrowth”maybedefinedasthepositiveimprovementinareliabilityparameteroveratimeperiodbecauseofchangesmadetotheproductdesignorthemanufacturingprocess[7].Similarly,theterm“reliabilitygrowthprogram”maybedescribedasastructureprocessfordiscoveringreliabilityrelateddeficienciesthroughtesting,analysisofsuchdeficiencies,andimplementationofcorrectivemeasurestolowertheiroccurrencerate.Theseriousthinkingconcerningreliabilitygrowthmaybetracedbacktothelate1950s.In1964,apopularreliabilitygrowthmonitoringmodelwaspostulatedbyDuane[8].Comprehensivelistsofpublicationsupto1980onreliabilitygrowtharegiveninRefences9and10.Thischapterdiscussesreliabilitytestingandgrowth.11.2RELIABILITYTESTINGAwiderangeoftopicscanbecoveredunderthissubject,includingclassificationofreliabilitytests,successtesting,confidenceintervalestimatesformeantimebetweenfailures,andacceleratedlifetesting.Eachofthesetopicsispresentedbelow.©1999byCRCPressLLC11.2.1RELIABILITYTESTCLASSIFICATIONSBasically,reliabilitytestsmaybegroupedunderthreeclassifications:(1)reliabilitydevelopmentanddemonstrationtesting,(2)qualificationandacceptancetesting,and(3)operationaltesting[11].Reliabilitydevelopmentanddemonstrationtestingisconcernedwithmeetingobjectivessuchastoindicateifanydesignchangesarerequired,todetermineifthedesignistobeimprovedtosatisfythereliabilityrequirement,andtoverifyimprove-mentsindesignreliability.Thenatureofthistypeoftestingdependsonfactorssuchasthelevelofcomplexityunderconsideration,andthetypeofsystem/subsystembeinginvestigated.Forexample,inthecaseofelectronicparts,thereliabilitydevel-opmentanddemonstrationtestingcouldtaketheformoflifeteststoevaluateifthepartcansatisfyitsreliabilitygoals;ifnot,whatmeasuresarenecessary?Inordertomeetreliabilitydevelopmentanddemonstrationobjectiveseffectively,theaccumulatedtestdatamustbeofakindthatallowsinsightintothefailureprobabilities/failureeffectsforacertaindesignunderconsideration.Furthermore,thesedataserveasagoodbasisforreliabilityanalysisandassessmentfortwoparticularitems:designunderconsiderationandsubsequentrelatedprograms.Qualificationandacceptancetestingisconcernedwithmeetingtwobasicobjec-tives:toarriveatdecisionifapart/assembly/enditemistobeacceptedornot,andtodetermineifacertaindesignisqualifiedforitsprojecteduse.Theseobjectivesdifferfromtheobjectivesofotherreliabilitytests,particularlywithregardtotheaccept/rejectmechanism.Qualificationandacceptancetestingincorporatestheusualtestingandscreeningthequality-controlfunctionofincomingparts.Withrespecttothematerialsandcomponentstobeusedinthesystem/equipmentunderdevel-opment,thequalificationandacceptancetestingstartsearlyintheprogram.Operationaltestingisconcernedwithobjectivesthatincludeverifyingtheresultsofreliabilityanalysisconductedduringthesystem/equipmentdesignanddevelop-ment,providingdataindicatingdesirablechangestooperatingpoliciesandproce-dureswithrespecttoreliability/maintainability,andprovidingdataforsubsequentactivities.Allinall,theoperationaltestingprovidesthefeedbackfrompracticetotheory.11.2.2SUCCESSTESTINGThistypeoftestingissometimespracticedinreceivinginspectionandinengineeringtestlaboratorieswhereano-failuretestisspecified.Usually,themaingoalforthistestistoensurethatacertainreliabilitylevelhasbeenachievedatagivenconfidencelevel.Inthiscase,forzerofailures,thelower100(1–α)%confidencelimitonthedesiredreliabilitylevelcanbewrittenas[12](11.1)RLowk=α1©1999byCRCPressLLCwherekisthenumberitems/unitsplacedontest.αisthelevelofsignificanceorconsumer’srisk.Thus,with100(1–α)%confidence,itmaybestatedthat(11.2)whereRTisthetruereliability.TakingthenaturallogarithmsofthebothsidesofEquation(11.1)leadsto(11.3)Thus,fromEquation(11.3),weget(11.4)Thedesiredco