Placeyourimageontopofthisgraybox.Ifnographicisapplicable,deletegrayboxandnotch-outbehindgraybox,fromtheTitleMasterDesignFMEA简介2010/11/152•DFMEAisanacronymfor:DesignFailureModesandEffectsAnalysisWhatisDFMEA?2010/11/153FEMA的好处•FEMA更关注于预防而不是补救措施:在生产之前的设计阶段,只消耗了15%的成本,但决定了95%的价值。•在设计阶段越早发现问题,采取纠正措施的成本就越低。•FEMA可以记录并跟踪采取的措施以降低风险。2010/11/154High-LevelProcessMapIdentifyFMEAProjectScopeUnderstandtheDevScopeDesignFEMA:12stepsprocessWhichhighriskareasshouldtheteamfocustheFMEAefforts?Whataretheinputs,outputs,controls,noise,andunintendedresultsofthefocusedscope?Whichdesignfailuremodesaretheriskiest?Whatactionswilltheteamtaketopreventthemfromoccurring?2010/11/155•Boundaryorextentoftheanalysis•Defineswhatisincludedandwhatisnot•Understandsystemfunctionalrequirements•UnderstandvoiceofcustomerWillavoid:•Analyzeintoareasthatarenotimportant•EstablishingthewrongteamIdentifyingtheProjectScope2010/11/156UnderstandtheDevScope–P-DiagramFunctionality/System/sub-system/ComponentSignalCausesthesystemtodelivertheuserintentNoiseFactors(UnexpectedVariationduetoexternalenvironment,internal,piece-to-piece,customerusage,wearout)ControlFactors(Designparameterswhosenominalvaluescanbeadjustedbytheuserorprogrammingmodesinthesoftware)IdealFunction(Systemoutputthatdeterminestheperceivedresult)UnintendedResults(FailureModes)2010/11/157•ListFunction:UsethefunctionfromtheFunctionalDiagramandtheP-Diagram.•ListPotentialFailureModes:LookattheunintendedresultontheP-Diagram.Thefailuremodeisusuallytheoppositeoftheintendedfunction.Ask–Ifthefunctionhastoomuch,toolittle,none,wrong,toolateortooearly,changesovertime?Brainstormotherfailuremodes.12StepstotheDFMEA2010/11/158Consider,ataminimumfivecommonfunctionalfailuremodes:•–FailuretoExecute•–IncompleteExecution•–InterruptTimingFailures-blockedinterrupt,priorityconflicts,etc…•–Deliversincorrectresult•–ExecutesatincorrecttimeConsider,ataminimumfourcommonfailuremodesappliedtoeachinterface•–Failuretoupdateavalue•–Incompleteupdateofavalue•–Incorrectinputvaluestotheinterface•–Updateexecutesatincorrectrate12StepstotheDFMEA(cont.)2010/11/159•ListPotentialEffects:If(potentialfailuremode),Then(what?).Theeffectiswhatthecustomerwillseeifthefailuremodeoccurs.•AssignSeverityRatingtoPotentialEffects:Eachbusinesswillhavespecificdefinitionsforallscales.Makesureyourteaminterpretsthemthesametoproduceconsistentscoring.12StepstotheDFMEA(cont.)2010/11/1510AssignSeverityRatingtoPotentialEffects(Cont.):•MakesureyourteaminterpretsallscalesthesametoproduceconsistentscoringinordertomakeyourFMEAscomparable.12StepstotheDFMEA(cont.)SeverityofEffectRankHarmToEndUser10Beillegal9Renderproductorserviceunfitforuser8ExtremeCustomerDissatisfaction7Partialmalfunction6Alossofperformancewhichislikelytoresultinacompliant5Aminorperformanceloss4Aminornuisancebutcanbeovercomewithnoperformanceloss3Onlyminoreffectonperformance2Nodiscernableeffect12010/11/1511•ListPotentialCauses:Avoidstatementslike“operatorerror”or“productmalfunction”Toolsusedtofindcauses:5Why,FishboneNote:Trytofindallcausesforthefailuremodebeforemovingtoanotherfunction.Note:What’sthedifferencebetweenafailuremode,effectandcause?Thinkofthe‘FailureMode’aswhatyouobservewhenafailureoccurs.The‘Effect’ishowthefailureimpactsthecustomer.Causeisthemostimportantofthesethree,findthecauseandtakeactiontocontrolthemisthemainfocusofDFMEA.12StepstotheDFMEA(cont.)2010/11/1512•AssignOccurrenceRatingDONOTguess,usehistoricaldatatodeterminetheoccurrencerating.12StepstotheDFMEA(cont.)FailureProbabilityRank1in2101in391in881in2071in8061in40051in100041in1000031in10000021in100000012010/11/1513•ListCurrentControls:Preventioncontrolsareproactivewaystopreventfailure–simulationmodels,designspecifications,finiteelementanalysisDetectioncontrolsdetectthecauseoccurringandwarnustostopanescapingdefectfromreachingthecustomer12StepstotheDFMEA(cont.)2010/11/1514•AssignDetectionRating:Iftherearepreventionanddetectioncontrols,usethe“worst”control,orhighestnumber.12StepstotheDFMEA(cont.)ChanceofEscapeDescriptionRate100%Nodetectionmethodinplace,escapeisalmostcertain1090%Unprovenorunreliabledetectionmethod980%Veryslighteffectivenessindetection870%Slighteffectivenessindetection760%Loweffectivenessindetection650%Mediumeffectivenessindetection540%Moderatelyhigheffectivenessindetection430%Higheffectivenessindetection320%Veryhigheffectivenessindetection210%Designcontrolwillalmostcertainlydetectthepotentialcauseofthefailuremode.12010/11/1515CalculateRiskPriority:ReserveSeverity9and10forthemostcriticalfailuremodesVoting–trytoobtainconsensusUsethemean(couldalsousemedian)Deferto“expert”Takethehighestrating12StepstotheDFMEA(cont.)2010/11/1516•UseRPNstoHelpDecideonHigh-PriorityFailureModesAllhighRPNnumbersMUSThaveaction!AlwaysworkthetopRPNs(100ortop35%)Givespecialattentiontooccurrenceof7andaboveGivespecialattentiontoseverityof8andabove12StepstotheDFMEA(cont.)2010/11/1517•TakeActiontoReduce/EliminatetheRisk:Focusactionsonthegoalofreducing:Occurrence,thenDetection,thenSeverity.Note:ADFMEAisnotcompleteduntiltheproduc