6西格玛(英文PPT53页) 53页

6OverviewSixSigma:-ADefinition-AppliedtoGE-GEQualityInitiative-WhyThisApproach?-OriginofSixSigma-The“BreakthroughStrategy”-ArrivingatSigmaSixSigmaStructureKeyConcepts&ToolsAPracticalExampleAnOverview....NotalotofDetails!! 6Overview“SixSigma”Ifwecan’texpresswhatweknowintheformofnumbers,wereallydon’tknowmuchaboutit.Ifwedon’tknowmuchaboutit,wecan’tcontrolit.Ifwecan’tcontrolit,weareatthemercyofchance.MikelJ.HarryPresident&CEOSixSigmaAcademy,Inc.ARigorousMethodforMeasuring&ControllingOurQuality“...willbringGEtoawholenewlevelofqualityinafractionofthetimeitwouldhavetakentoclimbthelearningcurveonourown.”JohnF.Welch,Jr.1995GEAnnualReport 6OverviewWhatDoes“Sigma”Mean?SigmaisaMeasureoftheConsistencyofaProcessIt(isAlsothe18thLetterintheGreekAlphabet! WhyDoesGENeedAQualityInitiative?GERaisingTheBarNewGoaltobe“BestintheWorld”vs.#1or#2CustomersareExpectingMore,weMustDeliver“Ship-and-fix”ApproachnoLongerToleratedintheMarketAimtoSpeedPastTraditionalCompetitorsin5YearsGoalConsistentwithReducedTotalCostsWeMustAcknowledgeOurVulnerabilitiesPoorQualityThatImpactsCustomersProblemswithNPITooHighInternalCosts6OverviewWeNeedaMajorInitiativetoMoveFromWhereweAretoWhereweWanttobe 6OverviewWhyDoesGENeedAQualityInitiative?40%35%30%25%20%10%15%5%CostofFailure(%ofSales)DefectsperMillion3.4233621066,807308,537500,000Sigma654321EstimatedCostofFailureinUSIndustryis15%ofSales;TakingGEFroma3toa6CompanyWillSave~$10.5BillionperYear! Why“SixSigma”?ProvenSuccessfulin“Quality-Demanding”Industriese.g.,Motorola,TexasInstruments(manyprocessstepsinseries)ProvenMethodtoReduceCostsHighlyQuantitativeMethod–ScienceandLogicInsteadofGutFeelIncludesManufacturing&Service(closetocustomer)andProvidesBridgetoDesignforQualityConceptsHasSupportandCommitmentofTopManagementItWorks!!! 6OverviewSigma3456SpellingMoneyTime1.5MisspelledWordsperPageinaBook1MisspelledWordper30PagesinaBook1MisspelledWordinasetofEncyclopedias1MisspelledWordinalloftheBooksinaSmallLibrary$2.7MillionIndebtednessper$1BillioninAssets$570Indebtednessper$1BillioninAssets$63,000Indebtednessper$1BillioninAssets$2Indebtednessper$1BillioninAssets31/2MonthsperCentury21/2DaysperCentury30MinutesperCentury6SecondsperCentury6isSeveralOrdersofMagnitudeBetterThan3!!!Sigma:AMeasureofQuality 6OverviewWhereDoes“SixSigma”ComeFrom?MikelJ.HarryoneoftheOriginalArchitectsPreviouslyHeadedQualityFunctionatABBandMotorolaNowPresident/CEOofSixSigmaAcademyinPhoenix,ArizonaHasConsultedforTexasInstruments,AlliedSignal(andothers)CurrentlyRetainedbyGEtoTeachtheImplementation,DeploymentandApplicationofSixSigmaConcepts&ToolsLearningfromThoseWhoHavehadSuccessWith6WillAccelerateitsImplementationatGE 6OverviewSo...WhatisSixSigma?AMeasurementSystemAProblem-SolvingApproachADisciplinedChangeProcess“THESIXSIGMABREAKTHROUGHSTRATEGY”MeasureAnalyzeImproveControl 6OverviewHowDoWeArriveatSigma?Measuring&EliminatingDefectsisthe“Core”ofSixSigmaMeasurementSystemIdentifytheCTQsLookforDefectsinProductsorServices“CriticaltoQuality”CharacteristicsortheCustomerRequirementsforaProductorServiceCountDefectsorfailurestomeetCTQrequirementsinallprocessstepsDefineDefectOpportunitiesAnystepintheprocesswhereaDefectcouldoccurinaCTQArriveatDPMOUsetheSIGMATABLEConvertDPMOtoSigmaDefectsPerMillionOpportunities23456308,53766,8076,2102333.4PPMDefectsperMillionofOpportunitySigmaLevel 6OverviewMeasurementSystem23456308,53766,8076,2102333.4PPMSIGMALEVELDEFECTSperMILLIONOPPORTUNITYIRSTaxAdviceBestCompaniesAirlineSafetyAverageCompanyGEAirlineBaggageDoctor’sPrescriptionRestaurantBillsAverageCompanyin3to4RangeSomeSigma“Benchmarks” 6OverviewMeasurementSystemAGraphic/QuantitativePerspectiveonVariationAverageValueManyDataSetsHaveaNormalorBellShapeNumberofPeopleArrivingatCRDTime7:007:157:307:458:008:158:308:459:009:15 6OverviewProblemSolvingApproachCenterProcessReduceSpreadXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXOff-TargetUnpredictableOn-Target6HelpsusIdentifyandReduceVARIATIONdueto:-InsufficientProcessCapability-UnstableParts&Materials-InadequateDesignMargin TargetUSLLSLTargetUSLLSLTargetUSLLSLCenterProcessReduceSpreadOff-TargetUnpredictableOn-TargetDefects6OverviewProblemSolvingApproach“LowerSpecificationLimit”“UpperSpecificationLimit”LessVariationMeansFewerDefects&HigherProcessYields 6OverviewProblemSolvingApproachKeyComponentsof“BREAKTHROUGHSTRATEGY”MeasureAnalyzeImproveControlIdentifyCTQ&CTP(CriticaltoProcess)VariablesDoProcessMappingDevelopandValidateMeasurementSystemsBenchmarkandBaselineProcessesCalculateYieldandSigmaTargetOpportunitiesandEstablishImprovementGoalsUseofParetoChart&FishboneDiagramsUseDesignofExperimentsIsolatethe“VitalFew”fromthe“TrivialMany”SourcesofVariationTestforImprovementinCenteringUseofBrainstormingandActionWorkoutsSetupControlMechanismsMonitorProcessVariationMaintain“InControl”ProcessesUseofControlChartsandProceduresAMixofConceptsandToolsWillAlsoIntegratewithNPIProcess 6OverviewDisciplinedChangeProcessANewSetofQUALITYMEASURESCustomerSatisfactionCostofPoorQualitySupplierQualityInternalPerformanceDesignforManufacturabilityWillApplytoManufacturing&Non-ManufacturingProcessesandbeTracked&ReportedbyEachBusiness 6OverviewStructureQualityCouncilMembers:Labs&Functions“Pipeline”&BBProjectPrioritiesTraining&CertificationMeasurements&RewardsCommunicationsChampionsLeadership:OverallInitiativeProjectFundingHR:Training&RewardsBlackBeltsLead6ProjectTeams“Measure/Analyze”“Improve/Control”OutwithBusinessesHereatCRDMasterBlackBeltsTeach6MentorBlackBeltsMonitorBBProjectsWork“Pipeline”ProjectsAResourcePoolTeamMembersLearn/Use6ToolsWorkonBBProjectsPartofTheJobOutwithBusinesses6ProjectswiththeGEBusinesses TabulationofGESixSigmaResults BenefitTarget&UpdateCurrentbenefitslevel@10.865MMQPIDloading:Carryoverfrom1999:4.059CompletedProjects2000:3.313ActiveProjects2000:3.285Total:10.865MM KeyConcepts&Tools6Overview 6OverviewChangingFocusFromOutputtoProcessYDependentOutputEffectSymptomMonitorX1...XNIndependentInput-ProcessCauseProblemControlIdentifyingandFixingRootCausesWillHelpusObtaintheDesiredOutputf(X)Y= ProcessCapability6OverviewSustainedCapabilityoftheProcess(longterm)USLTTime1Time2Time3Time4InherentCapabilityoftheProcess(shortterm)LSLTargetOverTime,a“Typical”ProcessWillShiftandDriftbyApproximately1.5 6Overview“ShortTermCentered”versus“LongTermShifted”SixSigmaCenteredLSLUSLTProcessCapabilitySHORTTERM.001ppm.001ppm+6LONGTERMLSLUSLT3.4ppmSixSigmaShifted1.5ProcessCapabilityHigherDefectYieldinLongTermProcessCapabilitythanShortTermProcessCapability-64.51.5 6OverviewTyingitAllTogethershiftCDAB0.51.01.52.02.5123456CONTROLPOORGOODTECHNOLOGYPOORGOODABCDGoodControl/PoorTechnologyPoorControl/PoorTechnologyPoorControl/GoodTechnologyWORLDCLASS!!!shorttermProblemCouldbeControl,TechnologyorBoth 6OverviewShortTermCapabilityShortTermCapabilityRatio(Cp)Cp=LSL-6USLExampleUSLLSL3.0==-3.063.0-(-3.0Cp=Cp=1LSLUSL2.50.53.0ProcessMeanTTargetA3ProcessThePotentialPerformanceofaProcess,ifitWereonTarget 6OverviewLongTermCapability(Cpk)CpCpk=LongTermCapabilityRatioExampleCp=1(previouschart)Target=-0.5=0Cpk1-(-0.5-03=Cpk=0.83-Off-TargetPenaltyTarget-3ThePotentialPerformanceofaProcess,CorrectedforanOff-TargetMeanLSLUSL2.50.53.0ProcessMeanTTargetA3Process 6OverviewZ-ScaleofMeasureZ=AUnitofMeasureEquivalenttotheNumberofStandardDeviationsthataValueisAwayfromtheTargetValue-3.0-0.53.0Z-ValuesUSLLSL2.50.53.0=ProcessMeanZTTarget0A3Process TheDefinitionsofYieldFinalTestProcess(Process4)PassProcess3Process1Process2100(UnitsTested)65708291Yield1Yield2Yield3Loss1Loss3RejectsLoss299125FirstTimeYield(Yft)=UnitsPassedUnitsTested=6570=0.93RolledThruputYield(Yrt)=(Yield1)(Yield2)(Yield3)....=91826570(((())))=0.65100917082NormalizedYield(Ynm)==1/n(Yrt)(0.65)1/4=0.89(n:TotalNumberofProcesses)6OverviewYieldExclusiveofReworkProbabilityofZeroDefectsAverageYieldofAllProcesses 6OverviewTheImpactofComplexityTheImpactofComplexityRolledRolledYieldYieldNumberofOperationsNumberofOperations1.001.000.900.900.800.800.700.700.600.600.500.500.400.400.300.300.200.200.100.100.000.001101001,00010,000100.0001,000,0001101001,00010,000100.0001,000,000ProcessMeanCenteredonEachOperationProcessMeanCenteredonEachOperation1101001,00010,000100.0001,000,0001101001,00010,000100.0001,000,000RolledRolledYieldYieldNumberofOperationsNumberofOperations1.001.000.900.900.800.800.700.700.600.600.500.500.400.400.300.300.200.200.100.100.000.00AstheNumberofOperationsIncreases,aHighRolledYieldRequiresaHighforEachOperation54366543ProcessMeanShifted1.5atEachOperation 6OverviewBaselining&BenchmarkinganExistingProcessp(x)DefectsBenchmarkBaselineEntitlementBenchmark.....AWorld-ClassPerformanceEntitlement.....AchievablePerformanceGiventheInvestmentsAlreadyMadeBaseline.....TheCurrentLevelofPerformanceBaselining=CurrentProcess/Benchmarking=UltimateGoal SomeBasic6-RelatedTools6OverviewScatterDiagramOverSleptCarWouldNotStartWeatherFamilyProblemsOtherParetoDiagramFrequencyofOccurenceReasonsforBeingLateforWorkArrivalTimeatWorkTimeAlarmWentOff MaterialsPeopleTheHistogramControlCharts---------------------------------------------6OverviewSomeBasic6-RelatedToolsTheFishboneDiagramMeasurementsMethodsTechnologyStatementCause&EffectBeingLateforWorkPlotofDailyArrivalTime9:157:007:157:307:458:008:158:308:459:00AverageValueNumberofPeopleArrivingatCRDTime 6OverviewLCLUCLRangeChartROutofControlConditionLCLXUCLXBarChartSomeBasic6-RelatedToolsLCL=LowerControlLimitUCL=UpperControlLimitX=MeanR=AverageRangeMonitorsChangesinAverageorVariationOverTime DesignofExperiments6OverviewSCREENINGOPTIMIZATIONCHARACTERIZATIONForExperimentsInvolvingaLargeNumberofFactorsUsefulinIsolatingthe“VitalFew“fromthe“TrivialMany”ForExperimentsInvolvingaRelativelySmallNumberofFactorsUsefulWhenStudyingRelativelyUncomplicatedEffects&InteractionsForExperimentsInvolvingOnly2or3FactorsUsefulWhenStudyingHighlyComplicatedEffects&RelationshipsDOEisMoreEffectiveThanTestingOneFactorataTime 6OverviewUsingthe“OneFactorataTime”ApproachReduceCommutetoWorkto15Minutes(withoutworkinganabnormalworkschedule)TheGoalTheVariablesTimeofDeparturefromHome&RouteTakentoWorkTheApproachTry3PotentialRoutesatCurrentDepartureTime(7:45),SelecttheBest&VarytheDepartureTime‘tilwegetto15MinutesTimeofDeparture3217:157:307:458:008:15RouteCombinationSelectedTheResultUseRoute2andLeaveat7:15toReachGoal 6OverviewUsing“DesignofExperiments”(DOE)TimeofDepartureDOE(i)BetterAccountsforInteractiveVariablesMissedby“OneFactorataTime”,and(ii)EfficientlySearchesfor“SweetSpot”inParameterSpaceTheVariablesTimeofDeparturefromHome&RouteTakentoWorkTheApproachVarytimeofDepartureandRouteSimultaneously,inaSystematicFashionTheResultABetterCombinationAllowing15MoreMinutesofSleep!!!ActualCommutingTimeAverages(minutes)3217:157:307:458:008:15Route172023211915182019161215212018OriginalConclusionBestCombination“SweetSpot”ReduceCommutetoWorkto15Minutes(withoutworkinganabnormalworkschedule)TheGoal APracticalExample(The“Cookbook”)6Overview 6.....andBakingBreadYEASTFLOURUsinga12StepProcess6OverviewThe“BETTERBREAD”Company Step1.....Selecting“CriticaltoQuality”(CTQsorY)WhatisImportanttotheCustomer?RiseTextureSmellFreshnessTasteY=Taste!!6OverviewMeasure Step2.....DefiningPerformanceStandardsforCTQsorY6OverviewHowCouldWeMeasureTaste(Y)?PanelofTastersRatingSystemof1to10Target:AverageRatingat8Desired:NoIndividualRatings(“defects”)Below7Y=12345678910TargetDefectsWorstBestBut.....IsthistheRightSystem?Measure 6OverviewStep3.....ValidatingtheMeasurementSystemforYHowCouldWeApproachThis?BlindfoldedPanelRatesSeveralLoafSamplesPut“Repeat”PiecesfromSameLoafinDifferentSamplesConsistentRatings*onPiecesfromSameLoaf=“Repeatability”ConsistentRatings*onSamplesAcrossthePanel=“Reproducibility”“Repeatability”&“Reproducibility”SuggestValidMeasurementApproachPanelMemberLoaf1Loaf2Loaf3A589B491C492D898E482F591G892*WithinOneTasteUnitMeasure 6OverviewStep4.....EstablishProductCapabilityforY(Taste)Thisisa3Process!7Defects(ratingsbelow7)24Ratings(fromourpanel)=.292292,000Defectsper1,ooo,oooLoavesOR765432112345678910#ofRatingsRating64321143Defects<7Target=8AnalyzeHowDoWeApproachThis?BakeSeveralLoavesUnder“Normal”ConditionsHaveTasterPanelAgainDotheRatingAverageRatingis7.4ButVariationistooGreatfora6Process3x10+4x9+6x8+4x7+3x6+2x5+1x4+1x31+1+2+3+4+6+4+3 6OverviewStep5.....DefineImprovementObjectivesforY(Taste)HowdoweDefineImprovement?BenchmarktheCompetitionFocusonDefects(i.e.tasterating<7)DetermineWhatisan“AcceptableSigmaLevel”SetImprovementObjectivesAccordinglyMaybea5ProcessWillSuffice!1,000,000-100,000-.............................10,000-.............................1,000-.............................100-.............................10-.............................1-234567“BETTERBREAD”BakingProcessBestCompetitorRangeforImprovementDefectsPerMillionSigmaScaleFreihoferWONDERPepperidgeFarmSunbeamAnalyze 6OverviewStep6.....IdentifySourcesofVariationinY(Taste)HowdoweDeterminethePotentialSourcesofVariation(Xs)?HavetheChefsBrainstormSomeLikelyOnesMightbe:-AmountofSaltUsed-BrandofFlour-BakingTime-BakingTemperature-BrandofYeastYEASTFLOURMultipleSources:Chefs,Suppliers,ControlsAnalyze 6OverviewStep7.....ScreenPotentialCausesofVariation(Xs)HowdoweScreenforCausesofVariation(Xs)?DesignanExperimentUseDifferentSourcesofPotentialVariationHavePanelRatetheBreadUsedintheExperimentResultsLeadtothe“VitalFew”CausesYEASTFLOURSourceConclusionNegligibleMajorCauseNegligibleMajorCauseNegligibleFocusonThe“VitalFew”Improve 6OverviewStep8.....DiscoverVariableRelationshipsBetween“VitalFew”(Xs)andYHowdoweFindtheRelationshipBetweenthe“VitalFew”(Xs)andTaste(Y)?ConductaMoreDetailedExperimentFocus:OvenTemperaturefrom325to375and3BrandsofFlourRUN#TEMPBRAND1325A2325B3325C4350A5350B6350C7375A8375B9375CFLOURFLOURFLOURBrandABrandBBrandCImproveResults:350&BrandAisBestCombinationofTemperature&FlourNote:TimeisaFactorOnlyifTemperatureChangesSignificantly Step9.....EstablishToleranceson“VitalFew”(Xs)HowdoweEnsureOvenTemperatureisControlled?DataSuggests350(5)isbestTemperaturetoReduceTasteVariationBrandAFlourtobeUsedExceptinCaseofEmergency“BETTERBREAD”toSearchforBetterAlternativeSupplierofFlourJustinCaseFLOURBrandABut.....IsOurMeasurementSystemCorrect?Improve 6OverviewStep10.....ValidatetheMeasurementSystemforXsHowCouldWeApproachThis?NeedtoVerifytheAccuracyofOurTemperatureGaugesNeedfor“Benchmark”InstrumentationforComparisonRentSomeOther“HighEnd”GaugesComparetheResultsVerifythatourInstrumentsareAccurateControl 6OverviewStep11.....DetermineAbilitytoControlVitalFewXsHowCouldWeApproachThis?CheckANumberofOvensMonitorTemperaturesOverTimeFocusontheProcessCapabilityLookforDegreeofVariationVariationOKBut...AverageisHigh(andthealgorithmshouldbechecked)30345#ofOvensTemperature346357347348349350351352353354355356252015105Control 6OverviewStep12.....ImplementProcessControlSystemonXsWhatdowedoGoingForward?CheckOvensDailyforTemperatureLevelsAuditUsageFrequencyofAlternativeFlourSupplier(e.g.,BrandC)PeriodicallyReassemblethePaneltoTestTasteCharttheResultsAnd.....PlottheDataOverTimeFLOUR“BrandC”354353352351350349348135791113151719212325Control