11 Six Sigma Analysis

IntroductiontoSixSigmaPatHammettTQM-UniversityofMichigan11IntroductiontoSixSigmaSeeCourseWebsiteforthefollowingReferencePaper:AnIntroductiontoSixSigmawithADesignExample,RobertWhite,IEEE1992.2OutlinenSixSigma-HistoryandPhilosophynSixSigma-StatisticalFoundationnSixSigmaMethodologyIntroductiontoSixSigmaPatHammettTQM-UniversityofMichigan23SixSigma-ShortHistorynPioneeredbyMotorolain1980snMethodologyusedtofocusresourcesonsolvingproblemsandincreasingcustomersatisfactionwithanemphasisonprofits(bottomline).nSomeKeyPlayersinSixSigmaHistory:nMikelHarry-oneoftheoriginalMotorolaarchitectswholaterfoundedSixSigmaAcademy.nJackWelch,CEOGeneralElectricnLarryBossidy,CEOAlliedSignalnTheseCEOsarelargelyresponsibleforthespreadofSixSigmabyembracingthemethodologyanddemonstratingitspotentialintheirrespectivecompanies.4SixSigmaPhilosophynKeytobusinesssuccessistodriveoutproductvariation.nNeedbetterthan99%qualitylevelsnConsidersomeoutcomesfor99%levelsn20,000lostarticlesofmailperhourn5000incorrectsurgicaloperationsperweekn200,000wrongprescriptiondrugseachyearnNoelectricityforalmost7hourseachmonthIntroductiontoSixSigmaPatHammettTQM-UniversityofMichigan35SixSigmaandBottomLinenMuchofsixsigmabuzzresultsfromexecutivesclaiminghugesavings.nGE-JackWelchnOperatingMarginBeforeSixSigma13.6%(‘95);afterSixSigma16.7%(‘98)~$600Million*.nAlliedSignal-LarryBossidynOperatingMarginfrom12%to14.1%in‘99.nCumultativeSixSigmaSavings‘94-’99=~$2billionindirectcosts*.*Harry,M.andSchroeder,S.(2000).SixSigma:TheBreakthroughManagementStrategy,Currency.6SixSigma&TQMnFromaphilosophicalstandpoint,SixSigmastrategydifferslittlefromTQM.nSamebasictenetsforsixsigmaasTQM.nFromanimplementationstandpoint,however,SixSigmahasbeenmoresuccessfulby:nPlacingagreateremphasisonsolvingproblemsthatresultinsignificantsavings.nProvidingcommonlanguageandmetrics.nProvidinganimplementationstrategy.IntroductiontoSixSigmaPatHammettTQM-UniversityofMichigan47SixSigmaProblemSolvingnDefine-nimprovementopportunitywithanemphasisonincreasingcustomersatisfaction.nMeasure-ndetermineprocesscapability(Cp/Cpk)&dpmo(defectspermillionopportunities).nAnalyze-nidentifythevitalfewprocessinputvariablesthataffectkeyproductoutputvariables(“Findingtheknobs”).nImprove-nMakechangestoprocesssettings,redesignprocesses,etc.toreducethenumberofdefectsofkeyoutputvariables.nControl-nImplementprocesscontrolplans,installreal-timeprocessmonitoringtools,standardizeprocessestomaintainlevels.8CommonLanguagenKeymetricsarereferredtoas:nKPOVs-KeyProductOutputVariablesnCritical-to-QualityCharacteristics(CTQ)nCritical-to-CostCharacteristics(CTC)nCritical-to-DeliveryCharacteristics(CTD)nKPIV-KeyProcessInputVariablesnDefectsPerMillionOpportunitiesnRolledThroughputYieldIntroductiontoSixSigmaPatHammettTQM-UniversityofMichigan59DefectsPerMillionnSixSigmaGoal3.4DPMopportunitiesnProcessCapabilityMetricsnCp=2.0;Cpk1.5;3.4DPMOnSo,ifCp=2.0,meanmaydeviateupto+/-1.5σfromnominalandstillachieve3.4DPMO.LSLUSLTarget(Nominal)Pr(defect)=0.00000343.4DPMOCp=2.0Allowupto1.5σshift(Cpk1.5)10OpportunitiesnCharacteristicLevel-#defectsforsomecomponentcharacteristic(e.g.,tirepressure)nInthisclass,wehavebeenexaminingdefectsonacharacteristicorfeaturelevelbasis.nBut,acomponenthasmultiplecharacteristicsandasystemhasmultiplecomponents.nDefinitionofOpportunity–anychanceforerrornSo,Total#Opportunitiesinasystem=n(Σ#characteristics/component)*#produced.nIfpartA~10characteristics,B~50,C~30characteristicsnThen,SystemABChas90Mopportunities/1MproducedIntroductiontoSixSigmaPatHammettTQM-UniversityofMichigan611YieldVs.RolledYieldnFirstTimeYield-#AcceptableUnits(w/oscraporrework)/TotalUnitsProduced.nRolledThroughputyieldistheprobabilityofbeingabletopassaunitofproductthroughentireprocess-defectfree.nItismoresensitivetoproductcomplexity.nRolledYieldYRT=0.93*0.89*0.84*0.9=0.6310708393100units63unitsScrap10Scrap393/100=.9383/93=.8970/83=.8463/70=.977Op1Op2Op3Op412RolledYieldCalculationsnRolledThroughputYield(YRT)decreasesasaddoperations(unlessoperationyieldis100%)Yields00.20.40.60.811234Operation#YieldProportionOperationYieldCumRolledYieldProcessOperation#Units#GoodUnitsOperationYield(perunit)CumRolledYieldperunitOp1100930.930.93Op293830.890.83Op383700.840.70Op470630.900.63RolledThroughputYield0.63IntroductiontoSixSigmaPatHammettTQM-UniversityofMichigan713ConcernswithYieldbyUnitnSuppose:nCaseA:10partsareinspected,part#3has1defectandpart#4has1defect--yieldis80%.nCaseB:10partsareinspected,part#3has2defectsandpart#4has5defects--yieldis80%.nArethesetwoprocessesperformingthesame?14RolledYieldw/MultipleDefectOpportunitiesperunitnRolledthroughputyieldonaunitbasisisnotdefectsensitive.nAnalternativeapproachistoconsiderdefectspermillionopportunities(DPMO).ProcessOperation#Units#Rework/Scrap#Opportunities/unit#DefectsTotal#OPDPO(defectsperopportunity)Op110074104000.03Op293105104650.02Op383135154150.04Op470710107000.01Sum4519800.02DPMO22,727IntroductiontoSixSigmaPatHammettTQM-UniversityofMichigan815AssessingPerformanceofMultipleProductsnSimilarly,rolledyieldmaybecalculatedusingdefectper