MSA測量系統(tǒng)分析英文版培訓(xùn)教材

1MeasurementSystemsAnalysisLearningObjectivesTo:UnderstandtheneedforaMeasurementSystemsAnalysis(MSA)UnderstandthecomponentsoftheMeasurementSystemUnderstandthetestsinvolvedinaMSAToexecuteaVariablesGageRepeatability&Reproducibility(VGR&R)Isoplot(continuousdata)AttributeGageRepeatability&Reproducibility(AGR&RLearnhowtofixapoorMeasurementSystemUnderstandhowapoorMeasurementSystemimpactsCapabilityMeasurementSystemAnalysisQuantifywasteIdentifypossibleBoB/WoWleversExperimentIsolateandverifytherootcausebyfocusing&splittingdictionariesusingdataConcludewhatconditionshavetobeimproved/redesignedUpdateMFRGeneratesolutionsVerifysolutionsDefinepermanentcountermeasuresRiskanalysisDescribethemodifiedprocess:newprocessmapUpdateMFRDesigntheimplementationplanDefine/trackactivitiesTrackcountermeasuresimplementationCommunicateandclarifytotheorganisationImproveTrackperformanceUpdateMFREnsurechangesareanchoredintheorganizationTrainallrelevantpeopleStandardizeModifyauditchecklistsTrackimprovedperformanceUpdateMFRAcommonviewofthe‘a(chǎn)sis’processandhowwellitisperformingIdentifycriticalcausesand/orlevers(Thevitalfew)DescribeandDesigntheimprovedprocessDrawuptheimplementationplanandintroduceconfirmresultsEstablishcontrolandhandovertoprocessowner.MapMeasureExplore

EvaluateDefine

DescribeImplementImproveControlConformDescribeproblem(oropportunity)andobjectiveMeasurehistoricalperformancetrendsSetupmeasure-mentsystemMapcurrentprocessShareavailableknowledgeandgetacommonviewManagebyfacts,notopinionsStartMFRReportWhyPerformaMeasurementSystemAnalysis(MSA)?Toensurethatthemeasurementsystemisnotasignificantsourceofvariability.TodeterminewhetheractionsarenecessarytorepairorreplacethemeasurementsystemToaccuratelyandpreciselydeterminethetruecapabilityofourprocess.WhatisaMSA?Asetofdesignedexperimentstodeterminewhetherthemeasurementsystemisbothaccurateandprecise.AccuracycomparesthemeasurementsystemaveragetoastandardPrecisionisameasureofthemeasurementsystemvariabilityThereisaMSAforthetwotypesofdataVariables Agageisusedbymanyoperatorstomeasureapartcharacteristictosatisfyacustomerstandard.Attribute Peopleareinspectingpartsoritemstodeterminewhetherthecolorofapart,ascreenprint,fieldsinaform,timeorsomeotherattributeisacceptablecomparedtoacustomerstandard.SomeprojectsrequirecreativitytoconductanMSAOperationalDefinition

EXTREMELYIMPORTANTInanyMSAitisABSOLUTELYCRITICALtowritedownanOPERATIONALDEFINITION:AStandardMethodInmanycasestheOperationalDefinitionincludesanUpperandLowerSpecificationLimitInothercasesitisawrittendefinitionthatdescribesexactlyhowthemeasurementsaretobemade,forexampleAcompanywantsanitemtobedeliveredin3days.Whendoestheclockstartandwhendoestheclockstopt0–CustomerplacesorderwithCustomerServiceRepresentative(CSR)t1–CSRentersorderinsystemt2–Orderisscheduledtobemanufacturedt3–Orderismadet4–Orderisstockedt5–Orderisattheshippingdockreadytobepickedupt6–Orderispickedupt7–Orderisdeliveredt8–Orderisstockedt0t1t2t3t4t5t6t7t8Whoisresponsiblefordeterminingthestandard?Continuous/VariablesMSAWhatistheTrueDPMO?ABlackBeltisinterestedinknowingtheDPMOofthediameterofabatchoflampfilamentscomparedtocustomerspeclimitsShesamples30filamentsandplotsthediametersinanHistogramLSLUSLHowdoweknowthatourmeasurementsystemcanproperlyrejectbadpartsandproperlyacceptgoodparts,suchthatwecanevaluateourtrueDPMO?SourcesofVariabilityObjective-toensurethevariabilityfromthemeasurementsystemislessthansomeacceptablevaluesuchthatwecanassessourtrueprocesscapability.AccuracyandPrecisionwithintheMeasurementSystemSourcesofVariabilitywithintheMeasurementSystemcanbeattributedtoeitherAccuracy–MeasurementSystemaveragesoutlinedbelowinredDiscrimination–MeasurementSystemresolutionoutlinedinblackPrecision–MeasurementSystemvariabilityoutlinedbelowinblueAccuracy

ExplanationAccuracyisameasureofcentrallocationwithrespecttoaknownreferencestandardKnownReferenceStandardProcessAverageAccuracy

BiasMeasurementSystemBias:-Todeterminewhetheranoffsetexistsinthegage,determinedviaa“calibrationprogram”:Amasterpartisidentified10measuresofthemasterpartaremadeandtheaveragevalueiscalculatedTodeterminethebiastheaveragevalueissubtractedfromthemastervalueThegageisoffsetbytheamountofthebias.TruevalueMeasuredvaluesmeasurementbiasAccuracy

BiasExampleABlackBeltneedstoconductaMSAandshefirsttestsforBias.Sheidentifiesamasterpartandthegageusedtomeasurethispart.Themasterpartisknowntobe0.350”.Sheasksanoperatortomeasurethepart10timesandrecordsthesevalues,whichareshownbelow.ShethendeterminestheMeasurementSystemBiasbysubtracttheaverageofthese10measuresfrom0.350”0.350”0.341”0.343”0.330”0.340”0.358”0.352”0.329”0.330”0.348”TheseresultsindicatethatonaveragethemeasurementsystemisBiasedby-0.007”.Thereforesheneedstooffsetthegagebypositive0.007”Accuracy

LinearityMeasurementSystemLinearity:-TheabilityoftheMeasurementSystemtomeasureoveritsoperatingrangewithminimalBiasIdentifymasterpartsthatspantheoperatingrangeofthegageanddeterminethebiasoftheseparts.PlotthedatainascatterplotwheretheX-axisarethemasterpartsandtheY-axisaretherespectivebiasesFitaregressionlinetothedataLinearityResults:R2oftheregressionlinetobeascloseto100%aspossibleTheslopeoftheregressionlinetobeascloseto0aspossibleindicatingnobiasacrosstheoperatingrangeAccuracy

MinitabKeystrokes-LinearityExampleABlackBeltcontinueswithherMSAandconductsalinearitystudy.Sheidentifies5masterpartsthatspantheoperatingrange(2”to10”)ofhergage.Oneoperatormeasuredeachpart12timesandrecordsthedatainMinitab.Gagelin.mtwStat>QualityTools>GageLinearityStudySelectPartandenterin“PartNumbers”SelectMasterandenterin“MasterMeasurements”SelectResponseandenterin“MeasurementData”Determineyourlong-termhistoricaldeviation,multiplyby6andenterinProcessVariationAccuracy

MinitabGraphOutputforLinearitySlopeoftheregressionline*(6*s)100*{Slopeoftheregressionline*(6*s)}R2oftheRegressionlineAveragebiasAveragebias/(6*s)Theslopeoftheregressionlineisdeterminedbydividingthe%linearityby100inthiscaseitisrelativelylowat0.1317TheR2isveryhighindicatingthesystemislinearAccuracy

StabilityStabilityResultsIncontrolcontrolchartIfadatapointisoutofcontrolgagemayrequirecalibration.MeasurementSystemStability–TheamountofvariabilityintheBiasovertime.OnadailyoraweeklybasisstabilityismeasuredbyplottingtheresultsofyourbiasstudyinaControlChart.Accuracy

StabilityExampleTheBlackBeltretrievesthelastthreemonthsofbiasdataasmeasuredonadailybasisbytheQualityControldepartment.SheplotsthisinformationinanIndividualsandMovingRangeChart(I-mR).Theseresultsareshownbelow.Accuracy

StabilityExampleAnoutofcontrolconditionononeofthecontrolchartsisanindicationthatthemethodtocalibratetheMeasurementSystemneedstobeevaluatedDiscriminationDiscrimination–TheabilityoftheMeasurementSystemtodetectadequatechangesinprocessvariationAtaminimumthemeasurementsystemshouldbeabletodiscriminateto1/10thetolerance(UpperSpecification–LowerSpecificationLimit)Ideallywedesiretomeasure1/10theprocessvariationLSL=0.200”LSL=0.300”LSL=0.200”LSL=0.300”DiscriminationItcanbeverycostlytohaveameasurementsystemdiscriminate1/10theProcessVariationLSL=0.200”LSL=0.300”Howexpensivewoulditbeifwewereabletomeasure1/10thelong-termprocessvariationshownabove?PrecisionPrecisionisameasureofvariabilityTestforBias,Linearity,andStabilityensurestheMeasurementSystemisontargetPrecisionensuresthereisminimalvariabilityinthemeasuresAmuchmoredesirablestateWhatistheTrueVariabilityofthePartsbeingMeasured?ABlackBeltneedstoknowtheamountofvariabilityinthediameterofabatchoffilaments30filamentsdiametersareplottedinaHistogramThewidthofthehistogramisassumedtheresultsofthevariabilityinthediameterofthefilaments.Doothersourcesofvariabilityinfluencethehistogram?Precision

Part&MeasurementVariabilityAGageRepeatability&Reproducibility(GR&R)StudyisadesignedexperimentthatpartitionssourcesofvariabilitywithintheMeasurementSystemAGR&RcanbeconductedforbothAttributeandVariablesdataPrecision

GageRepeatability&ReproducibilityTheMeasurementSystemvariabilitycanbefurtherpartitionedintoRepeatability&ReproducibilityPrecision

Repeatability&ReproducibilityRepeatability–DetermineswhetherthevariabilityofthegageisconsistentTheabilityofthegagetoachievethesamemeasuredvaluewhenoneoperatormeasuresthesameparttwice.Reproducibility–Determineswhetherthevariabilitybetweenoperatorsisconsistent.Theabilityofmultipleoperators,whentheytakemultiplemeasuresononepart,toachievethesameaveragevalues.UnderstandingthecontributionfromRepeatability&ReproducibilitycanassistinresolvingsomemeasurementsystemissuesHowMuchVariabilityisAcceptableintheMeasurementSystem?WhenconductingaVGR&RthefollowingratiosareusedtodeterminewhethertheMeasurementSystemisacceptable

%Contribution*

%Study*

%Tolerance*Unacceptable >10% >30% >30%Marginal 3–10% 10–30% 10–30%Excellent <3 <10% <10% *Note:Intheactualformulathevariancesorstandarddeviationsaremultipliedby5.15,whichrepresents99%oftheareaunderanormalcurveHowistheVariabilityinaVGR&RPartitioned?ToconductaVGR&Rthefollowingareidentified:10Partsthatspantherangeofthelong-termvariability2or3OperatorswhousethegageTheGageusedtotakethemeasuresTheExperimentisdesignedasfollows-1123121212PartsOperatorsRepeatMeasures2123121212312312121210partsmeasuredtwicebythreeoperators=60measuresConductingtheVGR&R

SettingUptheMinitabWorksheetAVGR&Rwillbeconductedwith10parts,3operators,2repeats.OpenaMinitabworksheetandtitle3columns,Parts,Operators,andMeasuresForthe“Parts”columnweneedtolist1through10,6timesfor60measuresCalc>MakePatternedData>SimpleSetofnumbersandenterthedataasfollowsConductingtheVGR&R

SettingUptheMinitabWorksheetForthe“Operators”columnweneedtolist1through3,10timeseachandthenlistthesequencetwiceCalc>MakePatternedData>SimpleSetofnumbersandenterthedataasfollowsConductingtheVGR&R

9StepMethodThefollowingstepsareusedtoconductaVGR&R:CalibratethegageSelect10partsthatspantherangeofthelong-termvariabilityoftheprocess.Ifnotpossible,collectpartsoverseveraldays.Identify2or3operatorswhousethegage.ConducttheVGR&RthegageisusedHavethefirstoperatormeasurethepartsinrandomorderandrecordthemeasuresinMinitab.Havethesecondandthirdoperatorsmeasurethepartsinrandomorder.Repeatsteps5and6suchthatalloperatorshavemeasuredeachparttwiceinrandomorder.AnalyzetheresultsinMinitabDrawyourconclusionsandifrequiredtakecorrectiveaction.Let’sassumewehaveconductedaVGR&RandweneedtoanalyzethedataOpenfileVGR&RExample.mtwAnalyzingtheVGR&RData

MinitabKeystrokesToanalyzethedataclickonStat>QualityTools>GR&RStudy(Crossed),thenentertheParts,Operators,andMeasuresinformation.AlwayshaveANOVAclickedonClickon“Options”andenter.2fortolerance,asthatistheUSL–LSLforthischaracteristic.AnalyzingtheVGR&RData

MinitabSessionWindowGageR&RStudy-ANOVAMethodGageR&RforMeasuresTwo-WayANOVATableWithInteractionSourceDFSSMSFP

Parts90.00918160.0010202289.4510.00000Operators20.00011050.000055315.6790.00011Operators*Parts180.00006340.00000351.5420.14308Repeatability300.00006860.0000023Total590.0094242GageR&R %ContributionSourceVarComp(ofVarComp)

TotalGageR&R5.49E-063.14Repeatability2.29E-061.31Reproducibility3.21E-061.83Operators2.59E-061.48Operators*Parts6.20E-070.35Part-To-Part1.69E-0496.86TotalVariation1.75E-04100.00

StdDevStudyVar%StudyVar%ToleranceSource(SD)(5.15*SD)(%SV)(SV/Toler)

TotalGageR&R2.34E-031.21E-0217.726.03Repeatability1.51E-037.79E-0311.433.89Reproducibility1.79E-039.22E-0313.544.61Operators1.61E-038.28E-0312.164.14Operators*Parts7.87E-044.05E-035.952.03Part-To-Part1.30E-026.70E-0298.4233.52TotalVariation1.32E-026.81E-02100.0034.06NumberofDistinctCategories=8Whatdoesallofthismean? ANOVAtable

VarianceComponents PartitioningtheVariability Thissectionderived fromVarianceComponents123Weareinterestedinsections2and3AnalyzingtheVGR&RResults

VarianceComponents %ContributionSourceVarComp(ofVarComp)

TotalGageR&R5.49E-06

3.14

Repeatability2.29E-061.31Reproducibility3.21E-061.83Operators2.59E-061.48Operators*Parts6.20E-070.35Part-To-Part1.69E-0496.86TotalVariation1.75E-04100.001.75E-04=1.69E-04+5.49E-06100%=96.86%+3.14%2.29E-06+3.21E-06

1.31%+1.83%2.59E-06+6.20E-07

1.48%+0.35%* WhentheP-valueintheANOVAtableforReproducibilityis<0.25MinitabwillfurtherpartitionReproducibilityintothatfromtheOperatorandOp*PartInteraction.%ContributionAnalyzingtheVGR&RResults

StandardDeviation %ContributionSourceVarComp(ofVarComp)

TotalGageR&R5.49E-063.14Repeatability2.29E-061.31Reproducibility3.21E-061.83Operators2.59E-061.48Operators*Parts6.20E-070.35Part-To-Part1.69E-0496.86TotalVariation1.75E-04100.00

StdDevStudyVar%StudyVar%ToleranceSource(SD)(5.15*SD)(%SV)(SV/Toler)

TotalGageR&R2.34E-031.21E-0217.726.03

Repeatability1.51E-037.79E-0311.433.89Reproducibility1.79E-039.22E-0313.544.61Operators1.61E-038.28E-0312.164.14Operators*Parts7.87E-044.05E-035.952.03Part-To-Part1.30E-026.70E-0298.4233.52TotalVariation1.32E-026.81E-02100.0034.06NumberofDistinctCategories=812345Variance(s2)StandardDeviation(s)5.15*St.Dev.

(99%areaundernormalcurve)RatiosofSourcesvs.TotalVariationRatiosofSourcesvs.Tolerance%Study%ToleranceAnalyzingtheVGR&RDataThemeasurementsystemisdeemedtobemarginalUnacceptable >10% >30% >30%Marginal 3–10% 10–30% 10–30%Excellent <3 <10% <10%

%Contribution

%Study

%ToleranceSincetheGageismarginal,whatactionshouldwetaketoimprovethegage?AnalyzingtheVGR&RData

MinitabGraphsAnalyzingtheVGR&RGraphswillhelpdetermineareasforimprovement.Wewilllookateachgraphindividually.AnalyzingtheVGR&RResults

MinitabGraphs–ComponentsofVariationThisgraphisabarchartofthedataintheMinitabSessionWindowThisgraphdoesnotprovideuswithanyquantitativeinformationDONOTreadthisgraphtodeterminewhetherthemeasurementsystemisacceptable.READthenumbersdirectlyfromthesessionwindow.AnalyzingtheVGR&RResults

MinitabGraphs–RChartbyOperators

Eachdatapointrepresentstherange(max–min)foreachpartmeasuredtwicebyeachoperator.ThedatapointsarerunintheordertheyareenteredwithintheMinitabworksheet,1-10.Thisgraphrepresentstherepeatabilityofthegage.Wedesireallofthepointsonthisgraphtobewithinthetworedlinesasthisindicatesarepeatablegage.AnalyzingtheVGR&RResults

MinitabGraphs–X-barChartbyOperatorsEachdotrepresentstheaverageofthetwomeasuresbyeachoperatorTheredlines(UpperandLowerControlLimits–calculatedfromtheRchartbyOperators)representsthevariabilityinthegage.Thevariabilityobservedinthisgraphrepresentsthevariabilityintheparts.Wedesirethesedotstobeoutsidetheredlines.ThisindicatesthevariabilityinthepartsisgreaterthanthevariabilityinthegageAnalyzingtheVGR&RResults

MinitabGraphs–ByPartsTheX-axisrepresentsthe10partsthatweremeasuredEachdotrepresentsameasureforthatpart.Eachpartwasmeasuredtwicebythreeoperatorsthereforetherewouldbe6dotsperpartTheredcrossesrepresenttheaverageofsixmeasuresforeachpartTheperfectgraphwouldcontainoneblackdotforeachpartindicatingtheexactsamemeasuresbyeachoperator.AnalyzingtheVGR&RResults

MinitabGraphs–ByOperatorsTheX-axisrepresentsthe3OperatorsEachdotrepresentsameasurebyanoperatorforeverypart.Thereare20measuresperoperator(10partsmeasuredtwice).TheredcrossesrepresentstheaverageofallmeasuresforeachoperatorThevariabilityobservedwithinanoperatorshouldbeduetothevariabilityofthe10partsTheidealgraphwouldhaveallofthedotsforeachoperatoralignedandtheredaveragelinewouldbehorizontal.AnalyzingtheVGR&RResults

MinitabGraphs–Operators*PartsInteractionTheX-axisrepresentsthe10partsThedotsrepresentseachoperatorsaveragemeasureforthe10partsTheidealplotiswhenallthreelines(operators)overlaponeanotherindicatingtheexactsameaveragemeasureforeachpart.Whenthelinesdivergearoundonepartthatisanindicationthattheoperatorsarehavingdifficultymeasuringthatpart.VGR&RConclusionsBasedonthedataintheMinitabSessionwindowthegageismarginallyacceptable.ThevariabilityintheMeasurementSystemappearstobeevenlydividedbetweenRepeatability(1.31%)andReproducibility(1.83%),basedon%ContributionThegreatestopportunityforimprovementappearstobeinReproducibilityspecificallytheOperators(1.48%).ThisisvalidatedbyobservingtheOperators*PartsInteractiongraph.OnaverageOperator3appearstobemeasuringconsistentlylowerthanOperators1and2.NextStepsObserveanddocumentthedifferenceinmethodsbetweenOperators1and2vs.3andwiththeoperatorsdeterminewhoismeasuringmoreaccurately.Identifybestmethodandstandardize.RepeatVGR&RtovalidateVGR&RExerciseObjective–Toconduct,analyze,anddrawconclusionsinaVGR&R.Assembleingroupsof4andidentifyanAdministratorand3Operators.TheVGR&Rwillconsistof3operators,7parts,3repeatmeasures,unlessotherwisenoted.Theadministratorwillcollectfromtheinstructor7partstobemeasuredandthegage.AllparticipantsinthegroupbeginonSlide29andcreatetheMinitabworksheetinordertoconducttheVGR&RandcontinuefromthereDuringtheVGR&R,theadministratorneedstohidethesamplesfromtheoperatorsandtheoperatorsaretomeasuretheminrandomorder.Asoneoperatorismeasuringtheotheroperatorsarerequiredtobeoutoftheroom.Afterallofthedatahasbeencollected,acopyneedstobegiventoeachpersoninthegroupsotheycananalyzethedataindividual.ThendiscussasagrouptodrawconclusionsWhenyouhavecompletedyouranalysiscalltheInstructorovertoreviewyourresultsFixesforaPoorMeasurementSystemDONOTTHROWOUTANYOFTHEPARTSINTHEMSAReviewthedataenteredintoMinitabanddetermineifanynumbersweretransposedorenteredincorrectly,correct,andre-analyzethedata.Ifthereappearstobeoneortwooutliersinthedatasetandnumberswerenottransposedhavetheoperatorsre-measurethepartandreanalyze.ReproducibilityContactSupplieroftoidentifypropermethodologyofusingthegageTraintheOperatorsinacommonmethodologyEnsureanyfixturesrequiredforthetestarecorrectandanynecessarytoolsareavailableEnsurelightingintheareaisadequateRepeatabilityContactSupplierofgageastocorrectSetupofgagePropersoftwaresettingsEnsureyouareusingthepropergageforthecharacteristictobemeasuredRepairthegageBuynewgageFixesforaPoorMeasurementSystem

StopGap-CentralLimitTheorem(CLT)IfafixcannotbeimplementedimmediatelyusetheCLTBytakingmultiplemeasuresandaveragingthesemeasuredwecanreducethevariabilityinthemeasurementsystembyoneoverthesquarerootofthenumberofmeasures–Example–Ameasurementsystemhasbeendeemedunacceptableandanimmediatefixcannotbeimplemented.Haveeachoperatormeasureeachsamplefourtimesandtaketheaverageofthefourmeasures.EntertheseaveragesintoMinitabandanalyzetheVGR&R.IfacceptabletheoperatorsonthefloorwillthenberequiredtotaketheaverageoffourmeasuresandreportthisvalueTheCLTisnotasolutiontoapoorMeasurementSystemItneedstobefixed!!!WhatiftheMSAisaDestructiveTest?Batchesofpartswillbepreparedthatrepresentthelong-termvariabilityoftheprocess.ItisassumedthatthevariabilityofthepartswithinabatchareconsistentAtestisconductedwith3Operators,10parts,2measures,thestepsareasfollows:Prepare10batchesofparts,wherethebatchesspanthelong-termvariabilityoftheprocess.Forthistest6partsshouldbepreparedperbatch.ThefirstoperatormeasuresarandomsamplefromarandombatchandproceedtomeasureonesamplefromeachbatchThesecondandthirdoperatorsmeasureonesampleeachfromeachbatchThefirstoperatorrepeatsbydrawinganothersamplefromeachbatch,asdothesecondandthirdoperators.AnalyzethedatainMinitabusing

Stat>QualityTools>GR&RStudy(Nested)Readsessionwindowandgraphsthesame asGR&R(Crossed)anddrawconclusions inthesamemanner11234562123456BatchesSamplesSamplesPoorMSAResults-ImpactonDPMO

BiasIssuesProblemswithBiasScenario1BiashasnotbeencorrectedandweareacceptingBadPartsBiasCorrectedLSLUSLLSLUSLAcceptingmanybadpartsthatshouldhavebeenrejected.FalselyunderestimatedDPMOBiasisanissuePoorMSAResults-ImpactonDPMO

BiasIssuesProblemswithBiasScenario2BiashasnotbeencorrectedandwearerejectingGoodPartsBiasCorrectedLSLUSLLSLUSLRejectinggoodpartsthatshouldhavebeenaccepted.FalselyoverestimatedDPMOBiasisanissuePoorMSAResults-ImpactonDPMO

Repeatability&ReproducibilityIssuesProblemswithRepeatabilityandReproducibilityScenario1

WhenaVGR&RhasnotbeenconductedandwearefalselyacceptingBadPartsVGR&RCorrectedLSLUSLLSLUSLAcceptingmanybadpartsthatshouldhavebeenrejected.FalselyunderestimatedDPMOPoorMSAResults-ImpactonDPMO

Repeatability&ReproducibilityIssuesProblemswithRepeatabilityandReproducibilityScenario2WhenaVGR&RhasnotbeenconductedandwearefalselyrejectingGoodPartsVGR&RCorrectedLSLUSLLSLUSLRejectinggoodpartsthatshouldhavebeenaccepted.FalselyoverestimatingDPMOPoorMSAResults-ImpactonDPMO

GeneralCommentsWhatactionswouldwetakegiventhefollowingFixtheMeasurementSystem,orFixCapabilityLSLUSLLSLUSL*%ContributionMSAScenarios

3%Contribution–10%ToleranceSamplesaredrawnfortheVGR&RthatspanthelongtermvariabilityoftheprocessActionNoactionisrequiredLSLUSLMSAScenarios

45%Contribution–10%ToleranceSamplesaredrawnfortheVGR&Rthatspanthelongtermcapabilityoftheprocess.ActionNoactionisrequiredThereisnoadvantagetoimprovingthisMeasurementSystemasthecapabilityisOutstanding!!!OnlyneedtocalibrateLSLUSLMSAScenarios

50%Contribution–50%ToleranceSamplesaredrawnfortheVGR&Rthatspanthelongtermcapabilityoftheprocess.ActionNeedtofixboththeMeasurementSystemandtheCapabilityFixMeasurementSystemfirstthenCapabilityLSLUSLMSAScenarios

3%Contribution–50%ToleranceSamplesaredrawnfortheVGR&Rthatspanthelongtermcapabilityoftheprocess.ActionNeedtofixboththeMeasurementSystemandtheCapabilityFixCapabilityfirstthenbegintoimprovetheMeasurementSystemLSLUSLAttributeMSA(AGR&R)WhyanAGR&R?Whenthecharacteristicbeingmeasureisnotcontinuousbutdiscrete,forexampleBinary–2levelsPassorFailGoodorBadNominal–NonaturalorderingofthelevelsBlue,black,red,yellowScratch,dent,pitsOrdinal–naturalorderingofthelevelsNone,mildsevereExcellent,aboveaverage,average,belowaverage,poorIfpossibledetermineifacontinuousmeasureisavailableandusethattomeasurethedefect!ObjectiveofanAGR&RToassessyourinspectionorworkmanshipstandardsagainstyourcustomer’srequirementsTodetermineifinspectorsacrossallshifts,allmachines,etc…usethesamecriteriatodetermine“good”from“bad”Toquantifytheabilityofinspectors