07 - DOE Intro 1_1 实验设计

整理文档很辛苦,赏杯茶钱您下走!

免费阅读已结束,点击下载阅读编辑剩下 ...

阅读已结束,您可以下载文档离线阅读编辑

资源描述

DOEIntro.1ProprietarytoGeneralElectricCompanyRev.1.1IntroductiontoDesignofExperimentsDesignofExperiments(DOE)isapowerfultoolthatletsyouobtaininformationabouthowfactors,aloneandincombination,affecttheoutputofaprocess.Itisanefficientmeansofdatacollectionthatallowssimultaneoustestingofmultiplefactorstodeterminetheirimpactontheaverageprocessoutputorthevariationinprocessoutput.Aftercompletingthischapter,youwillbeableto:•setupfullfactorialdesigninMinitab•useMinitabtoanalyzethemaineffectsandinteractionplotstodeterminetheeffectofX’sontheresponsevariable•UnderstandthebenefitsandlimitationsoffractionalfactorialdesignsDOEIntro.2ProprietarytoGeneralElectricCompanyRev.1.1DesignofExperiments:OverviewWHATisDesignofExperiments(DOE)?Designofexperiments(DOE)isatoolthatallowsyoutoobtaininformationabouthowfactors(Xs),aloneandincombination,thataffectaprocessanditsoutput(Ys).WHYisDOEanimportanttool?DOEusesanefficient,costeffective,andmethodicalapproachtocollectingandanalyzingdatasothattheXsareindependenttooneanothertodetermine•amathematicalmodelthatdescribestherelationshipbetweenprocessinputsandoutputsor•thebestconfigurationorcombinationofXs.Bytestingmorethatonefactoratatime,DOEisabletoidentifyallfactorsandcombinationsoffactorsthataffecttheprocessYs.DOEIntro.3ProprietarytoGeneralElectricCompanyRev.1.1WhatTypeofQuestionsDoesDOEAnswer?Beginwith...Whatdoyouwanttoknow?TypeofExperimentalStrategy?FactorialDesignsTypicalQuestionsAddressedbyFactorialDesigns•WhatarethemostsignificantX’s?HowcanInarrowmylistofpotentialX’stothetrulyvitalfew?•HowdotheX’saffecttheY’s?CanIdescribetherelationshipquantitatively?•ArethereanyinteractionsamongtheX’sthataffecttheY’s?WhatshouldIdoaboutthese?•WhatarethebestsettingsandtolerancesformyX’stooptimizemyY’s?DOEIntro.4ProprietarytoGeneralElectricCompanyRev.1.1TypesofDesignedExperimentsTherearemanytypesofDesignedExperiments.FullFactorial,FractionalFactorialandScreeningdesigns.FullFactorialDesigns:(Focusofthistraining)•Comprehensiveapproach•Mostinformation•Higheroutlayofresources•TypicallyusedwithlessthanfivefactorsFractionalFactorialDesigns:(seeappendix)•Includesselectedcombinationsoffactors•Lessinformationonimpactofcombinationsoffactors•Lessresources•UsefulwithhighnumberoffactorsScreeningDesigns:•UsedasfirststeptoreducelargenumberoffactorstomoremanageablenumberDOEIntro.5ProprietarytoGeneralElectricCompanyRev.1.1WhatCanYouDoWithThisEquation?•SetX’sToAssureThatYourYHitsTarget•Trade-OffX’sToReduceCostButStillHitTargetY•FindTolerancesonX’sToKeepYinSpecWhatkindofKNOWLEDGEcanweobtainfromadesignedexperiment?Y=157.3+28.5X1-102.2X2+53.1X1X2OutputsProcessX’sorFactorsPROCESSX1X2X3X4YDOEIntro.6ProprietarytoGeneralElectricCompanyRev.1.1ComparisonofOne-Factor-At-A-TimeExperimentationwithDOEMethods:Problem:ImproveGasMileageApproach:EvaluateGasTypeandTimingSettingOne-Factor-At-A-TimeApproach:(1)SetTimingatT1,EvaluateMPGofGasTypesG1andG2(2)G1GivesBestMPG.HoldGasType=G1Constant(3)EvaluateMPGforTimingT1andT2.DetermineBestValueBestValueG1,T1GasTypeTimingSet-UpMPGT1T1G1G23020GasTypeTimingSet-UpMPGT2T1G1G13025DOEIntro.7ProprietarytoGeneralElectricCompanyRev.1.1MissedThisPointUsingOne-At-A-TimeApproach!GasType&TimingInteract:EffectofOneFactorDependsontheLevelofTheOtherFactor50403020T1T2G2G1MPGComparisonofOne-Factor-At-A-TimeExperimentationwithDOEMethods:Problem:ImproveGasMileageApproach:EvaluateGasTypeandTimingSettingDesignofExperimentsApproach:GasTypeTimingSet-UpMPGT2T1G1G13025G2G2T1T22045FullFactorialDesign:DOEIntro.8ProprietarytoGeneralElectricCompanyRev.1.1Let’sStartWithAVerySimpleDesignABLOW-1,-1HIGH1,-1HIGH-1,1HIGH1,10,022Design1-1-121-13-11411RunABPros:Cons:•Simpledesign•4teststocontrol•1interaction•Designisorthogonal•Usingonly2levels,youassumesaffectsarelinearDOEIntro.9ProprietarytoGeneralElectricCompanyRev.1.1SomeDesignConsiderations:BalanceandOrthogonalCopyright1995SixSigmaAcademyInc.Balance:AbalancedDOEhasthesamenumberofrunsatthelowlevelasatthehighlevelforeachfactor.Forexample,ina22design,thereare4runs--youwouldhave2atthelowsettingoffactorA,and2atthehighsettingoffactorA.Ifarun4wasnotcompletedintheexperiment,itwouldnotbebalanced.ThesameholdstrueforfactorB.1-1-121-13-11411RunABLow“A”SettingsHigh“A”SettingsDOEIntro.10ProprietarytoGeneralElectricCompanyRev.1.1SomeDesignConsiderations:BalanceandOrthogonalBalanceandOrthogonalconditionsarebuiltintothestandarddesigns.Usethestandarddesignsandyoudon’thavetoworryabouttheseconditions!!Orthogonal:AnorthogonaldesignmeansthatforeveryleveloffactorA,therearerunswithboththelowandhighsettingoffactorB.Anorthogonaldesignwillguaranteethefactors,Xs,areindependentofoneanother1-1-121-13-11411RunABForthelowsettingof“A”,thereisarunwith“B”atit’slowsettingandonewith“B”atit’shighsettingDOEIntro.11ProprietarytoGeneralElectricCompanyRev.1.1StepstoperformingaDesignedExperimentPreparingforaDOE1.SelectyourYs.2.SelectyourXs.3.SelectlevelsforeachX.4.PerformariskassessmentofyourXlevels.5.Selectthedesign.PerformingtheDOE6.SetupthedesigninMinitab.7.Runtheexperimentandcollectthedata.8.EnterthedataintoMinitab.An

1 / 65
下载文档,编辑使用

©2015-2020 m.777doc.com 三七文档.

备案号:鲁ICP备2024069028号-1 客服联系 QQ:2149211541

×
保存成功