工程实验设计第10部分：嵌套与裂区设计及其工业应用.pptxVIP

DOX6EMontgomery1DesignofEngineeringExperiments

Part10–NestedandSplit-PlotDesignsTextreference,Chapter14,Pg.525ThesearemultifactorexperimentsthathavesomeimportantindustrialapplicationsNestedandsplit-plotdesignsfrequentlyinvolveoneormorerandomfactors,sothemethodologyofChapter13(expectedmeansquares,variancecomponents)isimportantTherearemanyvariationsofthesedesigns–weconsideronlysomebasicsituations

DOX6EMontgomery2Two-StageNestedDesignSection14-1(pg.525)Inanesteddesign,thelevelsofonefactor(B)issimilartobutnotidenticaltoeachotheratdifferentlevelsofanotherfactor(A)ConsideracompanythatpurchasesmaterialfromthreesuppliersThematerialcomesinbatchesIsthepurityofthematerialuniform?ExperimentaldesignSelectfourbatchesatrandomfromeachsupplierMakethreepuritydeterminationsfromeachbatch

DOX6EMontgomery3Two-StageNestedDesign

DOX6EMontgomery4Two-StageNestedDesign

StatisticalModelandANOVA

DOX6EMontgomery5Two-StageNestedDesign

Example14-1(pg.528)Threesuppliers,fourbatches(selectedrandomly)fromeachsupplier,threesamplesofmaterialtaken(atrandom)fromeachbatchExperimentanddata,Table14-3DataiscodedMinitabbalancedANOVAwillanalyzenesteddesignsMixedmodel,assumerestrictedform

DOX6EMontgomery6MinitabAnalysis–Page530

DOX6EMontgomery7PracticalInterpretation–Example14-1Thereisnodifferenceinpurityamongsuppliers,butsignificantdifferenceinpurityamongbatches(withinsuppliers)Whatarethepracticalimplicationsofthisconclusion?Examineresidualplots–pg.532–plotofresidualsversussupplierisveryimportant(why?)Whatifwehadincorrectlyanalyzedthisexperimentasafactorial?(seeTable14-5,pg.529)Estimationofvariancecomponents(pg.532)

DOX6EMontgomery8VariationsoftheNestedDesignStaggerednesteddesigns(Pg.533)Preventstoomanydegreesoffreedomfrombuildingupatlower