mechanical performances of dies

Upload: jaya-vijayan

Post on 03-Jun-2018

215 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/12/2019 Mechanical Performances of Dies

    1/284

    Final Technical Report

    ESMART Subtask 5.8: Mechanical Performance of Dies

    AwardNumberDEFC3604GO14230

    CoOpAgreementNo.2005307

    OSURFProjectNumber746200

    ProjectPeriod: January2004June2011

    R.AllenMiller,PrincipalInvestigator

    Phone: 6145811754

    Email: [email protected]

    OhioStateUniversity

    210BakerSystemsEngineering

    1971NeilAve.

    Columbus,OH43210

    Contributors:

    KhalilKabiri Bamoradian,ResearchEngineer

    AbelardoDelgadoGarza,PhDStudent

    KarthikMurugesan,PhDStudent

    AdhamRagab,PostdoctoralResearcher

    PartnerOrganization: NorthAmericanDieCastingAssociation

    September13,2011

  • 8/12/2019 Mechanical Performances of Dies

    2/284

    i

    Acknowledgement/Disclaimer

    Acknowledgement: ThisreportisbaseduponworksupportedbytheUS.Departmentof

    Energy

    under

    Award

    No.

    DOE

    award

    DE

    FC36

    04GO14230.

    Disclaimer: Anyopinions,findings,andconclusionsorrecommendationsexpressedinthis

    materialarethoseoftheauthoranddonotnecessarilyreflecttheviewsoftheDepartment

    ofEnergy.

    ProprietaryDataNotice: Thisreportdoesnotcontainanyproprietarydata.

    DocumentAvailability: ReportsareavailablefreeviatheU.S.DepartmentofEnergy(DOE)

    InformationBridgeWebsite:http://www.osti.gov/bridge

    ReportsareavailabletoDOEemployees,DOEcontractors,EnergyTechnologyDataExchange

    (ETDE)representatives,andInformationalNuclearInformationSystem(INIS)representatives

    fromthefollowingsource:

    OfficeofScientificandTechnicalInformation

    P.O.Box62

    OakRidge,TN37831

    Tel:(865)5768401

    FAX:(865)5765728

    Email:[email protected]

    Website:http://www.osti.gov/contract.html

  • 8/12/2019 Mechanical Performances of Dies

    3/284

    ii

    TableofContentsAcknowledgement/Disclaimer................................................................................. i

    Table

    of

    Contents

    ....................................................................................................

    ii

    ListofFigures......................................................................................................... iv

    ListofTables............................................................................................................ v

    ListofAcronyms..................................................................................................... vi

    ListofAppendices..................................................................................................vii

    ExecutiveSummary............................................................................................... viii

    1. Introduction...................................................................................................... 1

    2. Background....................................................................................................... 2

    2.1 SummaryofPreviousWork.............................................................................. 2

    2.2 SummaryoftheStateoftheArt................................................................... 10

    2.3 Objectives....................................................................................................... 11

    2.4 TasksandApproach........................................................................................ 12

    3. ResultsandDiscussion.................................................................................... 14

    3.1 AnalysisofDiePositionEffects....................................................................... 14

    3.1.1 ComputationalExperiments 16

    3.1.2 DimensionalAnalysis 17

    3.1.3 ModelAdequacy,FEAResults 21

    3.1.4ModelAdequacy,ExperimentalMeasurements 22

    3.2 DieFailureCaseStudy.................................................................................... 27

    3.3 CavityPressureModelingMethods...............................................................29

    3.3.1 FluidStructureInteraction(FSI)Model 30

    3.3.2 AlternativetoFSI 32

    3.3.3 TrackingofCavityDistortion 33

    3.4 EvaluationofEjectorSideandSlideDesign...................................................35

    3.4.1 DesignofExperiments 35

    3.4.2

    Dimensional

    Analysis

    37

    3.4.3 CoverSidePartingSurfaceSeparation 43

    3.5 ModelingandDesignGuidelines.................................................................... 44

    4. BenefitsAssessment....................................................................................... 45

    5. Commercialization.......................................................................................... 46

    6. Accomplishments............................................................................................ 46

  • 8/12/2019 Mechanical Performances of Dies

    4/284

    iii

    7. Conclusions..................................................................................................... 48

    8. Recommendations.......................................................................................... 48

    9. References...................................................................................................... 49

    10.AppendicesAE............................................................................................. 55

  • 8/12/2019 Mechanical Performances of Dies

    5/284

    iv

    ListofFiguresFigure1 TypicalDieCastingSystemModelUsedbyOSU...........................................10

    Figure2 FreeBodyDiagramoftheMechanicalLoads................................................15

    Figure3 DepictionofSystemResponse...................................................................... 15

    Figure4 CoordinateSystemandTiebarLabels..........................................................17

    Figure5 Schematicofthetestdieonthemachineplatens.......................................23

    Figure6 Schematicofstraingaugesandcoordinatesystem......................................23

    Figure7 TiebarLoadMeasurementsvs.Predictions.................................................26

    Figure8 TheCasting.................................................................................................... 28

    Figure9 Defects........................................................................................................... 28

    Figure10 FSIcavitydisplacementpredictions............................................................31

    Figure11Castingfiniteelementmesh........................................................................ 33

    Figure12 Shellelementmesh..................................................................................... 35

    Figure13 SchematicofPillarPatternsusedintheStudy...........................................37

    Figure14LengthScalesRepresentingUnsupportedSpan..........................................42

  • 8/12/2019 Mechanical Performances of Dies

    6/284

    v

    ListofTablesTable1 DescriptionofVariables................................................................................. 17

    Table2:ParameterEstimatesforTopTieBarModelFit.............................................19

    Table3:ParameterEstimatesforBottomTieBarModelFit.......................................20

    Table4:SummaryofFiniteElementModels...............................................................21

    Table5:ComparisonofModelPredictionsfora3500TonMachine..........................21

    Table6:ComparisonofModelPredictionsfora1000TonMachine..........................22

    Table7:ComparisonofModelPredictionsfora250TonMachine............................22

    Table8:ExperimentalArray......................................................................................... 24

    Table9:ComparisonofMeasurementsandPredictions.............................................25

    Table10:DifferencebetweenMeasurementsandModelPredictions.......................27

    Table11 FactorsusedinDesignofExperiments........................................................36

    Table12NonDimensionalStructuralDesignParameters..........................................39

    Table13ParameterEstimatesforEjectorSideFit......................................................41

    Table14 ParameterEstimatesforCoverSideFit.......................................................44

  • 8/12/2019 Mechanical Performances of Dies

    7/284

    vi

    ListofAcronymsNADCA NorthAmericanDieCastingAssociation

    DOE DepartmentofEnergy,alsoDesignofExperiments

    FEA FiniteElementAnalysis

    FSI FluidStructureInteraction

    SPSS Statisticalanalysissoftware

  • 8/12/2019 Mechanical Performances of Dies

    8/284

    vii

    ListofAppendicesAppendixA: K.KabiriBamoradianDieandDieCastingMachineForceandDeflection

    Predictions,,ReportpreparedforNADCA,July2010.

    AppendixB:A.Ragab,K.KabiriBamoradian,DoorCloserDieCaseStudy,May2004.

    AppendixC: R.A.Miller,K.KabiriBamoradian,andA.Garza,FiniteElementModelingof

    CastingDistortioninDieCasting,NADCATransactions2009,NorthAmerican

    DieCastingAssociation,Wheeling,Illinois,April2009

    AppendixD: AbelardoGarzaDelgado,K.KabiriBamoradian,R.A.Miller,Determinationof

    ElevatedTemperatureMechanicalPropertiesofanAluminumA380.0Die

    CastingAlloyintheAsCastCondition",NADCATransactions2008,North

    AmericanDieCastingAssociation,Wheeling,Illinois,May2008.

    AppendixE: DieandDieCastingMachineComputerSimulations: Modeling,Meshing,

    BoundaryConditions,andAnalysisProcedures

  • 8/12/2019 Mechanical Performances of Dies

    9/284

    viii

    ExecutiveSummaryAsanetshapeprocess,diecastingisintrinsicallyefficientandimprovementsinenergy

    efficiencyarestronglydependentondesignandprocessimprovementsthatreducescrap

    ratessothatmoreofthetotalconsumedenergygoesintoacceptable,usablecastings. A

    castingthatisdistortedandfailstomeetspecifieddimensionalrequirementsistypically

    remeltedbutthisstillresultsinadecreaseinprocessyield,lostproductivity,andincreased

    energyconsumption. Thisworkfocusesondeveloping,andexpandingtheuseof,computer

    modelingmethodsthatcanbeusedtoimprovethedimensionalaccuracyofdiecastingsand

    producediedesignsandmachine/diesetupsthatreducerejectionratesduetodimensional

    issues.

    Amajorfactorcontributingtothedimensionalinaccuracyofthecastingistheelastic

    deformationsofthediecavitycausedbythethermomechanicalloadsthediesaresubjected

    to

    during

    normal

    operation.

    Although

    thermal

    and

    die

    cavity

    filling

    simulation

    are

    widely

    usedintheindustry,structuralmodelingofthedie,particularlyformanagingpartdistortion,

    isnotyetwidelypracticed.Thismaybedueinparttotheneedtohaveathorough

    understandingofthephysicalphenomenoninvolvedindiedistortionandthemathematical

    theoryemployedinthenumericalmodelstoefficientlymodelthediedistortion

    phenomenon.Therefore,twoofthegoalsofthisworkaretoassistineffortstoexpandthe

    useofstructuralmodelingandrelatedtechnologiesinthediecastingindustryby1)

    providingadetailedmodelingguidelineandtutorialforthoseinterestedindevelopingthe

    necessaryskillsandcapabilityand2)bydevelopingsimplemetamodelsthatcapturethe

    resultsandexperiencegainedfromseveralyearsofdiedistortionresearchandcanbeused

    topredictkeydistortionphenomenaofrelevancetoadiecasterwithaminimumof

    backgroundandwithouttheneedforsimulations. Theseobjectivesweremet. Adetailed

    modelingtutorialwasprovidedtoNADCAfordistributiontotheindustry. Powerlawbased

    metamodelsforpredictingmachinetiebarloadingandforpredictingmaximumparting

    surfaceseparationweresuccessfullydevelopedandtestedagainstsimulationresultsfora

    widerangeofmachinesandexperimentaldata. Themodelsprovedtoberemarkably

    accurate,certainlywellwithintherequirementsforpracticalapplication.

    Inadditiontomakingdiestructuralmodelingmoreaccessible,theworkadvancedthestate

    oftheartbydevelopingimprovedmodelingofcavitypressureeffects,whichistypically

    modeledasahydrostaticboundarycondition,andperformingasystematicanalysisofthe

    influenceofejectordiedesignvariablesondiedeflectionandpartingplaneseparation. This

    cavitypressuremodelingobjectivemetwithlessthancompletesuccessduetothelimitsof

    currentfiniteelementbasedfluidstructureinteractionanalysismethods,butanimproved

    representationofthecasting/dieinterfacewasaccomplishedusingacombinationofsolid

  • 8/12/2019 Mechanical Performances of Dies

    10/284

    ix

    andshellelementsinthefiniteelementmodel. Thisapproximationenabledgoodprediction

    offinalpartdistortionverifiedwithacomprehensiveevaluationofthedimensionsoftest

    castingsproducedwithadesignexperiment. Anextradeliverableoftheexperimentalwork

    wasdevelopmentofhightemperaturemechanicalpropertiesfortheA380diecastingalloy.

    Theejectorsidedesignobjectivewasmetandtheresultswereincorporatedintothemeta

    modelsdescribedabove.

    Thisnewtechnologywaspredictedtoresultinanaverageenergysavingsof2.03trillion

    BTUs/yearovera10yearperiod.Current(2011)annualenergysavingestimatesoveraten

    yearperiod,basedoncommercialintroductionin2009,amarketpenetrationof70%by

    2014is4.26trillionBTUs/yearby2019. Alongwiththeseenergysavings,reductionofscrap

    andimprovementincastingyieldwillresultinareductionoftheenvironmentalemissions

    associatedwiththemeltingandpouringofthemetalwhichwillbesavedasaresultofthis

    technology.TheaverageannualestimateofCO2reductionperyearthrough2020is0.085

    MillionMetricTonsofCarbonEquivalent(MMTCE).

  • 8/12/2019 Mechanical Performances of Dies

    11/284

    1

    1. IntroductionDiecastingisanearnetshapemanufacturingprocessinwhichpartswithcomplex

    geometriesareproducedbyinjectingmoltenmetalintosteelmolds/diesunderhigh

    pressure.Themoltenmetalisheldinthediecavityuntilitsolidifiesandpartiallycooled,the

    dieisopenedandthepartisejected,andtheprocessisrepeatedthousandsoftimes. The

    processofferscompetitiveadvantagesovermanyothernetshapemanufacturingprocesses

    suchasforgingandstampingwithitsabilitytoproducepartswithcomplexgeometric

    features,highsurfacefinishandtightdimensionaltolerances.

    Asanetshapeprocess,diecastingisintrinsicallyefficientandimprovementsinenergy

    efficiencyarestronglyrelatedtodesignandprocessimprovementsthatreducescraprates

    sothatmoreofthetotalconsumedenergygoesintoacceptable,usablecastings. Acasting

    thatisdistortedandfailstomeetthespecifieddimensionalrequirementsisscrappedand

    remeltedbutthisstillresultsinadecreaseinprocessyield,lostproductivity,andincreasedenergyconsumption. Thisworkfocusesondevelopingandexpandingtheuseofcomputer

    modelingmethodsthatcanbeusedtoimprovethedimensionalaccuracyofdiecastingsand

    producediedesignsandprocesssetupsthatreducerejectionratesduetodimensional

    issues.

    Oneofthemajorfactorsthatcontributetothedimensionalinaccuracyofthecastingisthe

    elasticdeformationsofthediecavitycausedbythethermomechanicalloadsthediesare

    subjectedtoduringnormaloperation.Diecastingdiesareexpectedtorunseveralmillion

    cyclesduringtheirlifetime.Highmanufacturingcostsprohibitprototypingandanyserious

    diedeformationordiefailureproblemsarenotnoticeduntilthefirstproductionrun.Adie

    cancostanywherebetween$50,000to$1,000,000andthedeliverytimesrangesfrom312

    monthsdependinguponthecomplexityofthedies.Thereforeitisextremelyimportantthat

    thediedistortionbepredictedandcontrolledatthedesignstage.

    Althoughthermalanddiecavityfillingsimulationarewidelyusedintheindustry,structural

    modelingofthedie,particularlyformanagingpartdistortion,isnotyetwidelypracticed.

    Thismaybedueinparttotheneedtohaveathoroughunderstandingofthephysical

    phenomenoninvolvedindiedistortionandthemathematicaltheoryemployedinthe

    numericalmodelstoefficientlymodelthediedistortionphenomenon.Therefore,twoofthegoalsofthisworkaretoassistineffortstoexpandtheuseofstructuralmodelingand

    relatedtechnologiesinthediecastingindustry. Specificallyto

    1. Provideadetailedmodelingguidelineandtutorialforthoseinterestedindevelopingthe

    necessaryskillsandcapabilityand

  • 8/12/2019 Mechanical Performances of Dies

    12/284

    2

    2. developsimplemetamodelsthatcapturetheresultsandexperiencegainedfromseveral

    yearsofdiedistortionresearchandcanbeusedtopredictkeydistortionphenomenaof

    relevancetoadiecasterwithaminimumofbackgroundandwithouttheneedfor

    simulations.

    Inadditiontheworkisintendedtoadvancethestateoftheartbydevelopingimproved

    modelingofcavitypressureeffectswhichistypicallymodeledasahydrostaticboundary

    conditionandtoperformasystematicanalysisoftheinfluenceofejectordiedesign

    variablesondiedeflectionandpartingplaneseparation.

    Intermsofcommercialization/disseminationofthemethodology,thetutorialmentioned

    aboveisnowavailableinelectronicformfromtheNorthAmericanDieCastingAssociation

    (NADCA)websiteandislinkedtoothermaterialrelatedtodistortion. Twometamodels

    (onefortiebarbalanceusedformachinesetuptominimizedistortionandoneforparting

    planeseparation(flash)prediction)havealsobeentransferredtoNADCAandavailableto

    theindustry. Themetamodelsaredimensionlesspowerlawmodelsthatencapsulate

    resultsfromsimulationsperformedinthisprojectandinprecedingprojectsinaformthat

    canbeusedwithouttheneedforacomplexFEAsimulation. Eachmetamodelhasbeen

    convertedtoasimpleapplicationthatcanbedownloadedfromtheNADCAwebsite. Thetie

    barbalanceresultshavealsobeenimplementedasamodulewithinversion4ofthePQ2

    softwareprogramthatisdistributedbyNADCA. Thissoftwareisusedtohelpfindthebest

    matchofdie,machine,andprocesssetupwithpartdesignrequirements. Includingthetie

    barbalancemodelinthissoftwareenablesbasicdistortionconsiderationstobeconsidered

    alongwithmachinepoweranddiefillingconditions.

    2. BackgroundThepurposeofthissectionistoprovideasummaryofpreviousworkperformedonthe

    mechanicalperformanceofdiesandtooutlinetheobjectivesofthisproject.

    2.1 SummaryofPreviousWorkTodateonlytwogroupsintheworldhavesystematicallyaddressedthemechanical

    performanceofthedieandmachinesystemindiecasting.OneofthegroupsisOSU,

    supportedthroughpreviousDOEfundedwork.ThesecondisDaveCaulkandhiscolleagues

    atthe

    General

    Motors

    Tech.

    Center

    when

    they

    developed

    the

    Die

    Cast

    software

    (e.g.,

    [25]).

    Othersaroundtheworldhaveperformedexperimentalworkandgenerallyoverlysimplified

    staticanalyses,butonlythesetwogroupshavelookedatthefundamentalprinciplesthat

    underliethebehaviorofthedieinresponsetothemachineclampingforces,cyclicheat

    loads,andhighcavitypressuresduringfillandintensification.Perhapsbecauseonlyhigh

    pressureprocessesareimpactedbytheseadditionalloads(sand,investment,permanent

    moldsystemsarenotsubjecttothesameclampingandpressureloadsandonlypermanent

  • 8/12/2019 Mechanical Performances of Dies

    13/284

    3

    moldissubjecttocyclicheatloads),theproblemsarefarlesswidelystudiedthanthermal

    andfillrelatedissuesforcastingprocesses.

    Thefocusofthediecastingresearchcommunityhasbeenmostlyondeveloping

    numerical/computationalmethodstosolveheattransfer,fluidflow,solidificationand

    thermaldistortionrelatedproblems.Veryfewresearchershavepaidattentiontotheroleof

    mechanicalloadssuchasclampingforceandcavitypressureindieandcastingdistortion.

    Eveninthemodelswheremechanicalloadsareconsidered,notallofthemaccountforthe

    stiffnessofthemachinepartssuchastheplatens,tiebarsandthetogglemechanism.

    AhuettGarza[21]and[22],conductedanelaboratestudyoftheloadsinvolvedindiecasting

    processandheconcludedthatdiedeflectionsimulationswithreasonableresolutioncanbe

    carriedoutbyaccountingfortheclampingload,heatreleasedduringsolidification,the

    intensificationpressure,andtheheatremovedduringlubricantspray.Hisstudyshowedthat

    theheat

    released

    during

    fill,

    the

    momentum

    during

    filling

    and

    the

    pressure

    surge

    atthe

    end

    offillcanbeignoredinthediedeflectionsimulationsandstillresultswithreasonable

    accuracyandresolutioncanbeachieved.Byanorderofmagnitudeanalysisitwasshown

    thattheheatreleasedduringfillcanbeignoredwhentheratiobetweenhalfthethicknessof

    thepartandthefilltimeisgreaterthanorequaltooneseventh.Thiscorrespondstoacase

    wherethesolidificationtimeisatleastanorderofmagnitudegreaterthanthefilltime.The

    detailsofthescaleanalysisarealsoprovidedin[23].

    Basedontheresultsofhisstudyaninitialfiniteelementmodelingprocedurewasdeveloped

    andtested[21],[24]and[25].Thepreliminarymodelconsistedofonlythecoverdie,ejector

    dieandtheejectorsupportblock.Firstathermalanalysiswascarriedouttoobtainthe

    nodaltemperaturevaluesonthedies,whichwerelaterusedinthestressanalysis. The

    cavitypressurewasmodeledasahydrostaticpressurewithmagnitudeequaltothatof

    intensificationpressure.Theclamploadwasmodeledasapressureboundarycondition

    behindtheejectorsupportblock.Arigidsupportwasassumedbehindthecoverdieand

    nodesonthebacksurfaceofthediewereconstrainedinalldirections.Thestiffnessofthe

    machinewasnotaccountedforinthismodel.

    InasubsequentstudyDedhia[26]comparedthepartingplaneseparationpredictionsofa

    modelwithrigidsupportbehindthecoverdieversusthepartingplaneseparationpredictionsfromamodelthataccountedforthemachinestiffness.Springelementswere

    usedtoaccountforthestiffnessoftheplatensandthetogglemechanism.Theclampload

    wasmodeledbyapplyingappropriatedisplacementboundaryconditionstothespring

    elementsthatrepresentedthetogglemechanism.Theseparationvaluesatseverallocations

    alongtheedgesofthecavitywereusedasameasureofdiedeflection.Themaximum

  • 8/12/2019 Mechanical Performances of Dies

    14/284

    4

    separationvaluewasabout12%to20%higherinthemodelwithspringelementsthanthe

    valuesfromthemodelwithrigidsupport,dependinguponthedesignfeaturesofthedie.

    Choudhary[27]developedafiniteelementmodelinwhichthethreemachineplatens,theC

    frameandthetiebarsweremodeledexplicitly.Thediewasadummystructurethat

    consistedoftwoparallelplatesconnectedbypillarsonthefourcornersoftheplates.A

    rollersupportwasmodeledatthebottomofcoverandejectorplatens.Asupportblockwas

    modeledatthebottomoftherearplatentopreventdisplacementintheverticaldirection.A

    smallslidingcontactwasdefinedatallinterfaces.Thenodesontheeitherendofthetiebars

    weretiedtocorrespondingnodesontheejectorandrearplaten.Theclamploadwas

    appliedasapressureboundaryconditiononthetoggleblocksonthecoverandrearplatens.

    Thermalloadsandcavitypressurewereignoredinthemodel.Thedeflectionofthecover

    platenwaspredictedateightdifferentlocationsbehindthecoverplatenandtheresults

    werecomparedwithcorrespondingvaluesfromthefielddata.Thedeflectionpatternfrom

    thesimulationswassimilartothepatternobservedonthefielddata.Buttheindividual

    deflectionvaluesfellintherangeof10%to15%oftheobservedfielddata.Thismodelwas

    fairlyaccurategiventhevariousapproximationstotheboundaryconditionsinthemodel

    andtheprocedurefollowedtomodeltheclampload.

    Inanotherdiedistortionmodelingstudy,Chayapathi[28]usedafiniteelementmodelin

    whichthetiebarswereexplicitlymodeledandthetogglemechanismwasrepresentedby

    linearspringelements.Thenodesononeendofthetiebarthatareincontactwiththe

    nodesinthecoverplatenweretiedtothecorrespondingnodesonthecoverplaten.The

    nodes

    on

    the

    other

    end

    of

    the

    tie

    bar

    were

    constrained

    in

    all

    six

    degrees

    of

    freedom.

    The

    cornernodesonthebottomofthecoverplatenwereconstrainedinverticaldirectionto

    preventrigidbodymotion.Theclamploadwasappliedbyspecifyingdisplacementsonthe

    freeendofthespringelements.Theintensificationpressureloadwasappliedasapressure

    boundaryconditiononthecavitysurfaces.

    Ragabetal[29]experimentallyverifiedtheadequacyofthisfiniteelementmodelfor

    predictingthecontactloadsbetweenthediesandplatensonthecoverandejectorsides.

    Thecontactloadbetweentheplatensandthediesweremeasuredusingatotalof35load

    cells,18loadcellsonthecoversideand17loadcellsontheejectorside.Thecontactload

    wasmeasuredundertwodifferentloadingconditions,underclamploadonlyandduring

    actualcastingoperation.Theloadcellsandthefixturesusedintheexperimentswerealso

    explicitlyincludedinthefiniteelementmodel.Thesummationofcoversideloadcell

    measurementsdecreasedby7%afterintensificationwhereasthesummationofcoverside

    loadcellpredictionsfromsimulationremainedconstant.Ontheejectorsidethesummation

    ofloadcellmeasurementsandpredictionsremainedconstant.Thedifferencebetween

  • 8/12/2019 Mechanical Performances of Dies

    15/284

    5

    modelpredictionsandmeasurementsonthecoversidewereattributedtothefactthatthe

    modelisstifferthanthedie/machineactuallyis.

    Toaddresstheseobserveddifferencesbetweenthemodelpredictionsandmeasurements

    onthecoversidefurthermodelingimprovementsweretestedbyArrambide[30].Various

    machinecomponentswereincludedinthefiniteelementmodelsandthepredictionswere

    comparedagaintotheexperimentalloadcellmeasurements.Fourdifferentmodelswere

    tested.Thefirstmodelincludedthecoverandejectorplaten,dies,inserts,theloadcellsand

    fixturesallofthemmodeledusingquadratictetrahedralelements.Thetiebarswere

    modeledusingbeamelements,withtheoneendofthebeamelementsconstrainedtothe

    coverplatenandtheotherendwasfixedinspace.Severalnodesonthebottomofthecover

    platenwasconstrainedinverticalandtiebardirections.Theclamploadwasappliedasa

    pressureboundaryconditionbehindtheejectorplaten.Inthesecondmodeltherearplaten

    wasalsoincludedandthetwoendsofthetiebarswereconstrainedtothecoverandrear

    platens.Thetogglemechanismwasrepresentedbybeamelementsandtheclamploadwas

    appliedbyspecifyingappropriatetemperatureonthesebeamelements.Inthethirdand

    fourthmodelsthefrontsupportframewasaddedtotheprevioustwomodels.

    Comparisonbetweenloadcellmeasurementsandloadcellpredictionsbetweensimulations

    showedthatthefrontsupportframedidnothaveanyeffectonthecontactloadbetween

    thediesandtheplatens.Includingtherearplatenandtogglemechanisminthemodel

    alteredtheloaddistributiononvariousloadcellpredictionsby2 34%,withanaverageof

    11%.Alsothefullmodelshowedgoodcorrelationwiththeexperimentalmeasurements.To

    test

    the

    adequacy

    of

    the

    full

    model

    to

    predict

    parting

    plane

    separation,

    a

    simplified

    model

    withdies,insertsandloadcellsonlywasbuilt.Theclamploadwasappliedbehindtheload

    cellsdirectlyusingthepredictionsfromthefullmodelandalsotheloadsfromthe

    experimentalmeasurements.Themaximumseparationshowedadifferenceof0.001which

    fallswithintheresolutionofthenumericalsimulation.

    Inallofthediedistortionmodelingstudiesdiscussedabove,theintensificationpressurewas

    assumedtobehydrostaticanditwasmodeledasapressureboundaryconditiononthe

    cavitysurfacesinthefiniteelementmodels.Inrealitytheliquidmetalcarriesthehydrostatic

    pressurefromtheplungermechanismandtransfersittothecavitysurfaces.Butthesolid

    elementsusedinthestructuralfiniteelementmodelscannotcarrythishydrostaticpressure

    tothecavity.

    GarzaDelgado[31]developedatwodimensionalfluidstructureinteraction(FSI)finite

    elementmodelusingADINAtopredictdiedistortion.Itwasafullycoupledthermo

    mechanicaltestmodelthatwasdevelopedtogainanunderstandingofthecapabilityofthe

    fluidstructureinteractionmodeltopredictdiedistortion.AnFSIboundaryconditionwas

  • 8/12/2019 Mechanical Performances of Dies

    16/284

    6

    definedattheinterfacebetweenthesolidandliquiddomain.Asoliddomainwasusedto

    representthediesandaliquiddomainwasusedtorepresentthecastingandthepressure

    loadwassimulatedbyspecifyinganodalpressureboundaryconditioninthegateareaofthe

    fluiddomain.TheFSImethodusesaconjugateheattransfertocalculatetheheatfluxes

    acrosstheinterfaceandhencenointerfacialheattransfercoefficientshadtobedefined

    betweentheliquidandthesoliddomain.Latentheateffectswereincludedinthemodelby

    specifyingtemperaturedependentspecificheatcurve.Thismodelwasdevelopedfor

    demonstrationpurposesanditisyettobeimplementedincomplexdiedistortion

    simulations.

    Anotherimportantdynamicloadthathasbeenignoredindiedistortionsimulationsisthe

    impactloadcausedbythesuddendecelerationoftheplungermechanismattheendoffill.

    Xueetal[32],attemptedtopredictthepressuredistributioninthediecavityduetothis

    impactloadingusingaCFDmodel.Thegoalwastousethepressurepredictionsfromthis

    CFDmodeltoapproximatethedynamiccavitypressureinstructuraldiedistortion

    simulations.FLOW3Dwasusedtosimulatethemetalflowintheshotsleeve,runnerand

    thedies.Themoltenaluminumwastreatedasslightlycompressiblefluidwithtemperature

    independentmaterialpropertiesandaK turbulentmodelwasusedinthesimulation.Heat

    transferbetweenthemetalandthedieswasincludedinthemodelandtheheatconduction

    inthediewasneglected.Pressurehistoryatdifferentcavitylocationswasinvestigated.The

    pressurespikewasfoundtobemorethantwiceoftheintensificationpressurethatis

    normallyusedintheproductionofthisexperimentalcastpartusedinthisstudy.Itwasalso

    observedthatthepressurewithinthecavitywasalmostuniformandthemaximumpressure

    differenceinthecavitywasalsoverysmall(about50PSI)attheinstanttheimpactoccurs. It

    wasalsoobservedfromthepredictionsthatthepressureinthecavitywaszeroduringthe

    slowshotphaseanditreachedthepeakatdifferentlocationsatdifferentinstantsoftime

    duringthefastshot.Themaximumpressureoccurredduringthedecelerationphasethrough

    outthecavity.

    Milleratal[33],developedafiniteelementmodeltopredictthedeflectionoftheslidesin

    nonopenclosediesandtheresultsfromthemodelwerecomparedwithfielddata.Thefield

    dataconsistedoftheslideblowbackandtiltvaluesfromnominalpositionunderdifferent

    pressureloadsfordifferentslidedesign.Thesimplefiniteelementmodelassumedarigid

    supportbehindthecoverdie.Themodelpredictionsshowedagoodcorrelationwiththe

    experimentaldataexceptforthehighpressurecases.Thiswasduetotheassumptionof

    rigidsupportbehindthecoverdie.

    Vashist[34]studiedtheeffectofdifferentsupportstructuresforthedieonthepartingplane

    separation.Thegoalwastostudythepartingplaneseparationpatternsonaproductiondie

  • 8/12/2019 Mechanical Performances of Dies

    17/284

    7

    thatflashedseverelyafteritwasmovedfroma1000tonmachinetoa2500tonmachine.

    Thediehadtobemountedloweronthe2500tonmachineduetothelocationoftheshot

    hole.Therefore,toevenlyspreadouttheclampload,thediefootprintwasincreasedonthe

    coveronthemachineplatenbyaddingsupportstructurestothedies.Thisstudyanalyzed

    theeffectofdifferenttypesofsupportstructuresanddifferentclamploads.Theresults

    showedthattheaddedsupportsstiffenedthedie/machinestructureandincreasedthe

    platencoveragearea,buttheydidnotaidintransferringtheclamploadtothediefaces.

    Flynnetal[35]measuredthedeflectionofthesameproductiondieatvariouslocationsof

    thedieusingLVDTsandtheyreportedagoodmatchbetweenthemodelpredictionsby

    Vashist[34]andtheexperimentalmeasurements.

    GarzaDelgado[36]studiedthefailureoftieboltsthatoccurredonahotchambermachine,

    usingasequentiallycoupledthermomechanicalmodel.Thiscasestudyshowedthatnon

    uniformheatgrowthonthepartingsurfaceofthedieresultedinunequaldistributionof

    loadsonthetieboltsandresultedintieboltsfailure.Themachineframe,shankandbracket

    weremodeledexplicitlyinthestructuralmodel.Thetogglemechanismandthetiebolts

    weremodeledusing3Dbeamelements.

    BaroneandCaulk[39]presentedamethodtopredicttheultimatedistortionofboththe

    castingandthedieduetothermalandmechanicalloadsinthediecastingprocess.They

    formulatedthediedistortionproblemasanonlinearthermoelasticcontactproblemsolved

    byiterativeboundaryelementmethod.Butthecastingdistortionwasanalyzedasan

    unconstrainedthermoelasticshrinkageusingfiniteelementmethod.Theirmodelincluded

    the

    dies,

    the

    ejector

    support,

    and

    cover

    and

    ejector

    platens.

    The

    tie

    bars

    and

    toggle

    mechanismswererepresentedbyspringelementsbehindtheejectorandcoverplaten

    respectivelywithappropriatestiffnessvalues.Suitabledisplacementconstraintswere

    providedonselectednodesonthebottomofthecoverplatentopreventrigidbodymotion

    oftheentirestructure.Thisboundaryconditionisanapproximationofanchoringthe

    machinetothebase.Theunevencontactonthepartingsurfacewashandledusing

    contact/gapelementswhoseformulationprovidesforloadtransferbetweenthedie

    componentsonlywhenthemutualsurfacetractioniscompressive.Frictionbetweenthe

    contactsurfaceswereignoredinthismodel.Thecavitypressurewasmodeledasa

    hydrostaticpressurewithamagnitudeequaltothatoftheintensificationpressure.The

    modelingapproachwastestedonafrontdrivetransmissioncasedieandtheresultswere

    presented.Theadvantageofthismethodasclaimedbytheauthorsisthatthecastingand

    diedistortioncanbeanalyzedsimultaneouslyandtheshrinkageallowanceforthediecavity

    canbeestimated.

  • 8/12/2019 Mechanical Performances of Dies

    18/284

    8

    Ragabetal[40]studiedtheeffectofcastingmaterialconstitutivemodelonthedeflection

    andresidualstresspredictionsonthecasting.Thecoverandejectorplatenswereincludedin

    themodelandthetiebarswererepresentedusingspringelements.Theclamploadwas

    appliedbyspecifyingdisplacementboundaryconditiononthespringelementsrepresenting

    thetogglemechanism.Thecavitypressurewasmodeledasapressureboundarycondition.

    Acontactconstraintwasusedbetweenthedieandthecasting.Afullycoupledthermo

    mechanicalanalysiswasconducted.Threedifferentmaterialmodelswereconsideredforthe

    casting,viz.,elastic,elasticplasticandelasticviscoplastic.Theresidualstresspredictions

    wereaffectedsignificantlybythematerialmodels. Thedistortionpredictionswereless

    affected.Theelasticmodeloverestimatedthestressesandtheviscoplasticmodellacked

    therequiredmaterialpropertydata.ThereforeinafurtherstudyRagab,[41]usedthe

    elasticplasticmodeltopredicttheeffectofkeymodelingfactorsoncastingdistortion

    predictions.Thefactorsconsideredweretheyieldstrengthandstrainhardeningofthe

    castingmaterial,theheattransfercoefficientbetweenthedieandthecastingandthe

    injectiontemperatureofthemetal.Thestudyconcludedthattheyieldstrengthhadamajor

    effectonresidualstressesatejectionwhiletheinjectiontemperatureandtheheattransfer

    coefficienthadamajoreffectontheresidualstressesatroomtemperature.The

    disadvantageofthemodelusedinRagabsstudywasthatthesolidcastingcouldnotfollow

    thedistortedshapeofthecavityduetotheuseofsolidelementsforthecastingsandhence

    itmightaffectthecastingdistortionprediction.

    GarzaDelgado[42]addressedthisissuebyusingashellmeshrepresentingthecasting

    surfaceinthesequentiallycoupledthermomechanicalmodelthatincludedtheclamploads,

    intensificationpressureloadandthethermalload.Thenodaldistortionvaluesoftheshell

    meshwerethenmappedontothesurfacenodesofasolidmeshforthecasting.Thenthe

    solidcavitywastiedtothedistortedshapeofthedieusingtiedcontactandthecooling

    stagesofthecastingweresimulatedusingafullycoupledelasticplasticthermomechanical

    model.Modelingthetiebarsexplicitlyandapplyingtheclamploadsthroughspring

    elementscausedproblemsinestablishingcontactbetweenthedieandthecasting.

    Thereforetheclampforcewasmodeledasapressureboundaryconditionbehindtheejector

    plateninthiswork.

    NumerousparametricdiedesignstudieshasbeenconductedattheCenterforDieCastingat

    OhioStateUniversitytounderstandtheroleofstructuraldiedesignparametersondie

    deflection.Someoftheseparametricdiedesignstudiesarereviewedhere.

    Jayaraman[43],conductedaparametricstudyoftheslidedesignvariablesforaninboard

    lockdesignandtheworkwascontinuedbyChakravarti[44,45]usingarefinedmodel.The

    variablesconsideredwerethepreload,theangleofthelockingsurfaceandthepivot.Results

  • 8/12/2019 Mechanical Performances of Dies

    19/284

    9

    suggestedthatthepreloadhadnoeffectontheblowbackandtiltvaluesbutitaffectedthe

    fatiguelifeoftheslides.Thetrendsalsoshowedthatthetipseparationincreasedwith

    lockingfaceangleandslideswithhighpivotwerebettersupportedbytheejectorplaten.

    Therewasnegligibleeffectonthepartingplaneseparationwithintherangeofdesign

    variables.

    TheeffectofproudinsertsonthepartingplaneseparationwasfirstanalyzedbyDedhia[26],

    [46].Thestudyconcludedthatusingproudinsertsresultedinalowerseparationduringthe

    initialcyclesbutatthelaterstagesasthediewarmsupandgrowstheseparationvalues

    weresimilartothecaseswithflushinserts(insertpartingsurfaceinlinewiththedieparting

    surface).AmuchrefinedmodelwaslaterusedbyTewari[3],[45]tostudytheeffectof

    proudinsertsonthepartingplaneseparation.Thecontactpressurebetweenthediesand

    theplatensandthecompressivestressesinthediepocketwerealsostudied.Itwas

    observedthatthepartingplaneseparationwasreducedaroundthecavitybyhavingaproud

    insertandthecontactpressurebetweenthediesandtheplatenswasnotaffecteddueto

    theuseofproudinserts.Theproudinserthadnegligibleeffectonthecompressivestresses

    inthediepocket.

    Tewari[3],[45],analyzedtheeffectofaddingabackplatebehindtheejectorsupportbox.

    Resultsfromhisstudyindicatethatthebackplatehasnegligibleeffectonboththeparting

    planeseparationandthecontactpressurebetweentheejectordieandtheplaten.

    AparametricstudyconductedbyDedhiaetal[26],[46],showedthatusingproudpillarshad

    noeffectonthepartingplaneseparation.Thedifferenceinpartingplaneseparationvalues

    betweenthecaseswithproudpillarsandflushpillarswaslessthan5%.

    Aseriesofparametricstudies[13],[28]wereconductedtogainunderstandingoftheeffect

    ofimportantstructuralvariablesofthediesandthemachineonthediedistortion.The

    summaryoftheworkwaspublishedin[4],[45].

    Thevariablesinvestigatedwerediesize(%ofplatenareacovered),insertthickness,

    thicknessofdiesteelbehindtheinsertanddielocationontheplaten.Responsesurface

    modelsbasedondesignofexperimentswereusedtostudytheinteractionbetweenthe

    variablesandtheireffectofpartingplaneseparation.Theapproachforthesensitivity

    analysiswasdevelopedbyChayapathi[1],[28]andtheinitialexperimentaldesignarray

    consistedof15experimentalruns. Anadditional16runswerefurtheraddedbyKulkarni

    andTewari[2],[3]toascertaintheresults.Thestudyhasshownthatthedominantfactor

    thataffectsthediedistortiononthecoversideisthecoverplatenthickness.Thindies

    performedbetterthanthickdies.Themorethesteelbehindthediesthelesserwasthe

    partingplaneseparationobserved.Smallormediumsizeddies(covering40%to50%of

  • 8/12/2019 Mechanical Performances of Dies

    20/284

    10

    platenarea)performedbetter.Kulkarni[2]attemptedtostudythecoverandejectorside

    performancesseparately.Buttheejectorsidedesignvariablessuchastherailsize,number,

    sizeandlocationofthesupportpillarswerenotcontrolledinthecomputational

    experiments.Thereforethestudywasinconclusiveaboutthecontributionofejectorside

    design

    to

    the

    ejector

    side

    separation.

    2.2 SummaryoftheStateoftheArtThestateoftheartinmodelingthediecastingdie/machinesystemisbrieflysummarizedin

    thissectiontoplacetheworkperformedinthisprojectincontext.

    Thevastmajorityofdiecastingmodelingworkperformedinindustryaddressesflow

    modelingofdiefillingandthermalmodelingofthepartanddieasthediecools. Invirtually

    everycasethemachineisignoredexceptasthesourceofthemetalspeedduringfillingand

    thedieisassumedtobecompletelyrigidexceptpossiblyforthermalgrowth. Thereis

    generallyno

    consideration

    of

    the

    mechanical

    forces

    and

    pressures

    at

    work

    in

    the

    system.

    Asdescribedintheprevioussection,theworkatOhioStateoverthepastseveralyearshas

    addressedanddevelopedstructuralmodelingtechniquesandboundaryconditionssuitable

    tomodelmanyaspectsofthesystemresponsetothemachineclampingforces,cavityand

    injectionpressure,aswellasthethermalloadsthatmostmodelingconsiders. Adepictionof

    themajorcomponentsofthesystemastypicallymodeledisshowninFigure1.

    Figure1 TypicalDieCastingSystemModelUsedbyOSU

  • 8/12/2019 Mechanical Performances of Dies

    21/284

    11

    Themachinebase,allthreeplatens(assumingathreeplatensystem),thecoverandejector

    diecomponents,andthetiebarsareexplicitlymodeled. Themachineclampingmechanism

    isrepresentedwithbeamelementswhoselengthisadjustedtoprovidetheproperclamp

    magnitude.

    Thepartanddiecavityaretypicallymodeledindependentlyandthesystemisanalyzedin

    quasisteadystatebymodeling50ormorethermalcyclesofthediesothatthetimevarying

    dietemperaturefieldisconsistentfromonecycletothenext. Thestructuralanalysisis

    staticandthecavitypressureisnotexplicitlymodeledbutaddressedwithahydrostatic

    boundarycondition. Theinabilitytoexplicitlymodelthepressureattheinterfacebetween

    thecastingandthedieisoneofthelimitingfactorsindevelopingrobustmodelsofpart

    distortion. Overall,themodelingapproachhasproventobequiterobustandusefulbutit

    doesrequireconsiderableunderstandingofthemechanicsofthesystemandofstructural

    modeling. Consequently,structuralmodelingofthesystemisnotyetwidelypracticedin

    industryasadesigntool. Itis,however,beginningtobeusedfortroubleshootingsystem

    failures.

    2.3 ObjectivesIngeneralterms,theobjectivesofthisprojectaretomaketheprinciplesandmethodsof

    systemstructuralmodelingmoreaccessibletotheindustrythroughthedevelopmentof

    detailedtutorialmaterialandthroughtheuseofmetamodels(modelsofmodels)that

    capturetheresultsofsimulationsinrelativelysimpledimensionlessequationsthatcanbe

    programmedonaspreadsheetorinasimplecomputerapplication. Inaddition,theissueof

    thepressure

    boundary

    condition

    mentioned

    above

    isexplored

    insome

    detail

    and

    areas

    of

    themodelingthatarelesswelldeveloped,specificallythedesignoftheejectorsideofthe

    die,areexplored.

    Severaloftheindustrialcasestudiessummarizedintheprevioussectionshowthateven

    veryexperiencedprocessengineersandtoolingdesignersoftenincorrectlypredicthowthe

    diewillrespondtochangestothestructuralelementsinthedieassembly. Industrydoesnot

    alwaysapplybasicengineeringprincipleswhenconsideringdiemechanicalperformance.As

    aconsequenceconsiderabletimeandmoneyisspenttryingoutdiedesignoptionsthat

    ultimatelydonotwork.Thisprocessresultsinverylongtryoutswithmanyscrapped

    castings.Theseproblemscanbeavoidedoratleastminimizedwithcorrespondingsavingsof

    wastedenergy.

    Specificgoalsincluded:

    Systematicallyaddressingdesignoftheejectorsideofthediecastingdie.

    Developparametricinformationabouttherelationshipbetweendieshoedesign,

    slidecarrierdesign,andthemachine.

  • 8/12/2019 Mechanical Performances of Dies

    22/284

    12

    Analyzingtherelationshipbetweenthediecenterofpressure,diegeometriccenter,

    andplatencenterofpressureinthepursuitofbettersetupguidelines.

    Examinethepossibilityofexplicitlymodelingthecasting/dieinterfacewiththe

    objectiveofeliminatingtheneedtouseapressureboundarycondition

    Providedesign

    guidelines

    for

    the

    industry

    addressing

    the

    mechanical

    design

    ofdies

    andtherelationshipbetweendieandmachine.

    2.4 TasksandApproachThetaskstomeetthesegoalsinclude:

    1. AnalysisofDiePositionEffects

    ApproachAcollectionofcomputationalexperimentsdesignedusingdesignof

    experimentsprincipleswasusedtodevelopFEAmodelsofdiesinvariousconfigurations

    onthediecastingmachine. Thenthroughtheapplicationofdimensionalanalysis,the

    dataproducedalongwithdatafromotherexperimentsperformedbythegroup,werethenusedtoconstructadimensionlesspowerlawmodeltopredictingtiebarbalanceas

    afunctionofdieconfigurationparameters. Thistaskiscloselyrelatedtotask4

    describedbelow.

    2. DieFailureCaseStudy

    ApproachAstudywasperformedinconjunctionwithanindustrypartnerthatanalyzed

    therootcauseofprematurecrackingofadieusedtoproduceadoorcloserhousing.

    Thecrackresultedinasurfaceblemishthatresultedincastingsbeingrejected. The

    primarypurposeofthistaskwastogainmoremodelingexperiencewithfailuresandto

    extendthemodelingprocedurestoconsidertheseissues.

    3. CavityPressureModelingMethods

    ApproachAcomprehensiveFEAmodelingstudysupportedwithresultsfromcasting

    experimentsandaseparatesetofexperimentstoobtainhightemperaturestressstrain

    propertieswasused.

    4. EvaluationofEjectorSideandSlideDesign

    Approach Acollectionofcomputationalexperimentsdesignedusingdesignof

    experimentsprincipleswasusedtodevelopFEAmodelsofvariousdieshoeandpillar

    configurations. Thenthroughtheapplicationofdimensionalanalysis,thewereusedto

    constructadimensionlesspowerlawmodeltopredictingmaximumpartingplane

    separationasafunctionofdesignparameters. Thistaskiscloselyrelatedtotask1

    describedabove.

    5. ModelingandDesignGuidelines

  • 8/12/2019 Mechanical Performances of Dies

    23/284

    13

    ApproachDrawingonexperienceswithmodelingperformedaspartofthisprojectand

    withthepreviousworksummarizedintheprevioussection,acomprehensivemodeling

    tutorialwasdeveloped. Inaddition,averyshortsummaryofkeyprincipleswas

    developedasanintroductiontothearea.

  • 8/12/2019 Mechanical Performances of Dies

    24/284

    14

    3. ResultsandDiscussion3.1 AnalysisofDiePositionEffects

    Abriefdescriptionofthemodelingandanalysisperformedforthistaskispresentedbelow.

    AsummaryreportthatdescribesthediepositionanalysisworkispresentedinAppendixA.

    ThisreportwasprovidedtoNADCAfordistributiontotheindustryandalsocontainsthe

    summaryfortheejectorsidedesignwork.

    Asimplifiedrepresentationofadiecastingmachineanddieusedformodelingwasshownin

    Figure1previously.

    Thedieisclampedandheldclosedbythecombinationofatogglemechanism(represented

    asbeamelementsinthefigure)andtiebarsthatareanchoredtothecoverandrearplaten.

    Theejectorplatenmoveswiththetoggletoopenthedie. Thetiebarsactmuchlikeboltsor

    aclampinprovidingtherestrainingforceneededtokeepthedieclosed. Thedegreeto

    whichthedieisheldclosedandpressureonthepartingsurfacesofthediearerelatively

    uniformdependsonthethermalgrowthofthedie,thecavityshapeandtheplacementof

    thepositionofthedieontheplatens.

    AsimplifiedfreebodydiagramofthemechanicalloadsinthesystemisshowninFigure2. In

    acompletelybalancedcase,eachtiebarcarries1/4thoftheclampforceandprovidesthe

    restrainingforcetokeepthedieclosed. Fromtheclampforceappliedatthebackofthe

    ejectorplaten,theloadpathisthroughtheejectorplaten,throughtheejectordie,across

    thediepartingsurface,throughthecoverdietothecoverplatenandtothetiebars. Again

    inaperfectlybalancedsituation,asthecavitypressureisapplied,theforcesacrossthepartingsurfacedropwhiletheforcesbetweenthediesandplatensandtheforcesinthetie

    barsremainunchanged.

    Figure3illustratesthecharacteristicsoftheresponseofthesystemtotheclampand

    pressureloads(theblacklinesdenotethesystempriortotheapplicationofloads,thegreen

    linesareamagnifieddepictionofthesystemafterloading). Notethattheplatens(anddie)

    distortandthetiebarsstretchresultinginapseudorigidbodymotion. Theactualdistortion

    patterninaparticularcasedependsonthemagnitudeoftheloadsandthegeometryofthe

    systemandthedistortionpatterncanchangesignificantlyifthediepositionontheplaten

    changes.

    Inunbalancedsituationswherethedieand/orthediecavityareoffcenter,thedistortionof

    thesystemismoresevereandthetiebarloadingwillnotbebalancedwithsomebars

    carryingmorethanthenominalandsomecarryingless.

  • 8/12/2019 Mechanical Performances of Dies

    25/284

    15

    Figure2 FreeBodyDiagramoftheMechanicalLoads

    Figure3 DepictionofSystemResponse

    Ifthefourtiebarsdonotcarryequalloads,thediescloseunevenlyatthediepartingsurface

    anddieflashingmayoccur.Inextremecases,poorlybalancedtiebarloadscouldalsoleadto

    tiebarfailure.Thecommonpracticetoovercomethetiebarloadimbalanceproblemisto

    adjustthelengthofthetiebarsbetweentheplatenssothatallthefourtiebarscarryequal

  • 8/12/2019 Mechanical Performances of Dies

    26/284

    16

    loads.Insuchacasetheminimumclamploadrequiredtoholdthediestogetherwillbe

    higherthantheonethatwouldbeneededifthedieswerecenteredontheplatenthus

    limitingthecapacityofthemachine.

    Thecurrentapproachusedinindustrytopredictthetiebarloadsbalancesthemomentsdue

    tothecavitypressurebutignoresthelocationofthediewithrespecttotheplatencenter

    anditassumesthatthemachineandthediesareperfectlyrigidandconsequentlycanbe

    quiteinaccurate.

    Toaddresstheseissues,anonlinearpowerlawmodelwasdevelopedtopredictthetiebar

    loadsofthediecastingmachinebasedonthelocationofthedieandcavitycenterof

    pressurewithrespecttothetiebars.Themodelwasobtainedbycurvefittotiebarload

    predictiondatafromcomputationalexperiments.Thecomputationalexperimentswere

    conductedusingthefiniteelementmodeling.Anexperimentaldesignwasdevelopedbased

    onthe

    horizontal

    and

    vertical

    dimension

    ofthe

    die,

    the

    locations

    ofthe

    die

    and

    cavity

    center

    ofpressurewithrespecttotheplatencenterandthemagnitudeofcavitycenterofpressure.

    Dimensionalanalysiswasusedtoincorporateotherimportantscalefactorsandobtainthe

    nondimensionalparameters.Thenonlinearmodelwasthenfittothenondimensional

    formofthelocation,scaleandloadvariables.Experimentaltiebarloadmeasurementswere

    thencomparedtothepowerlawmodelpredictionstochecktheadequacyofthepowerlaw

    models.

    3.1.1 ComputationalExperimentsThefactorsthatwereconsideredarethedielength,diewidth,locationofthediewith

    respecttotheplatencenterandthelocationofthecavitycenterofpressurewithrespectto

    theplatencenterandthemagnitudeofthecavitypressure.Thedescriptionofthevariables

    andtheirlevelsareshowninTable1.Thelocationvariablesaredefinedwithrespecttoa

    coordinatesystemwithoriginonthecenteroftheplatenareabetweenthetiebars.The

    schematicofthecoordinatesystemandthetiebarlabelsareshowninFigure4.

  • 8/12/2019 Mechanical Performances of Dies

    27/284

    17

    Factor FactorDescription Level1

    Level

    2Level

    3Level

    4Level

    5LX DieWidth(inches) 26.49 30.4 32.6 34.6 36.5

    LY DieHeight(inches) 26.49 30.4 32.6 34.6 36.5

    DPX Dielocation

    inX

    direction(inches)4 2 0 2 4

    DPYDielocationinY

    direction(inches)4 2 0 2 4

    CPX

    Locationofcenterof

    pressurein X

    direction(inches)

    4 2 0 2 4

    CPY

    Locationofcenterof

    pressurein Ydirection

    (inches)

    4 2 0 2 4

    CPR CavityPressure(KSI) 2 4 6 8 1

    Table1 DescriptionofVariables

    Figure4 CoordinateSystemandTiebarLabels

    Afewadditionalcaseswerealsoaddedtotheexperimentalarraytoincludecaseswitha

    cavityloadequalto100%ofclamploadandafewcaseswithnocavitypressureload.

    3.1.2 DimensionalAnalysisDimensionalanalysisandananalysisofthephysicalphenomenainvolvedleadtoapower

    lawformshownin(1)

  • 8/12/2019 Mechanical Performances of Dies

    28/284

    18

    c1 c2

    tb tb

    tb tb

    Tie bar load DPX DPY=c0 1 1+

    Nominal Load 0.5L 0.5L

    CPRA CPX CPY CPRA 1+c3 exp 1 +c4 1+

    CLAMP 0.5L 0.5L CLAMP

    CPRA CPRA 1-c5 exp

    CLAMP

    CLAMP

    (1)

    Thisformwasassumedforeachindividualtiebarandparametersselectedtofitthepower

    lawtothepredictedloadsforeachofthefourtiebarsforallofthe70casesinthe

    experimentalarray. TheSPSSstatisticalsoftwarepackagewasusedtoperformthenon

    linearregressionandthesequentialquadraticprogrammingalgorithminSPSSwasusedto

    solvethenonlinearregressionproblem.

    Thesamemodelformwasobtainedforallthefourtiebarsandthemagnitudesofthe

    coefficientsandexponentsforthetwotoptiebarswereapproximatelyequalwithdifferent

    signs.Similarlythemagnitudesoftheexponentsandcoefficientsforthetwobottomtiebars

    werenearlythesame.Thereforetheloaddataforthetoptiebarswerepooledandthedata

    forthebottomtiebarswerepooledandthesamemodelformwasfittothepooleddata.

    Theresultsaregivenbyequations(2)and(3)respectively.Thetermswith+/ signsare

    reversedbetweenthetwoequations.Theseparametersrepresenttheverticallocationof

    thedieandcenterofpressurerespectivelyandhencetheirsignsarepositiveforthetoptie

    barsandnegativeforthebottomtiebars.

    Theparameterestimates,thestandarderrorintheestimatesandtheconfidenceintervals

    fortheestimatesforthetopandbottomtiebarsareprovidedinTable2andTable3

    respectively. Thequalityofthefitisclearlyverygood.

    Theresultingequationsareasfollows:

    0.354 0.303

    tb t b

    tb t b

    Top Tie bar load DPX DPY=1.005 1 1+

    Nominal Load 0.5L 0.5L

    CPRA CPX CPY CPRA 1+0.063 exp 1 +0.886 1+

    CLAMP 0.5L 0.5L CLAMP

    1-0.

    CPRA CPRA098 exp

    CLAMP CLAMP

    (2)

  • 8/12/2019 Mechanical Performances of Dies

    29/284

    19

    0.294 0.256

    tb t b

    t b tb

    Bottom Tie bar load DPX DPY=1.04 1 1-

    Nominal Load 0.5L 0.5L

    CPRA CPX CPY CPRA 1+0.062 exp 1 +1.03 1-

    CLAMP 0.5L 0.5L CLAMP

    1-0

    CPRA CPRA.106 exp

    CLAMP CLAMP

    (3)

    ParameterEstimates

    Parameter Estimate

    Std.

    Error

    95%Confidence

    Interval

    Lower

    Bound

    Upper

    Bound

    c0 1.005 0.002 1.001 1.009

    c1 0.354 0.011 0.333 0.374

    c2 0.303 0.012 0.279 0.327

    c3 0.063 0.005 0.052 0.074

    c4 0.886 0.045 0.796 0.976

    c5 0.098 0.006 0.109 0.086

    AdjustedRsquare=0.96

    Table2:ParameterEstimatesforTopTieBarModelFit

  • 8/12/2019 Mechanical Performances of Dies

    30/284

    20

    ParameterEstimates

    Parameter Estimate

    Std.

    Error

    95%Confidence

    Interval

    Lower

    Bound

    Upper

    Bound

    c0 1.04 0.002 1.036 1.044

    c1 0.294 0.012 0.271 0.318

    c2 0.256 0.014 0.228 0.284

    c3 0.062 0.005 0.051 0.073

    c4 1.025 0.052 0.922 1.128

    c5 0.106 0.005 0.117 0.095

    AdjustedRSquare=0.95

    Table3:ParameterEstimatesforBottomTieBarModelFit

    Thetermnominalloadonthelefthandsidesoftheequationsisdefinedasonefourthofthe

    totalclampload.ThesignbeforethetermsinvolvingDPXandCPXindicatesthatthesigns

    ofthesevariablesdependonthelocationofthedie/cavityandalsoonthetiebarforwhich

    theloadiscalculated.Ifthedieand/orcavityispositionedtowardsthetiebarforwhichthe

    loadistobepredicted,apositivesignshouldbechosenforthecorrespondingvariablesand

    ifthedieand/orcavityispositionedawayfromthetiebarforwhichtheloadistobepredictedanegativesignshouldbechosen.Forexample,ifthedieishorizontallyoffcenter

    towardsthehelperside,apositivesignshouldbeusedbeforethevariableDPXtopredict

    theloadsonthehelpersidetiebarsandanegativesignshouldbeusedbeforeDPXto

    predicttheloadsontheoperatorsidetiebarsandviceversa.Thesamesignconvention

    appliestothetermsinvolvingthehorizontalcavitylocationCPX.

    Thethirdtermintheequationrepresentsthemomentscausedbythecavitypressureload.

    Themomenttermsandtheloadtermappearasexponentialtermsinthemodel.This

    indicatesthatthetiebarloadsincreaseordecreaseinanexponentialfashionasthecavity

    centerofpressureismovedtowardsorawayfromtherespectivetiebars. Theparameters

    involvingthediedimensions,Lx/AandLy/Awerefoundtohaveanegligibleeffectonthe

    modelpredictionsandhencetheywereignoredinthemodel.Thesenondimensional

    parameterswereincludedintheinitialmodelfitting.Buttheerrorinvolvedintheir

    estimatesforthecorrespondingexponentsweremuchhigherthanthevalueofthe

    exponentsitself.Thereforethismodelbehavesasalumpedmodelwheretheclampand

  • 8/12/2019 Mechanical Performances of Dies

    31/284

    21

    pressureloadsareapproximatedbypointloadsactingonthediecenterandcavitycenterof

    pressurerespectively.

    3.1.3 ModelAdequacy,FEAResultsThepowerlawmodelsshowninequations(2)and(3)wereobtainedbycurvefittingtotie

    datafroma800tonfourtogglemachine. Tostudytheadequacyofthemodeltopredictthe

    tiebarloadsonmachinesofotherdesignsandclampingcapacity,thepowerlawmodel

    predictionswerecomparedagainstthefiniteelementmodelpredictionsoftiebarloadon

    machinesofotherdesignsandtonnages.Threedifferentmachinefiniteelementmodels

    wereconsidered,viz,a3500tonfourtogglemachine,1000tonfourtogglemachineanda

    250tontwotogglemachines.Thedielocation,cavitylocation,theclamploadand

    magnitudeofcavitypressureforthesethreecasesaresummarizedinTable4. TheFEAand

    powerlawpredictions,presentedasthedifferencefromnominal,forthesamefourcases

    areshowninTable5,Table6,andTable7respectively.

    Machine

    Design

    DPX

    (in)

    DPY

    (in)

    CPX

    (in)

    CPY

    (in)

    CPR

    (PSI)

    Ltb

    (in)

    Cavity

    Load

    (tons)

    Clamp

    Load

    (tons)

    3500ton4

    toggle4 0 4 0 0 84.25 0 3500

    1000ton

    4toggle0 1.25 0 3.63 10000 44 602 722

    250ton2

    toggle

    0 3.14 0 0.423 10000 21.75 135 250

    Table4:SummaryofFiniteElementModels

    TieBarLoad/NominalLoadPrediction

    FEA PowerLaw

    TopTieBar1 3.8% 3.0%

    TopTieBar2 2.8% 3.8%

    BottomTieBar1 2.9% 2.6%

    BottomTieBar2 4.0% 3.1%

    Table5:ComparisonofModelPredictionsfora3500TonMachine

    (DPX=4,DPY=0,CPX=4,CPY=0,CPR=0PSI)

  • 8/12/2019 Mechanical Performances of Dies

    32/284

    22

    TieBarLoad/NominalLoadPrediction

    FEA PowerLaw

    TopTieBar1 4.1% 6.7%

    TopTieBar2 4.1% 6.7%

    BottomTieBar1 4.1% 2.1%

    BottomTieBar2 4.1% 2.1%

    Table6:ComparisonofModelPredictionsfora1000TonMachine

    (DPX=0,DPY=1.25,CPX=0,CPY=3.63,CPR=10000PSI)

    TieBarLoad/NominalLoadPrediction

    FEA PowerLaw

    TopTieBar1 8.2% 8.3%

    TopTieBar2 8.2% 8.3%

    BottomTieBar1 8.2% 14%

    BottomTieBar2 8.2% 14%

    Table7:ComparisonofModelPredictionsfora250TonMachine

    (DPX=0,DPY=3.14,CPX=0,CPY=0.423,CPR=10000PSI)

    The1000and250tonmachineshaveverydifferentdesignscomparedtothe800and3500

    tonmachinesandtheFEAanalysesinthesecasesusedsimplifiedboundaryconditionson

    thecoverplaten. Thedifferenceinboundaryconditions,andnotthedifferenceindesign,is

    largelyresponsiblefortheconstantmagnitudeFEAresultsinthetwocasesinquestion.

    Theresultsshowreasonablygoodpredictivecapabilityacrossavarietyofdesignsas

    expectedofadimensionlessmodel.

    3.1.4ModelAdequacy,ExperimentalMeasurementsExperimentswereconductedona250tontwotogglemachinebyvaryingthedielocation

    andobtainingthetiebarloadsunderclamploadonly.Theschematicofthetestdieonthe

    machineplatensisshowninFigure5. Thetestdiemeasures13.38inchesby18inchesand

    thedistancebetweenthetiebarcentersis21.75inches.Fouruniaxialstraingaugeswere

    attachedtoeachtiebartomeasurethelongitudinalstrains.Thestraingaugeswere

    attachedtothetiebarsatadistanceof267mm(10.5in)fromtheinsidefaceofthe

    stationaryplatensothatthestraingaugesarehalfwaybetweenthestationaryandmovable

    platenswhenthetestdieisclosed.Theschematicofthestraingaugelocationsisshownin

  • 8/12/2019 Mechanical Performances of Dies

    33/284

    23

    Figure6. Thefourstraingaugesoneachtiebarare90apart.Thirteendifferentdiesetups

    werestudied,onewithadiecenteredontheplaten,fourcaseswithverticallyoffcenter

    dies,fourcaseswithhorizontallyoffcenterdiesandfourcaseswithdiagonallyoffcenter

    dies.ThethirteencasesaresummarizedinTable8.Notallcombinationsofdiagonallyoff

    centerdiescouldbestudiedduetothelimitationofthespaceavailablebetweenthetie

    bars.

    Figure5 Schematicofthetestdieonthemachineplatens

    Figure6 Schematicofstraingaugesandcoordinatesystem

  • 8/12/2019 Mechanical Performances of Dies

    34/284

    24

    RunDPX

    (inches)DPY(inches)

    1

    0 02 0 2

    3 0 4

    4 2 0

    5 4 0

    6 0 2

    7 0 4

    8 2 0

    9 4 0

    10 2 1.87511 4 1.875

    12 2 1.875

    13 4 1.875

    Table8:ExperimentalArray

    Eachcasewasrepeatedthreetimesandaconstantclamploadof2500KNwasappliedinall

    cases.Thoughthemachinewasprogrammedtoapplyaclampingloadof2500KN,theactual

    clamploadappliedbythemachinevariesslightlyfromthenominal.Themetalinjection

    stagewasignoredintheseexperimentsduetothepracticaldifficultyofmovingtheshot

    sleeveforeachtest.Thestrainsoneachtiebarwereobtainedunderclamploadonly.The

    averageofthestrainsmeasuredbythefourstraingaugesoneachtiebarwascalculatedto

    obtainthestrainoneachtiebar. Thetiebarloadswerethencalculatedfromthestrain

    values.Thenominalclamploadpertiebarwasassumedtobetheaverageofthefourtiebar

    loadsandtheloadoneachtiebarwasestimatedastheratiobetweenthetiebarloadand

    thenominalload.

    Theexperimentalmeasurementsandpowerlawpredictionsoftiebarloadsforthethirteen

    casesareshowninTable9andFigure7. Ideallytheloadsonallfourtiebarsshouldbeequal

    forcase1wherethediesarecenteredontheplaten.Howeverthemeasurementsshowthattheloadsonbottomtiebarsarelowerthanonthetoptiebars.Thiscouldbedueto

    inaccuracyinpositioningthediesontheplatencausingthemeasurementstobebiased

    towardsthetoptiebars. Lackofsquarenessandperfectflatnessoftheplatenscouldalso

    contributetothisobserveddifference.

  • 8/12/2019 Mechanical Performances of Dies

    35/284

    25

    Case DPX DPYExperimental

    Measurements PowerLawPredictionsT1 T2 B1 B2 T1 T2 B1 B2

    1 0 0 1.03 1.02 0.99 0.97 1.00 1.01 1.04 1.04

    2

    0

    2

    1.10 1.09 0.92 0.90 1.05 1.06

    0.99

    0.993 0 4 1.16 1.15 0.85 0.83 1.10 1.11 0.93 0.93

    4 2 0 0.99 1.09 0.90 1.02 0.94 1.07 0.98 1.09

    5 4 0 0.94 1.14 0.85 1.07 0.86 1.12 0.91 1.13

    6 0 2 0.96 0.95 1.05 1.04 0.94 0.95 1.08 1.08

    7 0 4 0.92 0.90 1.10 1.08 0.87 0.88 1.12 1.12

    8 2 0 1.09 0.98 1.03 0.91 1.06 0.94 1.09 0.98

    9 4 0 1.16 0.92 1.09 0.83 1.11 0.86 1.13 0.92

    10 2 1.875 1.06 1.15 0.84 0.95 0.98 1.12 0.93 1.04

    11 4 1.875 1.03 1.22 0.76 1.00 0.90 1.17 0.87 1.0812 2 1.875 1.02 0.90 1.11 0.97 1.00 0.89 1.13 1.02

    13 4 1.875 1.08 0.84 1.17 0.91 1.05 0.82 1.18 0.95

    Table9:ComparisonofMeasurementsandPredictions

    Thedatashowthatthemodelpredictionsareconsistentlyslightlylowerthanthe

    experimentalmeasurementsforthetoptiebarsandtheyareconsistentlyslightlyhigher

    thanthemeasurementsforthebottomtiebarsinallofthe13cases. Thiscanbeattributed

    totheconstrainttypeusedbetweenthecoverplatenandthemachinebasefortheFEA

    modelusedtoconstructthemetamodel.Theedgenodesofthecoverplatenandthebase

    weretiedusingamultipointconstraintinthecomputational(FEA)experimentsthatcreated

    thedataforthemetamodel.Thoughthe250tonmachineusedfortheexperimentshasa

    weldedjointthemultipointconstraintusedintheFEAmightbestifferthantheactual

    weldedjointonthemachine.Thelackofflatnessandsquarenessofthediescouldalsohave

    contributedtosomeofthesedifferences.

  • 8/12/2019 Mechanical Performances of Dies

    36/284

    26

    Figure7 TiebarLoadMeasurementsvs.Predictions

    ThedifferencesbetweenthemeasurementsandmodelpredictionsareshowninTable10.

    Thedifferencesbetweenthemodelpredictionsandtheloadmeasurementsvaryfrom0.1%

    to12%dependingonthedielocation.Asexpected,theworstcasesarethediagonallyoff

    centercases,case10andcase11,wherethemeasurementsshowthattheloadonthetop

    tiebar1(T1)ishigherthanthenominalandthemodelpredictionsshowthattheloadson

    toptiebar1(T1)islowerthanthenominal.Similarlytheloadmeasurementsonbottomtie

    bar1(B2)arelowerthanthenominalincase10andcase11,andthemodelpredictions

    showthattheyarehigherthanthenominal.Overall,thepatternofresultsisverygood.

    Noteagainthatthepowerlawwasderivedfromsimulationdatabasedonan800ton,4

    togglemachineandtheexperimentaldataisfroma250ton,2togglemachineadding

    credibilitytotheclaimthatthepowerlawproducesreasonablepredictionsforawide

    varietyofmachinedesignsandsizes.

  • 8/12/2019 Mechanical Performances of Dies

    37/284

    27

    Case DPX DPYDifferencebetweenMeasurementsandModel

    Predictions(%)T1 T2 B1 B2

    1 0 0 2.54% 1.26% 4.83% 7.40%

    2 0 2 4.50% 2.84% 6.77% 9.17%

    3 0 4 6.52% 4.59% 7.20% 9.76%

    4 2 0 5.40% 2.03% 7.72% 6.95%

    5 4 0 7.63% 2.24% 6.75% 6.03%

    6 0 2 1.98% 0.10% 2.75% 4.90%

    7 0 4 4.28% 2.13% 1.80% 4.32%

    8 2 0 2.80% 3.37% 5.10% 7.81%

    9 4 0 5.19% 5.90% 3.89% 9.10%

    10 2 1.875 8.15% 3.25% 9.65% 9.23%

    11 4 1.875 12.33% 4.66% 11.54% 8.60%

    12 2 1.875 2.49% 1.02% 2.25% 5.41%

    13 4 1.875 3.39% 2.29% 0.91% 4.14%

    Table10:DifferencebetweenMeasurementsandModelPredictions

    3.2 DieFailureCaseStudyThedieusedtoproducethehousingofadoorclosingmechanismsufferedprematurecracksontheejectordiecavitysurface.Thesecracksledtovisiblemarksonthecastingsurface

    makingthecastingunacceptable. Theonsetofthecracksintheejectordieoccurredat

    approximately20,00040,000shots,muchbelowthedesignexpectationleadingto

    unexpectedcostelevation. Severaltechniquesweretriedbythediecastertofixthe

    problem,butwithnosuccess. Sincethecauseoftheproblemwasthoughttobeflexingof

    thedie,theCenterforDieCastingatOhioStateUniversitywasaskedtoanalyzethedie

    usingdiedistortionmodelingtechniquesdevelopedoverthepastseveralyears. The

    problemwasexpectedtoprovideagoodcasestudyfortestingtheanalysisprocedures.

    Figure8andFigure9illustratethecastingandthesurfacedefect. Severalmodifications

    weremadetothecastingandtheejectorinserttoavoidthecrackformationwithlittle

    success.Themodificationsincludedchangesinthecastinggeometryandaddingcooling

    linestotheejectorinsert.Thelastmodificationwastosplittheejectorinsertintotwopieces

    tominimizethebendingstrains.

  • 8/12/2019 Mechanical Performances of Dies

    38/284

    28

    Figure8 TheCasting

    Figure9 Defects

    ThisproblemwasprovidedtoOSUbyDCDTechnologiesandBlueRidgePressureCastings.

    Thegoalwastosimulatethedieandthecastingprocessusingthetechniquesdevelopedin

    theCenterforDieCastingattheOhioStateUniversityinordertoinvestigatethereasonfor

    thecracksandtoproposeappropriatediemodificationstominimizeit.

    Asequentiallycoupledthermomechanicalanalysiswasconductedandthemodelpredicted

    strainswere

    used

    to

    calculate

    the

    fatigue

    cycle

    life

    in

    the

    ejector

    insert

    cavity

    surface.

    The

    modelpredictedstrainscancausefatiguefailurewithin13,00023,000cyclesattheactual

    cracklocationssomewhatearlierthanhasbeenobserved.Thismaybeduetothelackoftwo

    featuresinthestructuralmodel,namely,thecastingandtheslides.Theinteractionbetween

    theinsertsandboththecastingandslidescanchangethestrainpatternanddecrease,to

    someextent,thetensilestrainsresultingfromthethermalloads.Includingthecastingand

  • 8/12/2019 Mechanical Performances of Dies

    39/284

    29

    slidesinthestructuralmodelwasnotpossibleduetotheneedtokeepcomputational

    resourcesmanageable.

    Theresultsshowthatthehighstrainsarelocalizedtoafewlocationssuggestingthat

    modifyingthediestructuredoesnotsolvetheproblem.Insteadchangingthepartgeometry

    toreducethestressconcentratoratthecracklocationscouldmakeadifference.The

    analysiswasrerunagainwithmodifiedfilletradiiatcrackslocationsandtheresultsshowed

    muchbetterresultsattwoofthefourlocations.

    Insummary,thecasestudyturnedouttobeausefultestofthemodelingprocedures,but

    sincetheproblemwasdeterminedtobethepartdesignandnotthedieormachine,the

    opportunitytorepresentdiechangesandperformbeforeandaftertestsdidnot

    materialize.

    AreportprovidingmoredetailaboutthisworkisfoundinAppendixB.

    3.3 CavityPressureModelingMethodsAsmentionedintheintroduction,ashortcominginthediestructuralmodelingprocedures

    currentlyusedisthepressureboundaryconditionusedtorepresentthemechanical

    interfacebetweenthecastingandthedie. Thistendsnottobeaproblemiftheissuesare

    structuraldesignofthediebutitmaybeanissueifaccuratepartdistortionmodelsare

    sought.

    Thedistorteddiecavityattheendoffillingrepresentstheinitialshapeconditionsforthe

    castingattheonsetofsolidification. Thisdistorteddiecavityshaperesultsfromtheeffects

    ofmainlythreeprocessloads:clamping,temperaturegrowthandintensificationpressure.

    Accuratepredictionsofcastingfinaldimensionsrequiremodelingthedistorteddiecavity

    shapeadequately,sincetheelasticdeflectionsexperiencedproduceddimensionalchanges

    inthediecavitythataffectthecastingdimensions.

    Indiedistortionmodelstheintensificationpressureeffectshavebeentraditionally

    representedbyusingaconstantpressureboundaryconditionappliedtothediecavity

    surface. Ideally,theintensificationpressureshouldbetheresultoftheloadingactionofthe

    pressurizedcastingactingontothediecavitysurfaces. However,thishydrostaticloading

    cannotbemodeledbecausecontinuumsolidelementslackahydrostaticpressuredegreeoffreedomandarerenderedinadequateforthesepurposes.

    Thelatestdevelopmentsinfiniteelementmodelingalgorithmsallowmodelingthe

    interactionoffluidandstructuralelements. Thesemodelingcapabilitiesrepresentthestate

    oftheartinfiniteelementcodesandwereinitiallyadoptedformodelingcastinganddie

  • 8/12/2019 Mechanical Performances of Dies

    40/284

    30

    distortion. ThefiniteelementpackageADINAwasselectedbecauseitallowsmultiphysics

    modelinginonesingleintegratedcode.

    Theadoptionofthismodelingtechniquewasthoughttoaugmentcastinganddiedistortion

    modelingeffortsduetothefollowing. First,sincethecastingcouldberepresentedusing

    liquidelementsthatpossessapressuredegreeoffreedom,modelingintensification

    pressureeffectscouldbereadilydonebyapplyingapressureloadtothebiscuitandletting

    thecastingfluidelementsloadthedeformabledie. Second,thedistorteddiecavityshapeat

    theendoffillingcouldbereadilyobtainedfromthedistortedcastingmeshbecausethe

    fluidstructureinteractionalgorithmrequirestheliquidelementstoalwaysfollowthe

    distortedshape. Therefore,theinitialcastingshapeattheonsetofsolidificationcouldbe

    readilyobtainedfromthedistortedfluidmesh.

    3.3.1 FluidStructureInteraction(FSI)ModelThe

    model

    was

    divided

    into

    solid

    and

    liquid

    domains.

    The

    solid

    domain

    was

    comprised

    of

    thestructuralelementswhichincludedtheinserts,die,ejectorsupportblock,ejectorand

    coverplatensandtiebars. Theliquiddomainwascomprisedofonlythecasting. Inthe

    structuralelementstheusualboundaryconditionsappliedtodiedistortionmodelswere

    considered. Contactbetweenallthedeformablebodieswasincorporatedaswell. Inthe

    fluiddomain,thecastingwasrepresentedwithliquidelements,whichhavepressureand

    velocitydegreesoffreedom. Afluidstructureinteractionboundaryconditionwasspecified

    forallthesurfacesinboththecastingandthediewheretheywereexpectedtointeract.

    Clampingandthermalforcewassimulatedbyapplyingapressureloadontheejectorplaten.

    Thermal

    load

    was

    modeled

    by

    prescribing

    a

    temperature

    load

    on

    the

    inserts

    and

    die.

    Intensificationpressurewasmodeledbynormalsurfacetractionloadonthebiscuitregionof

    thecasting. Theloadswereappliedsequentiallyinatotalofthreesteps.

    Thedisplacementsonthecastingmeshwereextractedattheendoftheanalysis. Figure10

    showsthepredictionsgivenbytheFSImodel. Theseresultsshowthatthecavitydistortion

    isnotnegligibleandthepredicteddisplacementmagnitudesemphasizetheimportanceof

    incorporatingthecontributionsofthediedeflectionsincastingdistortionanalyses.

    Thisdisplacementfieldrepresentedthedistortedcavityshapeattheendoffillingandwas

    consideredastheinitialcastingshapeforsubsequentthermalmechanicalsolidification/coolingmodeling. Thesamemeshforallthedifferentcomponentsusedinthe

    FSImodelwasusedinafullycoupledthermalmechanicalmodel. Themodelwascomprised

    ofthesamestructuralcomponentsasbeforewiththeadditionofthecasting. Thefinal

    coordinatesofthecastingintheFSImodelweretakenastheinitialcoordinatesforthe

    thermalmechanicalmodel.

  • 8/12/2019 Mechanical Performances of Dies

    41/284

    31

    Figure10 FSIcavitydisplacementpredictions

    Sincemodelingcastingsolidification/coolingmustconsidertheinteractionbetweenthe

    castingandthedie,thedeformedstatepredictedintheFSImodelhadtobereproduced

    usingthesameloadingconditions. Clampingandthermalloadsweremodeledasdescribed

    andtheintensificationpressurewasnowmodeledasapressureloadappliedtothedie

    cavitysurfaces. Contactbetweenthecastingandthediewasnotenableduntilthe

    deformationwasreproduced. Allthefiniteelementsusedtorepresentallthecomponents

    hadtemperatureanddisplacementsdegreesoffreedom. ThemodelwasruninAbaqus,

    usingtheC3D8Tcoupledtemperaturedisplacementthreedimensionalelements.

    Whilerunningthismodelconvergencedifficultieswereexperienced. Examinationofthe

    resultsshowedamismatchbetweenthecastingshapeanddiecavitysurfacesafter

    applicationoftheinitialloading. Thesemirigidbodymotionduetotiebarstretchingafter

    clampingwasidentifiedasthesourceofdivergenceinthethermalmechanicalmodel. The

    displacementpredictionsafterclampingbetweentheFSIandthethermalmechanicalmodel

  • 8/12/2019 Mechanical Performances of Dies

    42/284

    32

    differedbyasmuchas0.07mmandthisdifferencewaslargeenoughtopreventproper

    establishmentofcontact.

    Itwasconjecturedthatthedifferentcontactalgorithmsusedbyeachcodeproduced

    differentdisplacements. Sincecoupledthermalmechanicalmodelscanbeanalyzedin

    ADINAitwasdecidedtosetupthesolidificationmodelandrunitallinADINA. Thesame

    loadingandboundaryconditionsasdescribedbeforewerereproducedinanADINAthermal

    mechanicalmodel. Theresultsshowedthatthedisplacementpredictionsofthethermal

    mechanicalmodelweredifferentwhencomparedwiththeFSIpredictions. Themagnitude

    ofthesedifferenceswascloseto0.07mmaswell. Thesedifferencesindisplacement

    predictionswithinADINApreventedfurthermodelingeffortsusingthecodeanditwas

    abandoned.

    3.3.2 AlternativetoFSIThe

    modeling

    ofpart

    distortion

    was

    divided

    into

    three

    different

    models.

    The

    first

    model

    determinestheinitialcastingshapeafterthemetalhasfilledthediecavity. Thesecond

    modelsimulatesthecoolingofthecastinginsidethedie. Thethirdmodelsimulatesthe

    coolingofthecastingafterejection. Themodelsareruninthedescribedsequenceandthe

    resultsprovidedbyeachmodelaresubsequentlyusedbythefollowingone.

    Figure11showsthecastingusedforthisresearchproject. Thegeometryofthecastingwas

    designedforaprocesscontrolstudydonebyOsborne[42]aspartofhisresearchwork. All

    ofthecastingribsareformedintheejectorhalf,withthecovercontributingonlytoform

    thebackoftheplate. Thedesignofthiscastingwasdoneinordertousethedistance

    betweentheribsasameasureforincavitydistortion,whereasthedepthoftheribswas

    usedtoprovideameasureforacrosspartingplanedistortion.

  • 8/12/2019 Mechanical Performances of Dies

    43/284

    33

    Figure11Castingfiniteelementmesh

    3.3.3 TrackingofCavityDistortion

    Thepredictionsobtainedbythediedistortionmodelrepresentthefirststeptowardsthe

    modelingof

    casting

    distortion.

    As

    has

    been

    already

    described,

    the

    die

    experiences

    elastic

    deflectionsresultingfromthecombinedeffectsofthedifferentprocessloads. These

    deflectionscausesmallbutimportantdimensionalchangesinthediecavity,changesthat

    mustbecapturedsincetheyrepresenttheinitialshapetheliquidcastingacquiresattheend

    ofthefilling.

    Inordertoknowwhattheinitialcastingshapeispriortotheonsetofcooling,anaccurate

    descriptionofthedeformeddiecavityshapemustbeobtainedfirst. Ragab[35]

    experimentedtrackingthedistorteddieshapebytyingthecastingsurfacestothediecavity

    surfaces. Themodelincludedthecasting,die,platens,ejectorsupportblock. Allpartsinthe

    modelwere

    discretized

    using

    solid

    brick

    finite

    elements.

    It

    was

    reported

    that

    because

    the

    castingwasrepresentedbysolidelements,limitationsintheelementdeformations

    preventeditfromaccuratelytrackingthedieshapeasthedifferentloadswereapplied.

    Themethodologyusedinthisresearchworkreliesintheuseofashellmeshtotrackthe

    distortionsinthecavity. Thesamestructuralcomponentsforthediedistortionmodelare

    usedinthismodel. Theapplicationofthethreedifferentprocessloadsissequentiallydone

  • 8/12/2019 Mechanical Performances of Dies

    44/284

    34

    asdescribedbefore. Inordertoavoidhavingasolidcastingtrackingthecavitydistortions,a

    shellmeshisusedinstead. Theshellmeshisbuiltusingthesurfaceelementsofthecasting

    mesh,sharingthesamenodesasthecastingsurfaceelements. Bysharingthesamenodes,

    thedisplacementsobtainedfromtheshellmeshcanbereadilymappedontothecasting

    meshsurface.

    Thedimensionalchangesinthecavityaretrackedbytyingtheshellmeshtothediecavity

    mesh. Tyingtheshelltothediecavityprovidesadescriptionofthedistorteddiecavity

    shapeaftertheapplicationofthealreadymentionedstaticloads. Afterthediehasbeen

    distorted,thepredicteddisplacementsoftheshellmeshcanbeappliedtothecastingmesh,

    providingacastingshapethatmatchesthatofthedeformeddiecavity. Theunderlying

    assumptioninthisprocedureisthatthedistortionsinthecavityaresmallenoughthatthey

    canbemappedonlytothecastingsurfacewithoutaffectingitsinteriorstructure.

    Figure12

    shows

    the

    shell

    mesh

    used

    for

    this

    model.

    With

    athree

    dimensional

    mesh

    ofthe

    casting,theshellmeshcanbereadilyobtainedusingthemeshingcapabilitiesofanypre

    processor. Forthismodel,ashellthicknessof0.0254mmwasspecified. Toavoidany

    modelinglimitationsduetorigidityontheshell,theYoungsModulusoftheshellmaterial

    wasspecifiedtobethreeordersofmagnitudesmallerthanthatofregularsteel.

    Themodelisrunasastaticanalysis. Ashasbeenalreadydescribed,clamping,thermaland

    pressureloadaresequentiallyappliedinthreedifferentsteps. Attheendoftheanalysis,

    thedisplacementpredictionsfromtheshellareextractedandusedinthefollowingmodel.

    Thedescriptionformodelingthecoolingofcastinginsidethedieisprovidedinthenext

    section.

    Twopapers,onedescribingapartdistortionstudybasedonthisworkandonedetailing

    worktoobtainthehightemperaturepropertiesofthediecastingalloyA380arepresented

    inAppendicesCandDrespectively.

  • 8/12/2019 Mechanical Performances of Dies

    45/284

    35

    Figure12 Shellelementmesh

    3.4 EvaluationofEjectorSideandSlideDesignTheapproachtotheejectorsidedesigntaskcloselyfollowsthatusedfortiebarloading.

    Powerlaw

    models

    were

    developed

    to

    predict

    maximum

    parting

    plane

    separation

    on

    the

    coverandejectorside.Adesignofexperimentswasdevelopedbasedonthemajor

    structuralvariablesofthediecastingdieandthemachine.Astaticfiniteelementanalysis

    wasconductedateachdesignpointspecifiedinthedesignarrayusingourbasicstructural

    modelingmethodology. Thepredictedmaximumseparationofthecoverandejectorside

    partingsurfacesfromtheirnominallocationwasobtainedfromthefiniteelementmodels.

    Powerlawmodelswerefittothepartingplaneseparationdataandthenondimensional

    structuraldesignparameters.Thenondimensionalparameterswereobtainedusing

    dimensionalanalysisbasedonBuckinghampitheorem.Thedesignofexperiments,thenon

    dimensionalparameters

    and

    the

    power

    law

    models

    are

    summarized

    below.

    More

    details

    canbefoundinAppendixA.

    3.4.1 DesignofExperiments

    Thedesignvariablesincludedinthestudyarethediewidth,dielength,diethickness,

    thicknessofthediesteelbehindtheinsert(dieshoulderthickness),pillardiameterandthe

  • 8/12/2019 Mechanical Performances of Dies

    46/284

    36

    patternofejectorpillarsupports.Thedescriptionofthefactorsalongwiththeirhighandlow

    valuesisshowninTable11.

    Factor Description HighLevel LowLevelPt Platenthickness 9 13

    Lx Horizontaldimensionof 24 38

    LyVerticaldimensionofthe

    die24 38

    t Diethickness 5 10

    TR Diethicknessratio 0.4 0.5

    PD Pillardiameter 1.5 4

    X PillarPattern(discrete 4Levels/Patterns

    Table11 FactorsusedinDesignofExperiments

    Thesixthfactor,pillarpattern,isadiscretefactor.Theschematicofthefourdifferentpillar

    arrangementsbehindtheejectordiethatwereanalyzedisshowninFigure13whichshows

    therearviewofanejectordie.Thebacksurfaceoftherailsishatchedinthefigure.Therails

    are3incheswideinallofthecases.Therailsandpillarsare6incheslonginallofthecases.

    Inpillarpattern1thereareatotalofninepillars,onedirectlybehindthecenterofpressure

    andtheouterpillarsarelocatedataradialdistanceof6.75inchesfromthecenterpillar.In

    pattern1theouterpillarsare45apartfromeachother.Inpattern2,therearenopillars

    andtheejectordiehasonlyrailsupport.Inpattern3andpattern4therearefivepillars,one

    onthecenterandfourouterpillarseach90apart.Thedifferencebetweenpattern3andpattern4isintheorientationoftheouterpillars.A58runcentralcompositeresponse

    surfaceexperimentaldesignwaschosenbasedonthefivecontinuousfactors.Thenthe58

    runswererepeatedforeachlevelofthediscretefactor.

  • 8/12/2019 Mechanical Performances of Dies

    47/284

    37

    Figure13 SchematicofPillarPatternsusedintheStudy

    Pattern1,9pillars,(b)Pattern2,Nopillars,(c)Pattern3,FivePillars,(d)Pattern4,5pillars

    Themaximumcoverandejectorsidepartingplaneseparationwerepredictedforeachcase

    intheexperimentalarrayusingthefiniteelementmodelingmethodology

    3.4.2 DimensionalAnalysisTheobjectiveofthecomputationalexperimentsistodevelopempiricalcorrelationsto

    predictthemaximumpartingplaneseparationasafunctionofthedesignvariablesinvolved.

    Dimensionalanalysiswasusedtodeterminethenondimensionalparametersfortheempiricalcorrelations.Dimensionalanalysisisbasedontheprinciplethatanyvalidphysical

    relationshipshouldbedimensionallyhomogeneous.BasedonthisprincipleBuckingham[58]

    showedinhisfamousPitheoremthatanyequationdescribingthephysicalrelationship

    betweenthevariablescanbereformulatedasafunctionofdimensionlessproductsof

    variables.

  • 8/12/2019 Mechanical Performances of Dies

    48/284

    38

    Thechoiceofvariablestobeusedinthedimensionalanalysisandthegroupingofvariables

    thatformthenondimensionalgroupscanbedeterminedusingtheknowledgeofthe

    physicalphenomenonandengineeringjudgment.Thediesandinsertsareassumedaspre

    stressedflatplatesthatarerestingonelasticsupportsandsubjectedtouniformloading.The

    semianalyticalempiricalequationforpredictingmaximumdeflectionofaflatplatunder

    uniformloadingisgivenby[60].

    4

    3

    WLMaximum deflection

    Et (4)

    Wistheuniformlydistributedload,listhelengthoftheplate,Eistheyoungsmodulus,tis

    theplatethicknessandisaconstantthatdependsontheaspectratioandboundary

    conditions.Theunsupportedspanbehindthecoverdieischaracterizedbythelengthand

    widthofthedieandamodelforcoversideseparationwilltakeaformsimilartothe

    deflectionequationabove.Howeverontheejectorsidetheunsupportedspanischaracterizedbythedistancebetweenthepillarsandrailsupports.Jofreit[61]developeda

    semiempiricalequationtopredictthemaximumdeflectionofconcreteflatplatessupported

    onstraightbeamsaroundtheperipheryandagridofpillarsinbetweenthebeamsupports.

    Theequationtopredictthemaximumdeflectionwithinanyinnergridofpillarsisgivenby

    4

    3

    3

    4

    s nl lW

    Maximum deflectionEt

    (5)

    wherelsandlnarethelongspanandtheshortspanbetweenthepillarsupports.Itcanbe

    observedthatthespanvariableLinequation(4)isreplacedbytheweightedaverageofthe

    shorterandlongerspanbetweenthepillarsupports.Thereforeamodelfortheejectorside

    separationwilltakeasimilarform. Basedonthestructureoftheequations(4)and(5),the

    physicalrelationbetweenthemaximumpartingplaneseparationandthedesignvariables

    canbewrittenasfollows:

    max , , , , , , , , , , 0 tb nf Pt L t TR L l PD RW DistributedLoad E (6)

    where,max

    isthemaximumpartingplaneseparation,Ltbisthedistancebetweenthetiebar

    centers,Listhelengthscalerepresentingthelengthandwidthofthedieorthespan

    betweenthepillars,lnisthelengthscaleforspanbetweenpillars,PDisthepillardiameter,

    RWisthewidthofthedieandEistheyoung'smodulusofthediematerial.Theterm

    DistributedLoaddenotesthedistributedcontactloadatthepartingsurfaceanditis

    approximatesasfollows:

  • 8/12/2019 Mechanical Performances of Dies

    49/284

    39

    Pr Pr

    -

    x y x y

    TotalClamp Load Cavity essure ojectedAreaofCavityDistrubted Contact Load

    L L L L(7)

    ThenondimensionalparametersselectedaresummarizedasshowninTable12.Detailscan

    befoundinAppendixA.

    DimensionlessParameter

    Description

    1 (Ltb/t)or(Distancebetweentiebarcenters/Diethickness)

    2 (Pt/t )or(Platenthickness/Diethickness)

    3 (TR)or(Dieshoulderthickness/Diethickness)

    4 (L/t )or(Span/Diethickness)

    5 (MaxSeparation/t)(Youngs Modulus/Distributedcontactload)

    6 (Lx/Ly)or(Aspectratioofthedie)

    7 (Lcx/Lcy)or(Aspectratioofthecavity)

    Table12NonDimensionalStructuralDesignParameters

    Theparameter1,indicatesthatincreasingthedistancebetweenthetiebarcenterhasthe

    sameeffectasdecreasingthediethickness.Ifthedistancebetweenthetiebarsincreases

    foragivenplatenthickness,theplatenbendsmoreandthesupportavailableforthediesis

    reducedandthediesdeflectmore.Decreasingthediethicknesswillalsoresultinanincreaseinthedeflectionofthedie.ThoughthevariableLtbwasnotexplicitlyvariedinthe

    computationalexperiments,theparameter1variesamongtheexperimentalcasesdueto

    thevariationindiethicknessamongtheexperimentalcasesandtheeffectofLtbcanbe

    capturedfromthecomputationalexperiments.

    Thesecondparameter2canbecombinedwiththefirstparameter1,toobtainanewnon

    dimensionalparameter,say,12,whichisgivenby

    11 2

    2

    tb

    Pt

    L

    (8)

    Theparameter12,aboveshowsthatthereisalwaysacombinationofvaluesofplaten

    thicknessanddistancebetweentiebarcentersforwhichtheplatenstiffnessremainsa

    constant.Themostimportantnondimensionalparameteris4whichrepresentstheratio

    betweentheunsupportedspanandthediethickness.Thisparametersuggeststhatlarger

    thespanbehindthedie,thickershouldbethedie.Acombinationofthespananddie

  • 8/12/2019 Mechanical Performances of Dies

    50/284

    40

    thicknesscanalwaysbechosentoobtainadesiredmagnitudeofpartingplaneseparation.

    Ontheejectorside,theparameter4representstheratioofthespanbetweenthepillar

    supportstothediethickness.Inourexperimentsthepillarlocationswithrespecttothe

    centerofpressurewerefixed.Butthethicknessofthediewasvar