1Copyright1999byASMEProceedingsofMaterialsProcessingSymposium:1999ASMEInternationalMechanicalEngineeringCongress&ExpositionNovember14-19,1999,Nashville,TennesseeTOWARDSCONTROLLABILITYOFINJECTIONMOLDINGDavidKazmerDepartmentofMech.&Ind.EngineeringUniversityofMassachusettsAmherstDavidHatchDepartmentofMech.&Ind.EngineeringUniversityofMassachusettsAmherstABSTRACTProcesscontrolhasbeenrecognizedasanimportantmeansofimprovingtheperformanceandconsistencyofthermoplasticparts.However,nosinglecontrolstrategyorsystemdesignhasbeenuniversallyaccepted,andmoldingsystemscontinuetoproducedefectivecomponentsduringproduction.Thecapabilityoftheinjectionmoldingprocessislimitedbythethermalandflowdynamicsoftheheatedpolymermelt.Thispaperdiscussessomeofthedifficultiesposedbycomplexanddistributednatureoftheinjectionmoldingprocess.Theflowandthermaldynamicsoftheprocessareanalyzedwithrespecttotransportandrheology.Then,twonovelprocessingmethodsaredescribedtoenablein-cycleflow,pressure,andthermalcontrol.Simulationandexperimentalresultsdemonstrateeffectivenessoftheseinnovationstoincreasetheconsistencyandflexibilityinpolymerprocessing.Suchsystemdesignchangessimplifytherequisitecontrolstructureswhileimprovingtheprocessrobustnessandproductivity.INTRODUCTIONInjectionmoldingiscapableofproducingverycomplexcomponentstotightspecifications.Theprocessconsistsofseveralstages:plastication,injection,packing,cooling,andejection.Ininjectionmoldinganditsvariants(coinjection,injectioncompression,gasassistmolding,etc.),thermoplasticpelletsarefedintoarotatingscrewandmelted.Withahomogeneousmeltcollectedinfrontofthescrew,thescrewismovedforwardaxiallyatacontrolled,time-varyingvelocitytodrivethemeltintoanevacuatedcavity.Oncethemeltissolidifiedandthemoldedcomponentissufficientlyrigidtoberemoved,themoldisopenedandthepartisejectedwhilethenextcyclesthermoplasticmeltisplasticizedbythescrew.Cycletimesrangefromlessthanfoursecondsforcompactdiscstomorethanthreeminutesforautomotivecomponents.Controlofinjectionmoldingissignificantlychallengedbythenonlinearbehaviorofthepolymericmaterials,dynamicandcoupledprocessphysics,andconvolutedinteractionsbetweenthemoldgeometryandfinalproductqualityattributes.Arevisedsystemsviewofthemodernconventionalinjectionmoldingprocess1ispresentedinFig.1.Themachineparametersareindicatedontheleftsideofthefigure,andsomecommonmoldedpartmeasuresofqualityarelistedontheright.Inthisfigure,theprocessisdecomposedintofivedistinctbutcoupledstages.Theoutputofeachstagenotonlydirectlydeterminestheinitialconditionsofthenextstage,butalsoinfluencessomeofthefinalqualitiesofthemoldedpart.BarrelTemp1000PLASTICATIONINJECTIONPACKINGCOOLINGEJECTIONPROCESS/PARTQUALITYMeltPressureThermoplasticPelletsScrewPres0.02ScrewRPM0.5DistortionDimensionsClarityEconomicsResid.StressIntegrityEjectedPartRelaxationSolidifiedLayerDevelopmentStrengthAppearanceResidenceTimeMeltVolumeMeltTempMeltQualityInjectionVelocityProfile0.02MaximumInjectionPressure0.1PackingPressureProfile0.2PackingTime0.01MeltViscosityInletPressureFlowRateMoldCoolantTemperature200CoolingTime0.01MeltFrontVelocityMeltPresMeltDensityMeltTempSolidifiedLayerDevelopmentClampTonnageSolidifiedLayerDevelopmentCycleTimePartTempPartStrainPartStressEjectionStroke0.02EjectionVelocity0.01FlashMoldFailureShotSize0.02MACHINEINPUTSQUALITYATTRIBUTESSTATEVARIABLESFigure1:SystemsviewoftheinjectionmoldingprocessThincavityfillingofpolymermeltcorrespondstocreepingflow(Re1)surroundingahotcoreregion2.Asanexample,considerareferencevelocityof10cm/sec,referencethicknessof3mm,andaviscosityof100PaSeconds.TheReynoldsnumberbasedonthiscaseisverysmall,(10-3),indicatingthevalidityofthehighlyviscouscreepingflowassumption.Furthermore,theflowregionsareconsideredfullydeveloped,andboththeunsteadyandthegravitationalforceeffectscanbeignoredduetonegligiblelocalacceleration.Ontheotherhand,thethermaldiffusivity,2Copyright1999byASME=k/Cp,oftypicalpolymermeltsis(10-3)cm3/sec,andthekinematicviscosity,=/=103cm2/sec；hence,thePrandtlnumberisabout(106)andPecletnumber,Pe=Re*Pr,is(103).Usingtheseassumptions,themass,momentum,andenergyequationsreducetothefollowingformsintheCartesiancoordinatesystem:()()0=+wzvxtrrr(1)xPzvz=h(2)222ghr&+=+zTkxTvtTCp(3)wherezandxarethethicknessandstreamwisedirections；visthevelocitycomponent；Pisthepressure；histheshearviscosity；r,Cp,andkarethethermalproperties；g&istheshearrate,and2gh&istheviscousheatingterm.Thesolutionofthepressurefieldininjectionmoldingisobtainedbycouplingthemassandmomentumequations.Generally,themassequationprovidesaconvergencecriterionforflowrateaboutwhichthemomentumequationisiterativelysolvedtoproduceanaccuratepressurefield.Foreachinstantoftime,allthenodalpressuresonthemesharesolvedsimultaneously.Iterationisrequiredtoupdatetheshearrate,viscosity,andflowrateestimatesuntilfullconvergenceisachieved.Foracompressibleflow,thenetmassfluxmustequalanymassgainsorlosseswithintheelement3.Thenecessarysystemofequationscanbedeveloped,assembled,andsolvedusingaconventionalGalerkinformulationforafixedmeshandtransientmeltfront.Suchasimulationhasbeendeveloped,andwillbeutilizedinassessingstrategiesforprocessdevelopmentalongwithexperimentalvalidation.PROCESSDEVELOPMENTAnoverviewofinjectionmoldingcontrolisshowninFig.2.Attheinnermostlevel,onlythemachineactuatorsareregulated.Thislevelofcontrolwillensureproperexecutionoftheprogrammedmachineinputs(Fig.1).Atthesecondlevel,statevariablessuchasmelttemperatureandmeltpressurearecontrolledtotrackpre-specifiedprofiles.Thiswillprovidemoreprecisecontrolofthestateofthemelt.Attheoutermostlevel,themachineinputsareadjustedtoimprovethequalityofthepartthroughbettersetpointsgivenqualityfeedback.MachineActuatorsProcessMachineControlSetPointControlStateVariableControlMachineFeedbackQualityFeedbackStateVariableFeedbackPartAttributesMachineInputsFigure2:SystemdiagramofinjectionmoldingcontrolWhilemachinecontrolisimportant,itisthepolymerstate(pressure,temperature,andmorphology)whichdirectlydeterminesthemoldedpartquality4,5.Assuch,thispaperfocusesonclosingtheloopbetweenthemachineparametersandthepolymerstate.Ifachieved,theseadvancedcontrolstrategieswouldprovideincreasedmoldedpartqualityandconsistency.CavityPressureControlAfundamentalstatevariablethatcanberegulatedduringthemoldingcycleiscavitypressure.Closed-loopcontrolofcavitypressurecouldautomaticallycompensateforvariationsinmeltviscosityandinjectionpressuretoachieveaconsistentprocessanduniformsetofproductattributes6.Mannintroducedoneofthefirstpressurecontrolschemesbyusingmodulatedpressurereliefvalves7,andAbuFaradevelopedaprocesscontrolmodelbyrelatingthecavitypressureresponsetoopen-loopperturbations8.Srinivasanlaterusedthesemodelstoproposealearningcontrollerforclosed-loopcavitypressurecontrol9.Adaptivecontrolmethodshavealsobeenproposedtotrackcavitypressureprofileatusuallyonelocationinthemold10-12.Unfortunately,cavitypressurecontrolsuffersfromthelackofasystematicmethodofdeterminingthepressureprofile.Inaddition,itishandicappedbytheabsenceofappropriateactuatorsfordistributedpressurecontrol,asconventionalmoldingmachinesareequippedwithonlyoneactuator(thescrew)whichdoesnotallowsimultaneouscavitypressurecontrolatmultiplepointsinthemold.ConsiderthemelttransportsysteminaconventionalcoldrunnermoldasshowninFig.3.Itisevidentthatthegeometryis“hard-wired”intothemold.Therunnerlocationsarefixedandthegatedimensionsarealsofixed.Theresultingpressuredistributioncannotbecontrolledwithoutre-toolingmoldsteel.Figure3:TypicalPackingPressureDistributionToinvestigatethecontrollabilityoftheinjectionmoldingprocess,ahalf-factorialdesignofexperiments13wasperformedtodeterminethemaineffectsbetweenthecriticalprocessparametersandthepartdimensions:3Copyright1999byASME=ScrewSpeedeTemperaturVelocityPressureLLL10.018.005.023.000.029.018.051.002.043.010.057.0321(4)Inthisequation,themachineparametershavebeenscaledtotherangeof0to1,indicativeofthemaximumfeasibleprocessingrangeforthisapplication.Theresultingcoefficientsofthelinearmodelareactualchangeinpartdimensionsmeasuredinmm.Itshouldbenotedthatoncetoolingiscompleted,thedimensionalchangesavailablethroughprocessingarequitelimitedthoughfunctionallysignificant.Theprimaryconclusionthatshouldbedrawnfromeq.(4),however,isthatallthedimensionsreactsimilarlytochangesintheprocesssettings.Thus,themoldingprocessbehavesasaonedegreeoffreedomprocessinwhichonlyonequalityattributeiscontrollable.OneofNamSuhsaxioms14ofdesignstatesthat“independenceoffunctionalrequirementsshouldbemaintained.”Thisaxiomwasappliedtodevelopmultipledegreesoffreedomforcontrolofmeltflowandpressureinthemoldcavity.AsshowninFig.4,thevalvesmetertheflowofmeltfromtherunnersintothemoldcavity.Thepressuredropandflowrateofthemeltisdynamicallyvariedbytheaxialmovementofeachvalvestemwhichcontrolsthegapbetweenthevalvestemandthemoldwall.Byde-couplingthecontrolofthemeltatdifferentvalvestempositions,meltcontrolateachgatecanoverridetheeffectsofthemoldingmachineandprovidebettertimeresponseanddifferentialcontrolofthemelt.Eachvalveactsasanindividualinjectionunit,lesseningdependencyonmachinedynamics.Forclosedloopcontrol,manifoldpressuretransducerswereusedintherunnerdropsinsteadofinthecavity.Thisimplementationnotonlyprovideslowercostandgreaterreliability,butalsorendersaconventionalappearanceforthesystem.MeltInletValve1Valve2Cavity1Cavity2P1P2Figure4:DynamicFlowRegulationDesignTheresultingcontrollabilityoftheinjectionmoldingprocessisdemonstratedinFig.5wheremultiplepressureprofilescanbemaintainedinthemoldcavityofasinglepart.Inthesamecycle,threedifferentmagnitudesofmeltpressurewereexertedatdifferentgatesinthesamemoldcavity.ThecontrolpressurefortheholdingstageatGate1is41.4MPa(6000psi.),Gate2is41.4MPa(6000psi.),Gate3is20.7MPa(3000psi.),andGate4is62.1MPa(9000psi.).Inconventionalinjectionmolding,themeltpressurewouldbethesameatallgates.Thislevelofprocesscontrolhasnotpreviouslybeenachievedbyanymoldingtechnologythusfar.Eachgatecanexertaspecificholdingpressure.010203040506070024681012Time(sec)Figure5:DynamicFlowRegulationDesignThematerialshrinkageanddimensionschangeatdifferinglocationsinthepartbasedonthepressurecontoursandhistoriesaroundthegates.Theabilitytochangeindividualdimensionsorotherqualityattributeswithoutre-toolingmoldsteelprovidessignificantprocessflexibility.Itispossibletoaugmenteq.(4)withtheadditionaldegreesoffreedomandre-examinethecontrollabilityofthethreepartdimensions:+=P4P3P2P1ScrewSpeedeTemperaturVelocityPressureL3L2L121.000.002.000.016.000.017.010.000.060.031.000.001.003.002.001.000.005.009.003.001.008.005.002.0(5)Therearetwosignificantimplicationsofthisresult.First,theclosedloopcontrolofcavitypressureshassignificantlyreducedthedependenceofpartdimensionsonmachinesettings,asevidencedbythereductioninthemagnitudeofcoefficientsfortheprimarymachinesettings.Thiseffecthasalsobeenevidencedbyreductionsinthestandarddeviationsofmultiplepartdimensionsbyanaveragefactoroffive,resultinginanincreaseintheprocesscapabilityindex,Cp,fromlessthan1tofarbeyond2.Second,thesecondmatrixineq.(5)isevidenceoftheimproveddimensionalcontrollabilityprovidedbythedynamicregulationofthecavitypressuredistribution.Ingeneral,changingthecavitypressureatthegateclosesttoadimensionprovidesthemajoreffectonpartdimensions.Additionally,independentcontrolofthevalvestemsprovidesthecapabilitytovarydimensionsatonelocationwithoutinterferingwithdimensionsatanotherlocation.Thisflexibilitydoesnotexistin4Copyright1999byASMEconventionalmoldingbecauseholdpressurechangesintendedtoinfluenceoneareaofthepartcanbetransmittedtootherareasofthepartthroughthestaticfeedsystem.Itshouldbenoted,however,thatthetotalmagnitudeofdimensionalchangeavailablewithdynamicpressureregulationisapproximatelythesameasforconventionalmolding.Theseresultsmayhaveasignificantimpactontheproductandtoolingdevelopmentprocess.Currently,numericalmoldfillingsimulationsandexpertjudgmentsarecombinedtoestimatetheprocessbehaviorandmakecriticaldesigndecisions.Ifthesedecisionsareincorrect,thentoolingmodificationsmayberequired.Improvedcontrollabilityoftheinjectionmoldingprocesspermitscorrectionformanydesigninaccuraciesduringthemoldcommissioningstagewithoutretooling.Suchachangeinthedevelopmentprocesscouldsubstantiallyreducethetooldevelopmentcosts,shortenthedevelopmentcycle,andhastentimetomarket.Thedescribedprocessisalsosignificantinthatitmovespolymercontrolfromthemoldingmachinetothemolditself.Thisreducesthemoldingmachinetoapolymericpump.Variationsininjectionpressure,flowrates,packpressures,orpacktimesareallcompensatedthroughdynamicpressureandtemperaturecontrol.Themarketrepercussionscouldbesignificant,as1)anoldmachinewithoutclosedloopcontrolcanprovideconsistencyequaltomodernmachines,and2)amoldcommissionedonamoldingmachineintheUnitedStatesisensuredtoproduceconsistentpartsonamoldingmachineoverseas.Themoldbecomesitsownself-containedqualitycontrolmechanism.Assuch,thepotentialproductivityandqualitygainsaresubstantial.TemperatureControlThetypicalheatpathinthecoolingstageofinjectionmoldingisthatheatisconductedfromthehotpolymertothecomparativelycoldmold,thenconductedthroughthemoldtothecoolingline,whereitisconvectedawaybythecoolant.Recentresearchhasattemptedtodynamicallycontrolthethermalandfluidpropertiesofthemeltwithinthemoldingcycle.Whiledynamicpressurecontrolhasbeenprovenfeasible15andisbeingcommercialized,therelativelyslowthermaltransientshavepreventedsimilargainsinthermalmanagement.Thecoolingstageofinjectionmoldingcycleisnotidealforavarietyofreasonsimpactingboththeproductqualityandproductioneconomics.Theprocessphysicsdictatethatthemoldtemperaturemustbelessthanthepolymerheatdeflectiontemperaturesuchthatarigidpartisejected.However,thecoldmoldtemperatureconductsheatfromthehotpolymermelttothecoldmoldduringinjectioncausingthedevelopmentofaskinontheexteriorofthepartandpropagationoffrozenlayerstowardsthecoreofthepart.Thesefrozenlayersincreasetheflowresistance,makingthemoldcavitydifficulttofill.Sincefrozenlayersaredevelopedcontinuouslyduringinjectionandcooling,theylockinvaryinglevelsofstressandorientation.Thisvariationinpolymermorphologyasafunctionofthicknessreducesoptical,structural,andotherpartproperties16-19.Tocompensateforthenegativeeffectsofcoldmoldwalls,manufacturersmayrunthemoldathighermoldtemperatures,highermelttemperatures,higherinjectionpressures,andhigherinjectionvelocities20,21.Alternatively,alowerviscositypolymerorhigherpartwallthicknessmayberequiredwithcostand/orperformancedisadvantages.Alloftheseoptionsnegativelyimpacttheeconomicsofproduction.Infact,theeconomicdriversdictatehighermoldtemperaturesduringinjection(toallowthinpartwallthicknessesandlowinjectionpressures)butlowermoldtemperaturesduringcooling(toallowrapidsolidification).Thisoptimalmoldtemperaturecontrolstrategyisinfeasiblegivencurrentcontrolstrategiesandmaterialtechnologies.Thesizeofthemold,togetherwithitshighheatcapacityandthermalinertia,preventsdynamicclosedloopcontrolofthemoldsurface.Thisstatementisbasedonobjectiveanalysisaswellasobservationofprioracademicandindustrial22-34.Forinstance,Jansen35,Chen36,andotherresearchershaveutilizedathermoelectricdevicewithin