J.Cent.SouthUniv.(2012)19:174178DOI:10.1007/s1177101209883Multi-stageoptimumdesignofmagazinetypeautomatictoolchangerarmKIMJae-Hyun,LEEChoon-ManSchoolofMechatronics,ChangwonNationalUniversity,Changwon641-773,KoreaCentralSouthUniversityPressandSpringer-VerlagBerlinHeidelberg2012Abstract:Toenhancemachiningefficiency,toolchangetimehastobereduced.Thus,foranautomatictoolchangerattachedtoamachiningcenter,thetoolchangetimeistobereduced.Alsotheautomatictoolchangerisamainpartofthemachiningcenterasadrivingsource.Thestaticattributesoftheautomatictoolchangerusingthecommercialcode,ANSYSWorkbenchV12,weretriedtointerpret.Andtheoptimumdesignofautomatictoolchangerarmwasproposedbyperformingthemulti-stageoptimumdesign.Theshapeoptimizationoftheautomatictoolchangerwasproposedandtheresultwasverifiedtoobtainacceptableimprovements.Itispossibletoobtainanoptimizedmodelinwhichthemaximumdeformation,maximumstress,andmassarereducedby10.46%,12.89%and9.26%,respectively,comparedwiththoseoftheinitialmodel.Also,theresultsbetweenconventionalmethodbythedesignofexperimentsandproposedmethodbythemulti-stageoptimumdesignmethodwerecompared.Keywords:automatictoolchanger；optimumdesign；structuralanalysis；exchangearm1IntroductionRecently,inmachinemanufacturingindustries,moldsandmachinepartshavebeenchangedtosmallquantitybatchproductionsystem.Also,improvementsinproductivityandcuttingratearerequired.Whereas,itistruethathighqualityandlowcostaretobetargetedfromapracticalstandpoint.Therefore,themachinetoolsforsuchaimspursuetoachievehigh-speedprocessing,implementautomation,andreducedleadtime.Asaresult,itispossibletocheckthestatesoftoolsandworkpiecesusingpropersensorsinthemachinetools.Inaddition,amachiningcenterbasedonanautomatictoolchanger(ATC)andanautomaticpalletchanger(APC)aimstooperateanunattendedoperationfactoryfor24h.Theautomatictoolchangerstoresthetoolsusedinamachiningcentertoitsmagazineandchangesthetoolsautomaticallyasrequired.ThetoolchangedbysuchATCispreciselyequippedtoaspindle1.Also,itrepresentsanadvantagethatanoperatorofthemachiningcenterisabletoengageinotherworksduetothelessinterferenceforthemachinetools.Thatistosay,theoperatorcancontrolothermachinetoolsorpreparethenextworkpieces,whichleadstoreducedproductiontime.ThemagazinetypeATCusedinthisstudyrepresentsafeaturethatmanytoolsarestoredinthemagazine.Inthechangeoftools,twoarmsmovetochangetheequippedtooltothenexttoolbyrotatingthemby180inadirectlychangedmanner2.Thus,itisnecessarytoensurethetechnologiesforboththestructuralcharacteristicsofarmsandthedesignoflightweightsimultaneously.Inactualindustrialfields,designoptimizationisveryimportant.Therefore,variousoptimizationmethodsarepresentedfortheoptimizationofvariousmechanicalparts3.SONGetal4presentedoptimizationdesignoftheshortjournalbearingbyusingenhancedartificiallifeoptimizationalgorithm.ALLAIREetal5combinedthetopologicalandshapederivationsonthestructuraloptimization.BAGCIandAYKUT6presentedTaguchioptimizationtoverifytheoptimumsurfaceroughnessoftheCNCmilling.LAMBERTI7presentedadesignoptimizationalgorithmbasedonsimulatedannealingfortrussstructures.SEKULSKI8presentedthatthegeneticalgorithmcanbeanefficientmulti-objectiveoptimizationtoolforsimultaneousdesignofthetopologyandsizingofshipstructures.SEOetal9presentedshapeoptimizationanditsextensiontotopologicaldesignbasedonisogeometricanalysis.InoptimizingtheATCarm,thefactorsofthestructuralcharacteristicsandthelightweightarecontrarytoeachother10.Itshowsatrade-offthatifitpursuestoimprovethelightweightinstructures,thestructuralcharacteristicswillrepresentaweakness,andiftheFoundationitem:Work(RTI04-01-03)supportedbyGrantfromRegionalTechnologyInnovationProgramoftheMinistryofKnowledgeEconomy(MKE),KoreaReceiveddate:20110426；Accepteddate:20111010Correspondingauthor:LEEChoon-Man,Professor,PhD；Tel:+82552133622；E-mail:cmleechangwon.ac.krJ.Cent.SouthUniv.(2012)19:174178175structuralcharacteristicsareimproved,theachievementofthelightweightisdifficult.Therefore,forsatisfyingthesecontraryfactorsandoptimizingthem,theoptimizationofsucharmshapesindifferentwayispresentedbyusingthedesignofexperiments11.Inthisstudy,forachievingamoreimprovedoptimizationmodelthanthepreviousstudy11,amulti-stageoptimumdesignwasperformed.Theoptimumdesignwaspresentedusingthecommercialanalysisprograms,CATIAV5andANSYSWorkbench,andtheanalyticvaliditywasinvestigatedthroughcomparingtheinitialandconventionaloptimizedmodelswiththeoptimizedmodelimplementedinthisstudy.2StructureofATCATCconsistsofthreeelements,suchasmagazinepart,changerpart,andarmpart.Themagazinepartisadevicethatstoresmanytoolsandchangestoolsusingservomotors.Thechangerpartisequippedwithservomotors,whichrotatearms.Thearmpartshowsanarmshapeandchangestoolsbygearingthetoolsinthespindleandmagazineinamachiningcenterbyrotatingthemby180.Figure1illustratestheentirestructureoftheATCmodelledbyusingtheCATIAV5R17.Fig.1StructureofmagazinetypeATCThestructuralanalysisoftheinitialmodelofthearmwasperformed.Regardingthereferenceoftheperformedfiniteelementanalysis,thefiniteelementanalysisoftheinitialmodelwascarriedoutusingthecommercialanalysisprogram,AnsysWorkbenchV12.Theanalysiswasperformedbyminimizingtheadditionalpartemployedinthearm.Intheanalysismethod,ahexdominantmethodwasappliedinwhichafiniteelementanalysishadtotally51794nodesand13496elements.Figure2showstheinitialfiniteelementmodelofthearm.Fig.2InitialfiniteelementmodelofarmFortheboundaryconditionsintheanalysis,theholeatthecenteroftheATCarmwassupported,andthegravitationalaccelerationwasappliedtotheentirebody.Intheloadconditions,aloadof147Nwasappliedtotheclampsatbothendsforconsideringthemaximumweightofthetools.TheresultsofthestructuralanalysisarepresentedinFig.3.Themaximumdeformationoftheinitialmodelattheclampsis5.7487mandoccursatbothends.Also,themaximumstressisgeneratedattheedgeofthesection,whichpushestherearfingeroftheATCarm,andispresentedby4.1762MPa.Fig.3Structuralanalysisofarm:(a)Deformationdistribution；(b)Stressdistribution3Multi-stageoptimizationofarmThestaticcompliance,fx(=D/F),canbepresentedbyaninversenumberofthestaticstiffness.Inparticular,insomemachinestructureslikemachinetoolsandindustrialrobotsthatrequirehighaccuracyandmachiningefficiency,itbecomesthemostimportantstaticcharacteristicaswellasthestructureweightwherethesefactorsaretobecomprehensivelyandsimultaneouslyevaluated.Asmentionedabove,theoptimizationofthestaticissueisdeterminedasthestaticcharacteristicofthesetwoobjectivefunctionsandtheminimizationissueoftheweight12.J.Cent.SouthUniv.(2012)19:174178176Thus,inthisstudy,theoptimizationisperformedasamulti-stagemannerforsatisfyingeachobjectivefunction.Thefirststageisconfiguredasastagethatimprovesthestaticcharacteristics.Bydefiningdesignfactorsthatminimizethedeformation,anoptimummodelcanbeinduced.Thesecondstageisdeterminedasastageforimplementingitslightweight.Basedontheoptimummodelpresentedinthefirststage,theshapeoptimizationisperformedbyaimingareductioninitsweightby10%.3.1FirststageofoptimumdesignofarmInthefirststageoftheoptimumdesign,theoptimumdesignaimstominimizethedeformationofthearm.Figure4illustratesthedesignvariablesofthearm.Fig.4FactorsofATCarmThegeneralformalizationforthedimensionandtheoptimumshapedesigncanbepresentedbydefiningobjectivefunctionsandlimitationconditionfunctions1315.ForimplementingtheoptimumdesignfortheATCarm,theformalizationisdeterminedasfollows:FindXMinimizedeformation(X)SubjecttoaaLA,B,CU(=A,B,C)X=A,B,CwhereXrepresentsoneofthedesignvariables,andandshowthestressanddeformation,respectively.Also,aandashowtheallowancevaluesforthestressanddeformation,respectively.ThetermsofA,B,andCarethedesignvariables.Thedesignvariablesareconfiguredby30mminordernottopresenttheinfluencesofthecollisionandinterferenceinstructuresonthedesign.Intheoptimumdesign,theoptimumsolutioncanminimizethedeformationofthearmusingtheCATIAV5Productengineeringoptimizer.Table1givestheresultsoftheoptimization.Figure5illustratestheresultsofthestructuralanalysisoftheoptimaldesignedarm.Theboundaryconditionsintheanalysisareconfiguredasthesameastheexistinginitialmodel.Table1ResultsofoptimizationforreducingdeformationFactorInitialmodelOptimaldesignedmodelA/mm253.396B/mm7073.686C/mm2732.686Maximumdeformation/m5.74874.6683Maximumstress/MPa4.17623.6072Fig.5Structuralanalysisofoptimizedarmforreducingdeformation:(a)Deformationdistribution；(b)Stressdistribution3.2SecondstageofoptimumdesignofarmAchievingthelightweightofthearmisanimportantfactorforreducingthecostofworkpieces.Also,itispossibletoimprovetheeconomybyintroducingalightweightstructure16.Therefore,theoptimumdesignforimplementingthelightweightofthearmisperformedinthesecondstage.Thetargetinreducingthemassis10%ofthearmbasedonthemodelproposedinthefirststageoftheoptimumdesign.Forreducingthemassofthearm,theshapeoptimizationiscarriedoutusingtheANSYSWorkbenchshapeoptimizationfunction.Theformalizationfortheoptimumdesigncanbepresentedasfollows:FindZMinimizemass(Z)SubjecttoaaLrUZ=rwhereZisoneofthedesignvariables,andshowthestressanddeformation,respectively,andaandaaretheallowancevaluesforthestressanddeformation,respectively.Also,thedesignvariable,r,isconfiguredJ.Cent.SouthUniv.(2012)19:174178177tofindallsectionsinwhichthemassreductionispossibleexceptforthesections,whichhavesomelimitationsinthedesign.Figure6illustratestheresultsoftheoptimumsolutionthatminimizesthedeformationofthearm.AsshowninFig.6,thesectionpresentedbyRemoverepresentsamassreduciblesectionbyremovingit.Basedontheresults,thereduciblesectionsareremovedtoamaximumlevel.Figure7showstheproposedoptimumshapeforlightweightofthearmbasedontheresultsoftheshapeoptimization.Fig.6ResultofshapeoptimizationusingANSYSFig.7RedesignofarmThestructuralanalysisisperformedusingtheproposedoptimumdesign.Also,theboundaryconditionsintheanalysisareappliedasthesameastheexistinginitialmodel.Figure8showstheresultsofthestructuralanalysis,whichiscarriedoutthroughapplyingtheoptimumshape.Fig.8Structuralanalysisofoptimizedarmforlightweight:(a)Deformationdistribution；(b)StressdistributionThemaximumdeformationofthemodel,whichappliestheoptimaldesign,isreducedfrom5.7487mpresentedintheinitialmodelto5.1475mbyasmuchas10.46%andgeneratedattheendoftheclampasthesameastheinitialmodel.Also,themaximumstressisreducedfrom4.1762MPapresentedintheinitialmodelto3.6379MPabyasmuchas12.89%.Inaddition,themassisreducedfrom7.8712kgpresentedintheinitialmodelto7.1425kgbyasmuchas9.26%.Table2presentstheresultsofthecomparisonoftheoptimumdesign11usingthedesignofexperimentsperformedwiththemulti-stageoptimumdesignimplementedinthisstudy.Table2ComparisonofresultsPropertyInitialmodel(A)Conventionalmethod(B)Proposedoptimizationmethod(C)RatioofAtoC/%RatioofBtoC/%Maximumdeformation/m5.74875.21975.