Product modularity and assembly systems An automotive case study

CIRPAnnals-ManufacturingTechnology

journalhomepage:http://ees.elsevier.com/cirp/default.asp

Productmodularityandassemblysystems:AnautomotivecasestudyJ.Paralikas,A.Fysikopoulos,J.Pandremenos,G.Chryssolouris(1)*LaboratoryforManufacturingSystemsandAutomation,DepartmentofMechanicalEngineeringandAeronautics,UniversityofPatras,Patras26500,GreeceARTICLEINFOABSTRACTKeywords:ModulardesignFlexibilityAutomotiveBiWDesigningmodularproductsoffersanumberofadvantages.However,modularitymaybechallengingintermsofproductionandassembly.Inthispaper,theinfluenceofproductdesignmodularityonassemblysystemsisinvestigated.Amodulardesignfromtheautomotiveindustryispresentedandcomparedwithatraditionaldesign.Theinfluenceofmodularityontheassemblysystemisdiscussedincludingattributessuchasproductionrate,investmentcostandproduct’sdeliverydelay.ß2011CIRP.1.IntroductionIntegralproductarchitectureincludesacomplex,many-to-manymappingfromfunctionalelementstophysicalcomponentsand/orcoupledinterfacesbetweencomponents.Amodulararchitecturepresentsanone-to-onecorrespondencebetweenmodulesandfunctionsandspecifiesde-coupledinterfacesbetweencomponents[1].Productmodularityenablestheeasygenerationofproductfamiliesfromabasic-platformdesign,bysimplymixingandmatchingthevariousmodules.Viathisproductvariationahighdegreeofcustomizationmaybeachieved[2].Moreover,partsormodulescarryoverandreusearealsopossiblewithmodularity.However,theproduct’sstructurealsoinfluencesitsproduction[3].Manufacturingsystemscapableofproducinginanagilemanneralltheproductvariantsderivedfromamodulardesign,arerequired.DifferentdesignmethodologiestoderivesuchsystemshavebeendiscussedinRefs.[4–6].Thekeyelementofsuchsystemsistheintegrationoftechnologiessuchasflexiblefixturesandtooling,multi-skilledworkforce,redundantrobots,sensingtechniques,wirelesstechnology,fixturelessassembly,automatedrobotcalibration,flexibleshifts,andflexiblefloorspace[7].Thepurposeofthispaperistoinvestigatetheeffectsofproductmodularityonthedesign,configurationandoperationofassemblysystems.Asacasestudyanunder-bodystructureofanautomotiveBody-in-White(BiW)hasbeenselected.BiWintheautomotiveindustryreferstoacarbody’ssheetmetalcomponentsthathavebeenassembledtogetherbeforepainting.Theunder-bodystructure(Fig.1)isoneofthemostcriticalcomponentsofanautomotiveBiW.Twoalternativedesignsofanunder-bodystructure,amodularandanintegralone,aredescribed.Theyallowtheproductionofthreeunder-bodystructurevariants.Theassemblysystemsforbothdesignsaremodelledandsimulated.Thedifferentassemblyconfigurationsarethencomparedfromtheviewpointofproductionresponsivenesstodemand,cumulativedelays,estimatedutilizationratesandinvestmentcostperproduct.2.ModularversusintegraldesignTheunder-bodystructurecarriesandconnectsmanysignificantcarcomponentssuchasengine,transmission,andsuspension,*Correspondingauthor.0007-8506/$–seefrontmatterß2011CIRP.doi:10.1016/j.cirp.2011.03.009contributingsignificantlytothecarstiffness.Itdeterminesthelengthofthevehicleandtoagreatextentitsfinalshape.Theunder-bodystructurecanbedesignedineitherintegralormodularform.2.1.Integraldesignoftheunder-bodystructureAnunder-bodystructurewithintegraldesignarchitectureisgenerallycharacterizedbyhighcomplexity.Thismeansthatifachangeismadetoapartoftheunder-body,itwillaffectothersurroundingunder-bodypartstoo.Mostofthetimes,suchchangesrequirearedesigneffortbythedesignersandproductionplanners.Theintegralunder-body(Fig.2)duetoitsinflexiblearchitectureisnotkeeninderivingalternativedesignvariants.Therefore,onlyasinglevariantisconsideredtobeproducedbyanassemblylineconfiguration.2.2.Modulardesignoftheunder-bodystructureModulardesignoftheunder-bodystructuremayconsistofthreemainmodules:thefloor,thefrontendandrearendmodules(Fig.2).Throughthismodularity,alternativedesignvariantsmaybegeneratedbymixingandmatchingthedifferentvariantsofeachmodule.Furthermore,throughthesplit-upofthemodules,scalabilityinthelongitudinaldimensionoftheunder-bodystructurecanbeeasilyachieved,sinceonlyasmallportionoftheunder-bodypartswillbeinfluencedandwillrequireredesigning.Basedonthisscalabilityfeaturethreealternativedesignvariantsoftheunder-bodystructurecanbegenerated(Fig.3).Amodularunder-bodyisconsideredasaplatformsegmentfromwhichalternativeBiWvariantsintermsofshapeanddimensionscanbeproduced[8].3.AssemblysystemsconfigurationsWeassumeagenericvehicledemandprofile(Fig.4)foraperiodoftwoyears.Ahighpeakappearsafterproductionlaunch,whileaslowdownofdemandduetomarketcompetitionmayfollow.Amarketingcampaign,attheendofthefirstyear,cancreateasecondlowerpeak,beforetheendoftheproductionphase.Afterthesecondpeakanoverlapwiththenextgenerationofproductmayappear.Thetotalproductionvolumefortwoyearsis1,200,000vehicles,andisdistributedoversevenvehiclemodels(Fig.5).Coupe,166J.Paralikasetal./CIRPAnnals-ManufacturingTechnology60(2011)165–168Fig.1.BiWandtheunder-bodystructure(integraldesign).Fig.4.Vehiclesdemandprofileforaperiodoftwoyears.Fig.5.Under-bodystructuresvariantsandtwoyearsproductionvolume.Fig.2.Integralunder-body(left)andmodularunder-body(right)structures.󰀂Sub-Cell#3(s3):Rearframeassemblysub-cell:backpanel,railmemberandsiderails.󰀂Main-Cell#1(1):Frontstructure,floorreinforcement’sassemblyandmainfloorassembly.󰀂Main-Cell#2(2):OutputofMain-Cell#1,rearframeassembly.󰀂Main-Cell#3(3):A-pillars,sills,sparewheelwell,hill-kick,rearpanelandre-spottingofMain-Cell#2output.Fortheassemblyofthethreeunder-bodystructurevariantsfortheintegralproductdesign,threeindependentlinesarerequiredfortheproductionofthethreeunder-bodyvariants.Fig.3.Modularunderbodystructurevariantsdifferentlengthswithinfloor(dL)andrearend(dR).3.2.AssemblysystemforthemodularproductdesignTheassemblysystemconfigurationanddecompositionforthemodulardesignoftheunder-bodystructureisbasedonthefollowingassemblyconfiguration(Fig.7):󰀂󰀂󰀂󰀂󰀂

Sub-Cell#1(s1):Frontmoduleassemblyandre-spotsub-cell.Sub-Cell#2(s2):Floormoduleassemblyandre-spotsub-cell.Sub-Cell#3(s3):Rearmoduleassemblyandre-spotsub-cell.Main-Cell#1(1):Frontandfloormodulesassembly.Main-Cell#2(2):OutputfromMain-Cell#1,rearmoduleandre-spotting.cabrioleandthreedoorshatchbackvehiclesmodelsareproducedfromunderbodyvariant1(UV1).Fivedoorshatchbackandfourdoorssedanvehiclesmodelsareproducedfromunderbodyvariant2(UV2).Stationwagonandfivedoorsmini-vanvehiclesmodelsareproducedfromunderbodyvariant3(UV3).ProductionvolumedistributionpervehiclevariantisbasedoncurrentEuropeanautomotiveindustrytrends[9].Ingeneralthedemandforsport/highperformancevehiclevariantsislowerthanthedemandforfamily/largespace/utilityvehiclevariants.3.1.AssemblysystemfortheintegraldesignTheassemblysystemconfigurationanddecompositionfortheintegraldesignoftheunder-bodystructureisbasedonthefollowingassemblyconfiguration(Fig.6):󰀂Sub-Cell#1(s1):Frontstructureassemblysub-cell:frontsiderails,sidecrossmembersanddashpanel.󰀂Sub-Cell#2(s2):MainFloorassemblysub-cell:frontandrearcrossmembers,floorsidepanels,tunnel.Fortheproductionofthethreemodularunder-bodystructurevariantsoneassemblylinemayberequiredsinceallvariationsofthemodulardesigncanbeassembledbyoneline.3.3.DifferencesofthetwoassemblyconfigurationsSeveralbenefitsandchallengesaretobeconsideredaboutthetwodifferentassemblysystemsfortheintegralandthemodularunder-bodystructure.TheassemblylineforthemodulardesigncanproduceFig.6.Integralunder-bodyassemblyconfiguration.J.Paralikasetal./CIRPAnnals-ManufacturingTechnology60(2011)165–168167

Fig.8.Modularunder-bodystructuremainassemblylinesimulationmodel.Fig.7.Modularunderbodyassemblyconfiguration(UV3).allthreevariants.Ontheotherhand,inthecaseoftheintegralproductdesigneachassemblylinecanbuildonlyonevariant.Severalchallengesneedtobeaddressedforthecaseoftheassemblylineforthemodularproductdesign,assub-cellsshouldprovidethedesiredlevelofflexibilityfordifferentmoduleproduc-tions[10].Technologiessuchasvisionand/orRFIDidentificationandcalibrationsystemsforrobotguidanceshouldbeintegratedallowingtheassemblyofe.g.moduleswithdifferentlengths[7].Weassumethatinterfacesbetweenthemodulesremainthesameinordertoallowforthemain-cellstoproduceallthreeunder-bodyvariants.4.ResultsanddiscussionForanalysingthebehaviouroftheassemblylinesdiscreteeventsimulationwasusedwiththehelpofacommerciallyavailabletool[11].Twodifferentsimulationmodelswerecreated,eachonerepresentingtheintegral(Fig.6)andmodular(Fig.7),(Fig.8)assemblyconfigurations.ThedemandscenarioofFig.4wasusedasthedailyordersinputtobothmodels.Thescenarioalsoassumedthreeshiftproduction(7.5h/shift)andeachsimulationrunwasexecutedoveraperiodof240workingdays(1year).Someassumptionsthatweremadefortheimplementationofthesimulationmodelsare:(i)geometryofthepartsisthesameforbothintegralandmodularunder-body,(ii)numberofpartsforeachunder-bodystructureisthesame,(iii)thenumberofthespotweldsisthesameforbothintegralandmodularunder-bodiesvariants.Additionalspotweldsforextendedunder-bodiesareassumednegligibleincomparisonwiththetotalnumberofspotswelds,thuscycletimesremainalsothesame,(iv)materialsofpartsforbothintegralandmodularunder-bodyarethesame.Resultsobtainedfromthesimulationofmodularandintegralconfigurationsinclude(a)theassemblysystemresponsetothesamedemandprofileforeachline,(b)theproduct’scumulativedelayoverthesimulationperiod,(c)theproduct’sdelay(indays)withrespecttotheshortestprocessingtime,(d)theestimationoftheassemblylineutilizationrateandfinally(e)theinvestmentcostperproduct.Thediagrams(Figs.9and10)indicatethatthesystemwiththeintegrallinesshowsabetterresponsivenesstothedemandprofile.Thisisattributedtothehigherresourcesavailabilityduetotheexistenceofthreeparallelassemblylines.Inthecaseofthemodularlinethesameresourcesareusedtoproduceallthreevariantsandthereforethedemandprofileforeachvariantcannotbeeasilymet.InFig.9thelinesrepresentingtheinput(blue)andtheoutput(red)ofthemodularassemblysystemcoincide.Themaximumcapacityofeachassemblysystemwhichis1850and2700partsperdayfortheintegralandmodularrespectively,isthemainfactoraffectingthisresponsiveness.ThisisindicatedbythebluelineinFig.10.ThecumulativedelayfortheproductionofmodularandintegralofthetwodifferentassemblyconfigurationswascalculatedandisshownforaperiodoftwoyearsinFig.11.Thedelayofeachproducthasbeencalculatedwithrespecttotheminimumpossibleprocessingtimeforeachvariant.Indicativeaveragevaluesforthedelay(indays)perproductare0.1(integral)and34.15(modular)forthetwodifferentassemblyconfigurations.Ifasecondassemblylineforthemodularproductdesignisadded,thenthecorrespondingdaysofdelayperpartwillbesignificantlyreduced.AnestimationoftheaverageutilizationratesofthetwodifferentassemblysystemswascalculatedbasedontheproductionresponsesFig.9.Productionresponsesforeachunder-bodyvariantformodular(left)andintegral(right)assemblysystemsconfigurations.168J.Paralikasetal./CIRPAnnals-ManufacturingTechnology60(2011)165–168Fig.10.Overallproductionresponsesformodular(left)andintegral(right)assemblylines.Fig.11.Cumulativedelayforintegralandmodularassemblylines.Fig.12.Costperunder-bodystructureformodularandintegralassemblysystemsconfigurations.tothedemandprofile(Fig.9).Resultsshowed44.92%averageutilizationofthethreelinesfortheintegralproductdesign(13.5%forUV1,67.4%forUV2,and53.9%forUV3)and91.82%utilizationrateoftheassemblylineforthemodularproductdesign.Inthecaseofthesystemfortheintegraldesign,thisisexplainedbytheuseofparallellineswhichsharetheworkload.Assemblycostisdividedintofixedandvariablecosts[12,3].Fixedcostsincludetheinvestment,tooling,building,maintenanceandoverheadcosts.Variablecostsincludetherawmaterials,labourandenergycost.Anestimationofinvestmentcostperproductiscalculated(Fig.12).Thenumberoflines,thenumberofrobotsperline,thecostofeachrobotandthenumberofpartsproducedperlineweretakenintoaccount.Theinvestmentcostperproductoftheassemblylineforthemodularproductdesignwassignificantlyreduced(34.38Euro/part)comparedtotheoneofthethreeintegrallines(300Euro/partforUV1,60Euro/partforUV2and75Euro/partforUV3).Theresultscouldbeobtainedonlythroughsimulationbecausetheproductionequipmentexhibitsastochasticbehaviourwithrandomfailures,etc.Moreover,cumulativedelayresults(Fig.11)cannotbeestimatedwithoutsimulationmodels,asthedailyordersareassumedtobevariableandrandom.5.ConclusionsInthisworktheproductionofanautomotiveBody-in-White(BiW)under-bodystructurewasinvestigatedandmodelled,forbothmodularandintegralproductdesign.Multipleperformancecriteriawereusedfortheevaluationofthetwoassemblysystems.Assemblysystemsthatproducemodularproductspresenthighproductflexibilitybutthiscomesattheexpenseofproductdelayandlowresponsivenesstomarketfluctuations.Ontheotherhandsystemsproducingintegralproductscanabsorbfluctuationsmoreeasilybutaresignificantlylesscostefficientandarepronetounder-utilizationdependingonthedemandprofilethattheyhavetomeet.Theidentificationofabreakevenpointbetweenthedegreeofmodularityandtheassemblysystemprofitabilityisaresearchtopicforthefuture.Giventhecurrenttrendtowardsdevelopingmodularproducts,futureworkshouldfocusonthedevelopmentofmethodsfortheoptimizationofsystemscapableforproducingmodularproducts.Issuessuchaslinebalancing,optimization,energyconsumption[13]andproductionresponsivenesstodemandshouldbethemainareasofinvestigation.Acknowledgements

TheauthorswouldliketoexpresstheirthankstoG.Michalosforreviewingtheassemblylinesmodelling.TheworkreportedinthispaperwaspartiallysupportedbyCEC/FP6NMPProgramme,‘‘IntegrationMulti-functionalmaterialsandrelatedproductiontechnologiesintegratedintotheAutomotiveindustryofthefuture-FUTURA’’.References

[1]PandremenosJ,ChatzikomisC,ChryssolourisG(2009)OntheQuantificationofInterfaceDesignArchitectures.AsianInternationalJournalofScienceandTech-nologyProductionandManufacturingEngineering2(3):41–48.[2]ChryssolourisG,PapakostasN,MavrikiosD(2008)APerspectiveonManu-facturingStrategy:ProduceMorewithLess.CIRPJournalofManufacturingScienceandTechnology1(1):45–52.[3]ParthaP,DattaR,Roy.(2010)CostModellingTechniquesforAvailabilityTypeServiceSupportContracts:ALiteratureReviewandEmpiricalStudy.CIRPJournalofManufacturingScienceandTechnology3(2):142–157.[4]ChryssolourisG(2006)ManufacturingSystems:TheoryandPractice.2ndedi-tion.Springer,NewYork.[5]RoyS,Hinduja.TetiR(2008)RecentAdvancesinEngineeringDesignOptimisa-tion:ChallengesandFutureTrends.CIRPAnnals–ManufacturingTechnology57.[6]TomiyamaT,GuP,JinY,LuttersD,KindCh,KimuraF(2009)DesignMeth-odologies:IndustrialandEducationalApplications.CIRPAnnals–Manufactur-ingTechnology58:543–565.[7]MichalosG,MakrisS,PapakostasN,MourtzisD,ChryssolourisG(2010)AutomotiveAssemblyTechnologiesReview:ChallengesandOutlookforaFlexibleandAdaptiveApproach.CIRPJournalofManufacturingScienceandTechnology2(2):81–91.[8]PandremenosJ,ParalikasJ,SalonitisK,ChryssolourisG(2009)ModularityConceptsfortheAutomotiveIndustry:ACriticalReview.CIRPJournalofManufacturingScienceandTechnology1:148–152.10.1016/j.cirpj.2008.09.012.[9]ScavardaLF,SchafferJ,ScavardaAJ,ReisA,SchleichH(2009)ProductVariety:AnAutoIndustryAnalysisandaBenchmarkingStudy.BenchmarkingAnInternationalJournal16(3):387–400.10.1108/14635770910961399.[10]GuP,RaoHA,TsengMM(2001)SystematicDesignofManufacturingSystemsBasedonAxiomaticDesignApproach.CIRPAnnals–ManufacturingTechnology50(1):299–304.[11]JohanssonJ(2009)AnalysingDiscreteEventSimulationModellingActivitiesAppliedinManufacturingSystemDesign.Proceedingsofthe19thCIRPDesignConference–CompetitiveDesign,30–31.[12]MichalosG,MakrisS,ChryssolourisG(2008)AnApproachtoAutomotiveAssemblyCostModelling.2ndCIRPConferenceonAssemblyTechnologiesandSystems,.[13]HerrmannC,ThiedeS(2009)ProcessChainSimulationtoFosterEnergyEfficiencyinManufacturing.CIRPJournalofManufacturingScienceandTech-nology1(4):221–229.