NRELisanationallaboratoryoftheU.S.DepartmentofEnergy,OfficeofEnergyEfficiencyandRenewableEnergy,operatedbytheAllianceforSustainableEnergy,LLC.PredictiveModelsofLi-ionBatteryLifetimeKandlerSmith,Ph.D.EricWood,ShriramSanthanagopalan,Gi-HeonKim,YingShi,AhmadPesaranNationalRenewableEnergyLaboratoryGolden,ColoradoIEEEConferenceonReliabilityScienceforAdvancedMaterialsandDevicesColoradoSchoolofMines•Golden,Colorado•September7-9,2014NREL/PR‐5400‐628132KeyMessages•Semi-empiricalbatterylifetimemodelsaregenerallysuitableforsystemdesign&controloLong-termvalidationstillneededoStandardizationwouldbenefitindustryoCharacterizationrequiresexpensivecellagingexperiments•PhysicslifetimemodelsareneededtoreducetesttimeaswellasguidefuturecelldesignoOpenquestionsremainhowbesttomodelelectrochemo-thermo-mechanicalprocessesacrosslength-andtime-scalesLiuetal.,J.EchemSoc.(2010)NRELLife&MSMDModels3ThermalNRELElectrochemical/Thermal/LifeModelsMulti-ScaleMulti-Domain(MSMD)model•Inter-domaincouplingoffieldvariables,sourceterms•Efficient,flexibleframeworkforphysicsexpansion•Leadingapproachforlarge-cellcomputer-aidedengineeringmodelsKimetal.(2011)“Multi-DomainModelingofLithium-IonBatteriesEncompassingMulti-PhysicsinVariedLengthScales”,J.ofElectrochemistry,Vol.158,No.8,pp.A955–A969ElectrochemicalCurrentCollector(Cu)CurrentCollector(Al)pNegativeElectrodeSeparatorPositiveElectrode4Life-predictivemodel•Physics-basedsurrogatemodelstunedtoagingtestdata•Implementedinsystemdesignstudies&real-timecontrol•RegressiontoNCA,FeP,NMCchemistries15°C20°C25°C30°C10°CMinneapolisHoustonPhoenixLi-iongraphite/nickelatelife:PHEV20,1cycle/day54%∆DoDNocoolingAircoolingAircooling,lowresistancecellPhoenix,AZambientconditions33miles/daydriving,2trips/dayLiquidcooling,chilledfluidIllustrationbyJoshBauer,NRELNRELElectrochemical/Thermal/LifeModelsNCA=Nickel-Cobalt-AluminumFeP=IronPhosphateNMC=Nickel-Manganese-Cobalt5Outline•Background–Li-ionBatteriesoWorkingprinciplesoElectrochemicalwindowoDegradationmechanisms•LifePredictiveModeling•AutomotiveLifeStudies&Control6Li+VWorkingPrinciplesVNeg.ElectrodeGraphiteHardcarbonSiliconTitanateLifoilPos.ElectrodeLiXO2,X=NiMnCoCoNiCoAlLiMn2O4,LiFeP047ElectrochemicalOperatingWindowPotentialvs.Li(V)SOC(xinLixC6oryinLi1-0.6yCoO2)0%SOC100%SOCPotentialmeasuredatcellterminalschargedischargeFigurecredit:IlanGur(ARPA-E)&VenkatSrinivasan(LBNL),20078ElectrochemicalWindow–DegradationPotentialvs.Li(V)SOC(xinLixC6oryinLi1-0.6yCo)O2)Figurecredit:IlanGur(ARPA-E)&VenkatSrinivasan(LBNL)20079NegativeElectrodeDegradation(Graphite)GraphenelayerGraphiteexfoliation,cracking(gasformation,solventco-intercalation)ElectrolytedecompositionandSEIformationSEIconversion,stabilizationandgrowthSEIdissolution,precipitationPositive/negativeinteractionsLithiumplatingandsubsequentcorrosionDonorsolventSEILi+1)Manufacturingenvironment2)Applicationenvironmenti)gracefulfade(timeathighT,SOC)ii)suddenfade/damage(cyclingatlowT)Figurecredit:Vetteretal.,JournalPowerSources,2005Negative/electrolyteinteractions10PositiveElectrodeDegradation(MetalOxide)Positive/negativeinteractionsX.Xiaoetal.,EchemComm.,32(2013)31-34.Positive/electrolyteinteractionsFigurecredit:Vetteretal.,JournalPowerSources,200511MechanicalCoupledStress&Degradation•LeastunderstoodamongstECTMphysicsFigurecredit:Santhanagopalan,Smith,etal.,ArtechHouse(inpress)12MechanicalCoupledStress&Degradation•ExamplesCannarella,J.PowerSources(2014)Diercks,Packard,Smithetal.,J.EchemSoc(2014)Figurecredit:Santhanagopalan,Smith,etal.,ArtechHouse(inpress)13Outline•Background–Li-ionBatteries•LifePredictiveModelingoPhysics-basedoSemi-empirical•AutomotiveLifeStudies&Control14DegradationMechanismvs.LengthScale10-1010-810-610-410-210-0Chemistry•SEIgrowth•Liplating•Electrolytedecomposition•GasgenerationParticlescale•SEIμ-cracking•Fracture,damageoftransportpaths•Phaseevolution,voltagedroopElectrodescale•Electrodecreep,delamination,isolation•Separatorporeclosure•PorecloggingCellscale•3Delec,thermal,mech.inhomogeneities•Tabeffects•Stack/windModulescale•Thermal&mechanicalboundaryconditions15Macro-scaleStressModel10-1010-810-610-410-210-0•Stress/strainduetothermalandelectrodebulkconcentrationchanges•Coupledechem&thermalBehrou,Maute,Smith,ECSMtg.(2014)16Micro-scaleStress/DegradationModel10-1010-810-610-410-210-0DamageevolutionPerformanceimpactAn,Barai,Smith,Mukherjee,JES201417NRELLifePredictiveModel•Data:J.C.Hall,IECEC,2006.17RelativeCapacity(%)Time(years)r2=0.942Li-ionNCAchemistryArrhenius-Tafel-Wohlermodeldescribinga2(∆DOD,T,V)NCA•Correctseparationofcalendarvs.cyclingdegradationforextrapolationof½yeartestingto10+yearlife•Extensibletountesteddrivecycles,environments(stateform)RelativeResistanceRelativeCapacityQsites=c0+c2N+…R=a1tz+a2NCalendarfade•SEIgrowth•Lossofcyclablelithium•Coupledwithcycling•a1,b1=f(∆DOD,T,Voc,…)Q=min(QLi,Qsites)QLi=b0+b1tz+…Cyclingfade•Activematerialstructuredegradationandmechanicalfracture•a2,c2=f(∆DOD,T,Voc,…)NCA18KneeinFadeCriticalforPredictingEndofLifeExamplesimulation:1cycle/dayat25°C50%DOD:GracefulfadecontrolledbyLiloss~t1/280%DOD:Transitionstoelectrodesiteloss,N~2300cyclesLifeover-predictedby25%withoutaccountingfortransitionfromLiloss(~chemical)tositeloss(~mechanical)NCATime(Days)19ElectrodeSiteLoss–CellAgingData•Graphite/iron-phosphatemeta-datasetfrommultiplelabsFeP20ElectrodeSiteLoss–CellAgingData•Graphite/iron-phosphatemeta-datasetfrommultiplelabs•13of50+testconditionsshowapparent“knee”incapacityfadecurveFeP21ElectrodeSiteLossModel(graphite/ironphosphate)()()[]()()().expexp,,intercal.binder11R32111,22−+∆+−=−−refpulsepulserefrateraterefarefattCCTTETTRErefmTmDODmcc).,min(sitesLiqqq=NbtbbqzLi210++=Nccqsites20+=acceleratedpolymerfailureathighTbulkintercalationstrainbulkthermalstrainintercalationgradientstrain,acceleratedbylowtemperatureFePModelsuccessfullydescribes13agingconditionsfrom0°Cto60°C22Outline•Background–Li-ionBatteries•LifePredictiveModeling•AutomotiveLifeStudies&ControloTemperature(xEV)oChargecontrol(xEV/grid)oPrognostic/duty-cyclecontrol(xEV)23AmbientEffectsonBatteryAverageTemperature05101520253035Phoenix,AZLosAngeles,CAMinneapolis,MNTemperature(C)BEV75PassivecoolingAverageAmbientw/solareffectsw/drivingeffects05101520253035Phoenix,AZLosAngeles,CAMinneapolis,MNTemperature(C)PHEV35ChilledliquidcoolingAverageAmbientw/solareffectsw/driving+activeTMS•Ambientconditionsdominate•Thermalconnectionwithpassengercabin,parkinginshadedstructuresstronglyinfluencebatterylife•Batterytemperatureandlifetimeweaklycoupledtoambientconditions24OptimizedChargingStrategies•ReducetimespentathighSOC(delaycharging)•AvoidhighC-ratestolowerpeaktemperaturesCostfunctionCU-Boulder/Hoke(2014)25OptimizedChargingStrategies•DelayedchargingbestBeginchargeEndchargeA)Constantenergycost•Responsetopricesignals•NoV2Genergyexporteduntilelectricityprice$0.50/kWhBeginchargeEndchargeB)Variableenergycost26NRELARPA-EAMPEDProjectsinBatteryControlUtahState/FordProject:20%reductioninPHEVpackenergycontentviapowershuttlingsystemandcontrolofdisparatecellstohomogenousend-of-lifeNREL:Requirementsanalysis;lifemodelofFord/Panasoniccell;controlsvalidationofFordPHEVpacksEatonCorporationProject:DownsizedHEVpackby50%throughenablingbatteryprognostic&supervisorycontrolwhilemaintainingsameHEVperformance&lifeNREL:Lifetesting/modelingofEatoncells;controlsvalidationonEatonHEVpacksWashingtonUniv.Project:Improveavailableenergyatthecelllevelby20%basedonreal-timepredictivemodeling&adaptivetechniquesNREL:Physics-basedcell-levelmodelsforMPC;implementWUreformulatedmodelsonBMS;validateatcell&modulelevelAMPED=AdvancedManagementandProtectionofEnergyStorageDevices27NRELARPA-EAMPEDProjectsinBatteryControlUtahState/FordProject:20%reductioninPHEVpackenergycontentviapowershuttlingsystemandcontrolofdisparatecellstohomogenousend-of-lifeNREL:Requirementsanalysis;lifemodelofFord/Panasoniccell;controlsvalidationofFordPHEVpacks1CEnergy(Wh)Teardownanalysisofautomotivepackagedto70%remainingenergyshows±5.5%variationatEOL±5.5%variationatEOL•Activebalancingmostbenefitsenergyapplicationswithlargecell-to-celldisparity•Keyquestion:Howmuchdoescell-to-celldisparitygrowwithage?Extremedrivingconditions,largepackΔTcauseabnormallylargecellcapacityimbalancegrowth28Summary•MainfactorscontrollingbatterylifetimeoTimeathighT&SOC(weakcouplingwithDOD&C-rate)oCyclingathighDOD&C-rate;Low/highT&SOC•Semi-empiricalbatterylifetimemodelsaregenerallysuitableforsystemdesign&controloNRELmodelsdescribevariouscommercialchemistriesoLifeextensionsof20%to50%maybepossible•PhysicslifetimemodelstoprovidedesignfeedbackoElectrochemical/thermalprocesseswellunderstoodoMechanicscouplingunderwayatvariouslengthscales29Funding•USDept.ofEnergy–VehicleTechnologiesOffice:BrianCunningham,DavidHowell•USDept.ofEnergy–AdvancedResearchProjectsAgency:IlanGur,PatrickMcGrath,RusselRoss•USArmyTankAutomotiveResearch,Development&EngineeringCenter:YiDing,SonyaZanardelli,MattCastanierCollaborators•TexasA&MUniversity:ParthaMukherjee,PallabBarai,KaiAn•UniversityofColoradoatBoulder:KurtMaute,RezaBehrou,DraganMaksimovic,AndrewHoke•ColoradoSchoolofMines:CorinnePackard,DavidDiercks,BrianGorman•EatonCorporation:ChinmayaPatil•UtahState:ReganZane•FordMotorCompany:DycheAndersonAcknowledgements30Thankyou