Abstract — The paper demonstrates the model order reduction procedures applied to semiconductor devices with multiple heat sources. The approach is demonstrated for a device with nine heat sources where some of them are permanently active and other work u
ModelReductionforPowerElectronicsSystems
withMultipleHeatSources
A.Augustin ,T.Hauck ,B.Maj§,J.Czernohorsky§,E.B.Rudnyi andJ.G.Korvink
Semiconductor,Schatzbogen7,D-81829Munich,Germany
AutomotiveSystemsAG,P.O.Box900120,D-60441Frankfurt,Germany IMTEK,LaboratoryforSimulation,UniversityofFreiburg,Georges-Koehler-Allee103,D-79110Freiburg,GermanyEmail:adam.augustin@freescale.com,bartosz.maj@contiautomotive.com,korvink@imtek.de§Continental FreescaleAbstract—Thepaperdemonstratesthemodelorderre-
ductionproceduresappliedtosemiconductordeviceswith
multipleheatsources.Theapproachisdemonstratedfora
devicewithnineheatsourceswheresomeofthemareperma-
nentlyactiveandotherworkunderswitchingconditions.For
theorderreductionthesoftwarepackageMORforANSYS
isused,whichisbasedontheKrylovsubspacemethodvia
theArnoldialgorithm.
I.INTRODUCTION
Modernsemiconductortechnologiesallowthedevelop-
mentofpowerelectronicssystemswithmanifoldfunction-
alityintegratedononesingledie.Duetothiseverlasting
process,thedevicesincludemoreandmoreheatsources,
whichhavetobeconsideredinthermalsimulations.For
complexpackageandheatsourcecon gurationsaswell
asfordifferentapplications,accuratethermalmodelsare
needed.
Asmodulecon gurationisclosetoamultilayeredblock
structure,oneofthesimulationapproachesisbasedonan
analyticalsolutionoftheheattransferequation[1]–[3].
Obviously,realgeometryisdifferentfromtheidealized
geometryrequiredbytheanalyticalsolutionandthe nite
elementmethodseemstobetheonlyvalidalternative
fordetailedthermalsimulationsinthegeneralcase(see
forexample[4]).Unfortunately,thebiggestdisadvantage
ofthisapproachishighsimulationtimebecauseofhigh
dimensionalmodelswithhundredsthousandsdegreesof
freedom.Onepossiblewaytoovercomethisproblemisto
usemathematicalmethodsinordertoreducethesystem
orderwithaminimumlossofaccuracy.
Aformalmodelreductionapproach[5](overviewfrom
engineeringviewpointin[6])allowsustotakeahigh
dimensional niteelementmodelandgenerateitslow-
dimensionalapproximation.Assuch,itisanidealcan-
didateforthegoalabove.Severalresearchgroupshave
alreadydocumenteditssuccessfulapplicationtoathermal
problem[7]–[10].
However,theformalmodelreductionapproachalsoputs
somelimitsfortheoriginalthermalproblem.Forexample,
ithappensthatitsapplicationisnotstraightforwardfor
coolingofelectroniccomponentsasconsideredbythe
DELPHIandPROFITconsortium[11]–[14].Themainreasonisthattheuseof lmcoef cientsislimitedforforcedconvectionwithinavolumewithcomplexgeo-metricalconstraints.Inthiscase,onecannotavoidajointthermal-CFDsimulationandcompactthermalmodelsshouldpossessspecialproperties.Inourcase,thesituationisdifferent.85%ofheatisremovedbyconductionandthereisnoforcedconvection.Theremaining15%ofheatisremovedbynaturalcon-vection.Additionally, lmcoef cientshavebeenvalidatedbymeansofexperimentalmonitoringoftemperature.Asaresult,formalmodelreductioncanbeemployedforthethermalproblemwithoutanyfurthermodi cation.Finally,weshouldstressthatdynamicbehaviormustbepreservedinourcase.Thestructureofthispaperisasfollows.InsectionIIashortintroductionintothemodelorderreductiontheorywillbepresentedtogetherwiththereductionprocessfordeviceswithmultipleheatsources.InsectionIIIthenecessarypreparatorystepsforcalculationswithreducedmodelswillbeexplainedincludingtheconsiderationofsomepermanentactiveheatsources.InsectionIVthepresentedsimulationresultswillbediscussed.Finally,insectionVsomeconcludingremarkswillbemade.II.MODELREDUCTIONPROCESSUsingthe niteelementsimulationsoftwareANSYS,ageometricalandphysicalmodelisgenerated.Themodelshouldconcernallthermallyrelevantcomponentstorep-resenttherealdeviceasgoodaspossible.A.GeneralTheoryAfterthediscretization,the niteelementmodelisexpressedasfollows:EdT(t)
Model Reduction in Power Systems Using Krylov Subspace Methods_材料科学_工程...This argument is generalized for multiple interpolation points by taking the ...
(Heverlee) MODEL REDUCTION FOR CONTROL DESIGN Jury...(DT) systems the set of all FD LTI CT (DT)...6.4.7 Weighting functions with multiple zeroes. ...
Automatica 41 (2005) 273 – 279 www.elsevier.com/locate/automatica Brief paper Controller reduction for linear parameter-varying systems with a priori ...
1 Introduction Model order reduction of linear large-scale dynamic systems is...When multiple-input is to be preserved in Eq (1) or (3), a use ...
Model Reduction for Powe... 暂无评价 5页 免费 ...wiresizing for interconnects with multiple sources ...Power Optimization", IEEE interconnect trees and ...
Efficient Balance-and-Truncate Model Reduction for Large Scale Systems_电子/电路...To implement the inverse power iterations, we began with an LU factorizatio...
electronics and heat sinks: – Products requiring ...with multiple samples is not considered usable for...Limited reduction in uncertainty at 30 units and ...
Model Reduction at Leuven Measurement Systems(PPT)_电子/电路_工程科技_专业资料...Multiple D.o.F. System: f (t) 1 ground k1 m 1 c 1 x (t) 1 ...
Model Reduction of Large-Scale Systems(书)_中医中药_医药卫生_专业资料。Model Reduction of Large-Scale Systems CESAME (UCL) AND CSIT (FSU) ? Model Red...
Projection methods for model reduction of large scale dynamical systems(演示文稿...For multiple points ? Krylov based methods – Moment matching without moment...
热门文档