ABAQUS关于固有频率的提取方法.docx
《ABAQUS关于固有频率的提取方法.docx》由会员分享,可在线阅读,更多相关《ABAQUS关于固有频率的提取方法.docx(21页珍藏版)》请在冰豆网上搜索。
ABAQUS关于固有频率的提取方法
Abaqus固有频率提取
Naturalfrequencyextraction
Products:
Abaqus/Standard Abaqus/CAE Abaqus/AMS
References
∙“Procedures:
overview,〞
∙“Generalandlinearperturbationprocedures,〞
∙“Dynamicanalysisprocedures:
overview,〞
∙*FREQUENCY
∙“Configuringafrequencyprocedure〞in“Configuringlinearperturbationanalysisprocedures,〞
Overview
Thefrequencyextractionprocedure:
∙performseigenvalueextractiontocalculatethenaturalfrequenciesandthecorrespondingmodeshapesofasystem;
∙willincludeinitialstressandloadstiffnesseffectsduetopreloadsandinitialconditionsifgeometricnonlinearityisaccountedforinthebasestate,sothatsmallvibrationsofapreloadedstructurecanbemodeled;
∙willputeresidualmodesifrequested;
∙isalinearperturbationprocedure;
∙canbeperformedusingthetraditionalAbaqussoftwarearchitectureor,ifappropriate,thehigh-performanceSIMarchitecture(see “UsingtheSIMarchitectureformodalsuperpositiondynamicanalyses〞in“Dynamicanalysisprocedures:
overview,〞 );and
∙solvestheeigenfrequencyproblemonlyforsymmetricmassandstiffnessmatrices;theplexeigenfrequencysolvermustbeusedifunsymmetriccontributions,suchastheloadstiffness,areneeded.
Eigenvalueextraction
Theeigenvalueproblemforthenaturalfrequenciesofanundampedfiniteelementmodelis
where
isthemassmatrix(whichissymmetricandpositivedefinite);
isthestiffnessmatrix(whichincludesinitialstiffnesseffectsifthebasestateincludedtheeffectsofnonlineargeometry);
istheeigenvector(themodeofvibration);and
M and N
aredegreesoffreedom.
When
ispositivedefinite,alleigenvaluesarepositive.Rigidbodymodesandinstabilitiescause
tobeindefinite.Rigidbodymodesproducezeroeigenvalues.Instabilitiesproducenegativeeigenvaluesandoccurwhenyouincludeinitialstresseffects.Abaqus/Standardsolvestheeigenfrequencyproblemonlyforsymmetricmatrices.
Selectingtheeigenvalueextractionmethod
Abaqus/Standardprovidesthreeeigenvalueextractionmethods:
∙Lanczos
∙Automaticmulti-levelsubstructuring(AMS),anadd-onanalysiscapabilityforAbaqus/Standard
∙Subspaceiteration
Inaddition,youmustconsiderthesoftwarearchitecturethatwillbeusedforthesubsequentmodalsuperpositionprocedures.Thechoiceofarchitecturehasminimalimpactonthefrequencyextractionprocedure,buttheSIMarchitecturecanoffersignificantperformanceimprovementsoverthetraditionalarchitectureforsubsequentmode-basedsteady-stateortransientdynamicprocedures(see “UsingtheSIMarchitectureformodalsuperpositiondynamicanalyses〞in“Dynamicanalysisprocedures:
overview,〞 ).Thearchitecturethatyouuseforthefrequencyextractionprocedureisusedforallsubsequentmode-basedlineardynamicprocedures;youcannotswitcharchitecturesduringananalysis.Thesoftwarearchitecturesusedbythedifferenteigensolversareoutlinedin –1.
–1 Softwarearchitecturesavailablewithdifferenteigensolvers.
SoftwareArchitecture
Eigensolver
Lanczos
AMS
SubspaceIteration
Traditional
SIM
TheLanczossolverwiththetraditionalarchitectureisthedefaulteigenvalueextractionmethodbecauseithasthemostgeneralcapabilities.However,theLanczosmethodisgenerallyslowerthantheAMSmethod.TheincreasedspeedoftheAMSeigensolverisparticularlyevidentwhenyourequirealargenumberofeigenmodesforasystemwithmanydegreesoffreedom.However,theAMSmethodhasthefollowinglimitations:
∙AllrestrictionsimposedonSIM-basedlineardynamicproceduresalsoapplytomode-basedlineardynamicanalysesbasedonmodeshapesputedbytheAMSeigensolver.See “UsingtheSIMarchitectureformodalsuperpositiondynamicanalyses〞in“Dynamicanalysisprocedures:
overview,〞 ,fordetails.
∙TheAMSeigensolverdoesnotputepositemodaldampingfactors,participationfactors,ormodaleffectivemasses.However,ifparticipationfactorsareneededforprimarybasemotions,theywillbeputedbutarenotwrittentotheprinteddata(.dat)file.
∙YoucannotusetheAMSeigensolverinananalysisthatcontainspiezoelectricelements.
∙Youcannotrequestoutputtotheresults(.fil)fileinanAMSfrequencyextractionstep.
Ifyourmodelhasmanydegreesoffreedomandtheselimitationsareacceptable,youshouldusetheAMSeigensolver.Otherwise,youshouldusetheLanczoseigensolver.TheLanczoseigensolverandthesubspaceiterationmethodaredescribedin“Eigenvalueextraction,〞 .
Lanczoseigensolver
FortheLanczosmethodyouneedtoprovidethemaximumfrequencyofinterestorthenumberofeigenvaluesrequired;Abaqus/Standardwilldetermineasuitableblocksize(althoughyoucanoverridethischoice,ifneeded).Ifyouspecifyboththemaximumfrequencyofinterestandthenumberofeigenvaluesrequiredandtheactualnumberofeigenvaluesisunderestimated,Abaqus/Standardwillissueacorrespondingwarningmessage;theremainingeigenmodescanbefoundbyrestartingthefrequencyextraction.
Youcanalsospecifytheminimumfrequenciesofinterest;Abaqus/Standardwillextracteigenvaluesuntileithertherequestednumberofeigenvalueshasbeenextractedinthegivenrangeorallthefrequenciesinthegivenrangehavebeenextracted.
See “UsingtheSIMarchitectureformodalsuperpositiondynamicanalyses〞in“Dynamicanalysisprocedures:
overview,〞 ,forinformationonusingtheSIMarchitecturewiththeLanczoseigensolver.
Input File Usage:
*FREQUENCY,EIGENSOLVER=LANCZOS
Abaqus/CAE Usage:
Stepmodule:
Step
Create:
Frequency:
Basic:
Eigensolver:
Lanczos
ChoosingablocksizefortheLanczosmethod
Ingeneral,theblocksizefortheLanczosmethodshouldbeaslargeasthelargestexpectedmultiplicityofeigenvalues(thatis,thelargestnumberofmodeswiththesamefrequency).Ablocksizelargerthan10isnotremended.Ifthenumberofeigenvaluesrequestedis n,thedefaultblocksizeistheminimumof(7, n).Thechoiceof7forblocksizeprovestobeefficientforproblemswithrigidbodymodes.ThenumberofblockLanczosstepswithineachLanczosrunisusuallydeterminedbyAbaqus/Standardbutcanbechangedbyyou.Ingeneral,ifaparticulartypeofeigenproblemconvergesslowly,providingmoreblockLanczosstepswillreducetheanalysiscost.Ontheotherhand,ifyouknowthataparticulartypeofproblemconvergesquickly,providingfewerblockLanczosstepswillreducetheamountofin-corememoryused.Thedefaultvaluesare
Blocksize
MaximumnumberofblockLanczossteps
1
80
2
50
3
45
≥4
35
Automaticmulti-levelsubstructuring(AMS)eigensolver
FortheAMSmethodyouneedonlyspecifythemaximumfrequencyofinterest(theglobalfrequency),andAbaqus/Standardwillextractallthemodesuptothisfrequency.Youcanalsospecifytheminimumfrequenciesofinterestand/orthenumberofrequestedmodes.However,specifyingthesevalueswillnotaffectthenumberofmodesextractedbytheeigensolver;itwillaffectonlythenumberofmodesthatarestoredforoutputorforasubsequentmodalanalysis.
TheexecutionoftheAMSeigensolvercanbecontrolledbyspecifyingthreeparameters:
and
.Thesethreeparametersmultipliedbythemaximumfrequencyofinterestdefinethreecut-offfrequencies.
(defaultvalueof5)controlsthecutofffrequencyforsubstructureeigenproblemsinthereductionphase,while
and
(defaultvaluesof1.7and1.1,respectively)controlthecutofffrequenciesusedtodefineastartingsubspaceinthereducedeigensolutionphase.Generally,increasingthevalueof
and
improvestheaccuracyoftheresultsbutmayaffecttheperformanceoftheanalysis.
Requestingeigenvectorsatallnodes
Bydefault,theAMSeigensolverputeseigenvectorsateverynodeofthemodel.
Input File Usage:
*FREQUENCY,EIGENSOLVER=AMS
Abaqus/CAE Usage:
Stepmodule:
Step
Create:
Frequency:
Basic:
Eigensolver:
AMS
Requestingeigenvectorsonlyatspecifiednodes
Alternatively,youcanspecifyanodeset,andeigenvectorswillbeputedandstoredonlyatthenodesthatbelongtothatnodeset.Thenodesetthatyouspecifymustincludeallnodesatwhichloadsareappliedoroutputisrequestedinanysubsequentmodalanalysis(thisincludesanyrestartedanalysis).Ifelementoutputisrequestedorelement-basedloadingisapplied,thenodesattachedtotheassociatedelementsmustalsobeincludedinthisnodeset.putingeigenvectorsatonlyselectednodesimprovesperformanceandreducestheamountofstoreddata.Therefore,itisremendedthatyouusethisoptionforlargeproblems.
Input File Usage:
*FREQUENCY,EIGENSOLVER=AMS,NSET=name
Abaqus/CAE Usage:
Stepmodule:
Step
Create:
Frequency:
Basic:
Eigensolver:
AMS:
Limitregionofsavedeigenvectors
ControllingtheAMSeigensolver
TheAMSmethodconsistsofthefollowingthreephases:
Reductionphase:
InthisphaseAbaqus/Standardusesamulti-levelsubstructuringtechniquetoreducethefullsysteminawaythatallowsaveryefficienteigensolutionofthereducedsystem.Theapproachbinesasparsefactorizationbasedonamulti-levelsupernodeeliminationtreeandalocaleigensolutionateachsupernode.Startingfromthelowestlevelsupernodes,weuseaCraig-Bamptonsubstructurereductiontechniquetosuccessivelyreducethesizeofthesystemasweprogressupwardintheeliminationtree.Ateachsupernodealocaleigensolutionisobtainedbasedonfixingthedegreesoffreedomconnectedtothenexthigherlevelsupernode(thesearethelocalretainedor“fixed-interface〞degreesoffreedom).Attheendofthereductionphasethefullsystemhasbeenreducedsuchthatth