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

《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（25页珍藏版）》请在冰点文库上搜索。

ABAQUS关于固有频率的提取方法

Abaqus固有频率提取

6.3.5 Naturalfrequencye*traction

Products:

Abaqus/Standard  Abaqus/CAE  Abaqus/AMS  

References

∙"Procedures:

overview,〞 

∙"Generalandlinearperturbationprocedures,〞 

∙"Dynamicanalysisprocedures:

overview,〞 

∙*FREQUENCY

∙"Configuringafrequencyprocedure〞in"Configuringlinearperturbationanalysisprocedures,〞 Section14.11.2oftheAbaqus/CAEUser'sManual

Overview

Thefrequencye*tractionprocedure:

∙performseigenvaluee*tractiontocalculatethenaturalfrequenciesandthecorrespondingmodeshapesofasystem;

∙willincludeinitialstressandloadstiffnesseffectsduetopreloadsandinitialconditionsifgeometricnonlinearityisaccountedforinthebasestate,sothatsmallvibrationsofapreloadedstructurecanbemodeled;

∙willputeresidualmodesifrequested;

∙isalinearperturbationprocedure;

∙canbeperformedusingthetraditionalAbaqussoftwarearchitectureor,ifappropriate,thehigh-performanceSIMarchitecture（see "UsingtheSIMarchitectureformodalsuperpositiondynamicanalyses〞in"Dynamicanalysisprocedures:

overview,〞 ）;and

∙solvestheeigenfrequencyproblemonlyforsymmetricmassandstiffnessmatrices;theple*eigenfrequencysolvermustbeusedifunsymmetriccontributions,suchastheloadstiffness,areneeded.

Eigenvaluee*traction

Theeigenvalueproblemforthenaturalfrequenciesofanundampedfiniteelementmodelis

where

isthemassmatri*（whichissymmetricandpositivedefinite）;

 isthestiffnessmatri*（whichincludesinitialstiffnesseffectsifthebasestateincludedtheeffectsofnonlineargeometry）;

istheeigenvector（themodeofvibration）;and

Mand N

aredegreesoffreedom.

When 

ispositivedefinite,alleigenvaluesarepositive.Rigidbodymodesandinstabilitiescause 

tobeindefinite.Rigidbodymodesproducezeroeigenvalues.Instabilitiesproducenegativeeigenvaluesandoccurwhenyouincludeinitialstresseffects.Abaqus/Standardsolvestheeigenfrequencyproblemonlyforsymmetricmatrices.

Selectingtheeigenvaluee*tractionmethod

Abaqus/Standardprovidesthreeeigenvaluee*tractionmethods:

∙Lanczos

∙Automaticmulti-levelsubstructuring（AMS）,anadd-onanalysiscapabilityforAbaqus/Standard

∙Subspaceiteration

Inaddition,youmustconsiderthesoftwarearchitecturethatwillbeusedforthesubsequentmodalsuperpositionprocedures.Thechoiceofarchitecturehasminimalimpactonthefrequencye*tractionprocedure,buttheSIMarchitecturecanoffersignificantperformanceimprovementsoverthetraditionalarchitectureforsubsequentmode-basedsteady-stateortransientdynamicprocedures（see "UsingtheSIMarchitectureformodalsuperpositiondynamicanalyses〞in"Dynamicanalysisprocedures:

overview,〞 ）.Thearchitecturethatyouuseforthefrequencye*tractionprocedureisusedforallsubsequentmode-basedlineardynamicprocedures;youcannotswitcharchitecturesduringananalysis.Thesoftwarearchitecturesusedbythedifferenteigensolversareoutlinedin Table6.3.5–1.

Table6.3.5–1Softwarearchitecturesavailablewithdifferenteigensolvers.

SoftwareArchitecture

Eigensolver

Lanczos

AMS

SubspaceIteration

Traditional

SIM

TheLanczossolverwiththetraditionalarchitectureisthedefaulteigenvaluee*tractionmethodbecauseithasthemostgeneralcapabilities.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）fileinanAMSfrequencye*tractionstep.

Ifyourmodelhasmanydegreesoffreedomandtheselimitationsareacceptable,youshouldusetheAMSeigensolver.Otherwise,youshouldusetheLanczoseigensolver.TheLanczoseigensolverandthesubspaceiterationmethodaredescribedin"Eigenvaluee*traction,〞 Section2.5.1oftheAbaqusTheoryManual.

Lanczoseigensolver

FortheLanczosmethodyouneedtoprovidethema*imumfrequencyofinterestorthenumberofeigenvaluesrequired;Abaqus/Standardwilldetermineasuitableblocksize（althoughyoucanoverridethischoice,ifneeded）.Ifyouspecifyboththema*imumfrequencyofinterestandthenumberofeigenvaluesrequiredandtheactualnumberofeigenvaluesisunderestimated,Abaqus/Standardwillissueacorrespondingwarningmessage;theremainingeigenmodescanbefoundbyrestartingthefrequencye*traction.

Youcanalsospecifytheminimumfrequenciesofinterest;Abaqus/Standardwille*tracteigenvaluesuntileithertherequestednumberofeigenvalueshasbeene*tractedinthegivenrangeorallthefrequenciesinthegivenrangehavebeene*tracted.

See "UsingtheSIMarchitectureformodalsuperpositiondynamicanalyses〞in"Dynamicanalysisprocedures:

overview,〞 ,forinformationonusingtheSIMarchitecturewiththeLanczoseigensolver.

Input File Usage:

*FREQUENCY,EIGENSOLVER=LANCZOS

Abaqus/CAE Usage:

Stepmodule:

 Step

Create:

 Frequency:

 Basic:

 Eigensolver:

 Lanczos

ChoosingablocksizefortheLanczosmethod

Ingeneral,theblocksizefortheLanczosmethodshouldbeaslargeasthelargeste*pectedmultiplicityofeigenvalues（thatis,thelargestnumberofmodeswiththesamefrequency）.Ablocksizelargerthan10isnotremended.Ifthenumberofeigenvaluesrequestedis n,thedefaultblocksizeistheminimumof（7, n）.Thechoiceof7forblocksizeprovestobeefficientforproblemswithrigidbodymodes.ThenumberofblockLanczosstepswithineachLanczosrunisusuallydeterminedbyAbaqus/Standardbutcanbechangedbyyou.Ingeneral,ifaparticulartypeofeigenproblemconvergesslowly,providingmoreblockLanczosstepswillreducetheanalysiscost.Ontheotherhand,ifyouknowthataparticulartypeofproblemconvergesquickly,providingfewerblockLanczosstepswillreducetheamountofin-corememoryused.Thedefaultvaluesare

Blocksize

Ma*imumnumberofblockLanczossteps

1

80

2

50

3

45

≥4

35

Automaticmulti-levelsubstructuring（AMS）eigensolver

FortheAMSmethodyouneedonlyspecifythema*imumfrequencyofinterest（theglobalfrequency）,andAbaqus/Standardwille*tractallthemodesuptothisfrequency.Youcanalsospecifytheminimumfrequenciesofinterestand/orthenumberofrequestedmodes.However,specifyingthesevalueswillnotaffectthenumberofmodese*tractedbytheeigensolver;itwillaffectonlythenumberofmodesthatarestoredforoutputorforasubsequentmodalanalysis.

Thee*ecutionoftheAMSeigensolvercanbecontrolledbyspecifyingthreeparameters:

and

.Thesethreeparametersmultipliedbythema*imumfrequencyofinterestdefinethreecut-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.Ateachsupernodealocaleigensolutionisobtainedbasedonfi*ingthedegreesoffreedomconnectedtothene*thigherlevelsupernode（thesearethelocalretainedor"fi*ed-interface〞degreesoffreedom）.Atthe