Você está na página 1de 5

12/04/2016

Isthereawaytohandlesolverwarningslike&...| SOLIDWORKSForums

AllPlaces >Simulation>Discussions

5Replies Latestreplyon17Jun201510:12byShivaniPatel

HendriksHubbertz 08 Mar 201404:58

Isthereawaytohandlesolver
warningslike"zeropivot"problems?
ThisquestionisNotAnswered.

Helloeveryone,

isthereanywaytodecodesolverwarnings?Iwanttoknow,inwhichpartof
thegeometry(node/element)theproblemisexactly.
Forexampleawarning/messagecouldbelikethis:
PCGLSS0204:Negativeorzeropivot3.490117e+000atrow7284r)

OrcanIhavealookintotheresultfilesorlogfilestogetanidea,wherethe
problemisexactly?
UptonowIcanonlydorandomchangestothemethortheconstraintsuntil
thesimulationworkswithoutproblems.Itwouldbegreattogetsomeideas
whattochangetosavetimeiterating.

Bestregards

Ihavethesamequestion(0)

1548Views

Categories:Simulation

Tags:messages,solver,zero,pivot,logfile

MOSTLIKED

JaredConway 08 Mar 201416:37 (inresponsetoHendriksHubbertz)


Re:Isthereawaytohandlesolverwarningslike"zeropivot"problems?

norealwaytodigintothaterrorwithoutbeingadeveloper
butsomesuggestions:
1.lookitupinthesolidworksdiscussionforumandthesolidworksKb
2.reportittoyourreseller,atminimumtheycanreporttothedevelopersso
thatagooderrormessagecanbegenerated
3.useaniterativemethodtofindtheproblembutbeintelligentaboutit.errors
https://forum.solidworks.com/thread/79066

1/5

12/04/2016

Isthereawaytohandlesolverwarningslike&...| SOLIDWORKSForums

usuallycomeupfrombadgeometry,badmesh,orbadsetup.ifyoustart
simple(fixedbcandgravityandbondedcontacts)youwillbeabletoisolate
geometryandmeshrightawayforexample.

Actions

Like(0)

HendriksHubbertz 12 Mar 201404:59 (inresponsetoJaredConway)


Re:Isthereawaytohandlesolverwarningslike"zeropivot"problems?

Thanksfortheanswer.ItseemsthatIhavetokeepiterating(usinggravity
isagoodtip).

Actions

Like(1)

VincentJanzen 15 Jun 201513:21 (inresponsetoJaredConway)


Re:Isthereawaytohandlesolverwarningslike"zeropivot"problems?

Jared,ifyoudon'tmindmereopeningthisdiscussion,Iwonderifyouwould
givealittleclarificationofyouranswerabove?Wehavebeenfightingthese
errorsprettyextensively,onsomefairlylargeassemblies,andarelooking
foranythingthatmighthelptrackthemdown.Doyouhavesome
guidelinesorworkingdefinitionsyoucouldsuggestfor"badgeometry","bad
mesh"and"badsetup"?

Sofar,we'vebeenusinggeometryaslongasitdoesn'thaveanyofthe
verysmallfaces,etc.thatcausethemeshtofail.We'reusingameshifthe
aspectratioandJacobianratioseemreasonable.Setupismostlyamatter
ofvisualinspection,tomakesurethevariouscontactsetsseemcorrect,
andsuch.

Doyouknowofanyothergeneralcheckswecouldbedoing,especiallywith
regardtothegeometryandthemesh,thatmayhelpustogetourmodel
cleaner?

Thanks.

Actions

Like(0)

ShivaniPatel 17 Jun 201510:12 (inresponsetoVinc entJanzen)


Re:Isthereawaytohandlesolverwarningslike"zeropivot"
problems?

https://forum.solidworks.com/thread/79066

2/5

12/04/2016

Isthereawaytohandlesolverwarningslike&...| SOLIDWORKSForums

Thisisn'texactlyananswer,butit'stheprocessIgothroughwhen
dealingwithunstableassemblies.

Beforehand,ImakesureIhavenogaps,orImakenoteofgapsthat
shouldbewelded/bondedandcreatethosecontacts.
Ialsorecreatecontactsbetweenbeams/solids,beams/shells,and
solids/shells.

IfI'mgettingnosolutionduetoinstability,likezeropoint,Itry"soft
springs",thenafrequencystudywithsamesetup,thenselectmost
componentsinthetreeand"excludefromanalysis".Thethirdwayis
iterativeandcouldtakeawhile,butithelpsmenarrowdownwhichparts
arecausingtheinstability.UsuallyIgetluckywithsoftspringsandcan
seewhichpartsare"flyingoffintospace".

Actions

Like(0)

ShaunDensberger 12 Mar 201414:10 (inresponsetoHendriksHubbertz)


Re:Isthereawaytohandlesolverwarningslike"zeropivot"problems?

Inthefiniteelementmethod,youhaveageneralequation:

{f}=[K]{d}

where:

1.{f}isyourforcevector(yourappliedandunknownloads).
2.[K]istheglobalstiffnessmatrixofthesystem.
3.{d}isyourdisplacementvector(yourconstraintsandunknown
displacements).

Typically,asolverwilluseGaussianeliminationtouppertriangularizethe
stiffnessmatrixtofindyourdisplacementvectorsolutionthepivotterminthe
stiffnessmatrixisthefirstnonzeroterm.Azeropivotmeansthatyou'vegiven
thesolverasystemthatcannotbesolveddeterministically(mathematically,this
meansthatyourstiffnessmatrixisnotpositivedefinite).Ifwetakeasimple
springelementofstiffnesskN/manddonotapplyanyconstraintstothe
model,thenwe'llgetazeropivotafterdoingtheGaussianeliminationtoupper
triangularizethestiffnessmatrix.

https://forum.solidworks.com/thread/79066

3/5

12/04/2016

Isthereawaytohandlesolverwarningslike&...| SOLIDWORKSForums

Anegativepivotisusuallycalledasmallpivot,asthishappenswhenthepivot
indexissomeverysmallnumber(+/1e20),whichcomesaboutbecauseit's
incrediblyrareforacomputertocalculatezerowhendoingGaussianelimination.

Thatbeingsaid,thereareusuallytworeasonsforazeroorsmallpivot:

1.Yourmodelisnotconstrainedproperly.
2.Yourmodelhasaverylargedifferenceinstiffnessbetweenneighboring
elements(thisresultsinapivotthatgetsclosetotheroundofferrorof
thecomputer).

Thankfully,theaboveissuesare(usually)easytosolve:

1.Makesurethatyourmodelisconstraintagainstrigidbodymotion.Ifyou
haveanassembly,makesurepartsareconnected.
2.Don'taddinelementsthathaveverylargedifferencesintheirstiffness
whencomparedtootherelementsinthemodel(8ordersofmagnitudeor
so).Forexample,youdon'twanttoconnectaspringwithastiffnessof
1e1N/mtoanotherspringofstiffness1e12N/m.

"UptonowIcanonlydorandomchangestothemethortheconstraintsuntilthesimulationworks
withoutproblems."

Areyourandomlychangingyourconstraintsbecauseyoudon'tknowwhat
constraintsareneededtofixallDOF?Ifit'sanassembly,areyou
changing/addingconstraintstoadditionalparts?

Actions

https://forum.solidworks.com/thread/79066

Like(4)

4/5

12/04/2016

Isthereawaytohandlesolverwarningslike&...| SOLIDWORKSForums

2014DassaultSystemes
SolidworksCorp.

Followus
Home|TopofPage|Termsof
Use|PrivacyPolicy

Home | Topofpage | Help

https://forum.solidworks.com/thread/79066

2016JiveSoftware | Poweredby

5/5

Você também pode gostar