Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

Community>DiscussionForum>MeshQualitybasics

DiscussionForum
FilterbyTopic
Release
PopularTopics
MeshQualitybasics
AC/DCModule
Topics:4.0a 4.4
Threadindex|Previousthread|Nextthread|Startanewdiscussion
MEMS
4.3a
AcousticsModule
Mesh
RSSfeed|
4.3 Turnonemailnotifications|3RepliesLastpost:September10,20119:26amUTC
CFD
Microfluidics
4.2a
ChemicalReactionEngineering
Physics
4.2
Definitionsandoperators
September7,20108:02pmUTC
PorousMedia
4.1
FluidFlow
Resultsandvisualization
4.0a
Geometry
MeshQualitybasics
RFModule
4.0
HeatTransfer
Solving
3.5a
LiveLinkproducts
StructuralMechanics
AndrewPrudil
Hi, Materialsandfunctions
3.5

Listalldiscussions
I'vegotsomeprettybasicquestionsaboutmeshquality,butIcan'tfindanydirect
answerstothesequestions.
Whyisitimportant?Ifoundsourcesthatsaythatitisimportantforfast
convergenceofiterativesolvers,whataboutdirectsolvers?Arethereotherissues
suchassolutionaccuracy,stabilityandmeshinversion?Whatcauseslowquality
meshestobeproblematic?
Isitnecessarytoenforceaminimumelementquality?I'vereadyoushouldkeepit
above0.01butitdoesn'tsaywhathappensifyoudon't.
Dothesamemeshingrulesapplyfortriangularandquadrilateralmeshes?
Ifanyonecouldanswerthesequestionsitwouldbeofgreatassistance.Ihaven't
hadmuchluckfindingexplanationsaimedatthebeginnertointermediateusersof
FEA.
Reply|ReplywithQuote|Sendprivatemessage|ReportAbuse

September8,20105:32amUTCinresponsetoAndrewPrudil

Re:MeshQualitybasics
IvarKjelberg

Hi
meshingisanart:)andprobablyonecouldspendayearwithallspecificcases.
Inafewwords,generallyonewanttohavearegularmeshwithnodesroughlyon
thesurfaceofashereandevenlengthedges,suchthatsmalldisplacementsare
distributedamongallsubitemsinamoreorlessequalway,thisgivesbest

precision.Furthermore,thegradientofthevariablessolvedforshouldnotbetoo
largeovertheelementvolume,suchthatwegetagoodrepresentativityofthe
gradientanditschanges.ThinkothenyquistcriteriaforrepresentationofaSinus
wave,it'sverysimilar.
Inpracticeyoumeshbecomerathersqueezedor"distorted",ifyouapplytoolarge
deformationonanodeyoumighteven"invert"theelementthisisaserioustopology
changeasthesignofdisplacementschangesandtheresultsarewrong(atleast
locally).
Thenthemeshelementsareusedtomapshapefunctions(linearquadraticcubic...)
thehigherorderthemoreprecise,ifthemeshis"good",andthehigherorderthe
lessthemodelacceptspoormeshquality.ThatiswhyCOMSOlwarnsyouwhen
meshqualitybecomestoolow,itforcesfirstordershpaefunctionsonthese
elements.Orifthedisplacementsaretoolarge,thenittellsyouthatelementsare
inverted.
Thesewarningsaretheretotellyouthatthequalityoftheprepresentativityofyour
modelisindoubt,mostlyitsalocaleffect,butifitistherewhereyouhaveastress
concetrationeffectandwhereyouwouldexpectarupture,thenyouareinabad
situation,youmustdosomethingtoimprovethemeshquality(localhighermesh
density)
Triorquad,thetorbrick,thesearethenthegeometricalshapeoftheelements,
oftentheywerelinkedwiththeshapefunctionsappliedontheelements,COMSOL
treatsthisastwodifferentandindependentitems,whileclassicalprogrammeshave
atendencytolinktheshapefunctionstotheelementtypes.
PersonallyIusemostlytriangles(2D)andthethahedrals(3D)and2ndorder
elements(thedefault)anditworksfine,butwhenyouhavesymmetricparts
(typicallyrevolved)thenquad/brickscanbemadesmootherandmoresymmetrical
(typicallyforoptics,axesandothercylindricalsymmetryitemsetc).Envenin2D
axi,youmustusequadforanyelementtouchingthesymmetryaxis.
normallythehigherthenumberofnodesperelementsANDthehighershape
functions,ANDhigestmeshqualitygivesthemostpreciseresults,BUTittakesthe
longesttosolve(andsometimesoumightfindoutthatyouneedyourentirelifetime
togetthereonce!)
Sonaotherrule:usesymmetrytosimplifyyourmodel,thelesselementsthefasterit
solvesandthequickeryoucancheckyourmodel.Thereafteryoucanincreasethe
numberofmeshelementsforsomefinaltests,andcheckthatyougetamoreorless
constantresultout,independentofthemeshdensity
Hopethishelpsontehway

Goodluck
Ivar
Reply|ReplywithQuote|Sendprivatemessage|ReportAbuse

September9,20112:44pmUTCinresponsetoIvarKjelberg

Re:MeshQualitybasics
DanielNussbcher

ThxIvarforthisgreatdelineationofmeshing.
Ihaveanotherquestiononthistopic:
HowdoesabadelementqualityeffecttheaccuracyoftheresultsIobtain?
FromuniversityIknowyoushouldavoidmesheswithlowelementquality.
Usingthe"boundarylayer"or"refine"optionsInoticeabetterresolutionofthe
solutionbutwithasacrificeofmeshquality.
DoIgenerateworsevaluesthen???
Reply|ReplywithQuote|Sendprivatemessage|ReportAbuse

September10,20119:26amUTCinresponsetoDanielNussbcher

Re:MeshQualitybasics
IvarKjelberg

Hi
COMSOLusesfirstordershapeelementswhenthelowqualityorinvertedflag
appears,thismeansthattheseregionsmightbelessprecise,butthereisnotaclear
1:1rulehere.
Plottheelementquality,ifyourmin/maxdimensioningvaluesofyourdesigncomes
fromthesamecornerswherethemeshqualityisbad,thenIwouldnottrustthe
simulation,andwouldredoamshingcampaign
Ageneralcomment:usethelatestversionandnottheearly4.0,youwillbeless
frusttrated)

Goodluck
Ivar
Reply|ReplywithQuote|Sendprivatemessage|ReportAbuse

Rulesandguidelines

You might also like