Sie sind auf Seite 1von 23

Lecture9:

BestPracticeGuidelines
16.0Release

IntroductiontoANSYSFluent

1 2015ANSYS,Inc. April18,2016
Introduction
LectureTheme:
TheaccuracyofCFDresultscanbeaffectedbydifferenttypesoferrors.By
understandingthecauseofeachdifferenterrortype,bestpracticescanbedeveloped
tominimizethem.Meshingplaysasignificantroleintheefforttominimizeerrors.
LearningAims:
Youwilllearn:
Fourdifferenttypesoferrors
Strategiesforminimizingerror
Issuestoconsiderduringmeshcreationsuchasqualityandcelltype
Bestpracticesformeshcreation
LearningObjectives:
Youwillunderstandthecausesoferrorinthesolutionandhowtobuildthemeshand
performthesimulationinamannerthatwillminimizeerrors
Introduction ErrorTypes BestPracticesforMeshing Summary
2 2015ANSYS,Inc. April18,2016
MotivationforQuality
CFDResultsareusedformanydifferentstagesofthedesignprocess:
Design&optimizationofcomponentsandmachines
Safetyanalyses
Virtualprototypes

WhenundertakingaCFDmodel,considerationshouldbegiventothepurposeofthe
work:
Whatwilltheresultsbeusedfor?
Whatlevelofaccuracywillbeneeded?

Introduction ErrorTypes BestPracticesforMeshing Summary


3 2015ANSYS,Inc. April18,2016
DifferentSourcesofError
Thereareseveraldifferentfactorsthatcombinetoaffecttheoverallsolution
accuracy.Inorderofmagnitude:
Roundofferrors
Computerisworkingtoacertainnumericalprecision
Iterationerrors
Differencebetweenconvergedsolutionandsolutionatiterationn
Solutionerrors
Differencebetweenconvergedsolutiononcurrentgridandexactsolutionof
modelequations
Exactsolution Solutiononinfinitelyfinegrid
Modelerrors
Differencebetweenexactsolutionofmodelequationsandreality(dataoranalytic
solution)
Introduction ErrorTypes BestPracticesforMeshing Summary
4 2015ANSYS,Inc. April18,2016
RoundOffError
Inaccuraciescausedbymachineroundoff: Tip:Lookfor"dp"inthetitlebaroftheFluentwindowor
Highgridaspectratios thelowerrightcornerofthegraphicswindowtocheckif
Largedifferencesinlengthscales yourcurrentsessionisusingdoubleprecision

Largevariablerange

Howtoidentifyifroundofferrorisaproblem:
Calculatewithdoubleprecisionif yourcasemeets
theabovecriteria
Compareresultswithasolutionthathasbeen
calculatedwithsingleprecision
Ifimportantquantities(targetvariables)are
different,doubleprecisionshouldbeusedfor
subsequentcalculations

Introduction ErrorTypes BestPracticesforMeshing Summary


5 2015ANSYS,Inc. April18,2016
IterationError BestPractice
Definequantityorquantitiesofinterestforyoursimulation(targetvariables):
Headrise
Efficiency
Massflowrate

Selectconvergencecriterionfortheresiduals
Plotthevalueofthequantitiesofinterestasthesolutioniterates
Selectatighterconvergencecriterionandcontinueiteratingandplotting
Repeatuntilthevaluesofthequantitiesofinterestnolongerchange
Thiswillidentifywhatresiduallevelitisnecessarytoachieveinordertoensurethesolutionisfreefrom
iterationerror
Monotonicconvergenceofresiduals(nextslide)isdesirable,althoughnotalwayspossible
Reportmassandenergyfluxestoensurethesearebeingconserved

Introduction ErrorTypes BestPracticesforMeshing Summary


6 2015ANSYS,Inc. April18,2016
IterationErrorExample:2DCompressorCascade
Relativeerror:
Iterationerrors:

IsentropicEfficiency
0.01%
0.18% Differencebetween
convergedsolutionand
solutionatiterationn
(Residual)

Convergence
criterion

Checkformonotonicconvergence Rmax=102 Rmax=103 Rmax=104


Iteration35 Iteration59 Iteration132

IterationNumber
Introduction ErrorTypes BestPracticesforMeshing Summary
7 2015ANSYS,Inc. April18,2016
Discretization Error
Alldiscrete methodshavesolutionerrors:
Finitevolumemethods
Finiteelementmethods
Finitedifferencemethods
...

Differencebetweensolutiononagivengridandexactsolutiononaninfinitelyfinegrid

Exactsolutionnotavailable Discretization errorestimation

Introduction ErrorTypes BestPracticesforMeshing Summary


8 2015ANSYS,Inc. April18,2016
Discretization ErrorEstimation
Impingingjetflowwithheattransfer

2D,axisymmetric DD=26.5mmor101.6mm

ComparedGrids: H
50 50 800 800
SSTturbulencemodel r

Targetquantities:
Discretization schemes:
1st orderUpwind Heattransfer
2nd orderUpwind MaximumNusselt number

Introduction ErrorTypes BestPracticesforMeshing Summary


9 2015ANSYS,Inc. April18,2016
Discretization ErrorEstimation
1st order 2nd order
200
Theplotshows
Ifthegridisfineenough,
190 st nd
1 and2 ordersolutions
arethesame
Nu_max

180 Oncoarsermeshes,the2nd
ordersolutioniscloserto
thefinalsolution
170
Practicalalternativesfor
industrialcasesare:
160
Comparesolutionsfrom
differentorderschemes
150 Comparesolutionson
-3,47E-17 0,005 0,01 0,015 0,02 locallyorregionallyrefined
1/N_Cells meshes
Introduction ErrorTypes BestPracticesforMeshing Summary
10 2015ANSYS,Inc. April18,2016
ModelErrors
Inadequaciesof(empirical)mathematicalmodels:
Baseequations(Eulervs.RANS,steadystatevs.unsteadystate,)
Turbulencemodels
Combustionmodels
Multiphaseflowmodels

Discrepanciesbetweendataandcalculationsremain,evenafterallnumerical
errorshavebecomeinsignificant!

Introduction ErrorTypes BestPracticesforMeshing Summary


11 2015ANSYS,Inc. April18,2016
ModelError:ImpingingJet
SKE RNG KW

TKEContours

Nu* Results:H/D=2,RE=23000 Notehowthepredictionsdiffer


SKE dependingonwhichturbulencemodelis
used
RNG
KW Thek model(KW)performsbetterthan
thestandard(SKE)orRNGk modelsin
Modelerror thiscase
Thek basedmodelsoverestimatethe
productionofturbulenceatthestagnation
point,causingthepredictedNusselt number
tobetoohigh

Introduction ErrorTypes BestPracticesforMeshing Summary


12 2015ANSYS,Inc. April18,2016
SystematicErrors
Discrepanciesremain
Evenifnumericalandmodelerrorsareinsignificant
Systematicerrors:
Approximationsof:
Geometry
Componentvs.machine
Boundaryconditions
Fluidandmaterialproperties,

Trytounderstandapplicationandphysics

Documentanddefendassumptions!

Performuncertaintyanalysis
Introduction ErrorTypes BestPracticesforMeshing Summary
13 2015ANSYS,Inc. April18,2016
MeshingBestPracticeGuidelines
Choosingyourmeshstrategy
dependson

1.ACCURACY 2.EFFICIENCY 3.EASINESSTOGENERATE

Desiredmeshquality Desiredcellcount Timeavailable


Whatisthemaximum Lowcellcountforresolving FasterTetdominantmesh
skewness andaspect overallflowfeaturesvs High vs craftedHex/hybridmesh
ratioyoucantolerate? cellcountforgreaterdetails withlowercellcount

Goal:Findthebestcompromisebetweenaccuracy,efficiencyand
easinesstogenerate
Introduction ErrorTypes BestPracticesforMeshing Summary
14 2015ANSYS,Inc. April18,2016
Meshing:CaptureFlowPhysics
Gridmustbeabletocapture Recommendedmeshingguidelinesfor
importantphysics: boundarylayers
Boundarylayers Boththevelocityand thermalboundary
Heattransfer layersmustberesolved
Wakes,shock Thereshouldbeaminimumof1015
elementsacrosstheboundarylayer
Flowgradients
thickness
Themeshexpansionratiointhewall
normaldirectionshouldbemoderate:
1.21.3
y+ 1forheattransferandtransition
modeling
Introduction ErrorTypes BestPracticesforMeshing Summary
15 2015ANSYS,Inc. April18,2016
Meshing:CaptureFlowPhysics
Example:Velocityprofilesatairfoil
Bad Good

Introduction ErrorTypes BestPracticesforMeshing Summary


16 2015ANSYS,Inc. April18,2016
MeshQuality
Agoodmeshdependson: Good NotGood

Cellnottoodistorted

Cellnottoostretched

SmoothCellstransition

Introduction ErrorTypes BestPracticesforMeshing Summary


17 2015ANSYS,Inc. April18,2016
MeshQuality
Gridgeneration:
Scalablegrids
Skewness <0.95(accuracy,convergence)
alsoworstOrthogonalQuality>.01andaverage
valuemuchhigher
Aspectratios <100
Expansionratios<1.52
Capturephysicsbasedonexperience(shearlayers, Badcells
shocks)
Anglebetweengridface&flowvector
Concrete,quantitativerecommendationsforthese
NoBadcells
factorspresentedintheIntroductiontoANSYS
Meshingcourseareincludedintheappendixofthis
presentation
Gridrefinement:
Manual,basedonerrorestimate
Automaticadaptivebasedonerrorsensor Adaption

Introduction ErrorTypes BestPracticesforMeshing Summary


18 2015ANSYS,Inc. April18,2016
MeshQuality
Avoidsuddenchangesinmeshdensity

Notgood Good

Introduction ErrorTypes BestPracticesforMeshing Summary


19 2015ANSYS,Inc. April18,2016
Hexvs Tet Mesh:AccuracyComparison
Directionoftheflowwellknown
Quad/HexalignedwiththeflowaremoreaccuratethanTriwith
thesameintervalsize

U=0.1 Hex Tri


mesh mesh

U=1.0

Contoursofaxialvelocitymagnitudeforaninviscid coflowjet

Introduction ErrorTypes BestPracticesforMeshing Summary


20 2015ANSYS,Inc. April18,2016
Hexvs Tet Mesh:Accuracycomparison
Forcomplexflowswithoutdominantflowdirection,QuadandHex
mesheslosetheiradvantage
Quad&Triequivalent

qua

U=V=1.0,T=1
U=V=1.0, T=1

d tri

U=V=1.0, T=0 U=V=1.0, T=0


Contoursoftemperatureforinviscid flow
Introduction ErrorTypes BestPracticesforMeshing Summary
21 2015ANSYS,Inc. April18,2016
Summary
Trytounderstandapplicationandphysicsoftheapplication
Distinguishbetweennumerical,modelandothererrors
Documentanddefendassumptions
Geometry
Boundaryconditions
Flowregime (laminar,turbulent,steadystate,unsteadystate,)
Modelselection(turbulence,)
Sourcesofsystematicerror
Approximations
Data
Accuracyexpectationsvs.assumptions?
Introduction ErrorTypes BestPracticesforMeshing Summary
22 2015ANSYS,Inc. April18,2016
Resources
ERCOFTACSIG:QuantificationofUncertaintyinCFD
Roache,P.J.,VerificationandValidationinComputationalScienceandEngineering,
HermosaPublishers,1998
OntheANSYScustomerportal,searchfor"bestpractice"

Introduction ErrorTypes BestPracticesforMeshing Summary


23 2015ANSYS,Inc. April18,2016

Das könnte Ihnen auch gefallen