CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM (http://www.cfd-online.com/Forums/openfoam/)
-   -   High speed flow with Launder-Sharma-k-epsilon-Model (http://www.cfd-online.com/Forums/openfoam/96958-high-speed-flow-launder-sharma-k-epsilon-model.html)

sebastian February 6, 2012 06:33

High speed flow with Launder-Sharma-k-epsilon-Model
 
Dear All,

I need to run a compressible high speed nozzle flow in OF on a low-Re mesh (y+<1).
So far I did my calculations with the kOmegaSST turbulence model and it worked quite nice.
Now I have to use the k-Epsilon turbulence model on the same mesh.
The only low-Re k-epsilon turbulence for a compressible case available in OF is the one by Launder and Sharma.
I compared the code of the LaunderSharmaKE with the standard kEpsilon model and found some differences as additional source-terms as well as some turbulent Reynolds-number functions.

So I have some questions, hoping anybody can answer these:
  1. Is it okay to use the Launder-Sharma k-e-model for such a case?
  2. How does this turbulence model behave in regions far away from the wall? Does it differ from the standard k-epsilon model or are thes modifications only due to near wall regions?
  3. As my y+-values are everywhere below 1.0, I can not use wall functions. Any other suggestions to solve this flow problem with a k-epsilon model?


Thanks in advance!
Sebastian

vwibaut March 11, 2013 17:03

Hi sebastian,

Did you find something? I have to simulate a flow in a nozzle and I would like to use this turbulence model.
Other question: Could you post your boundary condition for the k-omega case? To compare with my boundary conditions.

Thanks
Valentin

immortality June 6, 2013 02:40

hello Sebastian
Did you find out the proper answers?
And how did you find out which method is suitable for low-Re compressible?

immortality June 6, 2013 02:49

hi Valentin
do you want to simulate a low-Re turbulence too?could you send your case to me?
Thanks.

vwibaut June 6, 2013 03:31

3 Attachment(s)
Hi eshan :)

Yes I would like to make a low-Reynolds simulation in a nozzle. I attached a case with k-omegaSST model with boundary conditions for low-Reynolds simulations (I think).

immortality June 6, 2013 08:10

thanks Valentine
could you tell me about how choose and which formulas(if they are any) you used for obtaining omega value on walls and also inlet values?
and why didn't you assign mut to be calculated on walls also?
thanks so much.

vwibaut June 6, 2013 11:06

omega on walls: although for a k-eps model you have to set k = 0 and eps=0 on walls. According to my understanding for a k-omega omega isn't 0 on walls. If you use the omega WallFunction it will assign the correct condition on walls (omega = omegavis + omegalog. Omegalog will be zero as you use a low-Re mesh).

for the inlet: first I impose a uniform value that I calculated with an approximate solution. But I don't think the profile has to be uniform. So I impose condition for k wich calculates the value for k depending on U (k = 3/2 (I U)^2). After that I choose an inlet function for omega wich uses the formula omega = sqrt(k) / (Cmu^0.25 * l). So wich depends on k. To be consistent with the value of k.

For mut: Set mut as calculated would be ok. But I prefer to set mut = 0 as you are in the laminar part of the turbulent boundary layer.

immortality June 6, 2013 12:38

thanks so much valentine.was very useful.
a question occurred to me.should internalField better to be equal to inflow values or be equal to values have set on walls?
and how can do about high-Re study?how should set the values on the walls(and also internalField)?
thanks.

immortality June 6, 2013 12:43

dear Sebastian
have a look into it:http://www.openfoam.org/mantisbt/view.php?id=835#c2167
it looks there isn't yet an low-Re kOmegaSST model in OF as Dr.Weller told me.

vwibaut June 6, 2013 12:46

for high Reynolds, you have to choose wallfunction for k, epsilon (or omega). The choice of wallFunction is made in the mut file. You can choose spaldingWallFunction which allows mesh with y+< 30 (for my case it was useful)

For internalField I set a approximate value. It's not very important as it is just a value to initialize your case.

vwibaut June 6, 2013 12:49

Quote:

Originally Posted by immortality (Post 432479)
dear Sebastian
have a look into it:http://www.openfoam.org/mantisbt/view.php?id=835#c2167
it looks there isn't yet an low-Re kOmegaSST model in OF as Dr.Weller told me.

I see this in an other post but it seems that in several cases, k-omegaSST (with low-Re) works fine

immortality June 6, 2013 13:03

thanks.
3 questions:
1)in low-Re as you said omega should not be zero(1e-10) but how can specify an appropriate value there?is it approximate?
2)what do you mean by using wallFunction for omega in low-Re case while in low-Re maybe we don't have to use wall functions afaik
Quote:

If you use the omega WallFunction it will assign the correct condition on walls
3)in high-Re case how can set the values of k/omega/epsilon?

vwibaut June 6, 2013 13:30

1) & 2) omegaWallFunction assigns the good value for omega at the wall regardless the type of turbulence model (low or high)

3) Where? (inlet, outlet, wall)?

immortality June 6, 2013 13:38

in 1 question I wonder how can set an omega value on wall?(what value?a value around inflow?)
in 3 I mean on the wall too.
thank you for help.

immortality June 6, 2013 14:10

and do you mean that main setting for wall function is done in mut not other variables?and no need to set other variables (k/omega/epsilon) to spaldingWallFunction?
Quote:

for high Reynolds, you have to choose wallfunction for k, epsilon (or omega). The choice of wallFunction is made in the mut file.
:)
------------------------------
also this error occurred whan I used spaldingWallFunction.maybe it has removed in 2.2.0 version I use?
Code:

[1] --> FOAM FATAL IO ERROR:
[1] Unknown patchField type spaldingWallFunction for patch type wall

Valid patchField types are :

124
(
MarshakRadiation
MarshakRadiationFixedTemperature
advective
alphatJayatillekeWallFunction
alphatWallFunction
atmBoundaryLayerInletEpsilon
buoyantPressure
calculated
codedFixedValue
codedMixed
compressible::epsilonLowReWallFunction
compressible::epsilonWallFunction
compressible::fWallFunction
compressible::kLowReWallFunction
compressible::kqRWallFunction
compressible::omegaWallFunction
compressible::thermalBaffle1D<hConstSolidThermoPhysics>
compressible::thermalBaffle1D<hExponentialSolidThermoPhysics>
compressible::turbulentHeatFluxTemperature
compressible::turbulentMixingLengthDissipationRateInlet
compressible::turbulentMixingLengthFrequencyInlet
compressible::turbulentTemperatureCoupledBaffleMixed
compressible::turbulentTemperatureRadCoupledMixed
compressible::v2WallFunction
convectiveHeatTransfer
cyclic
cyclicAMI
cyclicSlip
directionMixed
empty
energyJump
energyJumpAMI
epsilonLowReWallFunction
epsilonWallFunction
externalWallHeatFluxTemperature
fWallFunction
fan
fanPressure
fixedEnergy
fixedFluxPressure
fixedGradient
fixedInternalValue
fixedJump
fixedJumpAMI
fixedMean
fixedPressureCompressibleDensity
fixedRho
fixedUnburntEnthalpy
fixedValue
freestream
freestreamPressure
gradientEnergy
gradientUnburntEnthalpy
greyDiffusiveRadiation
greyDiffusiveRadiationViewFactor
groovyBC
groovyBCDirection
groovyBCFixedValue
groovyBCJump
inletOutlet
inletOutletTotalTemperature
kLowReWallFunction
kappatJayatillekeWallFunction
kqRWallFunction
mapped
mappedField
mappedFixedInternalValue
mappedFixedPushedInternalValue
mixed
mixedEnergy
mixedFixedValueSlip
mixedUnburntEnthalpy
multiphaseFixedFluxPressure
mutLowReWallFunction
mutURoughWallFunction
mutUSpaldingWallFunction
mutUWallFunction
mutkRoughWallFunction
mutkWallFunction
nonuniformTransformCyclic
nutLowReWallFunction
nutTabulatedWallFunction
nutURoughWallFunction
nutUSpaldingWallFunction
nutUWallFunction
nutkAtmRoughWallFunction
nutkRoughWallFunction
nutkWallFunction
omegaWallFunction
oscillatingFixedValue
outletInlet
outletMappedUniformInlet
partialSlip
phaseHydrostaticPressure
processor
processorCyclic
rotatingTotalPressure
sliced
slip
smoluchowskiJumpT
symmetryPlane
syringePressure
timeVaryingMappedFixedValue
totalFlowRateAdvectiveDiffusive
totalPressure
totalTemperature
turbulentHeatFluxTemperature
turbulentInlet
turbulentIntensityKineticEnergyInlet
turbulentMixingLengthDissipationRateInlet
turbulentMixingLengthFrequencyInlet
uniformDensityHydrostaticPressure
uniformFixedValue
uniformJump
uniformJumpAMI
uniformTotalPressure
v2WallFunction
variableHeightFlowRate
wallHeatTransfer
waveSurfacePressure
waveTransmissive
wedge
wideBandDiffusiveRadiation
zeroGradient
)
[1]
[1]
[1] file: /home/ehsan/Desktop/Central/Turbulent_WR/main_test2/processor1/0/mut.boundaryField.walls from line 36 to line 40.
[1]
[1]    From function fvPatchField<Type>::New(const fvPatch&, const DimensionedField<Type, volMesh>&, const dictionary&)
[1]    in file /opt/openfoam220/src/finiteVolume/lnInclude/fvPatchFieldNew.C at line 136.
[1]
FOAM parallel run exiting
[1]
[0]
[0]
[0] file: /home/ehsan/Desktop/Central/Turbulent_WR/main_test2/processor0/0/mut.boundaryField.walls from line 36 to line 40.
[0]
[0]    From function fvPatchField<Type>::New(const fvPatch&, const DimensionedField<Type, volMesh>&, const dictionary&)
[0]    in file /opt/openfoam220/src/finiteVolume/lnInclude/fvPatchFieldNew.C at line 136.
[0] [3]
FOAM parallel run exiting

[3]

-----------------------
another trouble I had for you is that
could you please describe me a bit whats the differences between mutUwallFunction and mutkWallFunction?
thanks so much.:)

vwibaut June 12, 2013 09:40

In my opinion, typically for high Reynolds number model with the use of the spalding wallFunction: kwall = kqRWallFunction
eps wall: epsilonWallFunction
muwall: mutUSpaldingWallFunction


When you use a k-omega model, regardless you use a low-Re or high-Re model, you have to put omegaWallFunction in the file 0/omega

immortality June 12, 2013 09:53

thanks.:)
when I was using mutkWallFunction,and by yPlusRAS -compressible in high-Re simulation I obtained y+ values in a range of 20-45.is it OK in your opinion?
if y+ be lower than 30 does the accuracy decrease?
and what does spalding method do?

vwibaut June 12, 2013 10:31

According to the Fluent user guide, the performance of a high-Re model decreases when you have y+<30. So I think it is the same with OpenFOAM.
Spalding wallFunction allows y+<30. In my case it is interisting because we have a big range of y+ and we have to ensure that y+ < 200 (so some y+ comes smaller than 30)


All times are GMT -4. The time now is 19:18.