LES kOmegaSSTSAS / looking for RASProperties
I'm trying to run a sim using LES kOmegaSSTSAS but facing this strange error / anybody experienced that before ? Thanks
Create mesh for time = 0
Reading field p
Reading field U
Reading/calculating face flux field phi
Selecting incompressible transport model Newtonian
Selecting turbulence model type LESModel
Selecting LES turbulence model kOmegaSSTSAS
Courant Number mean: 0.003328494907 max: 0.04563852373
Starting time loop
Time = 0.1
Courant Number mean: 0.003328494907 max: 0.04563852373
deltaT = 0.1
smoothSolver: Solving for Ux, Initial residual = 1, Final residual = 0.0001447730244, No Iterations 1
smoothSolver: Solving for Uy, Initial residual = 1, Final residual = 0.004463225608, No Iterations 1
smoothSolver: Solving for Uz, Initial residual = 1, Final residual = 0.003346085113, No Iterations 1
GAMG: Solving for p, Initial residual = 1, Final residual = 0.04815958026, No Iterations 4
GAMG: Solving for p, Initial residual = 0.05136590869, Final residual = 0.001241589961, No Iterations 1
time step continuity errors : sum local = 7.87914026e-05, global = -9.572829191e-06, cumulative = -9.572829191e-06
GAMG: Solving for p, Initial residual = 0.007659904341, Final residual = 0.0001102955053, No Iterations 2
GAMG: Solving for p, Initial residual = 0.006646878643, Final residual = 3.869521047e-07, No Iterations 6
time step continuity errors : sum local = 2.623671733e-08, global = -3.116971313e-09, cumulative = -9.575946162e-06
smoothSolver: Solving for k, Initial residual = 0.9999999999, Final residual = 0.0009974693262, No Iterations 1
--> FOAM FATAL ERROR:
request for RASModel RASProperties from objectRegistry region0 failed
available objects of type RASModel are
It's pretty weird. OF is asking RASModel type.
It might be worked in different machine. Anyway, it's a strange bug.
Did you solve this problem?
I encounter the same problem. ^_^
I think I found what's the problem.
Maybe you used the wrong wall treatment in omega as me, did you ues "omegaWallFunction"?
This BC isn't defined in LES models, I think.
Yes, this is because you have used wall functions in 0/k and/or 0/omega
So which patchField type (wall treatment) are we supposed to use for omega?
It isn't obvious to me which patchField types to use.
I checked the tutorials (OF2.0.0 and OF1.7.1) and couldn't find anything involving kOmegaSSTSAS.
I'm using "kqRWallFunction" for the "k" patchField type, and the solver (OF2.0.0 pimpleFoam and pisoFoam) successfully runs through the k-equation, but "bombs" on the omega-equation (as noted above by Pierre-Olivier Dallaire)
Did you ever resolve this problem? I am looking at it now because one of the students in our group is using this model and came to me with a question on this. I would recommend two possible solutions,
1. use fixedValue for k and omega and make sure that the mesh is sublayer resolved so that you can use classical values near wall values.
2. hack the code and adapt the RAS b.c. for omegaWallFunction to your kOmegaSSTSAS model.
Did anyone tried, as Eric suggested, to adapt the omegaWallFunction to the kOmegaSSTSAS turbulence model ?
As for now, I use a viscous-resolved grid and a fixedValue for omega, but it is quite tricky in some cases.
omegaWallFunction for kOmegaSSTSAS
I've tried to implement LES-version omegaWallFunction
for kOmegaSSTSAS model as follows.
1. copy RAS/derivedFvPatchFields/wallFunctions/omegaWallFunctions/omegaWallFunction to LES/derivedFvPatchFields/wallFunction
2. replace "RASModel" to "LESModel" in source code of omegaWallFunction
3. modify LESModel.H
// Member functions
//- Return the near wall distances
const nearWallDist& y() const
4. modify kOmegaSSTSAS.C
void kOmegaSSTSAS::correct(const tmp<volTensorField>& gradU)
volScalarField G("LESModel::G", nuSgs_*0.5*S2);
Am I correct?
I think your coding of the omegaWallFunction should be correct.
But how do you set your BC for omega in kOmegaSSTSAS.C (in fact, you don't)?
In the RAS-version this is done by:
and you set it to the matrix after the omega equation:
I am not sure if you can transfer the procedure from RAS-SST to SAS, because the arrangement of the omega and k-equation is not the same!
Has anybody an opinion about that?
BTW: I think the G is also wrong in version 2.0 (correct me if I am wrong)
G should be like this:
volScalarField G = nuSgs_*S2;
and not divided by 2
G should be the same as in RAS!
PS: @Masashi and all SAS-users: have you validated the SAS model against a test case?
Thank you for your reply.
I agree with you about the notation of G.
But I've modified these code without much thought.
And I've checked my implementation with attached example.
Strange behavior of k and omega
It's me again,
I've tested the kOmegaSSTSAS model on the channel flow Re_tau=395, initialized with a LES simulation.
I encountered a strange behavior of the turbulent fields k and omega, after ~10 iterations those fields do NOT change any more, they stay completely unchanged. Also nuSgs does not change a lot any more (as it should corresponding to the turbulent structure).
Any opinion on that?
thanks for all your inputs concerning the kOmegaSSTSAS model. If I understand correctly, the model is still not usable for grids with near wall resolutions of y+>30, at least not without implementing a modified wall function.. Is that right?
I have reported the problem and I got a very detailed explanation about the issue.
bug report - kOmegaSSTSAS requests RASModel RASProperties
Thank you. I had read your bug report, and it indeed pretty much confirms that there is no High Reynolds models available for KomegaSSTSAS as of now, right?
Yeah, that's my final conclusion as well.
In that bug report mentioned above, it is stated, that this issue should be fixed in version 2.1.1. I'm still not able to use omegaWallFunction in LES. Did anyone of you check?
I'm still wondering why OpenFOAM doesn't complain about kqRWallFunction which isn't part of LES either.
I don't think near wall treatment for the kOmegaSSTSAS is yet available.
Hi all, I hope some of you are still active in this forum.
The issue is still the same as I understand it. The SAS model indeed accepts the omega wall function nowadays (I think since 2.2) but doesn't evaluate it at all. This is really confusing.
But: The two models differ just by this one Fsas term, right? So wouldn't it be sufficient, if we just make a copy of the RAS model directory and add this one term? All wall function usage would be correct then, I guess.
|All times are GMT -4. The time now is 20:22.|