CFD Online Discussion Forums - Compressible WALE model

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)

- OpenFOAM Programming & Development (https://www.cfd-online.com/Forums/openfoam-programming-development/)

- - Compressible WALE model (https://www.cfd-online.com/Forums/openfoam-programming-development/96885-compressible-wale-model.html)

Here is an updated compressible dynamic smagorinsky LES model with the inclusion of a SGS K equation solved by feeding the locally dynamic muSgs coefficient into the K equation.

This code was utilized for OpenFOAM 1.7, and it needs to be slightly modified for OpenFOAM 2.1.

Cheers,
Deji

Thanks a lot. We implemented your old version in OpenFOAM 2.1 and it works great. We would surely modify it and use this new version. I undertand that this is exactly the same at the previous 'dynLocalAverageSmagorinsky' except for the change you specified. Is there something else that's new in this new version !

Thanks again for all the help.

Sagnik

Dear Deji,

Could you please also nicely share us the Make folder (files and options) for the dynamic Smag Model:

dynLocalAverageSmagorinsky.C

Thank you so much!

flames

Hello, everyone!

I tried to complied the dynLocalAverageSmagorinsky Model into openfoam 2.1.1. But I still met some problem when I run my case.

I put the .H and .C in the following path:

Code:

/home/user/OpenFOAM/user-2.1.1/src/turbulenceModels/compressible/LES/dynLocalAverageSmagorinsky

In the LES folder, the file structure is:

Code:

dynLocalAverageSmagorinsky (folder), Make (folder)

My Make/files is:

Code:

dynLocalAverageSmagorinsky/dynLocalAverageSmagorinsky.C



LIB = $(FOAM_USER_LIBBIN)/libcompressibleLESModels_user

Make/options:

Code:

EXE_INC = \

    -I$(LIB_SRC)/turbulenceModels \

    -I$(LIB_SRC)/turbulenceModels/LES/LESdeltas/lnInclude \

    -I$(LIB_SRC)/turbulenceModels/LES/LESfilters/lnInclude \

    -I$(LIB_SRC)/thermophysicalModels/basic/lnInclude \

    -I$(LIB_SRC)/meshTools/lnInclude \

    -I$(LIB_SRC)/finiteVolume/lnInclude \

    -I$(LIB_SRC)/turbulenceModels/compressible/LES/GenEddyVisc \

    -I$(LIB_SRC)/turbulenceModels/compressible/LES/LESModel \

    -I$(LIB_SRC)/turbulenceModels/compressible/LES/lnInclude



LIB_LIBS = \

    -lcompressibleTurbulenceModel \

    -lLESdeltas \

    -lLESfilters \

    -lfiniteVolume \

    -lmeshTools

The library libcompressibleLESModels_user.so is obtained in the following path (I confirmed that the .so file is successfully generated.):

Code:

/home/user/OpenFOAM/user-2.1.1/platforms/linux64IccDPOpt/lib

In the case, I add the following line in the system/control:

Code:

libs ("libcompressibleLESModels_user.so");

The file LESProperties is:

Code:

LESModel        dynLocalAverageSmagorinsky;    //Smagorinsky;    //oneEqEddy;



delta           cubeRootVol;



printCoeffs     on;



dynLocalAverageSmagorinskyCoeffs

{

     ce            1.048;

     filter        simple;

}

.........

When I run the case, I still had the problem as follows:

Code:

[0]

[0]

[0] --> FOAM FATAL ERROR:

[1] [2]

[2]

[2] --> FOAM FATAL ERROR:

[2] Unknown LESModel type dynLocalAverageSmagorinsky



Valid LESModel types:



15

(

DeardorffDiffStress

LRRDiffStress

Smagorinsky

SpalartAllmaras

SpalartAllmarasDDES

SpalartAllmarasIDDES

dynLagrangian

dynOneEqEddy

homogeneousDynOneEqEddy

homogeneousDynSmagorinsky

kOmegaSSTSAS

laminar

mixedSmagorinsky

oneEqEddy

spectEddyVisc

)

[2]

[2]

[2]     From function LESModel::New(const volVectorField&, const surfaceScalarField& ,transportModel&, const word&)

[2]     in file LESModel/LESModel.C at line [0] Unknown LESModel type dynLocalAverageSmagorinsky

It seems that the solver does not know the newly generated library, however I do not know which procedure I neglected to correctly use this turbulence model. Does anybody have any suggestions about the above implementations? Thank you in advance.

best regards,
f

++++++++++++++++++++++++++++++++++++++++++++++++++ +++++++++++++++++++++++++++++++++++++++++++++++++=
The above procedure should be correct. The reason why the error comes from is that I use pisoFoam to test the LES model, which obviously cannot be found by pisoFoam. A stupid error.

Dear Deji,

In the source code of dynLocalAverageSmagorinsky.C, the resolved turbulent stresses L_ij is calculated as:

Code:

volSymmTensorField Lij = filter_(rho()*sqr(U())) - sqr(filter_(rho()*U()))/filter_(rho());

Actually it is not completely equivalent to the equation in the paper by Martin, Piomelli and Canlder. Using the above equation, actually the following assumption is assumed:

$\overline{\rho f} = \overline{\rho}\overline{f}$

Did you test this assumption? I know that in compressible L_ij is easy to calculate in compressible LES, but as you know in openfoam it is not actually compressible. At least in my case it is Low Ma solver. I ask this just for curiosity. Thanks.

Nobody knows something about the errors which make the compiliation not successfully? Thanks.

Dynamic SGS model

Since it seems you are interested in more sophisticated SGS models, please have a look at the LEMOS extensions for OpenFOAM: Dynamic Mixed Model, Lagrangian Dynamic Mixed Model and Hybrid URANS-LES models with dynamic Smagorinsky :-)
http://www.lemos.uni-rostock.de/en/d...nfoam-content/

Thank you for your information. Are they for compressible or incompressible LES?

cheers,
f

sorry, my mistake, only incompressible as far as I know

Dear Deji,

About the calculating the dynamic parameters Cd and Ci for the Smagorinksy compressible LES SGS model, I think I have the following difference for my implementation:

First CI, the beta should be calculated as:

Code:

volScalarField beta = 2*sqr(2*delta())*filter_(rho())*sqr(mag(filter_(rho()*D)/filter_(rho())));

In the prediction of Cd, the strain rate tensor are treated not a Favre way. This is not consistent with the Paper by Martin, Piomelli and Candler 2000. In your code uploaded. They are calculated without the density (not a Favre way). Are these your simplifications?

thank you so much!

For clearance, I post the function I am using to predict Cd and Ci:

Code:

volScalarField dynLocalAverageSmagorinsky::cD_(const volSymmTensorField& D) const

{

    volSymmTensorField Dfilter = filter_(rho()*D)/filter_(rho());

    volScalarField magSijfilter = sqrt(2.0*(Dfilter && Dfilter));

    volScalarField magSij = sqrt(2*(D && D));



    volSymmTensorField Lij = filter_(rho()*sqr(U())) - sqr(filter_(rho()*U()))/filter_(rho());

//  volSymmTensorField Bij = -2*sqr(2*delta())*filter_(rho())*filter_(magSij)*(filter_(D)-filter_(tr(D))*I/3);

//  volSymmTensorField Bij = -2*sqr(2*delta())*filter_(rho())*magSijfilter*(Dfilter - Dtrfilter*I/3);

    volSymmTensorField Bij = -2*sqr(2*delta())*filter_(rho())*magSijfilter*dev(Dfilter);

    volSymmTensorField Aij = -2*sqr(delta())*rho()*magSij*dev(D);

    volSymmTensorField Mij = Bij - filter_(Aij);

    volScalarField LijMij = Lij && Mij;

    volScalarField MklMkl = Mij && Mij;



    // Performing local average on cell faces on return

    return fvc::average(LijMij)/fvc::average(MklMkl);

}

Code:

volScalarField dynLocalAverageSmagorinsky::cI_(const volSymmTensorField& D) const

{

    volSymmTensorField Dfilter = filter_(rho()*D)/filter_(rho());

    volScalarField magSij = sqrt(2.0*(Dfilter && Dfilter));



    volSymmTensorField Lij = filter_(rho()*sqr(U())) - sqr(filter_(rho()*U()))/filter_(rho());

    volScalarField Lkk = tr(Lij);

    volScalarField beta = 2*sqr(2*delta())*filter_(rho())*sqr(magSij);

    volScalarField alpha = filter_(2*rho()*sqr(delta())*sqr(magSij));



    // Performing local average on cell faces on return

    return (fvc::average(Lkk))/(fvc::average(beta - alpha));

}

Any comments is appreciated for the above code lines. Thank you.

Quote:

Originally Posted by deji (Post 375144)

Here is the locally dynamic Smagorinksy compressible LES SGS model. The LESProperties that I attached previously also has the necessary coefficients to run this SGS model. I ought to say that I have had problems running dynamic models in OpenFOAM at t=0. Therefore, run any static LES model such as Smagorinksy or the WALE model for one timestep, then you can proceed to utilize the dynamic model thereafter.

Cheers,
Deji

Problem in compiling dynLocalAverageSmagorinsky in OF 2.0.1

Hello everyone,

I am having problem in compiling the dynLocalAverageSmagorinsky model in openfoam 2.0.1. Following is the error trace

[narya@localhost dynLocalAverageSmagorinsky]$ wmake libso
SOURCE=dynLocalAverageSmagorinsky.C ; g++ -m64 -Dlinux64 -DWM_DP -Wall -Wextra -Wno-unused-parameter -Wold-style-cast -Wnon-virtual-dtor -O3 -DNoRepository -ftemplate-depth-100 -I/home1/narya/OpenFOAM/OpenFOAM-2.0.1/src/turbulenceModels -I/home1/narya/OpenFOAM/OpenFOAM-2.0.1/src/turbulenceModels/LES/LESdeltas/lnInclude -I/home1/narya/OpenFOAM/OpenFOAM-2.0.1/src/turbulenceModels/LES/LESfilters/lnInclude -I/home1/narya/OpenFOAM/OpenFOAM-2.0.1/src/thermophysicalModels/basic/lnInclude -I/home1/narya/OpenFOAM/OpenFOAM-2.0.1/src/turbulenceModels/compressible/LES/lnInclude -I/home1/narya/OpenFOAM/OpenFOAM-2.0.1/src/finiteVolume/lnInclude -I/home1/narya/OpenFOAM/OpenFOAM-2.0.1/src/meshTools/lnInclude -IlnInclude -I. -I/home1/narya/OpenFOAM/OpenFOAM-2.0.1/src/OpenFOAM/lnInclude -I/home1/narya/OpenFOAM/OpenFOAM-2.0.1/src/OSspecific/POSIX/lnInclude -fPIC -c $SOURCE -o Make/linux64GccDPOpt/dynLocalAverageSmagorinsky.o
dynLocalAverageSmagorinsky.C: In constructor âFoam::compressible::LESModels::dynLocalAverageSma gorinsky::dynLocalAverageSmagorinsky(const Foam::volScalarField&, const Foam::volVectorField&, const Foam::surfaceScalarField&, const Foam::basicThermo&, const Foam::word&, const Foam::word&)â:
dynLocalAverageSmagorinsky.C:142: error: no matching function for call to âFoam::compressible::LESModel::LESModel(const Foam::GeometricField<double, Foam::fvPatchField, Foam::volMesh>&, const Foam::GeometricField<Foam::Vector<double>, Foam::fvPatchField, Foam::volMesh>&, const Foam::GeometricField<double, Foam::fvsPatchField, Foam::surfaceMesh>&, const Foam::basicThermo&, const Foam::word&, const Foam::word&)â
/home1/narya/OpenFOAM/OpenFOAM-2.0.1/src/turbulenceModels/compressible/LES/lnInclude/LESModel.H:134: note: candidates are: Foam::compressible::LESModel::LESModel(const Foam::word&, const Foam::volScalarField&, const Foam::volVectorField&, const Foam::surfaceScalarField&, const Foam::basicThermo&, const Foam::word&)
/home1/narya/OpenFOAM/OpenFOAM-2.0.1/src/turbulenceModels/compressible/LES/lnInclude/LESModel.H:101: note: Foam::compressible::LESModel::LESModel(const Foam::compressible::LESModel&)
make: *** [Make/linux64GccDPOpt/dynLocalAverageSmagorinsky.o] Error 1..

anyone please help.

Kafka.

Quote:

Originally Posted by deji (Post 375869)

Hi deji,
I have never compiled a turbulence model before in openfoam. Can you tell me how should I compile your your code?
Regards,
Awais

The LES models I uploaded are for compressible LES solvers in OpenFOAM. You need to copy the code into the directory where the other SGS models are, go through the files in the MAKE directory and add the lines necessary for the compiler to know that it should compile the codes too. Then just type wmake libso, it is quite simple, but you should spend some time going through the directory that has the SGS models.

Cheers,
Deji

hello,
Dear Deji,
I tried to complied the dynLocalAverageSmagorinsky Model into openfoam 2.1.1. But I am having problem.Following is the error:
wmakeLnInclude: linking include files to ./lnInclude
SOURCE=dynLocalAverageSmagorinsky/dynLocalAverageSmagorinsky.C ; g++ -m64 -Dlinux64 -DWM_DP -Wall -Wextra -Wno-unused-parameter -Wold-style-cast -Wnon-virtual-dtor -O3 -DNoRepository -ftemplate-depth-100 -I/opt/openfoam211/src/turbulenceModels -I/opt/openfoam211/src/turbulenceModels/LES/LESdeltas/lnInclude -I/opt/openfoam211/src/turbulenceModels/LES/LESfilters/lnInclude -I/opt/openfoam211/src/thermophysicalModels/basic/lnInclude -I/opt/openfoam211/src/meshTools/lnInclude -I/opt/openfoam211/src/finiteVolume/lnInclude -I/opt/openfoam211/src/turbulenceModels/compressible/LES/GenEddyVisc -I/opt/openfoam211/src/turbulenceModels/compressible/LES/LESModel -I/opt/openfoam211/src/turbulenceModels/compressible/LES/lnInclude -IlnInclude -I. -I/opt/openfoam211/src/OpenFOAM/lnInclude -I/opt/openfoam211/src/OSspecific/POSIX/lnInclude -fPIC -c $SOURCE -o Make/linux64GccDPOpt/dynLocalAverageSmagorinsky.o
dynLocalAverageSmagorinsky/dynLocalAverageSmagorinsky.C: In member function ‘Foam::volScalarField Foam::compressible::LESModels::dynLocalAverageSmag orinsky::cQ_(const volSymmTensorField&) const’:
dynLocalAverageSmagorinsky/dynLocalAverageSmagorinsky.C:104:31: error: reference to ‘db’ is ambiguous
In file included from /opt/openfoam211/src/OpenFOAM/lnInclude/regIOobject.H:41:0,
from /opt/openfoam211/src/OpenFOAM/lnInclude/GeometricField.H:42,
from /opt/openfoam211/src/OpenFOAM/lnInclude/GeometricScalarField.H:38,
from /opt/openfoam211/src/OpenFOAM/lnInclude/GeometricFields.H:34,
from /opt/openfoam211/src/finiteVolume/lnInclude/volFields.H:37,
from /opt/openfoam211/src/thermophysicalModels/basic/lnInclude/basicThermo.H:39,
from /opt/openfoam211/src/turbulenceModels/compressible/turbulenceModel/turbulenceModel.H:51,
from /opt/openfoam211/src/turbulenceModels/compressible/LES/LESModel/LESModel.H:51,
from /opt/openfoam211/src/turbulenceModels/compressible/LES/GenEddyVisc/GenEddyVisc.H:42,
from dynLocalAverageSmagorinsky/dynLocalAverageSmagorinsky.H:53,
from dynLocalAverageSmagorinsky/dynLocalAverageSmagorinsky.C:24:
/opt/openfoam211/src/OpenFOAM/lnInclude/IOobject.H:246:35: error: candidates are: const Foam::objectRegistry& Foam::IOobject::db() const
/opt/openfoam211/src/OpenFOAM/lnInclude/IOobject.H:246:35: error: const Foam::objectRegistry& Foam::IOobject::db() const
dynLocalAverageSmagorinsky/dynLocalAverageSmagorinsky.C:104:63: error: expected primary-expression before ‘>’ token
In file included from /opt/openfoam211/src/turbulenceModels/compressible/LES/GenEddyVisc/GenEddyVisc.H:42:0,
from dynLocalAverageSmagorinsky/dynLocalAverageSmagorinsky.H:53,
from dynLocalAverageSmagorinsky/dynLocalAverageSmagorinsky.C:24:
/opt/openfoam211/src/turbulenceModels/compressible/LES/LESModel/LESModel.H: In instantiation of ‘static Foam::autoPtr<Foam::compressible::LESModel> Foam::compressible::LESModel::adddictionaryConstru ctorToTable<LESModelType>::New(const volScalarField&, const volVectorField&, const surfaceScalarField&, const Foam::basicThermo&, const Foam::word&) [with LESModelType = Foam::compressible::LESModels::dynLocalAverageSmag orinsky; Foam::volScalarField = Foam::GeometricField<double, Foam::fvPatchField, Foam::volMesh>; Foam::volVectorField = Foam::GeometricField<Foam::Vector<double>, Foam::fvPatchField, Foam::volMesh>; Foam::surfaceScalarField = Foam::GeometricField<double, Foam::fvsPatchField, Foam::surfaceMesh>]’:
/opt/openfoam211/src/turbulenceModels/compressible/LES/LESModel/LESModel.H:128:1: required from ‘Foam::compressible::LESModel::adddictionaryConstr uctorToTable<LESModelType>::adddictionaryConstruct orToTable(const Foam::word&) [with LESModelType = Foam::compressible::LESModels::dynLocalAverageSmag orinsky]’
dynLocalAverageSmagorinsky/dynLocalAverageSmagorinsky.C:39:1: required from here
/opt/openfoam211/src/turbulenceModels/compressible/LES/LESModel/LESModel.H:115:9: error: no matching function for call to ‘Foam::compressible::LESModels::dynLocalAverageSma gorinsky::dynLocalAverageSmagorinsky(const volScalarField&, const volVectorField&, const surfaceScalarField&, const Foam::basicThermo&, const Foam::word&)’
/opt/openfoam211/src/turbulenceModels/compressible/LES/LESModel/LESModel.H:115:9: note: candidates are:
dynLocalAverageSmagorinsky/dynLocalAverageSmagorinsky.C:115:1: note: Foam::compressible::LESModels::dynLocalAverageSmag orinsky::dynLocalAverageSmagorinsky(const volScalarField&, const volVectorField&, const surfaceScalarField&, const Foam::basicThermo&)
dynLocalAverageSmagorinsky/dynLocalAverageSmagorinsky.C:115:1: note: candidate expects 4 arguments, 5 provided
In file included from dynLocalAverageSmagorinsky/dynLocalAverageSmagorinsky.C:24:0:
dynLocalAverageSmagorinsky/dynLocalAverageSmagorinsky.H:91:9: note: Foam::compressible::LESModels::dynLocalAverageSmag orinsky::dynLocalAverageSmagorinsky(const Foam::compressible::LESModels::dynLocalAverageSmag orinsky&)
dynLocalAverageSmagorinsky/dynLocalAverageSmagorinsky.H:91:9: note: candidate expects 1 argument, 5 provided
In file included from /opt/openfoam211/src/turbulenceModels/compressible/LES/GenEddyVisc/GenEddyVisc.H:42:0,
from dynLocalAverageSmagorinsky/dynLocalAverageSmagorinsky.H:53,
from dynLocalAverageSmagorinsky/dynLocalAverageSmagorinsky.C:24:
/opt/openfoam211/src/turbulenceModels/compressible/LES/LESModel/LESModel.H: In static member function ‘static Foam::autoPtr<Foam::compressible::LESModel> Foam::compressible::LESModel::adddictionaryConstru ctorToTable<LESModelType>::New(const volScalarField&, const volVectorField&, const surfaceScalarField&, const Foam::basicThermo&, const Foam::word&) [with LESModelType = Foam::compressible::LESModels::dynLocalAverageSmag orinsky; Foam::volScalarField = Foam::GeometricField<double, Foam::fvPatchField, Foam::volMesh>; Foam::volVectorField = Foam::GeometricField<Foam::Vector<double>, Foam::fvPatchField, Foam::volMesh>; Foam::surfaceScalarField = Foam::GeometricField<double, Foam::fvsPatchField, Foam::surfaceMesh>]’:
/opt/openfoam211/src/turbulenceModels/compressible/LES/LESModel/LESModel.H:115:9: warning: control reaches end of non-void function [-Wreturn-type]
make: *** [Make/linux64GccDPOpt/dynLocalAverageSmagorinsky.o] Error 1

plz help me

Dear All,

In dynLocalAverageSmagorinsky and other dynamic LES model, when we need to calculate the model constants using a dynamic way, we will do the averaging, like i.e. <Lij*Mij>/<Mkl*Mkl>. In principle, this kind of averaging is performed in the homogeneous direction in the flows. I do not know how this is implemented in other flow solvers. In openfoam, this is done using fvc::average(), this operation is only based on weighting of areas of all the faces belonging to one cell. So this is LOCAL dynamic method. Could anyone say something about if such implementation is really reflecting the averaging in the original sense (i.e. averaging in one homogeneous direction)?

OFFO
Thank you

Hello, flames,

I also try to compile the dynLocalAverageSmagorinsky Model into openfoam 2.1.1. But when i use wmake libso, i see error message.
Could you please share with me your folders for the dynLocalAverageSmagorinsky Model.
Could you tell me whats that error?
Thank you

Hi All,

When I use this dynamic model, for the regions having large mesh size, I found there are negative muSGS appearing there. So I cannot confirm if this is acceptable results or not. The flow fields and pressure seem still fine even if the mySGS is negative there.

Anybody has such kind of experience using dynamic Smagorinsky model?

OFFO

deji, could you please share your literature reference for the compressible dyn smagorinsky with k equation (you call it hybrid-sgs)?

Thank you in advance!

got it:

local dyn compressible Smagorinsky:

Martin, M. Pino; Piomelli, Ugo; Candler, Graham V. -- Subgrid-Scale Models for Compressible Large-Eddy Simulations -- Theoret. Comput. Fluid Dynamics, 2000 13, p:361-376

And

Martin, M. Pino; Piomelli, Ugo; Candler, Graham V. -- A priori tests of SGS models in compressible turbulence -- 3rd ASME/JSME Joint Fluids Engineering Conference -- 1999 -- Army High Performance Computing Research Center

k-Equation is based on oneEqEddy:

k-Equation:
Yoshizawa, Akira; Horiuti, Kiyosi -- A Statistically-Derived Subgrid-Scale Kinetic Energy Model for the Large-Eddy Simulation of Turbulent Flows -- Journal of the Physical Society of Japan -- 1985 -- Vol. 54 -- Issue 8 -- pages 2834-2839 -- doi: 10.1143/JPSJ.54.2834 p, li

k_sgs-transport:
Menon, S; Yeung, P.-K.; Kim, W.-W. -- Effect of subgrid models on the computed interscale energy transfer in isotropic turbulence -- Computers & Fluids -- 1996 -- Volume 25 -- Issue 2 -- pages: 165-180 -- doi: 10.1016/0045-7930(95)00036-4

where the parameter c_k = 0.0094 is defined like in oneEqEddy by:
Nakayama, A.; Vengadesan, S. N. -- On the influence of numerical schemes and subgrid-stress models on large eddy simulation of turbulent flow past a square cylinder -- International Journal for Numerical Methods of Fluids -- 2002 -- Volume 38 -- Issue 3 -- pages: 227-253 -- doi: 10.1002/fld.214

Thank You, I was also looking for the source of literature. And also this is a good reference:
http://www.aem.umn.edu/~kmahesh/publ...aiaa10_les.pdf

Also which version of OpenFOAM could you compile the code with?

Cheers!

i compiled it with 2.3.x and it works well.
but you need to modify some small parts.

just diff the comp. Smagorinsky from 1.6 or 1.7 with the actual one. Thats all.

Thanks! Also you can have a look at following for other compressible LEs subgrid implementations: :)

https://github.com/fireFoam-dev/fire...mpressible/LES