[yekaniyasari]

Bug in LES Filter Codes:
Description:
1-The problem is about the laplace filter and anisotropic filter:
with running a case in LES solvers (oodles, channelOodles …) and using "laplace" filter or "anisotropic" filter with "dynamic or mix models" an error about laplace or anisotropic filtercoeffs occurred .for solving this problem we should add the following line in "turbulenceproperties" file in "constant" directory of relevant case.
For example:

mixedSmagorinskyCoeffs
{
ce ce [0 0 0 0 0 0 0] 1.05;
ck ck [0 0 0 0 0 0 0] 0.07;
filter anisotropic;
anisotropicCoeffs
{
widthCoeff 24; or 6 (for box or Gaussian filter)
}
}


2.The second problem was seen about definition of "widthcoeff" vector in "anisotropicfilter.C" file that was placed in"/src/lesmodel/lesfilter" .
In order to modify this problem, instead of the statement below:
"
for (direction d=0; d<vector::nComponents; d++)
{
coeff_.internalField().replace
(
d,
(2.0/widthCoeff_)*mesh.V()
/fvc::surfaceSum(mag(mesh.Sf().component(d)))().int ernalField()
);
}
"
We should use the following code:
"
for (direction d=0; d<vector::nComponents; d++)
{
coeff_.internalField().replace
(
d,
(4.0/widthCoeff_)*(mesh.V()/fvc::surfaceSum(mag(mesh.Sf().component(d)))().int ernalField())*(mesh.V()/fvc::surfaceSum(mag(mesh.Sf().component(d)))().int ernalField())
);
}
"
3. The third one , was about dimension problem : in "anisotropicfilter.C" file , in definition of filtration for tensor field , with substituting this part of code :
tmp<voltensorfield> tmpFilteredField
(
new volTensorField
(
IOobject
(
"anisotropicFilteredTensorField",
mesh().time().timeName(),
mesh(),
IOobject::MUST_READ,
IOobject::AUTO_WRITE
),
mesh(),
dimLength
)
);

With this form :
tmp<voltensorfield> tmpFilteredField
(
new volTensorField
(
IOobject
(
"anisotropicFilteredTensorField",
mesh().time().timeName(),
mesh(),
IOobject::NO_READ,
IOobject::AUTO_WRITE
),
mesh(),
dimLength*dimLength/(dimTime*dimTime)
)
);

the problem is solved.
You can download the correct file from here.

anisotropicFilter.C

Testcase:
All case for example: channel395

Version:
OpenFOAM version 1.3

[yekaniyasari]

in number 2 change " "widthcoeff" vector " to " "coeff_" vector "

[henry]

I think it would be better to use unFilteredField().dimensions() rather than hard-coding dimLength*dimLength/(dimTime*dimTime) as the dimensions of the tensor field resulting from the filtering of given field.

Henry

[yekaniyasari]

Dr Henry Weller,
Thanks for your attention.
Could you kindly comment about a problem in :

http://www.cfd-online.com/OpenFOAM_D...ages/1/40.html

By ehsan yasari on Tuesday, January 01, 2008 - 12:59 am
Best Regards

[HakikiCanakkaleli]

Hi,

Would you please mind to consider to read the following thread, if possible, as the fact that the proposed correction herein was not embedded into the current OpenFOAM made me think that you have some solid reference to rely on and keep the anisotropicFilter implementation as it was?

What is the academic reference of the implementation for Foam::anisotropicFilter?

Kind regards,