CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (http://www.cfd-online.com/Forums/openfoam-solving/)
-   -   LES SpalartAllmaras Simulation questions (http://www.cfd-online.com/Forums/openfoam-solving/109242-les-spalartallmaras-simulation-questions.html)

 Matt1986 November 13, 2012 09:13

LES SpalartAllmaras Simulation questions

Hi,

I am doing a LES simulation of a ball, which is flowed by air. I am calculating the draft coefficients of the ball with different Reynolds' numbers.
My SpalartAllmaras LES model is running, but I have a few questions regarding to the initial values.:confused:

In my 0-directory there is a nuSgs and a B file.
What are the functions of the files.
How can I calculate these values? Are there any formulas available?
Should the nutilda be calculated like the RAS SpalartAllmaras case with the well known formulas?
The nut is not used in the LES SpalartAllmaras case, am I right?

Regards,
Matthias

 Matt1986 November 13, 2012 10:18

I have only found the following formulas for B and nuSgs in the forum.

B = 2/3 rho*k*I-2*mu_t*D

But I am not sure if these are applicable and what the abbreviations mean.

Regards Matthias

 Matt1986 November 23, 2012 10:28

1 Attachment(s)
Hi,

My Simulation runs without B.
Now then I only need the definitions for the nuSGS and the nuTilda.
If the nuTilda is calculated as in the RAS Spalard Allmaras case, I only need the definition of nuSGS.
On http://foam.sourceforge.net/docs/cpp/a01896.html I found the following formula.
nuSgs = ck*sqrt(k)*delta
But ck is only described as a dimensioned skalar and delta isn't described anyway.
My Problem is, that the coefficients of drag of the ball are negative, so I think I have to set a value for nuSgs.
At the moment nuSGS is set to 1e-10.

Hopefully somebody is able to help me

Regards Matthias

 meindert November 23, 2012 10:55

Hi Matthias,

Have a look at http://foam.sourceforge.net/docs/cpp/a08492_source.html.

nuTilda is given by the solution to the equation on lines 297-311. The function on lines 45-49 is called to get nuSgs.

A discussion about the values of nut (or nuSgs in this case) and nuTilda came up not too long ago. You might want to have a look at that discussion: http://www.cfd-online.com/Forums/ope...t-nutilda.html.

 Matt1986 November 27, 2012 06:51

Hi,

Thank you for your help.
The following formula is given for fv1
x^3/(x^3 + cv1^3)
with x=nut/nu
cv1=7.1

According to
nuSgs=nuTilda*fv1
I should get my initial value for nuSgs.
Am I right?

Regards Matthias

 meindert November 27, 2012 07:55

Be aware that nut and nuTilda are not the same. Actually, nut is called nuSgs in this case. X is defined as nuTilda/nu (see lines 54-58 in foam.sourceforge.net/docs/cpp/a08492_source.html).

I do not fully understand your question about nuSgs. If you are asking whether you should initialise nuSgs, then the answer is: Yes, you should. Of course, you could test this very easily by removing nuSgs from your time directory. I expect your simulation will not run.

 Matt1986 November 27, 2012 17:04

Hi,

Thank you very much for your help.
I hadn't known, that nut is called nuSgs in this case.

Regards Matthias

 Matt1986 December 26, 2012 18:03

Hi,

I had a few problems with the 3D-case and the calculation time was very high.
So I set up a 2D case (extrudeMesh) of a long cylinder to get rid of the problems.
I chose a low Reynold number of 180. According to the book Large Eddy Simulation from J. Fröhlich, it is suitable to simulate 2D in case of low Reynold numbers.
My problem is, that I should determine the drag coefficent of the cylinder.
To get my drag coefficients, I inserted forceCoeffs in my control.dict file.

forceCoeffs
{
type forceCoeffs;
functionObjectLibs ("libforces.so");
patches (zylinder_region0); //change to your patch name
pName p;
UName U;
rhoName rhoInf;
rhoInf 1.225;
CofR (0 0 0);
liftDir (0 0 1);
dragDir (1 0 0);
pitchAxis (0 1 0);
magUInf 0.0243243243;
lRef 0.0555;
Aref 1.11e-3;
outputControl timeStep;
outputInterval 10;
}

To get the dragcoefficient cd of this instable case I made the average of the last 500 values. My problem is, that the figure of cd is in the 100Mio. range, where it should not be.
Is this the right way, to determine drag Coefficients of an LES case?
My yPlus Values of the cylinder are the following:
y+ : min: 0.652849 max: 59.6055 average: 22.989
Thank you for your time

Regards,
Matthias

 Matt1986 December 26, 2012 18:08

2 Attachment(s)
Here are 2 pics of the case

Regards Matthias

 Matt1986 January 3, 2013 16:17

Hi,

The problem with the high cd values is solved.
There were problems with the boundary conditions.

Thanks,
Matthias

 javidFoam May 24, 2015 07:26

«les

Hi
How do I set coefficients k and nusgs for LES-Spalart

 somFoam June 15, 2015 06:08

Quote:
 Originally Posted by Matt1986 (Post 399991) Hi, The problem with the high cd values is solved. There were problems with the boundary conditions. Thanks, Matthias
Can you please give some reference how to calculate the nuSgs for simple smagorinsky model.

 All times are GMT -4. The time now is 05:28.