Discretization settings Foam vs. Fluent
 

Dear Foamers,

I´m interested in the the performance of Fluent 6.3.26 in comparison with OpenFoam 1.5. I´m simulating a steady (simpleFoam solver in OF 1.5), isothermal, single phase problem using the realizable k-epsilon-model.
In order to get accurate results for my flow field, I chose the "second-order upwind" descritization scheme (for pressure, momentum, k, epsilon) in Fluent.

Here my Question:
How do I have to change the fvScheme-skript in OpenFoam to get the same settings as in Fluent?
I made different approaches using the "linearUpwind"-scheme. But it did´nt work, due to some errors. I´m not sure if "linearUpwind" is comparable to the "second-order upwind" scheme in Fluent.

I would appreciate any hints or help.
Thanks in advance.
wuppdupp

Among others I used this fvScheme script:
FoamFile
{
version 2.0;
format ascii;
class dictionary;
object fvSchemes;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

ddtSchemes
{
default steadyState;
}

gradSchemes
{
default Gauss linearUpwind;
grad(p) Gauss linearUpwind;
grad(U) Gauss linearUpwind;
}

divSchemes
{
default none;
div(phi,U) Gauss linearUpwind;
div(phi,k) Gauss linearUpwind;
div(phi,epsilon) Gauss linearUpwind;
div(phi,R) Gauss linearUpwind;
div(R) Gauss linearUpwind;
div(phi,nuTilda) Gauss linearUpwind;
div((nuEff*dev(grad(U).T()))) Gauss linearUpwind;
}

laplacianSchemes
{
default none;
laplacian(nuEff,U) Gauss linear corrected;
laplacian((1|A(U)),p) Gauss linear corrected;
laplacian(DkEff,k) Gauss linear corrected;
laplacian(DepsilonEff,epsilon) Gauss linear corrected;
laplacian(DREff,R) Gauss linear corrected;
laplacian(DnuTildaEff,nuTilda) Gauss linear corrected;
}

interpolationSchemes
{
default linearUpwind;
interpolate(U) linearUpwind;
}

snGradSchemes
{
default corrected;
}

fluxRequired
{
default no;
p;
}

// ************************************************** *********************** //

Dear Wuppdupp,

Maybe the scheme that you need is QUICK. Search the forum for related posts. I personally use GammaV for velocity.

Best Regards,

Paulo Rocha

Dear paulo,

I tried Quick and GammaV. (For the gradSchemes) Without success.
The only Schemes which work are linear an fourth.

It makes sense, since QUICK and GammaV are divSchemes. :)

Try it.

Best Regards,

Paulo Rocha

Hi Paulo,

thank´s for your advice. It works fine. Is there any possibility to use higher order discretization schemes also for the gradSchemes and the interpolationSchemes?
I was quite confused because in the OF documentation is mentioned, that you can use other Schemes also for gradSchemes and interpolationSchemes.

Again, thank you very much!

Wuppdupp

Yes,

if you write something wrong in the schemes text, OpenFOAM will show you all the possibilities.

The books of Ferziger & Peric / Versteeg & Malalasekra have good explanation about many schemes.

Best Regards,

Paulo Rocha

Yes, it is possible to use higher order than second schemes for gradients. In particular, if you want to be sure you always have second order accuracy, you should change "Gauss linear" into "leastSquares". If you want a fourth order scheme you can use "fourth". All schemes can be used in a "limited" version (limited leastSquares, for example), which reduces the accuracy locally to keep the solution bounded.

For interpolation schemes the choice is wider (see tab 4.6 in the manual). The "cubic" is fourth order central scheme, and it exists in the limited version "limitedCubic".

Best,