[preibie]

Hallo user ans developer of OpenFOAM 1.6-ext,

I want to simulate a room without any inlets and outlets only with a heat source on the ground. I choose the buoyantSimpleFoam solver. The steps I had done:

1. build and mesh the domain with Gambit
2. checkMesh OK (nonOrthogonality max = 63)
3. buoyantSimpleFoam: after 200 Iterations there are nonphysical oszillations (+-0.04 m/s) in the velocity fild (picture)
4. simpleFoam: same mesh the oszilations are very low (+- 1e-7) (picture)

Is this a problem with the mesh ore with buoyantSimpleFoam?

Preibie

[preibie]

Some more informations about the last hours:

We tryed something more. We changed the mesh (only the mesh) and used a hex grid. What I can say all was fine, no oscillations.
Why I'm confused: The first mesh passed checkMesh!
Next step: we convert the hex mesh with tetDecomposition into a tet Mesh. This mesh passed also checkMesh (nonOrthogonality max = 44 average = 18). It seems that this mesh is good enough, no oszillations!

Is the threshold (70) of nonOrthogonality to high?

[alberto]

Hi,
you should also list the schemes you are using. On tet meshes you often need to turn on limiters which are off by default in OpenFOAM.

[preibie]

fvSchemes:

ddtSchemes
{
    default         Euler;
}

gradSchemes
{
    default         Gauss linear;
}

divSchemes
{

    div(phiU,p)                    Gauss upwind;
    //div(phi,U)                     Gauss upwind;
    div(phi,U)                     Gauss reconCentral cellLimited leastSquares 1.0;
    div(phi,h)                     Gauss upwind;
    div(phi,k)                     Gauss upwind;
    div(phi,epsilon)               Gauss upwind;
    div((muEff*dev2(grad(U).T()))) Gauss linear;
}

laplacianSchemes
{
    default         Gauss linear corrected;
}

interpolationSchemes
{
    default         linear;
    interpolate(U)  reconCentral phi leastSquares;    //für tetraeder Gitter von Jasak
}

snGradSchemes
{
    default         corrected;
}

fluxRequired
{
    default         no;
    p               ;
}

[alberto]

I would try:

gradSchemes
{
    default         cellLimited leastSquares 1; 
}

divSchemes
{
    div(phi,U)                     Gauss linearUpwindV cellLimited leastSquares 1;
    div(phiU,p)                    Gauss upwind;
    div(phi,h)                      Gauss upwind;
    div(phi,k)                      Gauss upwind;
    div(phi,epsilon)               Gauss upwind;
    div((muEff*dev2(grad(U).T()))) Gauss linear;
}

Note that in your schemes, you are using inconsistent gradient formulations. Additionally, if leastSquares gives you problems, replace it with Gauss linear, which will reduce the accuracy, but tends to be more robust.

I hope this helps.