Unphysical phase volume fraction in FoamExtend3.1 using interDyMFoam
 

I have a layer A/R case where a cylinder sees two patches (along Z) add and remove layers while oscillating the enclosed cellZone. The case is run in parallel Hierarchical (4 4 1) xyz decomposition. The case runs fine in laminar mode. But as soon as I turn turbulence "on", it breaks down giving unphysical Alpha1 (liquid phase volume fraction) accompanied by increase in velocity. Here are a few other details:

--> CFL = 0.4
--> write in binary
--> time precision = 6
--> write precision = 6

fvSchemes:

ddtSchemes
{
default Euler;
}

gradSchemes
{
default Gauss linear;
}

divSchemes
{
div(rho*phi,U) Gauss vanLeerV;
div(phi,alpha) Gauss vanLeer;
div(phirb,alpha) Gauss vanLeer;//interfaceCompression;
div(phi,k) Gauss upwind;//limitedLinear 1;
}

fvSolution:

/*--------------------------------*- C++ -*----------------------------------*\
| ========= | |
| \\ / F ield | foam-extend: Open Source CFD |
| \\ / O peration | Version: 3.1 |
| \\ / A nd | Web: http://www.extend-project.de |
| \\/ M anipulation | |
\*---------------------------------------------------------------------------*/
FoamFile
{
version 2.0;
format ascii;
class dictionary;
object fvSolution;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

solvers
{
pcorr
{
solver PCG;
preconditioner DIC;

minIter 0;
maxIter 1000;
tolerance 1e-15;//1e-8, was initially 1e-2
relTol 0;
};

pd
{
solver PCG;
preconditioner DIC;

minIter 1;
maxIter 1000;
tolerance 1e-08;//was 1e-6
relTol 1e-2;
};

pdFinal
{
$pd;
tolerance 1e-08;//1e-6;
relTol 0;
};

"(U|k)"
{
solver PBiCG;
preconditioner DILU;

minIter 1;
maxIter 1000;
tolerance 1e-06;
relTol 0;
};
}

PISO
{
cAlpha 1;
}

PIMPLE
{
nOuterCorrectors 1;//was 4;
nAlphaCorr 2;
nAlphaSubCycles 3;
nCorrectors 4;
nNonOrthogonalCorrectors 1;
correctPhi yes;
}

relaxationFactors
{
U 0.7;
pd 0.3;
}

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

Any help to fix this problem is appreciated.

Wrong different Int scheme on div(phi,alpha).

Please use vanLeer01 or vanLeer01DC

Thank you Dr. Jasak. A follow up question:

Since my mesh motion is in the Z direction, it became clear to me that I cannot decompose my case in Z. So I decompose hierarchically into 16 domains in Xand Y. But, for a certain mesh size the layer AR works fine, while for a finer cell size, it does not (for the same deconposition method). Is there a general rule of thumb as to how the mesh should be decomposed so as to be able to run these layer AR cases in parallel?

Am I correct in my understanding that the latest version of movingBodyTopiFvMesh is unable to handle arbitrary mesh decompositions? If yes, can you provide an estimate of when this will be released (or how to how to fix this)?