[yambanshee]

Good day!


I am investigating the possible cause and implication of checkerboarding pressure in my solution. I am using the simpleFoam solver for a 3d steady state solution using the k-omega sst turbulence model. At this point, my solution has iterated through 47500 times, and have reached the following residuals:

smoothSolver:  Solving for Ux, Initial residual = 3.52304e-06, Final residual = 1.75112e-10, No Iterations 9
smoothSolver:  Solving for Uy, Initial residual = 1.29964e-05, Final residual = 7.31583e-10, No Iterations 9
smoothSolver:  Solving for Uz, Initial residual = 9.49277e-06, Final residual = 9.39161e-10, No Iterations 8
GAMG:  Solving for p, Initial residual = 0.00739052, Final residual = 7.33255e-06, No Iterations 49
time step continuity errors : sum local = 3.0622e-07, global = 1.03018e-09, cumulative = 1.84471e-05
smoothSolver:  Solving for omega, Initial residual = 1.55815e-08, Final residual = 8.38819e-11, No Iterations 3
smoothSolver:  Solving for k, Initial residual = 3.76601e-05, Final residual = 2.5968e-09, No Iterations 7
bounding k, min: -7.15622e-07 max: 186.885 average: 68.8195

my fvSchemes is as follows:

/*--------------------------------*- C++ -*----------------------------------*\
| =========                 |                                                 |
| \\      /  F ield         | OpenFOAM: The Open Source CFD Toolbox           |
|  \\    /   O peration     | Version:  5                                     |
|   \\  /    A nd           | Web:      www.OpenFOAM.org                      |
|    \\/     M anipulation  |                                                 |
\*---------------------------------------------------------------------------*/
FoamFile
{
    version     2.0;
    format      ascii;
    class       dictionary;
    location    "system";
    object      fvSchemes;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

ddtSchemes
{
    default         steadyState;
}

gradSchemes
{
    default         Gauss linear corrected 1;
}

divSchemes
{
    default                         none;
    div(phi,U)                      Gauss linearUpwind grad(U);
    div(phi,k)                      Gauss upwind;
    div(phi,epsilon)                Gauss upwind;
    div(phi,omega)                  Gauss upwind;
    div(phi,R)                      Gauss linear;
    div(R)                          Gauss linear;
    div(phi,nuTilda)                Gauss linear;
    div((nuEff*dev2(T(grad(U)))))   Gauss linear;
}

laplacianSchemes
{
    default         Gauss linear corrected 1;
}

interpolationSchemes
{
    default         linear;
}

snGradSchemes
{
    default         corrected;
}

wallDist
{
    method meshWave;
}

and my fvSolution:

/*--------------------------------*- C++ -*----------------------------------*\
| =========                 |                                                 |
| \\      /  F ield         | OpenFOAM: The Open Source CFD Toolbox           |
|  \\    /   O peration     | Version:  5                                     |
|   \\  /    A nd           | Web:      www.OpenFOAM.org                      |
|    \\/     M anipulation  |                                                 |
\*---------------------------------------------------------------------------*/
FoamFile
{
    version     2.0;
    format      ascii;
    class       dictionary;
    location    "system";
    object      fvSolution;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

solvers
{

    "(p|Phi)"
    {
        solver           GAMG;
        tolerance        1e-8;
        relTol           0.001;
        smoother         DIC;
        nPreSweeps       1;
        nPostSweeps      2;
        cacheAgglomeration on;
        agglomerator     faceAreaPair;
        nCellsInCoarsestLevel 3600;
        mergeLevels      1;
        maxIter      200;   
    }
    

    "(U|k|epsilon|omega)"
    {
        solver          smoothSolver;
        smoother        symGaussSeidel;
        tolerance       1e-10;
        relTol          0.0001;
    }

    "(U|k|epsilon|omega)Final"
    {
        $U;
        tolerance       1e-10;
        relTol          0;
    }
 }

SIMPLE
{
    nNonOrthogonalCorrectors 0;
}

relaxationFactors
{
    fields
    {
        p               0.3;
    }
    equations
    {
        U               0.7;
        k               0.6;
        "(epsilon|omega)" 0.6;
    }
}

cache
{
    grad(U);
}

As can be seen in the attached image, I am seeing pressure fluctuation of ~40 Pa on a simulation of a fan demonstrating a 200 Pa static pressure rise. As such, this could be a significant cause of error. I am mostly seeing the checkerboarding occur along a slip wall and near the outlet of the domain (zero pressure, zero velocity gradient), where the mesh also coarsens. The mesh used is pure hex.



Is it more likely to be mesh, boundary or interpolation scheme dependent? If it is the interpolation scheme, what measured can be used to remedy this? To my understanding, turbulence modelling is first order accurate at best, and so going beyond upwind is not needed

[mAlletto]

Is it a slip or no-slip wall? Is your outlet boundary far away enough it is consistent with your zero gradient B.C.?

[yambanshee]

Originally Posted by mAlletto

Is it a slip or no-slip wall? Is your outlet boundary far away enough it is consistent with your zero gradient B.C.?

It is a slip wall in this case, but I also see it occuring on non-slip walls.


My outlet is very far from the rest of the domain. It's consistent in distance to literature on similar simulations, and far enough that the pressure difference over the second half of the domain is less than one pascal