[WaterHammer1985]

I'm modeling water flow through a small opening (see attached image). Using snappyHexMesh, I've created refinement regions to transition the cell size around the opening. As flow approaches the opening, large waves form on the surface that aren't realistic. I'm unsure if I need to do something different with the mesh, or if I need to make changes to the fvScheme/fvSolution settings? Any insight is greatly appreciated.

fvScheme

ddtSchemes
{
    default         Euler;
}

gradSchemes
{
  default cellMDLimited Gauss linear 1;
  grad(k) faceLimited leastSquares 1;
  grad(omega) faceLimited leastSquares 1;
}

divSchemes
{
    div(rhoPhi,U)  Gauss limitedLinearV 1; //Gauss upwind;
    div(phi,alpha)  Gauss vanLeer;
    div(phirb,alpha) Gauss interfaceCompression 1;
    div(phi,p_rgh)  Gauss upwind;
    div(phi,k)      Gauss upwind;
    div(phi,epsilon) Gauss upwind;
    div(((rho*nuEff)*dev2(T(grad(U))))) Gauss linear;
    div(phi,omega) Gauss upwind;
}

laplacianSchemes
{
    default         Gauss linear uncorrected;
}

interpolationSchemes
{
    default         linear;
}

snGradSchemes
{
    default         uncorrected;
}

fluxRequired
{
    default         no;
    p_rgh;
    pcorr;
    gamma;
}

wallDist
{
    method meshWave;
}

fvSolution

solvers
{
    "alpha.water.*"
    {
        nAlphaCorr      1;
        nAlphaSubCycles 1;
        alphaOuterCorrectors yes;
        cAlpha          1;

        MULESCorr       yes;
        nLimiterIter    3;

        solver          smoothSolver;
        smoother        symGaussSeidel;
        tolerance       1e-8;
        relTol          0;
    }

   "pcorr.*" GAMG
    {
        tolerance        1e-7;
        relTol           0.0;

        smoother         DIC;//GaussSeidel;
        nPreSweeps       0;
        nPostSweeps      2;
        nFinestSweeps    2;

        cacheAgglomeration true;
        nCellsInCoarsestLevel 10;
        agglomerator     faceAreaPair;
        mergeLevels      1;
        maxIter 10;
    };

    p_rgh GAMG
    {
        tolerance        1e-7;
        relTol           0.0;

        smoother         DIC;//GaussSeidel;
        nPreSweeps       0;
        nPostSweeps      2;
        nFinestSweeps    2;

        cacheAgglomeration true;
        nCellsInCoarsestLevel 10;
        agglomerator     faceAreaPair;
        mergeLevels      1;
    };

    p_rghFinal GAMG
    {
        tolerance        1e-8;
        relTol           0.0;

        smoother         DIC;//GaussSeidel;
        nPreSweeps       0;
        nPostSweeps      2;
        nFinestSweeps    2;

        cacheAgglomeration true;
        nCellsInCoarsestLevel 10;
        agglomerator     faceAreaPair;
        mergeLevels      1;
    };



    "(U|k|omega|epsilon).*"
    {
        solver          smoothSolver;
        smoother        symGaussSeidel;
        nSweeps         1;
        tolerance       1e-6;
        relTol          0.1;
    };

    "(U|k|epsilon)Final"
    {
        $U;
        relTol          0;
    }


}


PIMPLE
{
    momentumPredictor no;
    nOuterCorrectors 2;
    nCorrectors     3;
    nNonOrthogonalCorrectors 3;
}

relaxationFactors
{
    fields
    {
    }
    equations
    {
        ".*" 0.1;
    }
}

[WaterHammer1985]

I restarted the simulation using interIsoFoam and the free surface looks much more stable (see attached image) but there are still some velocity discontinuities at the areas of mesh transition.

[akidess]

My experience is that interfoam is very sensitive to the mesh. Sudden changes in size are not good. Can you make a mesh with a smooth transition?

[WaterHammer1985]

Thanks for the suggestion. The mesh transitions are as slow/smooth as I know how to make in snappyHexMesh by slowly tapering refinement levels. I decreased the MaxCo to 0.2 (was at 1) and the velocity discontinuities washed out.

[aow]

Correct me if I'm wrong, but I think Akidess meant to intially setup your mesh better using blockMesh so that you have uniform grading instead of the sort of sudden jumps that are created using snappyHexMesh. Also, since you have only rectangular shapes in your problem as far as I can tell from your images, it would probably be best to just use blockMesh with a bunch of hexes instead of making a single one and then using snappyHexMesh. This will give you much better control over the quality of the mesh.

[WaterHammer1985]

Originally Posted by aow

Correct me if I'm wrong, but I think Akidess meant to intially setup your mesh better using blockMesh so that you have uniform grading instead of the sort of sudden jumps that are created using snappyHexMesh. Also, since you have only rectangular shapes in your problem as far as I can tell from your images, it would probably be best to just use blockMesh with a bunch of hexes instead of making a single one and then using snappyHexMesh. This will give you much better control over the quality of the mesh.

Using uniform grading near the free surface did the trick. Thanks for the info!

[aow]

Awesome! Glad to hear it worked out. Doing the messier blockMeshDict with multiple hexes and grading is a pain, but usually sorts out issues like this. Out of curiosity, what is the goal of your modeling?

[WaterHammer1985]

Agreed, more work upfront but better results in the end! Modeling is to look at flow conditions and hydrodynamic forces on a lock chamber closure section. The pressures seemed reasonable even with the velocity discontinuities but this condition was essentially hydrostatic so I'm not sure how it would have done with more velocity head.