[DOFuser]

Dear Foamers,

I have a problem when meshing a T-junction circular pipe. It is an extremely simple geometry, made by two circular cylinder, one normal to the other.

I use snappyHexMesh. The result of "castellating" and "snapping" procedures are ok. The problem arises when I try to add layers on all the walls, since the boundary layer collapses on one patch.

For illustrating my problem I attach the entire domain (Tjunction.jpg), an image showing a correctly meshed patch (correctPatch.jpg), and two images showing the patch where the boundary layer collapses (nonCorrectPatch.jpg and nonCorrectPatchHalf.jpg). Among the three patches, the two ones that are aligned result to be meshed correctly, while the patch of the normal duct is not. As it can be seen the boundary layer collapses in two points over the patch, but the collapsing takes place only just ahead the patch itself, since the layers are added correctly along the remaining part of the duct.

I attach also the settings I put in the SHMdict.

castellatedMesh true;
snap            true;
addLayers       true;

geometry
{
    Inlet.stl
    {
        type triSurfaceMesh;
        name Inlet;
    }
    Outlet01.stl
    {
    type triSurfaceMesh;
        name Outlet01;
    }
    Outlet02.stl
    {
    type triSurfaceMesh;
        name Outlet02;
    }
    MainDuct.stl
    {
    type triSurfaceMesh;
        name MainDuct;
    }
    SecondaryDuct.stl
    {
    type triSurfaceMesh;
        name SecondaryDuct;
    }
};

castellatedMeshControls
{
    maxLocalCells 4000000;
    maxGlobalCells 20000000;
    minRefinementCells 1;
    maxLoadUnbalance 0.10;
    nCellsBetweenLevels 4;
    features
    (
    {
        file "Inlet.eMesh";
            level 2;           
    }
    {
        file "Outlet01.eMesh"; 
            level 2;               
    }
    {
        file "Outlet02.eMesh";
            level 2;              
    }
    {
        file "MainDuct.eMesh"; 
            level 2;              
    }
    {
        file "SecondaryDuct.eMesh"; 
            level 2;                    
    }
    );

    refinementSurfaces
    {
    Inlet
        {
            level (1 1);
            patchInfo
            {
                type patch;
            }
        }
    Outlet01
    {
            level (1 1);
            patchInfo
            {
                type patch;
            }
        }
    Outlet02
    {
            level (1 1);
            patchInfo
            {
                type patch;
            }
        }
    MainDuct
    {
            level (2 2);
            patchInfo
            {
                type wall;
            }
        }
    SecondaryDuct
    {
            level (2 2);
            patchInfo
            {
                type wall;
            }
        }
    }

    resolveFeatureAngle 30;

    refinementRegions
    {
    }

    locationInMesh (0.5123456 0.12345 5.0111112);
    allowFreeStandingZoneFaces true;
}

snapControls
{
    nSmoothPatch 3;
    tolerance 2.0;
    nSolveIter 300;
    nRelaxIter 5;
    nFeatureSnapIter 10;
    implicitFeatureSnap false;
    explicitFeatureSnap true;
    multiRegionFeatureSnap false;
}

addLayersControls
{
    relativeSizes true;
    layers
    {
    MainDuct
        {
        nSurfaceLayers 8;
        }
    SecondaryDuct
        {
        nSurfaceLayers 8;
        }
    }
    expansionRatio 1.1;
    finalLayerThickness 0.4;
    minThickness 0.0000000001;
    nGrow 0;
    featureAngle 90;
    nRelaxIter 10;
    nSmoothSurfaceNormals 0;
    nSmoothNormals 30;
    nSmoothThickness 30;
    maxFaceThicknessRatio 1.2;
    maxThicknessToMedialRatio 0.9;
    minMedianAxisAngle 80;
    nBufferCellsNoExtrude 0;
    nLayerIter 50;
}

I would be pleased if anyone could help me.

Kind regards

[kishpishar]

Hello,

I've noticed similar behavior with non-adequate boundary layer coverage on some patches with tightened quality control parameters, eg. maxNonOrthogonality ~ 40. Perhaps you can examine this aspect. Also I think that nSmoothSurfaceNormals should be at least 3.

-kumar