CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   OpenFOAM Meshing & Mesh Conversion (https://www.cfd-online.com/Forums/openfoam-meshing/)
-   -   [snappyHexMesh] Protrusion cells at intersection of surfaces - snappyHexMesh (https://www.cfd-online.com/Forums/openfoam-meshing/193831-protrusion-cells-intersection-surfaces-snappyhexmesh.html)

Katt October 4, 2017 10:03

Protrusion cells at intersection of surfaces - snappyHexMesh
 
2 Attachment(s)
Hi;

-> I want to generate a model, having two cylinders intersecting orthogonally and at the point of contact, there is a sphere as shown in Figure1. [generated using "OpenSCAD"]

-> With reference through "motorbike" case and few other tutorials I did make progress extracting the geometry. The problematic region is around the location where the cylinder and sphere surface meet. Please see the outward protrusions in Figure2 after running sHM.

I used a refinement level of 1,2 and moved until 6,7 but still these protrusions do exist.

I tried several parameters but I do see them (protrusions) at all times. Is there a way to overcome this issue without tremendous rise in the cell count?

-> After running sHM, I check the mesh properties using "checkMesh". Every constraint is fine except the "skewness".

How is skewness of a mesh computed? Further the cell faces with skewness greater than the tolerance is written as output. How can we visualize these cell faces in paraview?

Ideally, lesser the skewness, better the accuracy for the numerical solver. I hit around "skewness: 4 - 4.2" with 4 being the default tolerance set by OpenFoam. Is there a parametric fix to this issue?

Katt

Katt October 6, 2017 05:26

Anyone interested in this topic...??

Ideally I would prefer to generate the geometry shown in Figure 1 using blockMesh (using arc edges and so on) and not using sHM.

Reason being, I would like to see more of a decent uniform/ structured sort of mesh rather than completely deformed pyramidal shaped control volumes which makes any sort of solver to struggle especially for interFoam solver where the interface smears terribly.

I ask this because I see few posts as here
https://www.researchgate.net/post/He...Foam_BlockMesh where they have a decent structure to their geometry and used blockMesh to generate cylinders and spheres.

C-L October 6, 2017 05:57

Have you used the surfaceFeatureExtract function that the motorBike tutorial uses? You should be able to identify the important features (the intersection lines) using this function and then tell SHM to maintain them. I'm pretty sure the motorbike tutorial does this for reference.

I haven't used blockMesh for anything other than simple cubes/cuboids, but if it is possible to generate that kind of mesh then it would take much longer to do than to use SHM!

Teosim October 9, 2017 05:55

2 Attachment(s)
Hi Katt,
to get rid of protrusions try to run surfaceFeatureExtract, like Charlie suggested.

Then you could increase the refinement level on extracted edges in snappyHexMesh dictionary (castellatedMeshControls.features):

Code:

    features
    (
       
        {       
        file  "mySTL.eMesh";
        level 4;
        //levels ((0.1 4));   
        }   
       
    );

This way you should be able to capture geometrical features near edges with a minimal impact on overall mesh cell count.

Katt October 12, 2017 08:20

Hi guys,

Thanks for the explanation. I could generate my mesh well with OpenFoam 3.x but I would like to have it in Foam-Extend 1.6.

I see that the "surfaceFeatureExtract" utility has been modified from OpenFOAM 2.2 version and exporting the final mesh from OpenFOAM3.x to FoamExtend1.6 is not straight forward.

Though there is surfacefeatureExtract in foam-extend-1.6, its behavior is quite weird. The .eMesh file is generated in an incompatible form that snappyHexMesh complaints.

Any advise here.


All times are GMT -4. The time now is 07:30.