[douglasx]

Hi Foamers,

I am currently using snappyHexMesh based on DTCHull tutorial in the 2.3 version.
In the tutorial, refinement in Z direction near the free surface is done by having small spacing in Z direction in blockMeshDict. Then some topoSet and refineMesh in X and Y directions are used to make sure block meshes near the ship hull is not having too big aspect ratio.

I am trying to add more refinement to this case by adding a cone shape refinement box (imported into SnappyHexMesh as stl file) behind the hull to capture Kelvin wake. However, in snappyHexMesh, it reports "cell *** of level * uses more than 8 points of equal or lower level".

After some tests I figured out that the problem is that the cone shape refinement box is intersecting with the refineMesh boxes. And when I split the cone into different parts, with each part fitted into each refineMesh box rather than making intersections, there is no more problem. However, splitting the cone left gaps in between.

I am wondering if this is the normal solution or there are more ways of going around the problem? So the regions refined by refineMesh cannot have any intersections with other refienment boxes?

I have attached a figure showing the basic configuration of my cone shape refinement box, ship hull and refineMesh boxes.

Any comment is appreciated.

Thanks,
Zeliang

[thg]

Hi Zeliang,

in principle there should be no problem with intersecting refinment regions or volumes.
But from the error message I would derive that you try to refine a mesh region (cells)
which was already deformed/snappped. Snappy can oly refine on pure hexyeder elemnts.
For me it look that you try to refine a mesh which is already processed with snappyHex. You should start from the blockMesh.

[douglasx]

Hi Thorsten,

Thank you for your reply. The procedure I took to building the mesh is blockMesh, refineMesh, snappyHexMesh, and the error message takes place. When I split my refinement regions to avoid intersections, the error message goes away.
I have also tried the suggestions of "blockMesh, snappyHexMesh, refineMesh" procedure. For this procedure, I do not have to worry about intersection of regions that upsets snappyHexMesh, but the resulting mesh quality is not as good (thousands of highly skewed cells).
So there is some kind of trade-offs.
Have you tried refineMesh+snappyHexMesh combination and have the ideal mesh?

Thanks,
Zeliang