[katiuskas]

Dear all,

I have been trying to refine my (2D) mesh in a certain region, let's say with a refinement level of 1. I have done so in two different ways: using blockMesh and then mergePatchPairs, and through snappyHexMesh. Both methods yield similar results: the "transition zones" between the different regions of the mesh use tetra elements (see first figure). In spannyHexMesh, I know you can control the "smoothness" of this transition with the nCellsBetweenLevels entry. So far so good...

My doubt comes from having seen meshes where no transition zone exists at all by using snappyHexMesh (see the second figure, taken from the wiki). Note that this is neither a non-conformal mesh nor a "hanging node" situation. It's just that some of the cells are now pentagons instead of squares (if you allow me to use the 2D jargon) and thus can have 5 neighbours instead of 4. Now, my question can be summarized into the following:

1. How can I create such a mesh? Every time I try, I always get tetra elements.
2. Would this "penta" mesh behave better than mine in terms of mesh quality and solver?

Thank you very much!

[Yann]

Hi Pere,

This is a matter of visualization in ParaView.
When loading your case in ParaView, untick the "Decompose polyhedra" option and you should get the same visualization as in the wiki.

Cheers,
Yann

[katiuskas]

Hi Yann,

Thank you very much for the rapid response! I would have never guessed it was a visualization thing. I did not know that some ParaView filters are not good at handling polyhedra and that's why it decomposes them into tetra and pyramids. Good to know!