Prismatic mesh - handling pyramids

[gbettega]

Hello,
I generated a prismatic mesh from a surface mesh patch.
Not all the elements are prisms (the major part is): the algorithm generates a small number of tets and pyramids, as usual.
Once (quasi) prismatic mesh is done, i would lilke to remesh the tetrahedral mesh connected to that, since some tets are very stretched. I would like to use mmg3D (as library).

I know experiments, and from:

Using MMG with prisms

The library in theory is compatible with the re-meshing of prisms (or wedges), but if I try to remesh a problem with just prisms it complains that there are not tetrahedra: – INPUT DATA %% out_step=5.mesh OPENED ** MISSING DATA. Check that your mesh contains points and tetrahedra. Exit program. prism_case.zip (15.1 KB) I am using the last version in master branch in GitHub compiled with GCC8.

that

“Mmg preserves the prisms entities but it doesn’t re-mesh it : it re-meshes the tetrahedral part of the mesh and preserves the interface with the prisms entities. This feature is useful for example if your initial mesh contains a prismatic boundary layer.”

however, unfortunately, as stated in

Handling pyramids in mmg3d

Hi Francesca, You have perfectly understood and yes, you will need to set required triangles (it should be sufficient) for triangles at the interface between tetra and other kind of elements (for prisms, Mmg do that automatically). Regards, Algiane

“Mmg doesn’t allow to preserve pyramids. Since the version [5.4.1] of Mmg you can provide and preserve “orphan points”, i.e. points that are not connected to the mesh if you specify these points as required […]”

Now, after the (quasi) prismatic mesh has been generated, I would like to preserve it during an overall re-meshing process, and this is possible only if the prismatic mesh is really fully prismatic (no pyramid inside).

Question:

1. when inizialiting the mesh, could I use

MMG3D_Set_prism

for a pyramid, by => defining <= is a degenerated prism?

2. will mmg3D “jump” over this degenerate prism == pyramid during remeshing and preserve it?

The same question for a tetrahedron initially belonging to the (quasi) prismatic mesh: could this be defined as a degenerated prism, so that mmg3D jumps over it and preserve it?

Grazie mille
Giovanni

[cprigent]

Dear Giovanni,

To both of your questions, I would answer yes: you can define a prism by calling the function MMG3D_Set_prism and put the same vertex twice to create a pyramid, or the same vertex three times to define a tetrahedron. Since MMG ignores the prisms during remeshing, they will be preserved.
However, for tetrahedra, you can also use the function MMG3D_Set_requiredTetrahedron to label a tetrahedron as required and ensure that it will not be modified by remeshing.
I would rather advise for this solution in general, because the output mesh containing degenerate entities might not be suitable for your applications.
Corentin