I wanted to experiment with other ways to build the mesh. I noticed that offsets <0 can cause unexpected problems with thick plates and small curvatures
Solo rondelle1.pmx (6.5 MB)
I think I’ve seen this behavior with other models in the past. I couldn’t find any existing reports of it on the forum, though.
Once you set the “First order elements” option to “None” you can extract them without any problem.
Solo rondelle 2.pmx (4.2 MB)
So it’s Gmsh’s shell mesh optimization breaking it.
I merged everything with the MERGE option and it is currently being calculated.
It may struggle more with such compounds having T connections.
With Mesh optimization set to Quasi-structured quad like in your model, the Mesh tab level Thicken Shell Mesh tool results in negative jacobians (all elements inverted in the transparent regions):
Without this setting, the mesh is correct:
Other optimization types also work. The Quasi-structured quad algorithm is known for some bugs in Gmsh.
I noticed a problem, so I first created a collection of elements, which I changed in the part. Only then did I extrude the surface, and everything looked fine. Your solid is probably faster, created directly from the shell after changing the surface orientation?
It’s pretty much the same tool, but I prefer the one in the Geometry tab because it’s part of the mesh setup items available in the tree. The version in the FE Model tab can be also used on imported meshes (when there’s nothing in the Geometry tab), though.
1 Like
Isn’t it easier to mesh a flat surface than a solid?
With the Thicken Shell Mesh tool in the Geometry tab, you still start from a shell mesh and just add this item to convert it to a solid mesh at the end of the meshing process. It’s the same workflow as with the Thicken Shell Mesh tool in the FE Model tab, but more automated.
1 Like
I cannot reproduce this problem. On my side, the shell mesh generated has a perfect shape, and if I thicken it, it retains it.
The mesh optimization in the Sehll Gmsh only works on 2D meshes, which is the basis for the thicken command. In the model, the mesh is deformed internally. That is strange for the thicken shell mesh command.
I can reproduce this. I will look into it since I think I could detect inverted elements and invert them again.