Hi,
I’m performing new tests in order to have s structured mesh: how do i’ve to proceed so that the mesh resmains coïncident in the current example?
Using Sweep, i cannot impose the number of layers as in Extrude, am i correct? such option will be usefull won’t be?
Additional question: how to have only quads on the triangle surface (as in “shell gmsh”)? options seem identical but i fail all the time.
Thanks for you feedback
Paul
test.pmx (438.7 KB)
It is when meshing a compound with the Transfinite algorithm. Otherwise, when there are multiple parts meshed with Extrude, Revolve or Sweep algorithms, you can use the Merge nodes tool after meshing.
Yes, that’s right.
Thanks Jakub for your quick answer,
that’s exactly what i would like to do, but since we cannot impose a discretization, submeshes are not coincident (i think i’ve used the same size)
is it a gmsh or a pmx limitation? is it on the roadmap to propose such (usefull) features?
In this case, you may have to disable Transfinite 3-sided faces and Transfinite 4-sided faces settings to achieve that but then the mesh won’t be structured.
Or use Sweep instead of Extrude and select Quasi-structured quad.
They don’t have to be perfectly coincident (the Merge nodes tool has a tolerance setting) but indeed they should be at least approximately coincident and you need to enforce matching numbers of elements on the common edges (using meshing parameters and local refinement).
Gmsh doesn’t have the sweep algorithm, it was developed independently in PrePoMax (although it’s based on Gmsh’s quad meshing) and it happened quite recently so it may still be improved in the near future. Feel free to create a feature request in the “green” section of the forum.
I know but nodes must be as close as possible; indeed one of the 2 nodes is simply moved are it may cause element quality loss
