Quasi-Stuctured Quad conformal with 1D submeshes

I just read the releases notes of salome 9.10 which has some interesting features, one is the Quasi-Stuctured Quad gmsh mesher which took my interest, nevertheless the gmsh algorithms looks like have a different behavior to the ‘classic’ ones of salome. As a windows user, I did not dig a lot in this algos as they were problematic in v9.9. in any case, now that they work correctly, i wanted to give a try to them,
and using this algos in a submesh (in a face), does not make a conformal mesh as can be seen in the image below:

my question is, how can i make that the 2D submesh (ie. the face in the upper face) is conformal to the rest of the body, where they 1D discretization have been already done. (i changed the order of the submeshes in case that it was that the issue, it did not change the resulting mesh.)
my workflow (that works with others algos such as netgen 2D) is:

  1. create a 3D mesh with extrusion 3D algo (not defining any 1D algo)
  2. create a 1D sub mesh for all edges wire discretization 15 elements
  3. create a 2D mesh of face use gmsh, and quasi struc algo
  4. comput it

the mesh will not be conformal.
here is the hdf file for the screenshot.
gmshConformal.hdf (291,5 KB)
best regards,

after some testing, i would like to rise this as a bug @NabilG of SMESH while using Gmsh algorithms in salome 9.10 (tested in windows 10 by using the .EXE installer).
here is the file to reproduce the issue:
bugGmsh.hdf (163,5 KB)


The Quasi-structured Quad is marked as an experimental algorithm in gmsh.
We sould change in smesh to display that information.
However it is weird because the results seems random.
I tried your example and some times it works…
It should be investigated further.

1 Like

hello yugi,
from my different tests, at least from what i observe, it is not that it works it is that when running the gmsh algo, it uses a 1D discretization of 1 or 2 elements only, in my example it is 2 so ‘sometimes’ it looks correctly, but if you change it to another value, lets say 15, there is no time where it works. in this case for the gmsh algo, it will produce a mesh where the number of elements will be always 2 (ignoring the setted up 15).
best regards.