[Gmsh] [PROBLEM] Elements normals

Christophe Geuzaine cgeuzaine at ulg.ac.be
Thu Mar 12 20:52:08 CET 2009


Xavier Mencaglia wrote:
> Dear,
> 
> I'm currently using the version 2.2.4 of GMSH to mesh a stp file which 
> was created by Catia V5 R18.
> When I check the normals of the surfaces of the step file, they are well 
> oriented.
> And then I mesh the step file (2D mesh - triangle elements only)  and I 
> check the normals of the elements. I can see some normals that were 
> inverted during the meshing operations. So some normals remain stable, 
> pointing in the same direction as the normals in the step file and some 
> normals are inverted.
> It's a little bit problematic for me because I'm working with composites 
> so I need to know and control the normals of the FE. The stacking 
> sequence is strongly dependent on normals orientation especially if 
> composites are not symmetric.
> 
> So I have 2 questions about that:
> 
> 1/ Is this problem well known and is it linked to the way the mesh is 
> performed ?
> 2/ If  this situation is normal, could you explain me why some normals 
> are inverted ? What is the rule that pilots this mechanism ?

Hi Xavier - The situation is (sort of) "normal". We make sure to orient 
the normals correctly for Gmsh models, but could not figure out a nice 
way to do it with OpenCascade models, due to the fact that the 
orientation of the bounding edges of a surface in OCC do not seem to 
match the orientation of the surface.

The routine that performs the reorientation is

   void orientMeshGFace::operator()(GFace *gf)

at the end of Mesh/meshGFace.cpp.

Let us know if you have any ideas on how to fix this: it is indeed a 
annoying problem...

Christophe


> 
> Thanks in advance for your answer,
> 
> Kind regards,
> 


-- 
Prof. Christophe Geuzaine
University of Liege, Electrical Engineering and Computer Science
http://www.montefiore.ulg.ac.be/~geuzaine