[Gmsh] Strange meshing behaviour

[Christophe Geuzaine]

Hi Peter,

I dug a bit deeper, and the issue is related to the "unification" of surfaces that is performed after a BooleanUnion operation, through the OpenCASCADE "ShapeUpgrade_UnifySameDomain" class.

This was added in Gmsh 4.5, and can be deactivated with 

   gmsh.option.setNumber("Geometry.OCCUnionUnify", 0)

With this your geometry behaves as in Gmsh 4.4.1. It would be nice to update your report to the OCC folks to let them know that the bad surface is generated by ShapeUpgrade_UnifySameDomain.

Thanks,

Christophe

-------------- next part --------------
A non-text attachment was scrubbed...
Name: cap_cut_fix.py
Type: text/x-python-script
Size: 2656 bytes
Desc: not available
URL: <http://onelab.info/pipermail/gmsh/attachments/20200312/4425c7b7/attachment.py>
-------------- next part --------------



> On 12 Mar 2020, at 07:14, Peter Johnston <p.johnston at griffith.edu.au> wrote:
> 
> Dear Christophe,
> 
> Thanks for your response. 
> 
> As suggested, I asked a question, using your data, on one of the OCC forums. I hope that was the correct thing to do?
> 
> However, I do have one further question for you. The python scripts that I wrote that use these OCC commands were written about six months ago and I was able to produce sensible meshes under gmsh 4.4.1. At least these must have been sensible, otherwise my numerical simulations would not have run and produced sensible output. Are you using a newer version of OCC than, say, when you released gmsh 4.4.1? I don't upgrade my gmsh all that often.
> 
> Thanks for your thoughts.
> 
> Kind regards,
> 
> Peter.
> 
> Associate Professor Peter Johnston (FAustMS, FIMA)
> School of Environment and Science
> Griffith University | Nathan | QLD 4111 | Technology (N44) Room 3.19
> T +61 7 373 57748| F +61 7 373 57656 Email p.johnston at griffith.edu.au
> From: Christophe Geuzaine <cgeuzaine at uliege.be>
> Sent: Wednesday, 11 March 2020 10:42 PM
> To: Peter Johnston <p.johnston at griffith.edu.au>
> Cc: gmsh at geuz.org <gmsh at geuz.org>
> Subject: Re: [Gmsh] Strange meshing behaviour
>  
> 
> Hi Peter,
> 
> Interesting. This seems to be a problem with the generation of the geometry, not the meshing.
> 
> Attached is the .brep file of the buggy surface (debugSurface.brep):
> 
> - the OpenCASCADE representation is wrong (see attached picture of the OpenCASCADE-produced STL triangulation), which seems to indicate that the parametric representation is wrong
> 
> - the surface also leads to an incorrect mesh with Gmsh 4.4.1
> 
> I would suggest to send the "debugSurface.brep" file to the OpenCASCADE developers for further analysis?
> 
> Christophe
> 
> <bug.png>
> 
> 
> 
> > On 11 Mar 2020, at 12:40, Peter Johnston <p.johnston at griffith.edu.au> wrote:
> > 
> > Hello,
> > 
> > I have just observed some strange behaviour with my python script running gmsh. Attached is a cut down version of a larger script that used to work under gmsh 4.4.1. This script creates a 3D volume using OCC and contains multiple surfaces. When I mesh the surface (ie use the 2D meshing button on the gui) surface number three gives some strange results; so strange that I cannot create a volume mesh. I have also attached a picture of the mesh on surface 3, indicating the places where the mesh is misbehaving. I am currently using gmsh 4.5.4.
> > 
> > (Please note that I had to change the file extension to escape institutional spam filters. It is really a python script.)
> > 
> > Any insights would be much appreciated.
> > 
> > Thanks very much,
> > 
> > Peter.
> > 
> > Associate Professor Peter Johnston (FAustMS, FIMA)
> > School of Environment and Science
> > Griffith University | Nathan | QLD 4111 | Technology (N44) Room 3.19
> > T +61 7 373 57748| F +61 7 373 57656 Email p.johnston at griffith.edu.au
> > <gmsh-prob.png><cap_cut.txt>_______________________________________________
> > gmsh mailing list
> > gmsh at onelab.info
> > https://protect-au.mimecast.com/s/HChgCNLwQEiBWGQEumqCoN?domain=onelab.info
> 
> ? 
> Prof. Christophe Geuzaine
> University of Liege, Electrical Engineering and Computer Science 
> http://www.montefiore.ulg.ac.be/~geuzaine
> 
> 
> 

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