Dear Sir or Madam, I am a new user of gmsh, and I have stubled across some possible issue with GMSH's algorithm. The version I used is the most recent binary (git). In my application of gmsh, I feed it a .geo-file, which already contains a partial triangulation of some of the model's surfaces. This seems to lead to very degenerate tetrahedra, that even the optimization algorithm cannot fix any more.
Having analyzed and simplified the issue, I found out that in my complex example, already after the 2D meshing step the mesh seems to feature bad angles. In the appended file "issue_gmsh_2018.geo" you find a very much simplified example, demonstrating the said behavior: With any 2D algorithm you support, I obtain very lengthy 2D triangles in the "border", as you will see. However, what I would expect is that the algorithm would subdivide the edges of the "border", resulting in a rather fine, and well-shaped (e.g. no small angles), triangle mesh in the "border". Moreover, I would expect that the inner quad is filled with fine triangles on the outside, and more coarse triangles in the inside. (See also: http://gmsh.info/gallery/gbr_small.png ). Having done some research, I speculate that your 2D delaunay mesher probably computes a "constrained Delaunay triangulation" (see also: https://www.cs.cmu.edu/~quake/triangle.defs.html ), and therefore does not dare to split any edge which I specify in the .geo file, unless the 1D mesher splits it. To me this seems to be the root of the problem. I propose the solution of my problem is to rather have computed the "conforming Delaunay triangulation" (see also: https://www.cs.cmu.edu/~quake/triangle.defs.html ). Could you please verify if this line of thought is right? If so, can you help me achieve what I want with gmsh? I already tried out several combinations of algorithms and parameters, which was was unsuccessful. Thanks and Best regards A. S
issue_gmsh_2018.geo
Description: Binary data
_______________________________________________ gmsh mailing list [email protected] http://onelab.info/mailman/listinfo/gmsh
