[ff3d-users] Suggestion for an adaptive meshing procedure using GMSH

[Dominique Deloison]

Hi Stéphane,

I am not sure I really need an adpative meshing procedure but I saw some 
messages on this topic.
Here is a suggestion to implement this.

Gmsh can generate a mesh using a background mesh. The background mesh is 
defined using the GMSH post-processing format. Besides the mesh 
definition, it requires for each cell (or vertice I am not sure about 
this point) the whished size. Gmsh can then define a new mesh respecting 
the size distribution defined by the background mesh.

To implement this with ff3d, it would be useful:
- to have acces for each cell to a value of its size,
- to save results using the GMSH post-processing data.

Then the procedure would be the following one:
- generate a coarse mesh with GMSH,
- solve the equations with ff3d,
- computes an error estimator and the wished size for each cell (a very 
basic criterion could be to half the size where the gradient  is high),
- save the new size distribution using the GMSH format.
- create a new mesh with GMSH using the results as background mesh.

Besides the fact that GMSH is a good pre/post, this application could 
justify the addition of its format to save results.
If you don't have the time to do this,  people in need for it  could 
probably create their own GMSH-compatible results with the "ofstream" 
command.

Best regards,

Dominique.