[Gmsh] Meshing a dam

[Jean-Francois Remacle]

At 11:54 14/06/2004, Jose Paulo Moitinho de Almeida wrote:

>Hello
>
>Some time ago we tried to use gmsh to obtain a tetra mesh of a dam, for which
>we had a "not perfect" hexahedral mesh.
>
>We used a semi-manual procedure, automatically extracting the points in the
>faces and the corresponding splines and then we defined by hand, one by one,
>the ruled surfaces which were finally merged. At least this is how I now
>remember the process...
>
>The resulting mesh had too much bad elements and we needed the mesh quickly,
>so we implemented "by hand" an hex to tetra conversion and left the geo file
>for future study.
>
>After another look at this file, I corrected some problems with inconsistent
>orientations, but cannot find further problems. Nevertheless we keep having a
>few unreasonable elements (slivers?), both with the large characteristic
>length (just dividing the hexs in tetras) and with small one.
>
>Alternatives:
>
>1) There are bugs in the geo file. This is quite probable. In that case could
>you please tell me where? Or even better: how can I help in implementing a
>geometry checker? Quite a few questions posted to the list are due to simple
>mistakes in the geometry file.


The geo file seems fine to me. I tryed to mesh the dam with elements
of size 5 and the resulting mesh has some bad elements, with aspect
ratios (gamma factor, see the book of P.L George and P. Frey) of
6*10e-4, which is bad. This is a VERY known (bad) feature of gmsh to
generate badly shaped tetraedron. It is in fact the case of ALL delaunay
based mesh generators. There should be a mesh enhancement
algorithm that eliminates the slivers by means of edge ans face
swappings. There is one in gmsh but completely bugged ... This
is one task very high in the TODO list, but disappointingly, we have
only a very little percentage of our time for gmsh...

I hope it answers your question.

JFR

>2) It is a problem with gmsh. Then I just hope that this file may help as a
>test case.
>
>Regards
>
>ZP
>
>_______________________________________________
>gmsh mailing list
>gmsh at geuz.org
>http://www.geuz.org/mailman/listinfo/gmsh


Prof. Jean-François Remacle
Dept. of Civil Engineering and
Center for Systems Engineering and Applied Mechanics
Bâtiment Vinci, Place du Levant 1
Catholic University of Louvain
B-1348 Louvain-la-Neuve, Belgium.
Tel : +32(0)10 472082
Fax: +32(0)10 472179