[Gmsh] An idea
CEM ALBUKREK
calbukrek at gmail.com
Fri Oct 24 22:00:58 CEST 2008
Hello Christophe,
I have been using Gmsh successfully to build complex tet meshes as input to
OpenFOAM. To give you a sense of the complexity attached is a picture of a
typical surface mesh, which is an assembly of NAS and STL files combined
into one closed volume using some scripts I developed. You have warned on
your website that GMSH is not so good at large meshes from STL's; what
threshold (# of tets) did you have in mind? I am able to go as far as the 32
bit memory allows me to at this point. May be the trick is the application
of effective surface mesh smoothing techniques before moving onto volume
meshing...
One issue with CFD is that one can march an unsteady (varying time)
simulation only as fast as the smallest tetrahedron allows; i.e.
velocity * time_step <= tet size
for each one of the tetrahedra.
Sometimes small elements are unavoidable no matter how great a surface mesh
one puts together. One way around the problem is to blindly erase the
tetrahedra smaller than a threshold and run the flow simulation around these
tiny blockages, which are too small to affect the bulk flow. I would like to
implement a little sub-routine to do the same thing with GMSH. For instance
we have the mesh quality filters for beta, gamma & theta. In a similar
fashion I would like to put one for the size (or volume) and also be able to
click delete. Do you see any problem with this? May be FEM solvers could
benefit from this kind of feature as well knowing tiny bubbles are not
uncommon in molded plastics and cast metals.
What do you think?
--
CEM M. ALBUKREK, PhD
Computer Aided Engineering Design Consultant
Tel: (857) 234-1035
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.geuz.org/pipermail/gmsh/attachments/20081024/ffa9eac7/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Case-SurfaceMesh.png
Type: image/png
Size: 229979 bytes
Desc: not available
URL: <http://www.geuz.org/pipermail/gmsh/attachments/20081024/ffa9eac7/attachment.png>