[Gmsh] GMSH for Thermal Codes?

[Christophe Geuzaine]

Joe Koski wrote:
> I have some old "legacy" thermal finite element codes that I use
> occasionally for consulting work, that would benefit from the use of a mesh
> generator. To date, I've been generating simple meshes by hand or with
> simple ad hoc Fortran codes. When I first found GMSH, I thought I had found
> the answer, but further examination of the manual has brought up some
> questions and issues. If any of you experienced GMSH users have some
> insights, I'd like to hear them.
> 
> First, thermal modelers spend about half of their mesh generation time
> attaching convection and thermal radiation boundary conditions to various
> surfaces of the model. It appears that GMSH has no facilities for tracking
> boundary conditions. Is this correct? Some commercial thermal mesh
> generators permit attachment of boundary conditions to the problem geometry,
> since entities such as surfaces are much more easily identified than a list
> of mesh nodes or element surfaces. Has this been thought about for GMSH? I
> realize that this is not usually an issue in structural models.

Joe - Could you give an example on how this "attachment of boundary
conditions" works? More specifically, how does it differ from the way we
define groups of geometrical entities in Gmsh (using the "Physical
Line", "Physical Surface", etc.,  commands)?

> 
> Second, while I think I could live most of the time with tetrahedrons and
> triangles, what is the user experience in "recombining" tets and triangles
> into hexahedra and rectangles? Does it take element by element inspection,
> or can it be somewhat automated?

There is no recombination algorithm implemented in 3D. In 2D, you can
recombine triangles into quadrangles with the "Recombine" command. Check 
the docs for the exact syntax.

Cheers,

Christophe

-- 
Christophe A. Geuzaine
Applied and Computational Mathematics, Caltech
geuzaine at acm.caltech.edu - http://geuz.org