[Gmsh] [Elmerdiscussion] interface region between two bodies

[Peter Raback]

Hi Shawn

The Gmsh2Elmer parser tries to locate the parent elements of lower dimensional elements. If they are not found then the remaining lower dimensional elements are assumed to be bulk elements. The reason for the problem could be that there are triangles that are not faces of any tets in gmsh file, or that the ElmerGrid parser cannot locate the parent elements properly (using the node numbering).

In ElmerGUI you can set the cli parameters of ElmerGrid in  Configure -> elmergrid -> String: 
In this case I would not try to merge any nodes but rather to remove some unused entities. You could try the "-autoclean" flag. This might, however, not mend the problem but instead remove some essential triangles from your mesh.

BR, Peter

PS. Elmer discussion has moved to http://www.elmerfem.org/forum and this Elmer list should be obsolite.

________________________________________
From: Shawn Fostner [sfostner at physics.mcgill.ca]
Sent: 04 December 2009 20:53
To: gmsh at geuz.org
Cc: elmerdiscussion at postit.csc.fi
Subject: [Elmerdiscussion] interface region between two bodies

Hi all,

I was working on a problem with this geometry a few weeks ago, and had
some suggestions from the list. In the end I think I found a geometry
that works for me (see the included file), mostly, but for 2 problems
and I can't tell if they are in elmer or gmsh. The initial mesh in gmsh
seems to work, and mesh (its included here), but when I load it into
elmer and run an electrostatic solver the following happens:

1) There is a phantom 3rd body somewhere, but for the life of me I can't
figure out what it is. It might be related to the next problem but I'm
not sure. I've tried to see it using the body coloring or labelling in
elmer, but in both cases I can't find it, and with the number of
elements the processing slows to a crawl, to the point where I can't
manipulate the model.

2) After running the solver, MOST stuff looks ok, but for a subtle
problem: Plane 18 (between the two regions), ends up with wildly varying
potentials (and hence electric fields) right at the interface. I suspect
thats where there is the phantom object, or something, but I can't tell
for sure. It does not appear to be two overlapping planes at that point
(though I've had a problem with that before as well, not in this geometry)

3) I have a number of what appear to be redundant elements, it complains
of it during both the solving stage, and when I look at the data later.
I suspect it is related to both 1 and 2.

Suggestions? I saw a suggested solution that MIGHT be similar in the
archives by Peter, but I can't for the life of me figure out how to do
this in the windows gui.

>/ If yes, then you could include the command line option
/>/ '-merge eps' in the gmsh import in ElmerGUI/ElmerGrid. This
/>/ will join nodes that are separated by a distance smaller than eps.


Thanks
Shawn
/