[Gmsh] double surfaces/lines between adjacent bodies

[Peter Raback]

Hi Matthias,

Do the lines share the same nodes?

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.

If no, this might be treated by ElmerSolver with the same functionality that is used for the periodic BCs, I think? But then you would have to have to have both lines existing so you can define a linear dependence between them.

BR, Peter

________________________________________
From: Zenker, Dr. Matthias [Matthias.Zenker at erbe-med.de]
Sent: 11 March 2009 16:21
To: elmerdiscussion at postit.csc.fi; gmsh at geuz.org
Subject: [Elmerdiscussion] double surfaces/lines between adjacent bodies

Hi Elmer and gmsh  list,

I use gmsh and Elmer to simulate an electrical current flowing through a conductive medium between two electrodes. I get the geometry from my CAD colleagues in STEP format, import it in gmsh, mesh it, read the mesh with Elmer and do the FEM calculation there. Most of the time, the calculations are 2D. This works in principle, but there is the following problem:

Electrode and medium have common surfaces (resp. lines in 2D). When I get the geometry in STEP format, every part (Electrodes and Medium) has its own outer surface (line). This leads to double surfaces (lines) between the adjacent bodies. For geometry and mesh import, this is no problem, neither for gmsh nor for elmer - the mesh is imported OK in Elmer. But when I do the simulation, I see that no current flows through this double surface (although only one of them is defined as physical in gmsh, and only that one is seen in Elmer).

Removing the double surface/line in gmsh is rather tedious (Identify the problematic regions where there is a double line between adjacent surfaces. Delete at least one of the adjacent surfaces. Delete the corresponding line at the interface. Delete its points. Redefine the surface(s) which was (were) deleted before. It gets more complicated in case of only partially overlapping lines which have to be partially reconstructed.).

Question: How do I deal with this problem? Is there a more elegant (and less time consuming) way than manual removal of the double lines? (The autocoherence function in gmsh which is intended to remove double duplicate entities in geometries is switched on, but does not seem to have any effect.)

I imagine that this must be a known problem, since doing simulations with geometries imported from 3D CAD systems is not so uncommon after all.

Further it is not totally clear to me where the problem really comes from. I suspect non-identical mesh nodes at the interface, which inhibit current flow across the border.

Ideas and suggestions are more than welcome...

Thank you,

Matthias

----------------------------------------------------------------
ERBE Elektromedizin GmbH
Dr.rer.nat. Matthias Zenker
Dipl. Phys.
Grundlagenentwicklung Technologie
Waldhoernlestrasse 17
72072 Tuebingen
Phone + 49 7071 - 755 226
Fax + 49 7071 - 755 5226
mailto:Matthias.Zenker at ERBE-med.de
www.ERBE-med.de<file://www.ERBE-med.de>
----------------------------------------------------------------




ERBE Elektromedizin GmbH
Firmensitz: 72072 Tuebingen
Geschaeftsfuehrer: Christian O. Erbe, Reiner Thede
Registergericht: Stuttgart HRB 380137