<html><body><div style="color:#000; background-color:#fff; font-family:arial, helvetica, sans-serif;font-size:10pt"><div>I recently constructed a 3D model in gmsh to investigate capacitive
coupling among a plurality of microstrip conductors oriented at various
angles to one-and-other. One model of interest contains three
dielectric regions. Each dielectric region is defined by a physical
volume entity in gmsh. Here is the problem encountered when using the
gmsh generated 3D mesh (V2 ascii .msh format) in Elmer:<br><br><span style="font-weight: bold">Although the original .msh file from gmsh has 3 dielectric regions, Elmer claims that it has 5 bodies. </span>
Since the mesh is unstructured and is moderately complex (~300K
tetrehedral elements), displaying the 5 bodies identified by Elmer is
intractable. I found that the following procedure results in a correct
translation of the mesh from gmsh to Elmer.<br><br>1) In gmsh, construct a .geo file, generate mesh and save as ascii V1 .msh file.<br>2) Import ascii V1 .msh file (from 1 above) into EnGrid.<br>3) Export from EnGrid as a gmsh ascii V2 .msh file.<br>4) Open gmsh ascii V2 .msh file (from 3 above) with Elmer.<br>5) Proceed to build Elmer .sif file.<br><br>So
far, only when I generate a V1 .msh file and convert is to a V2 .msh
format with EnGrid can I successfully use the mesh in Elmer. I don't
know if this is a bug in gmsh or Elmer, or perhaps an incompatibility
between versions. I am using Elmer version 6.1.0.svn.5272.dfsg-1ubuntu2
and gmsh version 2.5.1~beta1~svn9724~dfsg-1 in linux mint 12. Any
comments and/or suggestions regarding this issue are greatly
appreciated.<br><br>Dr. Jack Mandelman (aka "The Modelman")</div></div></body></html>