Elmer Discussion Forum

Hi,
I'm simulating a drift tube (with special cathode shape) with gmsh, then Elmer and garfield++, according with guide (https://garfieldpp.web.cern.ch/garfield ... er_doc.pdf).
Since drift tube has a big volume (r_wire = 0.015 mm, r_tube = 5 mm,l_tube is so minimal as possible), I have to use transfinite algorithm (transfinite lines, surfaces and volumes). So I have well structured mesh.
Garfield++ requires mesh elements to be only second-order (quadratic) tetrahedral, thats why I do not use "Recombine" option.
But when I import mesh to ElmerGrid or ElmerGUI, mesh reading seems to be bad, i.e. some surfaces disappear. And ElmerSolver results wrong.
But when I'm not using transfintie algorithm, everything is allrigth with mesh reading in Elmer.

Is it a convertation problem of mesh, that has been done with transfinite algorithm?
I attached t2.geo, gmsh log-file and ElmerGUI screenshot.

gmsh version is 2.10.1
Elmer version is 3.0 Mesa 17.2.8 (Vendor: X.Org, Renderer: AMD TURKS (DRM 2.43.0 / 4.4.0-127-generic, LLVM 5.0.0) )
Here what I'm done:
$ gmsh t2.geo -3 -optimize -order 2 > gmsh.log
$ ElmerGUI

I'll be glad for any help!

Hi

The keyline may be here: So ElmerGrid fails to find the side elements attached to the bulk elements. This happens both with linear and quadratic elements. Either there is something wrong in the way Elmer tries to locate the boundaries, or some of the boundaries are not attached to bodies i.e. they have separate numbering.

-Peter