Degenerate 2D elements, GMSH mesh

Numerical methods and mathematical models of Elmer
mustafaiqbal
Posts: 14
Joined: 07 Jan 2017, 02:05
Antispam: Yes

Re: Degenerate 2D elements, GMSH mesh

Post by mustafaiqbal »

Matthias,

On inspection I see that there are overlapping nodes. The black horizontal lines with 0 inner radius coincide with the green triangles in the region close to the apex. I am working in the xy plane already.

Any ideas on how to stop Elmer from struggling with my mesh?

Thanks,
Mustafa.
mzenker
Posts: 1999
Joined: 07 Dec 2009, 11:49
Location: Germany

Re: Degenerate 2D elements, GMSH mesh

Post by mzenker »

Hi,

since Elmer didn't generate the mesh, it seems to me that the problem doesn't lie on the Elmer side.
IMO you should find out what goes wrong in the meshing process, i.e. in gmsh. You could ask on the gmsh mailing list if you have difficulties.

HTH,

Matthias
mustafaiqbal
Posts: 14
Joined: 07 Jan 2017, 02:05
Antispam: Yes

Re: Degenerate 2D elements, GMSH mesh

Post by mustafaiqbal »

Matthias,

I have posted on a GMSH forum and will see if the issue can be resolved.

Please the the attached image which shows a zoomed region of the conical apex in ElmerGUI after importing the original .msh from GMSH. Running this I get the degenerate elements in the solver, but the elements seem fine from this image? It is only after reconverting to .msh from the Elmer mesh that we see those degenerate elements.

In addition I can look at the original .msh in GMSH and there are no degenerate elements, so it seems to me that the problem is with the way ElmerGrid imports and converts the mesh? See the image attached with the original .msh in Elmer.

Thanks,
Mustafa.
Attachments
msh before elmer.png
(322.15 KiB) Not downloaded yet
ElmerMesh.png
(92.04 KiB) Not downloaded yet
mzenker
Posts: 1999
Joined: 07 Dec 2009, 11:49
Location: Germany

Re: Degenerate 2D elements, GMSH mesh

Post by mzenker »

Hi Mustafa,

from the information I have I cannot tell where the problem is.
Normally the mesh itself should not be changed by the conversion with Elmergrid. Only the numbering is changed, which is why we need to reconvert the Elmer mesh to gmsh format in order to examine the elements which Elmer complains about.
If you post the .msh file as saved by gmsh and your Elmer sif file, I (or somebody else) could have a look at it. If your mesh is too big to post it here, or if it is confidential, then it would be a good idea to construct a small testcase (mesh and sif) reproducing the problem, and post that one.

Matthias
pavel
Posts: 34
Joined: 17 Apr 2014, 17:08
Antispam: Yes

Re: Degenerate 2D elements, GMSH mesh

Post by pavel »

Hi,

I have also a big mesh in MagnetoDynamics ~1M elements converted from external format. ElmerGrid does not give any warnings and converts it OK. During the solution ElmerSolver returns ERROR messages conserned of 20 or so elements with their numbers. Like this:

Code: Select all

...
ERROR:: ElementMetric: Degenerate 3D element: 40423
ERROR:: ElementMetric: Degenerate 3D element: 40423
ERROR:: ElementMetric: Degenerate 3D element: 39643
ERROR:: ElementMetric: Degenerate 3D element: 40423
ElementMetric: DetG:   0.000E+00
ElementMetric: Node: 1 Coord:   2.342E-01   2.127E-01   4.880E-03
ElementMetric: Node: 2 Coord:   2.342E-01   2.127E-01   4.205E-03
ElementMetric: Node: 3 Coord:   2.343E-01   2.112E-01   4.205E-03
ElementMetric: Node: 4 Coord:   2.343E-01   2.112E-01   4.880E-03
ElementMetric: Node: 2 dCoord:   0.000E+00   0.000E+00  -6.750E-04
ElementMetric: Node: 3 dCoord:   9.720E-05  -1.460E-03  -6.750E-04
ElementMetric: Node: 4 dCoord:   9.720E-05  -1.460E-03   0.000E+00
ERROR:: ElementMetric: Degenerate 3D element: 39643
ElementMetric: DetG:   0.000E+00
ElementMetric: Node: 1 Coord:   2.342E-01   2.127E-01   4.880E-03
ElementMetric: Node: 2 Coord:   2.342E-01   2.127E-01   4.205E-03
ElementMetric: Node: 3 Coord:   2.343E-01   2.112E-01   4.205E-03
ElementMetric: Node: 4 Coord:   2.343E-01   2.112E-01   4.880E-03
ElementMetric: Node: 2 dCoord:   0.000E+00   0.000E+00  -6.750E-04
ElementMetric: Node: 3 dCoord:   9.720E-05  -1.460E-03  -6.750E-04
ElementMetric: Node: 4 dCoord:   9.720E-05  -1.460E-03   0.000E+00
ERROR:: ElementMetric: Degenerate 3D element: 39643
...
Though solving process does not stop and results look qualitatively OK, though convergence tolerance is quite high.

I do not know whether this is related or not, just for reporting. And I am not really sure about the quality of original mesh.

- Pavel
mustafaiqbal
Posts: 14
Joined: 07 Jan 2017, 02:05
Antispam: Yes

Re: Degenerate 2D elements, GMSH mesh

Post by mustafaiqbal »

Dear Matthias,

Please find the meshes and Elmer files attached, thanks.

Pavel I've seen issues with degenerate elements in 3D meshes, have you tried using "Optimise 3D" and "Optimise 3D (Netgen)" in GMSH?

Thanks,
Mustafa.
Attachments
project.zip
Elmer files
(7.44 KiB) Downloaded 339 times
1_5_after_elmer.zip
msh->elmer->msh
(1014.42 KiB) Downloaded 352 times
1_5.zip
original msh
(1.37 MiB) Downloaded 344 times
mzenker
Posts: 1999
Joined: 07 Dec 2009, 11:49
Location: Germany

Re: Degenerate 2D elements, GMSH mesh

Post by mzenker »

Hi Mustafa,

I have run your case, and get no complaints about degenerate elements.
I have converted the mesh with

Code: Select all

ElmerGrid 14 2 1_5.msh -autoclean
and then run ElmerSolver using the sif file in your project directory. It seems that the result is a zero concentration in the whole geometry.

Is this really the case where you get "degenerate elements" messages?

Matthias
raback
Site Admin
Posts: 4812
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Degenerate 2D elements, GMSH mesh

Post by raback »

Hi

There is a rather fresh solver that could be used to analyze the mesh a little. It was written for much more complex meshes so it might not be optimal for the case, but still you at least should get the distribution of the aspect ratio of the edges.

Code: Select all

Solver next_free_solver_id
  Exec Solver = Before Simulation
  Equation = ElemInfo
  Procedure = "ElementStats" "ElementStats"
  Show Cohorts = Logical True
  Number Of Cohorts = Integer 100
End
-Peter
mustafaiqbal
Posts: 14
Joined: 07 Jan 2017, 02:05
Antispam: Yes

Re: Degenerate 2D elements, GMSH mesh

Post by mustafaiqbal »

Matthias,

The correct case does not have 0 concentration everywhere, there is a bulk concentration at the top boundary, 0 concentration at the slanted surface (conical electrode) and no flux on the other 3 boundaries. I am surprised you don't get the error, I just repeated the same steps you did (ElmerGrid to convert the mesh) and still get the degenerate mesh issue.

Thanks,
Mustafa.
mzenker
Posts: 1999
Joined: 07 Dec 2009, 11:49
Location: Germany

Re: Degenerate 2D elements, GMSH mesh

Post by mzenker »

Hi Mustafa,

I didn't use your ElmerGUI project file, but the 1_5.sif which was in the project folder. Could you post the sif which you use when you get the error?

Matthias
Post Reply