Personal tools
You are here: Home Forum Use NetGen3D fails seemingly random when meshing ellipsoids

NetGen3D fails seemingly random when meshing ellipsoids

Up to Use

NetGen3D fails seemingly random when meshing ellipsoids

Posted by Peter Toson at February 27. 2013

Hi!

 

I have a strange problem. I wrote a python script that reads a csv-File containing position, orientation and size of ellipsoids, creates the geometry and meshes it using wire discretisation, mephisto and netgen 3D.

When processing a file containing only (format: center-x;center-y;center-z;orientation-x;orientation-y;orientation-z;size-x;size-y;size-z)

+3.769233e+01;+4.474058e+01;-2.360167e+03;+9.941531e-01;+1.013820e-01;+3.716203e-02;+7.000000e+01;+1.000000e+01;+1.000000e+01

My script fails with the following error message

th. 140058766530304 - Trace /local01/salome/V6_5_0p1/modules/src/SMESH_V6_5_0p1/src/SMESH/SMESH_subMesh.cxx [1780] : NETGEN_3D failed on sub-shape #2 with error COMPERR_ALGO_FAILED "Exception in  netgen::OCCGenerateMesh() at Swap Improve"
"compound" has not been computed:
-  "NETGEN_3D_3" failed on "1". Error: Algorithm failed. Exception in  netgen::OCCGenerateMesh() at Swap Improve

 

However, only a small change to the input file makes the script work:

+3.769233e+01;+4.474058e+01;-2.360167e+03;+9.941531e-01;+1.023820e-01;+3.716203e-02;+7.000000e+01;+1.000000e+01;+1.000000e+01

 

Are there any "failsafe" settings for NetGen for meshing ellipsoids? I have multiple files containing ~100 ellipsoids and there are always 1-5 ellipsoids making problems. Should I try another 3D mesher (tetrahedronds are a requirement though)?

Re: NetGen3D fails seemingly random when meshing ellipsoids

Posted by Saint Michael at February 27. 2013

I can propose some variants to try.

1) Try to create greater toruses, say with scale factor 10, and after meshing is done, scale the mesh back to your desired size.

2) Try to use other algorithms e.g. (1) NETGEN 1D-2D-3D, or (2) NETGEN 3D + NETGEN 1D-2D, (3) NETGEN 1D-2D-3D + local Wire discretization (local == on a sub-shape), or etc.

3) Create a toroidal mesh by revolving a meshed disk. After revolution you are to (1) merge nodes and (2) split 3D elements into tetrahedra (menu Modification / Split into Tetrahedra).

St.Michael

Re: NetGen3D fails seemingly random when meshing ellipsoids

Posted by Saint Michael at February 27. 2013

Sorry, I read ellipsoid and think of torus. So the variant 3) is not feasible.

Re: NetGen3D fails seemingly random when meshing ellipsoids

Posted by Saint Michael at February 27. 2013


Some general notes.

3D optimization is the most unstable part of NETGEN 3D, so your results (~10% failure cases) is very testimonial.

MEFISTO makes pour mesh on spheres and very probably on your ellipses. Thus using other 2D algorithm could help.

St.Michael

Re: NetGen3D fails seemingly random when meshing ellipsoids

Posted by Peter Toson at February 28. 2013

Thanks for your input so far. For The record, Netgen1D-2D-3D did not work. I'm trying the other suggestions.

Re: NetGen3D fails seemingly random when meshing ellipsoids

Posted by Peter Toson at February 28. 2013

Okay, problem solved. Netgen 1D-2D-3D did the trick. Sorry about my previous post - wire discretisation - netgen 2D - netgen 3D did not work. That combination failed at creating 2D meshes (that's why i tried MEPHISTO in first place and avoided netgen 2D)

 

Thanks,

-- Peter

Powered by Ploneboard
Document Actions