The problem of meshing at edges

More
2 years 10 months ago - 2 years 10 months ago #4147 by VienC
You are not authorised to see this attachment.
You are not authorised to see this attachment.
You are not authorised to see this attachment.
You are not authorised to see this attachment.
You are not authorised to see this attachment.
You are not authorised to see this attachment.
Hi!

I created geometry and meshed it using Ngsolve
Code:
ngmesh = occgeo.GenerateMesh(meshsize.very_fine) [attachment=934]meshing_bug.PNG[/attachment] [attachment=935]FFFP_mesh.py[/attachment] [attachment=936]elektrode_single.stp[/attachment]
As you can see from the attached images, the edges at the bottom were meshed differently and wrongly. In my opinion, the meshing should be performed symmetrically at the bottom, which is not the case.
I was wondering what is the reason and how I can handle this problem. I attached the code to easily reproduce the result.
I ran Netgen on Window, The Output in the command line is given below:
Code:
NETGEN-6.2.2105 Developed by Joachim Schoeberl at 2010-xxxx Vienna University of Technology 2006-2010 RWTH Aachen University 1996-2006 Johannes Kepler University Linz Including OpenCascade geometry kernel optfile ./ng.opt does not exist - using default values togl-version : 2 OCC module loaded loading ngsolve library NGSolve-6.2.2105 Using Lapack Including sparse direct solver Pardiso Running parallel using 4 thread(s) importing NGSolve-6.2.2105 (should) load python file 'FFFP_mesh.py' Number of colours in STEP File: 0 Highest entry in topology hierarchy: 3 solid(s) Preparing visualization (deflection = 0.01) ... done Face 1 / 22 (parameter space projection) Face 2 / 22 (parameter space projection) Face 3 / 22 (parameter space projection) Face 4 / 22 (parameter space projection) Face 5 / 22 (parameter space projection) Face 6 / 22 (parameter space projection) Face 7 / 22 (parameter space projection) Face 8 / 22 (parameter space projection) Face 9 / 22 (parameter space projection) Face 10 / 22 (parameter space projection) Face 11 / 22 (parameter space projection) Face 12 / 22 (parameter space projection) Face 13 / 22 (parameter space projection) Face 14 / 22 (parameter space projection) Face 15 / 22 (parameter space projection) Face 16 / 22 (parameter space projection) Face 17 / 22 (parameter space projection) Face 18 / 22 (parameter space projection) Face 19 / 22 (parameter space projection) Face 20 / 22 (parameter space projection) Face 21 / 22 (parameter space projection) Face 22 / 22 (parameter space projection) Delaunay meshing Delaunay meshing Delaunay meshing start tetmeshing Success ! start tetmeshing Success ! Remove Illegal Elements Volume Optimization Curve elements, order = 2 ('plastics', 'wall', 'medium') ('bc_1', 'bc_2', 'bc_3', 'bc_4', 'Electrode_1', 'Electrode_1', 'Electrode_1', 'Electrode_1', 'Electrode_0', 'Electrode_0', 'Electrode_2', 'Electrode_2', 'Electrode_2', 'Electrode_2', 'Electrode_1', 'Electrode_0', 'Electrode_2', 'bc_18', 'bc_19', 'bc_20', 'bc_21', 'bc_22') Finished executing FFFP_mesh.py
Last edit: 2 years 10 months ago by VienC. Reason: forget attaching files
More
2 years 10 months ago - 2 years 10 months ago #4148 by VienC
Hi,

I couldn't attach the code and images.
Please find images in the link:
imgur.com/YNaSKTo
imgur.com/72JuTri
Update:
When executing on Ubuntu, there was an error message:
Code:
divide edge: local h too small CORRECTED
the reason is in
Code:
https://github.com/NGSolve/netgen/blob/master/libsrc/meshing/basegeom.cpp
In the function
Code:
DivideEdge
, there is a hardcoded variable
Code:
divide_edge_sections = 1000
I was wondering what the function means and why there are hardcoded variables.

Cheers,
Vien Che
Last edit: 2 years 10 months ago by VienC.
Time to create page: 0.105 seconds