[Gmsh] Doubt about meshing closed STL shell

Christophe Geuzaine cgeuzaine at uliege.be
Wed Nov 6 21:03:04 CET 2019 


> On 6 Nov 2019, at 09:24, MA Ariza-Gracia <miguel.ariza at artorg.unibe.ch> wrote:
> 
> Dear Dr Geuzaine,
> 
> I didn't check this example before sending the previous one.
> 
> I tried to adapt the previous python script to cope with the geometrical tolerance and the re-meshing. Now, it raises a "segmentation fault" error but it recognizes the nodes in the boundary and removes the duplicates.

Indeed, we will investigate why the surface classification crashes in this case.

In the meantime, in your particular case you can simply replace `classifySurfaces` by `createTopology` as explained in the comments, i.e. replace

gmsh.model.mesh.classifySurfaces(math.pi/6)

with

gmsh.model.mesh.createTopology()

This works fine over here.

Christophe


> 
> Info    : Reading 'inclusion-1.stl'...
> Info    : 15045 facets in solid 0 Created by Gmsh
> Info    : Done reading 'inclusion-1.stl'
> Info    : Reading 'inclusion-2.stl'...
> Info    : 15045 facets in solid 0 Created by Gmsh
> Info    : Done reading 'inclusion-2.stl'
> Info    : Removing duplicate mesh vertices...
> Info    : Found 255 duplicate vertices
> Info    : Removed 255 duplicate mesh vertices
> Info    : Done removing duplicate mesh vertices
> Info    : Classifying surfaces (angle: 30)...
> Info    : Found 1 model faces
> Info    : Found 0 model edges
> Info    : Done classifying surfaces (0.163405 s)
> Info    : Creating geometry of discrete curves...
> Info    : Done creating geometry of discrete curves (2e-06 s)
> Info    : Creating geometry of discrete surfaces...
> Segmentation fault (core dumped)
> 
> Enclosed a zip folder with the updated script
> 
> Best,
> 
> MA
> 
> On 05/11/2019 17:07, Christophe Geuzaine wrote:
>> I've added a small example in demos/api showing how I would solve this problem:
>> 
>> https://gitlab.onelab.info/gmsh/gmsh/blob/master/demos/api/glue_and_remesh_stl.py
>> 
>> Christophe
>> 
>> 
>> 
>>> On 5 Nov 2019, at 16:19, Christophe Geuzaine <cgeuzaine at uliege.be> wrote:
>>> 
>>> Can you send the 2 STLs (I don't see them in the zip)?
>>> 
>>> Christophe
>>> 
>>>> On 1 Nov 2019, at 11:30, MA Ariza-Gracia <miguel.ariza at artorg.unibe.ch> wrote:
>>>> 
>>>> Dear Dr Geuzaine, Dr Remacle, and the Gmsh community,
>>>> 
>>>> first of all, thank you for sharing the software with us! I'm a postdoc working (mostly) in optics and I need to reconstruct some free-shape surfaces.
>>>> 
>>>> My final goal is to create a volume mesh enclosed by two discrete STL surfaces (given by a point cloud). The pipeline to do so is as follows:
>>>> 1) Read point clouds and triangulate using Delaunay (scipy.spatial)
>>>> 
>>>> 2) Create a 2 discrete surfaces giving the node coordinates and the connectivity of the triangles
>>>> 
>>>> 3) Create a volume using a surface loop on both surfaces
>>>> 
>>>> 4) Mesh the volume
>>>> 
>>>> 5) Exporting the mesh to Abaqus/LS-Dyna (whatever software)
>>>> 
>>>> 
>>>> 
>>>> After reading the manual, tutorials and mailing list, the only approached that worked was the next one: having a single closed shell joining 'manually' the nodes in the boundary. I say manually because I needed to implement the substitution of the nodes myself and I couldn't do it using Gmsh using removeDuplicate and so on (i.e., I didn't know how to do it). After having a single closed shell without duplicate nodes and with the normals pointing outwards, I manage to mesh the volume inside the models (see manually-merged-thick.msh and manually-merged-thin.msh).
>>>> 
>>>> 
>>>> 
>>>> However, as this is my first approach to Gmsh, this is not efficient and I would like to know if:
>>>> 
>>>> 1) There is a more efficient procedure to, given two different STLs, mesh the volume enclosed by the two of them (as I said, I tried may approaches including the ones regarding Earth topology reconstruction in the mail list, but I couldn't find a proper solution).
>>>> 
>>>> 2) There is a way of both remeshing the STLs AND meshing the volume at the same time. In one of the approaches, I parameterized the discrete STL surfaces with classifySurfaces and createGeometry (see the Jupyter notebook, three first attempts). This allowed me to re-mesh the STL surfaces but, unfortunately, I could not mesh the volume after (complaining about 0 elements in the volume). As my Delaunay triangulation is quite ugly, this would be my preferred option if you could give me a hint on how to proceed.
>>>> 
>>>> 3) In the end, I will have two surfaces: one main surface with one inclusion inside itself. From what I've read, I know I cannot do Boolean operations (as I'm using meshes and discrete elements), so I would appreciate any hint on how to proceed in this regard.
>>>> 
>>>> 
>>>> 
>>>> Files enclosed in compressed folder 'inquiry-gmsh.zip':
>>>> 
>>>> 	• try-inquiry.ipynb: Jupyter notebook with the API code to create the surfaces
>>>> 	• {anterior-lenticle-surface.txt, posterior-lenticle-surface.txt}: files containing the Cartesian point cloud
>>>> 	• After running the code, it will generate:
>>>> 		• thin-1.stl, thin-2.stl: surfaces used to generate the convex-convex thin lens
>>>> 		• surface-1.stl, surface-2.stl: surfaces used to generate the concave-convex thick lens
>>>> 		• shell-thin-lens.msh: attempt to mesh the volume of the thin mesh using: i) Two different STL surfaces (gmsh.merge); ii) remove duplicate coincident nodes (gmsh.removeDuplicateNodes); and iii) re-parameterizing
>>>> 		• shell-concave2convex.msh: attempt to do the same as the previous one but with the thick lens (-> discard problems with sharp edges in the periphery)
>>>> 		• merged-shell-concave2convex.msh: attempt to mesh the volume of the thick lens using: i) One single STL surface with duplicated nodes in the boundary; ii) Remove duplicated nodes using removeDuplicateNodes; iii) and classify surfaces + creating geometry
>>>> 		• manually-merged-thick.msh and manually-merged-thin.msh: success BUT: i) I needed to remove myself the duplicated nodes; ii) If I try to parameterize the surface, I cannot mesh the volume inside but only the surfaces.
>>>> 
>>>> 
>>>> Versions used:
>>>> 
>>>> 
>>>> - Ubuntu 18.04
>>>> 
>>>> - Gmsh 4.4.1 through the implementation in conda-forge
>>>> 
>>>> - Python 3.7 with Anaconda
>>>> 
>>>> 
>>>> 
>>>> Best regards,
>>>> 
>>>> 
>>>> MA
>>>> 
>>>> <inquiry-gmsh.zip>_______________________________________________
>>>> gmsh mailing list
>>>> gmsh at onelab.info
>>>> http://onelab.info/mailman/listinfo/gmsh
>>> —
>>> Prof. Christophe Geuzaine
>>> University of Liege, Electrical Engineering and Computer Science
>>> http://www.montefiore.ulg.ac.be/~geuzaine
>>> 
>>> 
>>> 
>>> 
>>> _______________________________________________
>>> gmsh mailing list
>>> gmsh at onelab.info
>>> http://onelab.info/mailman/listinfo/gmsh
>> —
>> Prof. Christophe Geuzaine
>> University of Liege, Electrical Engineering and Computer Science
>> http://www.montefiore.ulg.ac.be/~geuzaine
>> 
>> 
>> 
> <mesh-not-working-alternative.zip>

— 
Prof. Christophe Geuzaine
University of Liege, Electrical Engineering and Computer Science 
http://www.montefiore.ulg.ac.be/~geuzaine

More information about the gmsh mailing list