[Gmsh] Question on creating meshes with layered elements

[Marco Antolovic]

 Hi Christiophe,
glad to see that this feature has been already formally requested.
>From my side I have spent some time thinking on how to solve, if possible, this problem with scripting tools only and I came out with the following idea
Assumptions:1. The ID of the volume to be meshed with layered extrusion is known2. The volume is extruded from a surface that belongs to a plane3. Embedded curves are present only on top and bottom surfaces of the volume4. The embedded curves on top does not match with embedded curves on bottom
I assume that if another entity is touching the volume (defined in points 1,2) on its top and bottom surfaces, due to conformal meshing requirements, this will create embedded curves on them hence assumption 3 and 4.
The procedure is as follows:
- Select the volume that requires layered extrusion- Duplicate its bottom surface (including embedded curves) and top embedded curves- Delete (recursive) the volume - Duplicate, translate and embed the top embedded curves on the bottom surface- Extrude the bottom mesh with Extrude{}{ Layers{...}}- Embed the top curve on top surface generated by Extrude- Run Coherence/BooleanFragmens
The attached script is a simplified version of the above procedure (the volume requiring extrusion is already deleted, top embedded curves, bottom surface and embedded curves already available).I wanted to prepare a simple script and hopefully get a feedback before I embark on further hours of scripting because there might be subtleties that only Gmsh developers are aware of which, if overlooked, could lead to a unicorn chase from my side :-)

The script builds a cylinder inside a parallelepiped.
The cylinder is the volume meshed with a layered extrusion (along z) and it features two embedded curves, one on top and one on bottom that are different in shape and location on XY plane. 
The mesh type in the parallelepiped is arbitrary, it only needs to guarantee that the total mesh (cube + cylinder) is conformal.
The results seem to give a conformal mesh and solve the extrusion problem.
Any feedback is much appreciated,
Many thanks,
Marco




    Il mercoledì 4 novembre 2020, 09:14:51 CET, Christophe Geuzaine <cgeuzaine at uliege.be> ha scritto:  
 
 

> On 22 Oct 2020, at 12:56, Marco Antolovic <marco_antolovic at yahoo.it> wrote:
> 
> Hi all,
> 
> I'm trying to find the best way to approach the following problem:
> 
> I have a parallelepiped with the edge alone z much smaller than that along x and y (this is typical, for example, 
> for PCB traces) and I would like to guarantee, in the volume, a predefined number of layers (np_layers) along z
> edge.
> 
> If I build the structure in gmsh this is quite simple to obtain by saying:
> 
> SetFactory("OpenCASCADE");
> thickness = 140e-6;
> nr_layers = 4;
> surf=newreg;
> Rectangle(surf)={.001,0,0,.001,.001,0};
> Extruded_surf() = Extrude { 0,0, thickness } { Surface{surf}; Layers{nr_layers}; Recombine;};
> 
> But how can I apply this in case the parallelepiped is coming from a step file? The command above
> (Extrude{...}{... Layers{}}) can't be used.

Indeed, we would need to "reverse engineer" the fact that the volume has indeed been created by extrusion. See

https://gitlab.onelab.info/gmsh/gmsh/-/issues/929

for the issue tracking this feature request.

Christophe


> 
> I was thinking about slicing the Volume with some planes (see example below)
> 
> SetFactory("OpenCASCADE");
> thickness = 140e-6;
> nr_layers = 4;
> surf=newreg;
> Rectangle(surf)={.001,0,0,.001,.001,0};
> Extruded_surf() = Extrude { 0,0, thickness } { Surface{surf};};
> vol() = Extruded_surf(1);
> For i In {1:nr_layers-1}
> surf1(i-1)=newreg;
> Rectangle(surf1(i-1)) = {.001, 0, thickness/nr_layers*i, .001, .001};
> EndFor
> results() = BooleanFragments{Volume{vol()};Delete;}{Surface{surf1()}; Delete;};
> 
> but when trying to recombine the triangles in the layers I get the following error:
> 
> Error  : Pyramid top vertex already classified on volume 2 (!= 3) - non-manifold quad boundaries not supported
> yet
> 
> Any suggestion on how to address the issue is much appreciated.
> 
> Regards,
> 
> Marco
> 
> _______________________________________________
> 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://people.montefiore.ulg.ac.be/geuzaine



  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://onelab.info/pipermail/gmsh/attachments/20201108/2ba07e56/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: structured_grids_with_step_files_test_V05.geo
Type: application/octet-stream
Size: 1332 bytes
Desc: not available
URL: <http://onelab.info/pipermail/gmsh/attachments/20201108/2ba07e56/attachment.geo>