Personal tools
You are here: Home Forum Use Re: Problem with simple tetrahedral mesh: Line within box

Re: Problem with simple tetrahedral mesh: Line within box

Up to Use

Re: Problem with simple tetrahedral mesh: Line within box

Posted by Theophile Guillon at June 04. 2019

Hi,

I'm trying to mesh a really simple geometry (box + 1 plane + 1 Line => Partition) using NETGEN 1D-2D-3D, but I'm getting a strange meshing pattern please see the attached file.

How can I handle this?

Thank you

 

Attachments

Re: Problem with simple tetrahedral mesh: Line within box

Posted by Christophe Bourcier at June 04. 2019

Hi,

Stange indeed!

What algorithm did you use?

Try Netgen 1D2D, it should give better result.

Christophe

Re: Problem with simple tetrahedral mesh: Line within box

Posted by Theophile Guillon at June 04. 2019

Thanks for your reply,

You can take a look at the script (example) may be you will find what I am doing wrong.

I tried Netgen 1D2D, it works to mesh surfaces but this does not mesh the volume...

Attachments

Re: Problem with simple tetrahedral mesh: Line within box

Posted by Theophile Guillon at June 05. 2019

OK ! problem solved using :

Algorithm 3D => NETGEN 3D

Algorithm 2D => NETGEN 1D-2D

 

 

Attachments

Re: Problem with simple tetrahedral mesh: Line within box

Posted by Gérald NICOLAS at June 05. 2019

Another solution, using "Automatic Tetrahedralization". Perhaps not the very best for the quality of the cells, but it is effective.

Note that the partition could be simplified.

Attachments

Re: Problem with simple tetrahedral mesh: Line within box

Posted by Theophile Guillon at June 07. 2019

Thanks Gérald for the suggestion!

Ok, now is level 2 of my problem: I would like to refine the mesh along the line.

So, I created a submesh then computed the whole thing (see file attached).

Problem: the mesh along the line is indeed refined BUT none of its vertices is connected to the mesh tetrahedra.

This can be seen using the mesh viewer, and is confirmed when exporting the mesh as a text file (additional nodes are connected to edges only).

I tried to predefine points along the line then force the mesh to pass through them, but I could not make it work (MG-Tetra 3D algorithm).

According to this thread, https://www.salome-platform.org/forum/forum_10/29053836, the solution should be to cut the line with planes for each additional node I want, but this would greatly complexify the geometry.

Any suggestion as to how to raise this issue?

Many thanks.

Théophile

Attachments

Re: Problem with simple tetrahedral mesh: Line within box

Posted by Theophile Guillon at May 12. 2020

Hi,

This issue is still relevant

Any suggestion ?

Many thanks

 

Re: Problem with simple tetrahedral mesh: Line within box

Posted by Daniel at May 17. 2020

Hello,

I believe that this occurs because you cannot refine the mesh along a single line that doesn't have defined "opposite edges'. Someone can correct me if I am wrong. 

When you create a sub-mesh you can add the option of " additional hypothesis" (see attached). This is easier to see if you have hexahedrons instead of tetrahedrons.  In geometry: Create a new line group with one of the edges of the cube. Then, add this edge as a sub-mesh and see how it changes by adding or removing this "additional hypothesis". That should show why you need additional edges to do what you want.

 

Best regards,

Daniel 

Attachments

Re: Problem with simple tetrahedral mesh: Line within box

Posted by Daniel at May 17. 2020

Re: Problem with simple tetrahedral mesh: Line within box

Posted by Theophile Guillon at June 05. 2020
Hi!
 
I didn't have a chance to work back on this problem since a couple of months, and I am sorry for remaining dead silent.
 
Anyway, thanks a lot, Daniel, for your suggestions.
Unfortunately, they do not solve the problem.
 
Actually, we have no problem creating a submesh on our inner edge and refining it.
As you can see on the attached file :
- left image: box edges are loosely meshed
- middle image: inner edge is properly refined.
 
The problem is: although the nodes on the submesh DO belong to the mesh, they are connected to no element at all in the mesh (whether 2D or 3D).
This can be seen on the right image: the top node belonging to the submesh (node #8) lies in the middle of one of the mesh top triangles.
This is no graphical bug, this is hard-confirmed in the mesh connectivity output file.
 
I guess the issue resides in the way our inner edge is created: it is inherited from a partition of the whole box for which the cut tool is a line. I think the inner edge is then somehow considered as a boundary of the model.
 
I will test other options (mesh sewing?), or even remeshing using other tools.
 
I will upload a solution when (if?) I find one.
 
Until then, any suggestion is welcome.
 
Thanks again, be safe during this period, and see you.
 
Regards,
 
Théophile
Attachments
Powered by Ploneboard
Document Actions