[Gmsh] Error : Self intersecting surface mesh, computing intersections (this could take a while)

Tue Mar 18 10:57:43 CET 2014

Hi Steffen,

there are still ill-defined triangles, e.g. 2319, 2320, 2321. You can find that out by yourself by inspecting your mesh with gmsh in the way I described in my last mail.
Remark: checking "surface faces" can help to examine the surface.

HTH,

Matthias

Von: Steffen Becker [mailto:steffenbecker87 at yahoo.de]
Gesendet: Montag, 17. März 2014 22:39
An: Zenker, Dr. Matthias
Cc: Roman Fratczak; gmsh at geuz.org
Betreff: Re: [Gmsh] Error : Self intersecting surface mesh, computing intersections (this could take a while)

Hi Matthias,

thank you for your help!
I think I successfully deleted all the triangles with zero surface. Unfortunately the same error still appears (but at least with much less intersections). You find my new .msh file in the appendix. Do you have more ideas?

Regards,
Steffen

Am 17.03.2014 11:37, schrieb Zenker, Dr. Matthias:
Hi Steffen,

I think you have ill-defined triangles in your mesh. For example, there is a triangle 15328 which contains the nodes 7358, 7359, and 7360. But these nodes are aligned, so the triangle has zero surface. I would suspect this kind of problem to be the cause for the failure. I discovered that by examining your mesh after merging the intersect.pos, and by displaying node labels and surface labels in gmsh. I saw that this triangle was not visible, just its label, and searched for "15328" in your msh file to see what nodes it contains.
You should make sure that this kind of thing cannot happen when you define the surface mesh.

HTH,

Matthias

Von: Steffen Becker [mailto:steffenbecker87 at yahoo.de]
Gesendet: Montag, 17. März 2014 11:20
An: Zenker, Dr. Matthias
Cc: Roman Fratczak; gmsh at geuz.org<mailto:gmsh at geuz.org>
Betreff: Re: [Gmsh] Error : Self intersecting surface mesh, computing intersections (this could take a while)

Here it is.

Am 17.03.2014 11:14, schrieb Zenker, Dr. Matthias:
Hi Steffen,

I found out: Outlook is wrong.
I just discovered a small info saying that Outlook thinks that my_file.msh is insecure and that it has deleted the attachment (Grrrr). So the fault was on my side, sorry. Could you try to send it again with a changed extension, e.g. rename it to my_file.msh.txt?

Matthias

Von: Steffen Becker [mailto:steffenbecker87 at yahoo.de]
Gesendet: Montag, 17. März 2014 10:55
An: Zenker, Dr. Matthias
Cc: Roman Fratczak; gmsh at geuz.org<mailto:gmsh at geuz.org>
Betreff: Re: [Gmsh] Error : Self intersecting surface mesh, computing intersections (this could take a while)

Hi Matthias,

I don't know what's wrong here.
When I check my mails, there is a .msh file in the mail from 12.03.2014 18:21 and also one in the mail from 14.03.2014 12:33.
Finally, I sent you again a .msh file in this mail. Do you receive it?

Steffen

Am 17.03.2014 09:52, schrieb Zenker, Dr. Matthias:
Hi Steffen,

in the mail you mention there is only a geo file...

Matthias

Von: Steffen Becker [mailto:steffenbecker87 at yahoo.de]
Gesendet: Freitag, 14. März 2014 21:09
An: Zenker, Dr. Matthias
Cc: Roman Fratczak; gmsh at geuz.org<mailto:gmsh at geuz.org>
Betreff: Re: [Gmsh] Error : Self intersecting surface mesh, computing intersections (this could take a while)

Hi Matthias,

hm, I sent it on 12.03.2014 18:21, don't know what went wrong.
As you see, I created a ASCII .msh file as explained in the documentation, so I hope it's correct.
What can I do now?

Steffen

Am 14.03.2014 13:05, schrieb Zenker, Dr. Matthias:
Hi Steffen,

there was no msh file in any of your mails.
Depending on what is in the msh file, your steps seem correct to me.

Matthias

Von: Steffen Becker [mailto:steffenbecker87 at yahoo.de]
Gesendet: Freitag, 14. März 2014 12:33
An: Zenker, Dr. Matthias
Cc: Roman Fratczak; gmsh at geuz.org<mailto:gmsh at geuz.org>
Betreff: Re: [Gmsh] Error : Self intersecting surface mesh, computing intersections (this could take a while)

Hi Matthias,

I think you are talking about your mail from 12.03.2014 09:59 ?
In one of my last mails I already sent you a selfcreated .msh file - but the 3D mesh still failes. The intersect.pos shows me some
intersections, but when I check my nodes for duplicates I can't find any.
Hope my steps are correct:
- create "my_file.msh" with my python script and check it for nodes duplications -> none found
- open "my_file.geo" in Gmsh
- do "Modules" -> "Mesh" -> "3D"
- receive error message
- merge "intersect.pos" and see some intersections but I don't know why because the are no duplicated nodes

Steffen

Am 14.03.2014 09:50, schrieb Zenker, Dr. Matthias:
Hi Steffen,

you should be able to generate a filled volume mesh if you follow my suggestion from the other mail:

Write  the 2D mesh in gmsh format (e.g. as outerface.msh, gmsh documentation for more info on the msh format) and merge that one instead of the myfile.ply2 file, leaving the other command lines in your .geo file as they are. If you then do a 3D mesh, gmsh will take the 2D mesh as defined in your file. You can then refine the mesh if you wish.

HTH,

Matthias

Von: Steffen Becker [mailto:steffenbecker87 at yahoo.de]
Gesendet: Mittwoch, 12. März 2014 20:00
An: Roman Fratczak
Cc: Zenker, Dr. Matthias; gmsh at geuz.org<mailto:gmsh at geuz.org>
Betreff: Re: [Gmsh] Error : Self intersecting surface mesh, computing intersections (this could take a while)

Hi,

I'm so sorry for all time you spent with my problem, but now I see: We talk at cross. It's my fault, I didn't explained correctly what I need.
I want to receive a *filled* volume mesh.
I created 2 files (steffen.stl and steffen2.stl) with my python script.
With steffen.stl I successfully could create such a filled volume mesh as seen in steffen.msh.
When I try exactly the same with steffen2.stl then I get the error message in the subject (and the surface mesh doesn't get filled).
So that's why the .iges file you sent me isn't what I need.

Sorry for all this confusion,
Steffen

Am 12.03.2014 18:53, schrieb Roman Fratczak:
Steffen,

I did a volume based on your surface mesh - see attached IGES (compressed with 7-zip).
After reading it into GMSH, you can create the 1D, then 2D and finally 3D mesh with no errors.

You can do the same using GMSH volume definition (as already wrote)-> it will work, as the fundamental faces are plane triangles.

Regards
Roman

W dniu 12.03.2014 18:27, Steffen Becker pisze:
Wrong .geo file, sorry!

Am 12.03.2014 18:21, schrieb Steffen Becker:
Thank you so far,

I create both files: .msh and .geo, but I still get the same error. What's my mistake?

Steffen

Am 12.03.2014 11:48, schrieb Roman Fratczak:
Hi,

FYI - you can remove double vertices/edges (or repair) in mesh in FreeCad (for exampe).

regards
Roman

W dniu 12.03.2014 10:17, Steffen Becker pisze:
Hi Matthias & Roman,

I create a nodes- and elements-list like this:
# nodes:
N1 0,0,0
N2 0,1,0
N3 1,0,0
N4 0,0,1
# elements, which describe triangles:
E1 1,2,3
E2 2,3,4

So I do have double edges (here: between node 2+3). Didn't now that this is a problem for gmsh, sorry.

I will try Romans solution asap and give you a feedback.
Thank you both!

Steffen

Am 12.03.2014 09:59, schrieb Zenker, Dr. Matthias:
Hi Steffen,

I am not totally sure from what you wrote that there are really no duplicated nodes. In gmsh, you need to define the surfaces in a way that neighbouring surfaces/facets with a common border use the _same_ nodes to define the border. I could not see from your file if there are duplicated nodes, and would propose that you make sure that there are really no such duplications.

If the meshing still fails, then this is a question for the real gurus (Christophe?).

HTH,

Matthias

Von: Roman Fratczak [mailto:fratczakroman at o2.pl]
Gesendet: Mittwoch, 12. März 2014 09:51
An: Steffen Becker
Cc: gmsh at geuz.org<mailto:gmsh at geuz.org>
Betreff: Re: [Gmsh] Error : Self intersecting surface mesh, computing intersections (this could take a while)

Hi Steffen,

No it is more clear. If you are working with cloud of point, then you may use Meshlab for surface regeneration. There are many software that are able to handle in a resonable way the cloud of points (like rhino3D + resurf, other). But these are rather commercial.

>From the other side - try to do the simplest thing:

Adjust your python script for surface creation in order to save not as PLY but as GMSH "geo" file using points and lines

Point(1)={0,0,0} ;
Point(2)={0,1,0} ;
...

Line(1) = {1,2};
..
and.. Line Loop & Surface -> Volume ..

Then you will get a volume defined by geometric entities. In this case is shoul be easier to create a better mesh on surface and then on whole volume.

Regards
Roman

W dniu 11.03.2014 17:59, Steffen Becker pisze:
Hi,

@ Roman:
some more details:
I have a point cloud given and generate the surface mesh by myself (with python). I don't know a good algorithm to create a volume mesh out of this surface mesh - that's why I use gmsh for this task.
I also checked my surface mesh for double vertexes but I didn't find any. What do you mean with "double egdes"? Isn't there always a "double edge", because two neighbour facets always have one same edge?!

I will send you a .step or .iges file as soon as I figured out how to handle the "No OpenCASCADE model found" error.

I'm appending the other mailing history (with Matthias) to this mail - sorry for that "double mailing", I just forgot one time to "reply to all" -.-
Anyone else has an idea why the meshing fails?

Regards and thank you a lot,
Steffen

Hi Steffen,

I don't see either why the meshing fails, sorry.

Matthias

-----Ursprüngliche Nachricht-----

Von: Steffen Becker [mailto:steffenbecker87 at yahoo.de]

Gesendet: Dienstag, 11. März 2014 14:42

An: Zenker, Dr. Matthias

Cc: gmsh at geuz.org<mailto:gmsh at geuz.org>

Betreff: Re: [Gmsh] Error : Self intersecting surface mesh, computing

intersections (this could take a while)

Oh I'm really sorry, here is the file.

In addition I sent you also the intersect.pos. I don't understand why exactly

these surfaces are problematic and the "neighbour surfaces"

aren't - although they have the same geometry =/

Am 11.03.2014 12:07, schrieb Zenker, Dr. Matthias:

Hi Steffen,

if you don't provide the myfile.ply2, we cannot see your geometry...

 You can visualize intersect.pos by merging it to your geometry.

HTH,

Matthias

Am 11.03.2014 14:33, schrieb Roman Fratczak:
Hello,

It is possible, that you will get your geometry in other format?
In your "ply" there is already a surface mesh given. Try maybe use STEP, IGES,...  as input- then it is easer to recalculate the surface mesh.

In case of need of usage of "a dead mesh" - maybe try first to remesh it (i.e. in blender with modification Remesh) and/or clean it (removing double vertexes and edges).

regards
Roman

W dniu 11.03.2014 12:13, Steffen Becker pisze:
Hi,

in the appendix you find my model.
Thanks for  help!!

Regards,
Steffen

Am 11.03.2014 11:47, schrieb Roman Fratczak:
Hello Steffen,

Yes. You can try to the first Optimize quality of tetra, or all of them (just try).

I don't know how your model looks like. If the elements on the surface are too big, then the solid elements (probably tetragonal) may be with intersecting edges or faces. Therefore the reducing size of element may help (especially on the surfaces).

Or- just share your model (ply) -in order to be more precise.

regards
Roman

W dniu 10.03.2014 18:06, Steffen Becker pisze:
Hi Roman,
when you say "enable option: optimize", you mean" Tools -> Options -> Mesh -> Advanced -> Optimize quality of tetrahedra, Optimize quality of tetrahedra with Netgen, Optimize high order meshes (experimental)" ?
And why is reducing element size important? In my case it is important that the object has such "90 degree corners".

Regards,
Steffen

Am 10.03.2014 16:55, schrieb Roman Fratczak:
Hi Steffen,

If your volume is OK (is really a closed one) - try to:

1. Reduce element size (especially in a curved shapes)
2. Enable option "Optimize" - in mesh options

Regards
Roman

W dniu 10.03.2014 16:41, Steffen Becker pisze:
Hello everyone,

I'm a gmsh beginner and I want to create a 3D volume mesh out of a given surface-mesh but I got the error message as seen in the subject. You find my .geo file in the appendix.
Can anyone tell me what I have to do to create a 3D mesh? I even can't figure out where the intersection are located (don't know how to read the intersect.pos).
Sorry for my poor english.

Regards,
Steffen

_______________________________________________

gmsh mailing list

gmsh at geuz.org<mailto:gmsh at geuz.org>

http://www.geuz.org/mailman/listinfo/gmsh

_______________________________________________

gmsh mailing list

gmsh at geuz.org<mailto:gmsh at geuz.org>

http://www.geuz.org/mailman/listinfo/gmsh

_______________________________________________

gmsh mailing list

gmsh at geuz.org<mailto:gmsh at geuz.org>

http://www.geuz.org/mailman/listinfo/gmsh

_____________________________________________________________________
ERBE Elektromedizin GmbH
Firmensitz: 72072 Tuebingen
Geschaeftsfuehrer: Christian O. Erbe, Reiner Thede
Registergericht: Stuttgart HRB 380137

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.geuz.org/pipermail/gmsh/attachments/20140318/a4a394b4/attachment.html>