[Gmsh] RE : stl and prism extrusion

julien hoessler julien.hoessler06 at imperial.ac.uk
Mon Dec 14 08:47:58 CET 2009


Hi David,
The error message for the stl importation is the following:
Info    : Running '/Applications/Gmsh.app/Contents/Resources/bin/gmsh  
-2 reparam.geo'
Info    : Started on Mon Dec 14 07:31:09 2009
Info    : Reading 'reparam.geo'
Info    : Reading 'ChunkOfGeometry.msh'
Info    : 214164 vertices
Info    : Vertex numbering is dense
Info    : 426736 elements
Info    : Read 'ChunkOfGeometry.msh'
Error   : Incorrect face in compound surface -1000

So I may have to clean my stl a bit

ae-jh1406a:DUCT_VORTICES jh1406$

As far as the prisms are concerned, it is a little bit weird: if I run  
gmsh -3, I get the following errors:
Error   : Degenerated prism in extrusion of volume 5
Error   : Degenerated prism in extrusion of volume 5
Error   : Degenerated prism in extrusion of volume 5
Error   : Degenerated prism in extrusion of volume 5
Error   : Degenerated prism in extrusion of volume 5
Error   : Degenerated prism in extrusion of volume 5
Error   : Degenerated prism in extrusion of volume 5
Error   : Degenerated prism in extrusion of volume 5
Error   : Degenerated prism in extrusion of volume 5
Error   : Degenerated prism in extrusion of volume 5
Error   : Degenerated prism in extrusion of volume 5
Error   : Degenerated prism in extrusion of volume 5
Error   : Degenerated prism in extrusion of volume 5
Error   : Degenerated prism in extrusion of volume 5
Error   : Degenerated prism in extrusion of volume 5
Info    : Mesh 3D complete (0.018726 s)
Info    : 2083 vertices 5682 elements
Error   : ------------------------------
Error   : Mesh generation error summary
Error   :     0 warnings
Error   :  1698 errors
Error   : Check the full log for details
Error   : ------------------------------
Info    : Writing 'test_curved.msh'
Info    : Wrote 'test_curved.msh'
Info    : Stopped on Mon Dec 14 07:39:45 2009

If I run gmsh -2, I generate the same mesh without errors. I attached  
the test case.


I also get a lot of warnings if I try to switch to order 2 to get the  
curvature of the elements. It seems to work though as mesh file only  
contains 2d order prisms
Warning : We don't have a valid parameter on curve 23-529
Warning : We don't have a valid parameter on curve 529-530
Warning : We don't have a valid parameter on curve 530-531
Warning : We don't have a valid parameter on curve 531-532
Warning : We don't have a valid parameter on curve 532-533
Warning : We don't have a valid parameter on curve 533-534
Warning : We don't have a valid parameter on curve 534-21
Warning : We don't have a valid parameter on curve 22-535
Warning : We don't have a valid parameter on curve 535-536
Warning : We don't have a valid parameter on curve 536-537
Warning : We don't have a valid parameter on curve 537-538
Warning : We don't have a valid parameter on curve 538-539
Warning : We don't have a valid parameter on curve 539-540
Warning : We don't have a valid parameter on curve 540-24
Warning : We don't have a valid parameter on curve 24-541
Warning : We don't have a valid parameter on curve 541-542
Warning : We don't have a valid parameter on curve 542-543
Warning : We don't have a valid parameter on curve 543-23
Warning : We don't have a valid parameter on curve 14-544
Warning : We don't have a valid parameter on curve 544-545
Warning : We don't have a valid parameter on curve 545-546
Warning : We don't have a valid parameter on curve 546-23
Warning : We don't have a valid parameter on curve 13-547
Warning : We don't have a valid parameter on curve 547-548
Warning : We don't have a valid parameter on curve 548-549
Warning : We don't have a valid parameter on curve 549-24
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test_curved.geo
Type: application/octet-stream
Size: 982 bytes
Desc: not available
URL: <http://www.geuz.org/pipermail/gmsh/attachments/20091214/cf7067df/attachment.geo>
-------------- next part --------------


Best regards,
Julien

On 12 Dec 2009, at 17:09, David Colignon wrote:

>
> On 12/12/09 14:18, Hoessler, Julien wrote:
>> Hi David,
>> Thanks for the quick answer. The stl importation crashes with the  
>> stl file or a msh file obtained from the stl,
>
> Hi Julien,
>
> does it print an error message before crashing ?
>
>> but I guess that's because my stl is too fine (214K elements), I'll  
>> try again with a coarser mesh.
>> As far as the prism extrusion is concerned,it works well and after  
>> recombining, the prism look perfectly fine on a curved surface,  
>> except that I get errors telling me that the prisms are  
>> degenerated. Is it an unnecessary warning or should I be worried?
>
>
> Can you send a copy of these warnings ?
>
>
>> I also have a bonus question if you don't mind.
>
> You're welcome
>
>> When I generate the extrusion of the surfaces and put them on a  
>> list surf[], is it possible to access the edges and vertices of the  
>> top surfaces to define the volume above?
>
> No. But the Tools -> Visibility -> Tree browser can help tou to  
> quickly retrieve it.
>
>> I could extrude the edges and vertices too but it would mean that  
>> some entities would be duplicated.
>
> Not necessarily, after each extrusion operation, Gmsh issues the  
> "Coherence;" command which should eliminate duplicated
> entities.
>
> Regard,
>
> Dave
>
>
>> Best regards
>> Julien
>> ________________________________________
>> De : David Colignon [David.Colignon at ulg.ac.be]
>> Date d'envoi : vendredi 11 d?cembre 2009 14:40
>> ? : Hoessler, Julien
>> Cc : gmsh at geuz.org
>> Objet : Re: [Gmsh] stl and prism extrusion
>>
>> On 11/12/09 15:00, Hoessler, Julien wrote:
>>> Good afternoon all,
>>> First of all, thanks for writing this code, I'm using it a lot to  
>>> generate 2D/3D meshes for DNS and stability analysis. I was  
>>> wondering if it is possible to extrude a layer of prisms using the  
>>> surface normals instead of a specific direction. I need to  
>>> generate a layer of prisms out of a wall which is curved and there  
>>> is more than 80 degrees of angular difference between the two  
>>> extreme locations of the surface (it is smooth, but twists a bit  
>>> like a sinh).
>>
>>
>> Hi Julien,
>>
>> it is still experimental and quite hidden in the documentation
>>   ( http://www.geuz.org/gmsh/doc/texinfo/gmsh.html#Structured- 
>> grids ),
>> but you can try:
>>
>> Extrude { Surface { expression-list }; layers }
>>
>> which extrudes a boundary layer along the normals of the specified  
>> surfaces (if they are smooth and do not form acute
>> angles).
>>
>>> My second question is about the stl importation. I have a stl file  
>>> which is a 3D box that I can read and visualise without problems,  
>>> but I'd like to remesh it
>>
>> The undocumented "Compound Surface" command is also very  
>> experimental, but you can try the attached file (reparam.geo)
>> which here use a .msh file but should also work with a .stl file.
>>
>>
>> , mesh the volume inside  and split the box surface to assign  
>> different physical groups for boundary conditions (and add
>> a layer of prisms on one of the surfaces). Is it possible with  
>> Gmsh? I've searched through the questions in this mailing
>> list, do I need the opencascade support for this?
>>
>>
>> Regards,
>>
>> Dave
>>
>>
>> --
>> David Colignon, Ph.D.
>> Collaborateur Logistique du F.R.S.-FNRS
>> C?CI - Consortium des ?quipements de Calcul Intensif
>> ACE - Applied&  Computational Electromagnetics
>> Sart-Tilman B28
>> Universit? de Li?ge
>> 4000 Li?ge - BELGIQUE
>> T?l: +32 (0)4 366 37 32
>> Fax: +32 (0)4 366 29 10
>> WWW:    http://hpc.montefiore.ulg.ac.be/
>> Agenda: http://www.google.com/calendar/embed?src=david.colignon%40gmail.com
>>
>

Julien Hoessler
PhD Student
Department of Aeronautics
Imperial College London

Phone: +44 (0) 20759 45042
Email:   julien.hoessler06 at imperial.ac.uk