Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Independent Mesh (import) if not VMRL or STL

Please login with a confirmed email address before reporting spam


I want to create my own meshing but write into format that COMSOL works well with. I can write to STL
or VMRL but heard Parasolid is better?

Here's the two questions

1. Is there free-ware code that lets me write to Parasolid format. I found out that Parasolid is a proprietary format.


2. Also I heard that STL and VMRL are insufficient for mesh generation, They give faces and edges , but
can be missing something else that Parasolid helps with. I was wondering what it was missing? I don't

5 Replies Last Post 2010年7月9日 GMT-4 11:21
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2009年12月28日 GMT-5 10:36
Hi

In 2D you should use "dxf", there are free translaters/writer out on the web, even if its a propretarian Autocad format.

You have also GDS (a 2D or 2.5D file format) probably there aresome descriptions out there, havnt tried (used mostly by the electronics engineers)

for 3D you have IGES, STEP, VMRL and STL that are "open formats" I believe, documented on the web at least

"Parasolid" is the 3D core Uniraphics software is based on, this is a closed format, I do not believe you can find any free software for that, but there are several excellent CAD tools build upon parasolid

I beleive this is the same for ACES "sat" files, one ot the other major commercial 3D graphical core software tools

There is also the Netex-G format, but that one I have never used, cannot tell

But then, you have the COMSOL format, you can generate geometry from Matlab, linked to Comsol, and save as mphbin files

Hope this helps

Godd luck
Ivar
Hi In 2D you should use "dxf", there are free translaters/writer out on the web, even if its a propretarian Autocad format. You have also GDS (a 2D or 2.5D file format) probably there aresome descriptions out there, havnt tried (used mostly by the electronics engineers) for 3D you have IGES, STEP, VMRL and STL that are "open formats" I believe, documented on the web at least "Parasolid" is the 3D core Uniraphics software is based on, this is a closed format, I do not believe you can find any free software for that, but there are several excellent CAD tools build upon parasolid I beleive this is the same for ACES "sat" files, one ot the other major commercial 3D graphical core software tools There is also the Netex-G format, but that one I have never used, cannot tell But then, you have the COMSOL format, you can generate geometry from Matlab, linked to Comsol, and save as mphbin files Hope this helps Godd luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2010年2月24日 GMT-5 09:05
Hi,

I'm considering to use STL since it's very easy to generate this format for me. I understand that STL is not the best format and there is a bit of heuristics involved while importing. My question is then: Are there any risk of getting into problems if I convert my (large) 3D voxel-dataset into a STL-file where each object is floating, there are no ambigious triangles, and there are no shapes that isn't closed? Since the dataset is based on voxels, all edges are following the x, y, and z-axis.

My dataset has approximately 25 types of material, and I plan to make one STL file per material, and them import them sequentially while adding subdomain properties while immediately after each file import.

Best regards,
Fred-Johan Pettersen
Oslo university Hospital, Norway
Hi, I'm considering to use STL since it's very easy to generate this format for me. I understand that STL is not the best format and there is a bit of heuristics involved while importing. My question is then: Are there any risk of getting into problems if I convert my (large) 3D voxel-dataset into a STL-file where each object is floating, there are no ambigious triangles, and there are no shapes that isn't closed? Since the dataset is based on voxels, all edges are following the x, y, and z-axis. My dataset has approximately 25 types of material, and I plan to make one STL file per material, and them import them sequentially while adding subdomain properties while immediately after each file import. Best regards, Fred-Johan Pettersen Oslo university Hospital, Norway

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2010年7月8日 GMT-4 06:30
Hi Fred-Johan and all,
have you had any luck with importing STL? We are interested in doing something similar as far as I can see. The first trial .stl object reads in without complaints but I can't do anything with it, no subdomain settings, no boundary settings, no meshing etc. Obviously something went wrong while reading it in. The geometric properties look reasonable. But the error message when trying any of the above reads:
Failed to generate mesh for subdomain.
- Subdomain: 1
Internal error in boundary respecting.
- x coordinate: 1.00141
- y coordinate: 82.3333
- z coordinate: 4

If I create this point, it lies well within one face, not on an edge, not at a corner - no idea what the story is about this point. If I try to mesh after creating the point, the error message reads

Error: 4100
Failed to generate mesh for subdomain.
- Subdomain: 1
Internal error in boundary respecting.
- x coordinate: 1.0014
- y coordinate: 76
- z coordinate: 5.50657

So I would hazard a guess that there is a whole range of points that you discover one after the other, that seemingly are a problem - but why?

As regards error 4100 Ivar KJELBERG at some point suggested:
"often you can get around by meshing edges/faces by hand, removing "small" features in the geometry (if not relevant to the solution) or cut the volumes into smaller parts to allow the mesher to better adapt to complex geometry shapes.
Try also to change the default meshing size, or adapt it for the different parts
Try to understand where and what is getting wrong by studying the geometry around that coordinate (and any similar shaped element in your model).
Automatic meshing of complex geometries is not trivial, not even for a human ;)"

But as I said the point(s) in question are well within one face, nothing complicated going on around, no idea how to remove any small feature. The faces are all pretty much of the same size so I dont see why changing the default meshign parameters would change anything. And also, my error is not only about the meshing, since I cant even set boundary conditions!

Any ideas anyone?
Regards,
Susanne
Hi Fred-Johan and all, have you had any luck with importing STL? We are interested in doing something similar as far as I can see. The first trial .stl object reads in without complaints but I can't do anything with it, no subdomain settings, no boundary settings, no meshing etc. Obviously something went wrong while reading it in. The geometric properties look reasonable. But the error message when trying any of the above reads: Failed to generate mesh for subdomain. - Subdomain: 1 Internal error in boundary respecting. - x coordinate: 1.00141 - y coordinate: 82.3333 - z coordinate: 4 If I create this point, it lies well within one face, not on an edge, not at a corner - no idea what the story is about this point. If I try to mesh after creating the point, the error message reads Error: 4100 Failed to generate mesh for subdomain. - Subdomain: 1 Internal error in boundary respecting. - x coordinate: 1.0014 - y coordinate: 76 - z coordinate: 5.50657 So I would hazard a guess that there is a whole range of points that you discover one after the other, that seemingly are a problem - but why? As regards error 4100 Ivar KJELBERG at some point suggested: "often you can get around by meshing edges/faces by hand, removing "small" features in the geometry (if not relevant to the solution) or cut the volumes into smaller parts to allow the mesher to better adapt to complex geometry shapes. Try also to change the default meshing size, or adapt it for the different parts Try to understand where and what is getting wrong by studying the geometry around that coordinate (and any similar shaped element in your model). Automatic meshing of complex geometries is not trivial, not even for a human ;)" But as I said the point(s) in question are well within one face, nothing complicated going on around, no idea how to remove any small feature. The faces are all pretty much of the same size so I dont see why changing the default meshign parameters would change anything. And also, my error is not only about the meshing, since I cant even set boundary conditions! Any ideas anyone? Regards, Susanne

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2010年7月8日 GMT-4 08:50
...
...

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2010年7月9日 GMT-4 11:21
... actually, the stl object I read in has a NEGATIVE volume .. no wonder it doesnt become a subdomain .. any ideas on how to avoid the volume becoming negative? ...
... actually, the stl object I read in has a NEGATIVE volume .. no wonder it doesnt become a subdomain .. any ideas on how to avoid the volume becoming negative? ...

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.