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
7 mars 2011, 03:13 UTC−5
Hi
Indeed there are no "layers" for "eccentric cones" so you need to make several one on top of the other, slightly increasing in size. This should work, I do not see why the mesher should differ from both case. Are you sure its not another issue, of some small overlapping surfaces, or tangent touching surfaces, those are not liked by the mesher ?
--
Good luck
Ivar
Hi
Indeed there are no "layers" for "eccentric cones" so you need to make several one on top of the other, slightly increasing in size. This should work, I do not see why the mesher should differ from both case. Are you sure its not another issue, of some small overlapping surfaces, or tangent touching surfaces, those are not liked by the mesher ?
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
7 mars 2011, 06:23 UTC−5
Hi Ivar, thanks for your rapid response.
In fact, if I try to build a simple food packaging with one central ellipsoid and four eccentric cones at the corners, by making an union (without internal boundaries), rescaling it up (keeping the input object), and making a difference between the reescaled object and the original one, I can get a nice simple-layered geometry. But when I try to mesh it by mapping some lateral face and sweeping it to the entire geometry COMSOL begins complaining about "unsupported topology" or "collapsed sections are not allowed".
I've attached the geometry export file of 1/8th of my packaging. I'd be very thankful if you could show me some guidelines to sweep mesh my object.
Hi Ivar, thanks for your rapid response.
In fact, if I try to build a simple food packaging with one central ellipsoid and four eccentric cones at the corners, by making an union (without internal boundaries), rescaling it up (keeping the input object), and making a difference between the reescaled object and the original one, I can get a nice simple-layered geometry. But when I try to mesh it by mapping some lateral face and sweeping it to the entire geometry COMSOL begins complaining about "unsupported topology" or "collapsed sections are not allowed".
I've attached the geometry export file of 1/8th of my packaging. I'd be very thankful if you could show me some guidelines to sweep mesh my object.
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
7 mars 2011, 07:12 UTC−5
Hi
That geometry is too complex for a mesher to sweep mesh in one go automatically, it needs some "help".
When you construct it you should keep the interiour boundaries, you could even use a 1/6 shape and mirror/revolve it.
With the interiour boundaries you can start to mesh it by taking a surface and sweeping it.
Revolve mesh is not (yet ?) implemented in v4.1 (it was in 3.5a).
Note: COMSOL sweep meshes correct in some quadrant, but not in others, I haven't fully understood exactly how to get around that one, and Support has told me they are working on it to improve that for the future releases, so sometimes by changing workplane source plane (xy, yz, zx ...) it meshes more "as expected" or sweep meshing
Do you need the layers for fibre type structures, or just to get a nice structured mesh ?
Any how try to separate the two cone and ellipse with an interiour "natural " boundary.
If nothing works, you will have to use an external CAD tool
--
Good luck
Ivar
Hi
That geometry is too complex for a mesher to sweep mesh in one go automatically, it needs some "help".
When you construct it you should keep the interiour boundaries, you could even use a 1/6 shape and mirror/revolve it.
With the interiour boundaries you can start to mesh it by taking a surface and sweeping it.
Revolve mesh is not (yet ?) implemented in v4.1 (it was in 3.5a).
Note: COMSOL sweep meshes correct in some quadrant, but not in others, I haven't fully understood exactly how to get around that one, and Support has told me they are working on it to improve that for the future releases, so sometimes by changing workplane source plane (xy, yz, zx ...) it meshes more "as expected" or sweep meshing
Do you need the layers for fibre type structures, or just to get a nice structured mesh ?
Any how try to separate the two cone and ellipse with an interiour "natural " boundary.
If nothing works, you will have to use an external CAD tool
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
8 mars 2011, 18:18 UTC−5
Hi Ivar,
Thank you so much for bring me some good ideas.
These days I've been trying to "help" the mesher by spliting my object into three different domains with the aid of some interior boundaries. I still had problems with the mesher because it seems that COMSOL doesn't like that "artificial" boundaries, but I think I'm on the good way. I'll keep you updated.
At the beginning I tried to build a simple mesh of free tetrahedra for the entire geometry but when I ran it on my notebook it took forever and ended with some "out of memory" errors. I found the source of my problems in two thin layers of a few micrometers thick that needed a ridiculous amount of tetrahedra to be meshed correctly. So I did a bit of research and found an article in COMSOL Knowledge Base that recommended the sweep meshing to be used in that cases. Furthermore, and unfortunately, the main physics of my model takes place into these thin structures.
Thanks once more for your time.
Hi Ivar,
Thank you so much for bring me some good ideas.
These days I've been trying to "help" the mesher by spliting my object into three different domains with the aid of some interior boundaries. I still had problems with the mesher because it seems that COMSOL doesn't like that "artificial" boundaries, but I think I'm on the good way. I'll keep you updated.
At the beginning I tried to build a simple mesh of free tetrahedra for the entire geometry but when I ran it on my notebook it took forever and ended with some "out of memory" errors. I found the source of my problems in two thin layers of a few micrometers thick that needed a ridiculous amount of tetrahedra to be meshed correctly. So I did a bit of research and found an article in COMSOL Knowledge Base that recommended the sweep meshing to be used in that cases. Furthermore, and unfortunately, the main physics of my model takes place into these thin structures.
Thanks once more for your time.