MESHING 3D VOLUME
- Borja AP
- Topic Author
- Offline
- Junior Member
Less
More
- Posts: 21
- Thank you received: 0
14 years 7 months ago #4148
by Borja AP
MESHING 3D VOLUME was created by Borja AP
Hi folks!
I'm trying to mesh a 3D Bracket (imported as step and iges formats) with the maximum nomber of hexaedral finite elements, as the geometry has more than 4 sides, I followed the next process:
Global Mesh:
3d Algorithm: "3D Extrusion" (Prism_3D)
2d Algorithm: "Quadrangle 2d" (hypotesis, "Quadrangle Preference")
1d Algorithm: "Composite Discretization" (hypotesis, "Average length_1 = 5")
It does not mesh the hole geometry but the sides, being the thickness of the bracket, which are 4-sided.
So, I try to build a SUBMESH for both non-4-sided faces which in the geometry module have been set as a group under the main geometry.
SUBMESH parameters:
2d Algorithm: "Netgen 2D" with "Allow Quadrangles" checked.
1d Algorithm: "Composite side discretization" (hypotesis, "Average length_1 = 5")
After computing the mesh & submesh I get all the surfaces meshed but I recive a message that "Prism_3D" algorithm failed because the mesh on both faces of the submesh are different, so 3d prisms are not created.
I've tried the same process on a geometry build in salome, and it works correctly, but importing such geometry in both iges or step doesn't work.
I'am thinking on may be a precision/tolerance problem??
Any ideas?
You may find the images and iges/step files here
I'm trying to mesh a 3D Bracket (imported as step and iges formats) with the maximum nomber of hexaedral finite elements, as the geometry has more than 4 sides, I followed the next process:
Global Mesh:
3d Algorithm: "3D Extrusion" (Prism_3D)
2d Algorithm: "Quadrangle 2d" (hypotesis, "Quadrangle Preference")
1d Algorithm: "Composite Discretization" (hypotesis, "Average length_1 = 5")
It does not mesh the hole geometry but the sides, being the thickness of the bracket, which are 4-sided.
So, I try to build a SUBMESH for both non-4-sided faces which in the geometry module have been set as a group under the main geometry.
SUBMESH parameters:
2d Algorithm: "Netgen 2D" with "Allow Quadrangles" checked.
1d Algorithm: "Composite side discretization" (hypotesis, "Average length_1 = 5")
After computing the mesh & submesh I get all the surfaces meshed but I recive a message that "Prism_3D" algorithm failed because the mesh on both faces of the submesh are different, so 3d prisms are not created.
I've tried the same process on a geometry build in salome, and it works correctly, but importing such geometry in both iges or step doesn't work.
I'am thinking on may be a precision/tolerance problem??
Any ideas?
You may find the images and iges/step files here
Attachment images.zip not found
- Borja AP
- Topic Author
- Offline
- Junior Member
Less
More
- Posts: 21
- Thank you received: 0
14 years 7 months ago #4160
by Borja AP
Replied by Borja AP on topic Re:MESHING 3D VOLUME
Hi everybody,
I have been investigating a bit, it seems like it is all about tolerance/precision of the imported model, because building it on salome, doesn't give any errors when meshing though.
So, my question is, is it possible to set a Default Tolerance/Precision in the geometry module instead of 1e-7 which comes by default??
Thanks in advance,
BAP
I have been investigating a bit, it seems like it is all about tolerance/precision of the imported model, because building it on salome, doesn't give any errors when meshing though.
So, my question is, is it possible to set a Default Tolerance/Precision in the geometry module instead of 1e-7 which comes by default??
Thanks in advance,
BAP
- Borja AP
- Topic Author
- Offline
- Junior Member
Less
More
- Posts: 21
- Thank you received: 0
14 years 7 months ago #4167
by Borja AP
Replied by Borja AP on topic Re:MESHING 3D VOLUME
Hi everybody.
Is it possible that I am the unique person who experienced that problem with iges or step importation??
Come on Kees/Claus you may have some experience on it...
I am a Naval Architect and I will allways do the hull forms on specialized Naval Architecture Software so I'll allways deal with step or iges importation into salome.
Regards.
Salome-Meca_2009GPL
Salome v.4.1.4<br /><br />Post edited by: Borja AP, at: 2010/05/03 15:46
Is it possible that I am the unique person who experienced that problem with iges or step importation??
Come on Kees/Claus you may have some experience on it...
I am a Naval Architect and I will allways do the hull forms on specialized Naval Architecture Software so I'll allways deal with step or iges importation into salome.
Regards.
Salome-Meca_2009GPL
Salome v.4.1.4<br /><br />Post edited by: Borja AP, at: 2010/05/03 15:46
- kwou
- Offline
- Moderator
14 years 7 months ago #4169
by kwou
Interest: structural mechanics, solar energy (picture at 'my location' shows too little pv panels)
--
kind regards - kees
Replied by kwou on topic Re:MESHING 3D VOLUME
Hoi Borja
oeps, fjooeeoeo.
No, it did not check your problem. I rarely use step and iges files to read into Salome.
And I nearly always use netgen123 to mesh my constructions. Just because it is easy most of the time. Though whether this is useful for this stp/igs problem I cannot say of course.
But for once, tomorrow *) I *will* have a look at your step and iges files. So if you can bear at least one more night ....?
But maybe Claus will try tonight.
*) Tomorrow I will first install Ubuntu 10.04. So I probably have some issues for the rest of the month. So if you donot here from me the next couple of days, it is a non-operational pc. Unless someone convinces me not to upgrade, like eg:
caelinux.com/CMS/index.php?option=com_jo...=6&limitstart=24
or
caelinux.com/CMS/index.php?option=com_jo...&id=4161#msg4161
kind regard - kees<br /><br />Post edited by: Kees Wouters, at: 2010/04/30 00:19
oeps, fjooeeoeo.
No, it did not check your problem. I rarely use step and iges files to read into Salome.
And I nearly always use netgen123 to mesh my constructions. Just because it is easy most of the time. Though whether this is useful for this stp/igs problem I cannot say of course.
But for once, tomorrow *) I *will* have a look at your step and iges files. So if you can bear at least one more night ....?
But maybe Claus will try tonight.
*) Tomorrow I will first install Ubuntu 10.04. So I probably have some issues for the rest of the month. So if you donot here from me the next couple of days, it is a non-operational pc. Unless someone convinces me not to upgrade, like eg:
caelinux.com/CMS/index.php?option=com_jo...=6&limitstart=24
or
caelinux.com/CMS/index.php?option=com_jo...&id=4161#msg4161
kind regard - kees<br /><br />Post edited by: Kees Wouters, at: 2010/04/30 00:19
Interest: structural mechanics, solar energy (picture at 'my location' shows too little pv panels)
--
kind regards - kees
- Borja AP
- Topic Author
- Offline
- Junior Member
Less
More
- Posts: 21
- Thank you received: 0
14 years 7 months ago #4170
by Borja AP
Replied by Borja AP on topic Re:MESHING 3D VOLUME
Hi Kees,
Do not worry, I can wait, I will try building the geometry in rhinoceros differently and if it doesn't work may be i try Salome-Meca.2010 which I guess it won't be a problem there.
Good luck with your update and have a nice weekend!
Bap
Do not worry, I can wait, I will try building the geometry in rhinoceros differently and if it doesn't work may be i try Salome-Meca.2010 which I guess it won't be a problem there.
Good luck with your update and have a nice weekend!
Bap
- Claus
- Offline
- Moderator
Less
More
- Posts: 670
- Thank you received: 34
14 years 7 months ago #4171
by Claus
Code_Aster release : STA11.4 on OpenSUSE 12.3 64 bits - EDF/Intel version
Replied by Claus on topic Re:MESHING 3D VOLUME
Borja AP wrote:
Okay, so I played around with the bracket a bit while watching a horrible movie (The Happening, don't watch it) and my conclusion is that the netgen (with quads) is simply too flimsy to work with extrusion 3D.
I can get a pretty decent quality mesh by doing the following:
All the sides that has a normal parallel to XY plane i.e. the small faces incl. the holes are grouped and meshed with quads (as submesh). The two big surfaces are meshed with 2D tets and have 3D extrusion set for volume meshing.
That gives a reasonable mesh - combined with 'quadratic' i.e. 2nd order elements, it should be okay.
Another way, which is probably the easiest way now that I think about it, is to extrude one of the big surfaces after meshing it with netgen 2D-1D. Create a group on the lower big surface, select the group and create a new mesh (you can only select 2D now), define it as 2D-1D netgen with quads. Mesh it. Select the mesh and click extrude in the tool bar, tag 'select whole mesh/submesh' and select how many steps the extrusion should take, devide the Z-distance with this: 4 extrusion steps = Z-height of 0.0025 because you'd want a result of 0.01. Don't forget to switch to 'faces' instead of 'elements' in to top of the window, otherwise you'll just get a hollow shell.
In attachment: left is done with the first method, right, the second.
Note: Meshing on the Linux platform is shite atm., I can't find a good free mesher that'll create quality quad/hexa mesh - Salomé is actually the best one available, 'cept LS-DYNAs prepost, but that doesn't export the mesh, sadly.
I'll stop typing now.
Edit: to answer your question:
Hi every body.
Is it possible that I am the unique person who experienced that problem with iges or step importation??
Come on Kees/Claus you may have some experience on it...
I am a Naval Architect and I will allways do the hull forms on specialized Naval Architecture Software so I'll allways deal with step or iges importation into salome.
Regards.
Salome-Meca_2009GPL
Salome v.4.1.4
Okay, so I played around with the bracket a bit while watching a horrible movie (The Happening, don't watch it) and my conclusion is that the netgen (with quads) is simply too flimsy to work with extrusion 3D.
I can get a pretty decent quality mesh by doing the following:
All the sides that has a normal parallel to XY plane i.e. the small faces incl. the holes are grouped and meshed with quads (as submesh). The two big surfaces are meshed with 2D tets and have 3D extrusion set for volume meshing.
That gives a reasonable mesh - combined with 'quadratic' i.e. 2nd order elements, it should be okay.
Another way, which is probably the easiest way now that I think about it, is to extrude one of the big surfaces after meshing it with netgen 2D-1D. Create a group on the lower big surface, select the group and create a new mesh (you can only select 2D now), define it as 2D-1D netgen with quads. Mesh it. Select the mesh and click extrude in the tool bar, tag 'select whole mesh/submesh' and select how many steps the extrusion should take, devide the Z-distance with this: 4 extrusion steps = Z-height of 0.0025 because you'd want a result of 0.01. Don't forget to switch to 'faces' instead of 'elements' in to top of the window, otherwise you'll just get a hollow shell.
In attachment: left is done with the first method, right, the second.
Note: Meshing on the Linux platform is shite atm., I can't find a good free mesher that'll create quality quad/hexa mesh - Salomé is actually the best one available, 'cept LS-DYNAs prepost, but that doesn't export the mesh, sadly.
I'll stop typing now.
Edit: to answer your question:
No, or I don't know how to at least.<br /><br />Post edited by: Claus, at: 2010/04/30 02:11So, my question is, is it possible to set a Default Tolerance/Precision in the geometry module instead of 1e-7 which comes by default??
Code_Aster release : STA11.4 on OpenSUSE 12.3 64 bits - EDF/Intel version
Moderators: catux
Time to create page: 0.127 seconds