Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.
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.
The problem with interior boundaries
Posted Jun 12, 2012, 5:46 a.m. EDT Structural Mechanics Version 4.2, Version 4.2a 3 Replies
Please login with a confirmed email address before reporting spam
Hi,
I want to compute the mechanical properties for a structure by using the solid mechanics model of Comsol 4.2.0. This tridimensional structure contains some connected cylindrical bars and a sphere is placed at the intersection of these bars (so this structure contains some cylindrical bars and spheres). Only one material is used for all domains. To build this geometry, I have used some Boolean Operations like “Union” or “Difference” for which we have the possibility to “keep interior boundaries” or not. However, the final results (stress and strain) when I select the “keep interior boundaries” option are very different from the solution that this option is deselected. Knowing that I used a very fine mesh for both cases. Anyone please explain me this problem? What type of symmetry is more accurate in this case?
Thanks in advance.
I want to compute the mechanical properties for a structure by using the solid mechanics model of Comsol 4.2.0. This tridimensional structure contains some connected cylindrical bars and a sphere is placed at the intersection of these bars (so this structure contains some cylindrical bars and spheres). Only one material is used for all domains. To build this geometry, I have used some Boolean Operations like “Union” or “Difference” for which we have the possibility to “keep interior boundaries” or not. However, the final results (stress and strain) when I select the “keep interior boundaries” option are very different from the solution that this option is deselected. Knowing that I used a very fine mesh for both cases. Anyone please explain me this problem? What type of symmetry is more accurate in this case?
Thanks in advance.
3 Replies Last Post Jun 20, 2012, 4:28 p.m. EDT
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
the way you generate your geometry: 2-3 cylinders intersecting a sphere, or by specifically making the differences and unions should have no influence (apart they might slightly change the mesh densities locally if you remove many optional internal boundaries).
I seldom use differences in Union mode, while these are mandatory in Assembly mode to correctly make each region unique.
So it's strange, for me that you get different results.
--
Good luck
Ivar
the way you generate your geometry: 2-3 cylinders intersecting a sphere, or by specifically making the differences and unions should have no influence (apart they might slightly change the mesh densities locally if you remove many optional internal boundaries).
I seldom use differences in Union mode, while these are mandatory in Assembly mode to correctly make each region unique.
So it's strange, for me that you get different results.
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi Ivar,
Thanks very much for your answer. But until now I haven’t figured out why there is such a difference in results between different ways to model the geometry (“keep interior boundaries” or not). If I understand you correctly, the option for the interior boundaries doesn’t influent the final result (for very fine mesh) and a model without interior boundaries is preferred in the Union mode? To find a reason, I run four different computation cases concerning geometry:
(1) The geometry model has only cylinders; the option “keep interior boundaries” is selected.
(2) The geometry model has only cylinders; the option “keep interior boundaries” is deselected.
(3) The geometry model has cylinders and spheres at their connections; the option “keep interior boundaries” is selected.
(4) The geometry model has cylinders and spheres at their connections; the option “keep interior boundaries” is deselected.
For each case, I computed two parameters: solid.sx and solid.sy for all domains (these parameters are useful for the next computation, not detail here). Knowing that the radius of sphere is only 1.5 times bigger than the cylinder’s one. So physically the influence of spheres on the mechanical properties is not significant. The results show that the three first cases have approximately the same results but different from the last one ( case 4: geometry model with spheres and without interior boundaries). Probabilistically, we choice the case 4 like an error computation because it is different from three other computations. However, this computation results (case 4) seem to be comparable with the knowing analytical solution; and as you has previously recommended: a computation without interior boundaries in the Union mode is true (so case 2 and case 4 have normally the close results !?).
With these reasons, it is very difficult for me to know what the correct computation is. Could you help me to see more in details? I send you here both files for the last two computations. Please tell me if you need more information.
Thanks again.
Thanks very much for your answer. But until now I haven’t figured out why there is such a difference in results between different ways to model the geometry (“keep interior boundaries” or not). If I understand you correctly, the option for the interior boundaries doesn’t influent the final result (for very fine mesh) and a model without interior boundaries is preferred in the Union mode? To find a reason, I run four different computation cases concerning geometry:
(1) The geometry model has only cylinders; the option “keep interior boundaries” is selected.
(2) The geometry model has only cylinders; the option “keep interior boundaries” is deselected.
(3) The geometry model has cylinders and spheres at their connections; the option “keep interior boundaries” is selected.
(4) The geometry model has cylinders and spheres at their connections; the option “keep interior boundaries” is deselected.
For each case, I computed two parameters: solid.sx and solid.sy for all domains (these parameters are useful for the next computation, not detail here). Knowing that the radius of sphere is only 1.5 times bigger than the cylinder’s one. So physically the influence of spheres on the mechanical properties is not significant. The results show that the three first cases have approximately the same results but different from the last one ( case 4: geometry model with spheres and without interior boundaries). Probabilistically, we choice the case 4 like an error computation because it is different from three other computations. However, this computation results (case 4) seem to be comparable with the knowing analytical solution; and as you has previously recommended: a computation without interior boundaries in the Union mode is true (so case 2 and case 4 have normally the close results !?).
With these reasons, it is very difficult for me to know what the correct computation is. Could you help me to see more in details? I send you here both files for the last two computations. Please tell me if you need more information.
Thanks again.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
I'm on travel and not by my WS so I cannot look at your models, but the issue with interiour boundaries is mainly to avoid some small domains that appear might appear, these might force a very fine local mesh, in your example I suppose the difference is coming from the mesh density, with and without the internal boundaries these mesh densities will naturally change.
Normally, if you mesh fine enough, all similar geometries should show same results independently if the presence of interiour boundaries are there or not. The mesh insensitivity of a model is a first check one must do to be sure the results are valid w.r.t. the mesh density.
--
Good luck
Ivar
I'm on travel and not by my WS so I cannot look at your models, but the issue with interiour boundaries is mainly to avoid some small domains that appear might appear, these might force a very fine local mesh, in your example I suppose the difference is coming from the mesh density, with and without the internal boundaries these mesh densities will naturally change.
Normally, if you mesh fine enough, all similar geometries should show same results independently if the presence of interiour boundaries are there or not. The mesh insensitivity of a model is a first check one must do to be sure the results are valid w.r.t. the mesh density.
--
Good luck
Ivar