ElmerFEM is losing bits of multibody mesh made in Salome.

The graphical user interface of Elmer
raback
Site Admin
Posts: 4832
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: ElmerFEM is losing bits of multibody mesh made in Salome.

Post by raback »

Hi, would you have link to the initial UNV file. -Peter
ender_wiggin
Posts: 18
Joined: 21 Jun 2023, 22:56
Antispam: Yes

Re: ElmerFEM is losing bits of multibody mesh made in Salome.

Post by ender_wiggin »

kevinarden wrote: 08 Jan 2024, 23:14 Splitting is not necessary in ElmerGUI if you manage everything in Salome. I use partition instead of cuts, because with cuts you end up with multiple parts and have to worry about coincident nodes and mis-matching boundaries, etc.

Using partitions you end up with 1 part, 1 mesh, with multiple bodies, that have no interface issues.

I think you can add your salome file to the github repository, if you want me to look at it.
I have submitted a pull request that contains the most recent hdf file from my Salome project.

I did manage to find a way to get the mesh file from that Salome project to work. However, it still required the division of boundaries. The splitting angle had to be much larger than default though - 90 degrees on boundary property 2. Boundary property 1 is fine.

I would still be very grateful if you were to look at the Salome file so that, maybe, I won't have to go through the splitting process in the future. I am very appreciative for the effort you have made to help me.
kevinarden
Posts: 2313
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: ElmerFEM is losing bits of multibody mesh made in Salome.

Post by kevinarden »

Yes the splitting can be avoided by grouping all of the surface boundaries that you want to be boundaries in salome. You have 1 group the outer surface, so elmer defaults all of the remaining ones into the last boundary. To avoid splitting you must group all of your boundaries you need in salome just like you did bodies.
The steps are in geometry click the final part, create group
change from solid geometry to face, top of form
name the group
click on the outer boundary surface and select hide, so that you can see the inner bodies
select the surfaces on the inner bodies that you want to be a boundary, use add to add to group, or shift/select to select multiple
apply, will create a group and start a new form.
modeling.png
(274.41 KiB) Not downloaded yet
ender_wiggin
Posts: 18
Joined: 21 Jun 2023, 22:56
Antispam: Yes

Re: ElmerFEM is losing bits of multibody mesh made in Salome.

Post by ender_wiggin »

kevinarden wrote: 10 Jan 2024, 13:01 Yes the splitting can be avoided by grouping all of the surface boundaries that you want to be boundaries in salome. You have 1 group the outer surface, so elmer defaults all of the remaining ones into the last boundary. To avoid splitting you must group all of your boundaries you need in salome just like you did bodies.
The steps are in geometry click the final part, create group
change from solid geometry to face, top of form
name the group
click on the outer boundary surface and select hide, so that you can see the inner bodies
select the surfaces on the inner bodies that you want to be a boundary, use add to add to group, or shift/select to select multiple
apply, will create a group and start a new form.

modeling.png
I apologize for being a bit radio silent on this one. I had other things assigned to me including a different simulation, so had been set aside for a bit.

I have not yet tried adding the additional boundaries, but that is exactly what I did do for outside boundary. I may have misread your instructions posted on GitHub about which boundaries mattered, so that would explain why I didn't make a separate boundary group for the magnets.

In that other simulation I previously mentioned, I couldn't figure out how to make a structured mesh and boundary layer meshing wasn't behaving, so my quick and dirty fix was to force additional layers in the wall thickness. This resulted in me needing to perform this same kind operation being discussed in this forum thread except on the bodies.

I will report further when I have a chance to retry the mesh of the magnets.
Post Reply