I have confirmed that there is indeed a problem with sidesets on 3D meshes when
using AMR. Some of the sidesets are completely incorrect after a single
refinement step. The problem can be reproduced using some of the 3D examples
(ex14 or ex15). On a regular mesh some of the sidesets will form a "step"
pattern cutting into the mesh. Therefore some of the element sides that are
actually on the boundary are not in the sideset anymore which is what you are
seeing in your problem here. I worked with the problem a little before
Christmas after a postdoc reported difficulties to me running some of his 3D
problems with AMR and was unable to trace the origin of the problem but I'll
take a look at those new lines you mentioned to see if I can run our problems
when removing them as you did. What I do know is that the problem has been in
there since at least August but I wasn't "easily" able to go back before that
point yet. If I learn anything more I'll post to the list again.
Cody
On Dec 29, 2010, at 3:44 PM, Stefano Berrone wrote:
> I recently installed the release 0.7.0.3 (as well as the svn version). I
> tried to run a code developed for previous releases and I had some
> problems with boundary_id.
> The code
>
> if( elem->neighbor(s) == NULL )
> {
> short int side_id =
> mesh.boundary_info->boundary_id(elem,s);
> .....
> }
>
> correctly get me the boundary_id set on the boundaries in the initial
> mesh up to the first adaptive refinement, but after that it I get
>
> side_id =-1234
>
> on a boundary side.
> The same code was working with the previous releases.
>
> I have noticed that some new lines of code are present in the new
> version of BoundaryInfo::boundary_id(...). I guess that the problem
> comes from that code, in fact if I comment them the code behaves as
> expected.
> Has anybody had similar problems?
> Regards,
> Stefano
>
> ------------------------------------------------------------------------------
> Learn how Oracle Real Application Clusters (RAC) One Node allows customers
> to consolidate database storage, standardize their database environment, and,
> should the need arise, upgrade to a full multi-node Oracle RAC database
> without downtime or disruption
> http://p.sf.net/sfu/oracle-sfdevnl
> _______________________________________________
> Libmesh-users mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/libmesh-users
------------------------------------------------------------------------------
Learn how Oracle Real Application Clusters (RAC) One Node allows customers
to consolidate database storage, standardize their database environment, and,
should the need arise, upgrade to a full multi-node Oracle RAC database
without downtime or disruption
http://p.sf.net/sfu/oracle-sfdevnl
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users
