region mesh/tet implementation

Submitted by partnow on Sat, 11/21/2015 - 13:56

The LIGGGHTS 3.3.0 release announcement states: "Also made an important improvement to region mesh/tet. The all_in yes feature was implemented, and some algorithms necessary for particle insertions have been improved considerably. You should see a speed-up when using tet meshes for insertion."

Is this feature fully implemented ? All_in seems to be only an option for "fix insert".

Inspecting the code reveals that a mesh/tet region always enforces the caluclation of the corresponding surface of the region

tri_mesh.useAsInsertionMesh(false);
build_neighs();
build_surface();
tri_mesh.initalSetup();

which seems required for the use of the "all_in" feature for particle insertion. However, it would be desirable to prevent these calculations if a region mesh/tet is used without the intention to ever use "all_in yes".

Is it intended (and possible) to provide an additional option for the defintion of region mesh/tet that surpresses the use of "all_in yes" and hence to avoid the aforementioned calculations.

ckloss's picture

ckloss | Thu, 12/10/2015 - 23:13

Hi partnow,

I'll write it on my list of things to check for the next release!

Best wishes
Christoph

ckloss's picture

ckloss | Fri, 01/08/2016 - 11:29

Hi partnow,.

thanks for your feed-back. After reviewing your points, I'd leave it as is. The mesh topology is a general feature, which - at the moment - is only used by fix insert but it might be used for something else in the future. The runtime overhead introduced by calculating the topology should be marginal

Let me know what you think

Christoph

partnow | Fri, 01/29/2016 - 17:24

I use the region mesh/tet feature in an attempt to couple LIGGHTS with a FEM code. I obtain regions from the FE code as TET meshes which I can directly define as region with the "region mesh/tet" command. However, many of these meshes are very irregular disconnected volumes and I could use them in earlier versions when the mesh exterior surface was not calculated.
Since the surface calculation is enforced all the time (even if all_in is not required) many meshes are not suitable to be used with mesh/tets anymore. I have to check if this behaviour may changed in your latest release. Thanks for the frequent updating and improvements !
The fix insert is a powerfull and substantial feature and I reckon it'll be even more usefull if it is as robust as possible and tet meshes are just a very convenient way to introduce irregular volume regions to Ligghts simulations. (and they are used by FEM simulations anyway).

ckloss's picture

ckloss | Mon, 02/08/2016 - 14:58

Hi Partnow,

thanks for the info! I'll let you know in case of any updates relevant for this. By the way, if you want to share any of your results: Just email me a cool pic/video and I'll add it to the "featured work" section

Best wishes
Christoph

partnow | Fri, 03/04/2016 - 23:02

Hi Christop, thanks for your positive feedback. I'll get in touch with you as soon as we have something nice to show. We'll working on some geotechnical applications that utilize ligghts and finite element simulation.
Regards,
partnow