LIGGGHTS 3 - move/mesh for mesh/surface geometry

Apparently move/mesh crashes the run for the non-planar stl-file (an annular cylinder) in my problem, while this setup ran fine in LIGGGHTS 2.3.6. It consists of an annular cylinder, which is: 1. filled with particles 2. compacted by a piston, to achieve a high packing 3. moved in a cylindrical orbiting motion (composed by superposition of an eccentric rotation and a back-rotation around its local origin).

The relevant fixes are "move2" and "move4". Files are attached.

Thanks in advance.

Attachment	Size
test.tar_.gz	568.48 KB
input.txt	4.28 KB

Forums:

Bug Reports for CFDEM®coupling, LIGGGHTS®, and ParScale

ckloss | Tue, 04/08/2014 - 19:45

Hi Tijmus ,

your're using two fix wall/gran commands of mesh types in your simulation. This will be disallowed from LIGGGHTS 3.0.1 on because it can lead to crahes.
So please try with 3.0.1 (to be released on Thursday) and let us know if the problem persists!

thanks
Christoph

Tijmus | Thu, 04/10/2014 - 10:14

Hi Christoph,

Will do. What do you mean by disallowed exactly, will there be an alternative approach in 3.0.1? Otherwise it's hard to see if the problem persists.

Thanks.

Tijmus | Fri, 04/11/2014 - 15:31

Had a run with 3.01, which

Had a run with 3.01, which indeed gave an error concerning the double fix wall/gran mesh command. I then reduced this to 1 command, fixing the two meshes to one single wall. This however still leads to a segmentation fault when the orbital move-fixes are executed. It goes wrong in move2 and move4, which put the container in motion (only moving the piston does work).

The modified input-file is uploaded now as well.