Hi,
I am running a simple case of particles tumbling in a rotating drum using LIGGGHTS v2.3.7 and observed tiny explosions. I have run this case before using an earlier version of LIGGGHTS (v2.1.2 ) and did not see any such explosions. I am not sure what changed between the version to suddenly cause these explosions.
I have attached the input file and the drum stl file. I am using a stl file for the drum instead of a primitive because I already had a case setup for the stl file. I am also using the serial version but I believe the problem also occurs in parallel. I have tried reducing the time-step size to 1e-6 (instead of 1e-5 I am using now) but that does not seem to help as well. Recall, that I have run this case successfully before using a earlier version.
Here is a link to the video which shows the explosions.
http://www.youtube.com/watch?v=UqCEkuDMWHs
Can someone take a quick look at the input file and see if there is anything "obviously" wrong? Please also let me know if there are any suggestions to figure out what could be causing the explosions.
Thanks,
Rahul
| Attachment | Size |
|---|---|
 LIGGGHTS input file | 5.71 KB |
| STL file. Remove the .txt extension | 20.3 KB |
ckloss | Wed, 10/09/2013 - 10:17
Hi Rahul,
Hi Rahul,
the mesh import and handling changed between these version (in general, to the better) you have now better control e.g. over how edges are handled - see doc for details. However, having a quick look at your case it seems that we missed the case that adjacent triangles have different node orientations (which is usually not the case). I'll post a patch accounting for this for the next version. I tried your case with that patch and no problems occurred on my end
Cheers
Christoph
rkumar310 | Wed, 10/09/2013 - 18:12
Hi Christoph!
Hi Christoph!
Thanks for taking care of this so quickly! I will run this case in the next version when it becomes available.
Thanks,
Rahul
Lamers Jon | Thu, 10/10/2013 - 14:56
Hi Christoph and Rahul,
Hi Christoph and Rahul,
Thanks Rahul for bringing this up. I've noticed this has randomly in occured in past simulations I have conducted - and I assumed it was my own fault with timestep settings and such (this problem did not present in all of my simulations).
Would it be possible for you to show the necessary changes to the code to fix this (as a short term fix?). I don't want to waste your time so something as simple as the output of a diff command (or something else as quick?) would be really helpful. I'd imagine that this would just be for tri_mesh_I.h, but please include anything else you may have changed.
Thanks,
Jon