I encountered a possible bug while running a simulation wherein particles are falling onto the edge of a cylinder (modeled using mesh/surface). I mentioned this bug in the user forum thread titled "Older version of LIGGGHTS (2.0.6)" but I guess its better served in this forum. To illustrate the problem I made up a simple case of a single particle falling onto the edge of a cylinder. If the particle center is aligned with the edge, everything is fine, however if the particle center is offset a little bit, it gains a lot of energy almost immediately after it comes into contact with the edge and flies off. The only loading in this case is gravity and even for a small coefficient of restitution (0.5), the particle still manages to fly off.
I attach the two files needed for the simulation, the input file (Bug.txt) and the mesh file (Edge_cyl.txt). Please change the extension of Edge_cyl.txt to Edge_cyl.stl before running the simulation. In the thermo output you will notice that between steps 147900 and 148000 the kinetic energy of the particle jumps suddenly.
Thanks,
Tapan
| Attachment | Size |
|---|---|
 Bug .txt | 1.65 KB |
| Edge_cyl.txt | 51.9 KB |
ckloss | Mon, 08/27/2012 - 09:31
Hi Tapan, thanks - Philippe
Hi Tapan,
thanks - Philippe Seil and myself will have a look
Cheers, Christoph
ericparteli | Mon, 08/27/2012 - 17:20
comment on your case
Hi, Tapan,
great post, thanks so much! This problem is so interesting for me too and I would like to comment on it.
I run your case and I can reproduce the error too. What I see in the simulation is that the particle is penetrating into the cylinder (practically ignoring the stl surface). And then, after a small penetration, the deformation is so large (unrealistic for the material properties) that the velocity then gets unrealistically large, and that is why you see the "explosion", right? Just a note, I had a similar problem when running LIGGGHTS version 2.0.4 in parallel, whereas such "explosions" occurred only for specific processor choices. The problem seems to have been fixed in the subsequent versions - at least that problematic case I had before is running OK now. Indeed, your case problem doesn't seem to be connected to that bug (or maybe it is.. let us wait for Christoph's answer) because I run it with different processor choices and the explosion always occurs.
Is the thickness of the walls equal to 0.0?
All the best
Eric
tapsab | Mon, 08/27/2012 - 17:48
Eric, Thanks for your
Eric,
Thanks for your comment. Actually I think the problem I encountered is more general because I have seen the "exploding particle" for various different geometries where instead of a sharp edge I used a smooth fillet and yet the particle exploded. Maybe its related in general to particle interaction with corners that are inevitable in convex mesh surfaces.
To answer your question, the wall thickness is indeed zero, actually I am not aware that you can have a nonzero wall thickness.
Best,
Tapan
ericparteli | Mon, 08/27/2012 - 18:26
Hi, Tapan, thanks for your
Hi, Tapan,
thanks for your detailed explanations! In Inventor (the software that I use to model stl surfaces), I use the three-point-rectangle utility to construct the wall and then I do a revolution of 360 degrees, and that makes the cylindrical wall with thickness equal to the width of the original rectangle. My question was to see whether you encounter this error only if your wall has no thickness, but I have no clue as to which extent this might help Christoph and Philippe to localize the source of the error in your case.
Best
Eric