Hi guys,
I am now still testing the packing of particles and I used the packing example.
Differently, I used the pair_style of gran/hooke/history/stiffness 1 0 and tried to used slightly smaller particle size (I tried 3mm but failed), I also import mesh from stl files generated by Gmsh.
My problem is:
1. the Pour/dev function actually insert much less particles into the region, (less than 100), but the region is big enough to contain a lot more particles.
2. It's hard to set the size of particles and a suitable skin distance (I used 2mm, the same as example), say if I use smaller particle size (like R<3mm), LIGGGHTS will warn that Failed to allocate %d bytes for array %s
3. and even though I somehow select a fine combination of particles size and skin distance, the LIGGGHTS is so slow in running the simulation, while the example runs very quickly (my particle number is even less than that in the packing example). Sometimes, it takes a half day to run a 100+ particles simulation (using Hookean model and with interaction with walls), and sometimes, the computer just crash down completely......
4. After I tried bigger particle size (less particles, around 100 particles), it worked but it warned:
INFO: Maxmimum number of particle-tri neighbors >10 at step 1, growing array...done!
INFO: Maxmimum number of particle-tri neighbors >20 at step 1, growing array...done!
INFO: Maxmimum number of particle-tri neighbors >30 at step 1, growing array...done!
INFO: Maxmimum number of particle-tri neighbors >40 at step 1, growing array...done!
INFO: Maxmimum number of particle-tri neighbors >50 at step 1, growing array...done!
INFO: Maxmimum number of particle-tri neighbors >60 at step 1, growing array...done!
INFO: Maxmimum number of particle-tri neighbors >70 at step 1, growing array...done!
INFO: Maxmimum number of particle-tri neighbors >80 at step 1, growing array...done!
INFO: Maxmimum number of particle-tri neighbors >90 at step 1, growing array...done!
I really don't understand what that means, given that my simulation size is pretty small.
Attached is my input script that generated this warnings, the screen output as well as my mesh STL files.
Guys, please see if you can help. Also, (problem 2 & 3), why the simulation is so slow? something wrong with my script?
| Attachment | Size |
|---|---|
 in.test_v_0_7.txt | 3.41 KB |
| Screen Output.txt | 7.19 KB |
 walls_stl.rar_.gz | 317.94 KB |
Philippe | Mon, 02/27/2012 - 08:54
slow simulation
Hi,
I just took a look at your case, and I assume that the problems come from your very fine mesh. In LIGGGHTS, you do not gain anything from having a mesh where the triangles are significantly smaller than the particles unless you want to do eg. stress analysis. The geometry you want to import could also be done with ~30 triangles, which will be a lot faster.
Concerning insertion, you could try to increase the number of tries.
regards
Philippe
venes520 | Mon, 02/27/2012 - 09:39
Thanks Philippe, you are
Thanks Philippe,
you are right about the fineness of my mesh, I tried less triangles and now it works faster indeed.
However, the simulation still runs at a very low speed. any thoughts on that?
Also, do you know how to add a force onto the wall? so that the wall can compress the particles.
ckloss | Mon, 02/27/2012 - 11:14
>>However, the simulation
>>However, the simulation still runs at a very low speed. any thoughts on that?
I suggest you "debug" your simulation step by step
Christoph