Hello!
I am a new user to LIGGGHTS and I have been doing a research related to the centrifugal compaction of metal powder particles. I have set a simulation which is taking quite a long time on my laptop.
I have set the thermo command to show the status each 10000 steps (there are 8E7 steps in total). I have noticed that while in the beginning after 1 hour 1 million steps have been completed now the simulation is much slower and now in 1 hour there are about 300 000 steps completed.
I am wondering what might be the reason behind this. Is it just simply incapability of the hardware to perform under such conditions or there might be something in the code? I am attaching it just in case
Best regards,
Azamat
| Attachment | Size |
|---|---|
 in.txt | 5.89 KB |
richti83 | Tue, 04/07/2020 - 14:18
Simulation time increases
Simulation time increases with number of contacts.
I guess bc. of
fix gravi all gravity 9810 vector 0.0 0.0 -1.0
there are many contacts bc. I think particles will stick together on bottom layer. Is this gravity intended or by mistake ?
AzamatSalamat | Wed, 04/08/2020 - 13:35
Thank you
Thank you, richti83
Yes it is gravity intended, I am making a simulation of metal particles behaviour under artificial gravity of 1000g
Anyway I have optimized the simulation as much as I can
tjleps | Fri, 05/08/2020 - 03:56
Try using shrink-wrap
Try using shrink-wrap boundary conditions. If you have a lot of compaction, the processors at the bottom of the pile will have many more particles to integrate than the processors at the top. Shrink-wrap boundaries will eliminate any empty space at the top of the pile and better balance processor load. You should also check your neighbor length to make sure it's not too large. What is the computation time breakdown after a run?