Particles disappear during increase in gravity

Submitted by AzamatSalamat on Sat, 04/18/2020 - 07:47


I am doing simulations of powder compaction using artificial gravity. I have been simulating them under 2000g. So once this gravity is applied, particles start to disappear as in the screenshot attached. You can see that number of particles reduced from 7884 to 7508 and then it stays constant. Reducing the time step from 1e-6 to 1e-7 seconds has increased the final number of particles, but only slightly, so some particles are always disappearing.

I wanted ask is the problem only in the time step size? Do I need to reduce the time step further? The input file is also attached

Best regards,

EDIT: added the latest versions of files that worked for me

Binary Data gravity.tar_.gz1.41 MB
Plain text icon in.txt5.24 KB
Plain text icon in2.txt4.4 KB

AzamatSalamat | Mon, 05/04/2020 - 08:52

Decreasing time step further did not help much, but increasing the Young's modulus from 1e8 to 1e9 Pa did. I have set this value both for my particles and my mesh. Maybe particles were penetrating the geometry since it was not hard enough, idk.
Setting different Young's moduli or decreasing it made everything worse, destabilizing even more
But with higher Young's modulus for everything, I did not lose any particles and it seems ok.
Anyway I would suggest keeping time step as low as possible for high-speed/high-acceleration applications depending on your hardware of course

bernardnov | Fri, 05/29/2020 - 14:11

Hi AzamatSalamat,

I encounter the same problem with small spherical particles (50 micrometer diameter). I have not try to decrease the timestep, would you mind to share your timestep value for your application?


AzamatSalamat | Sun, 05/31/2020 - 18:22

Hello bernardnov,

I've separated my input on two files and have added their latest versions for this case in the first message. I used the timestep 1e-5 and increased Young's modulus.
For all cases, I would suggest using fix check/timestep/gran command to keep your timestep lower than 0.1 of Rayleigh and Hertz time


bernardnov | Sun, 05/31/2020 - 22:05

That's helping, thanks!