Two successive simulations with different gravity values, read_data Issue

Hi, everyone.

I want to execute two successive simulations with granular atoms. (LAMMPS-64bit-9Oct2020-MPI for windows)

First, I created granular atoms using the fix pour command. Due to the huge number of atoms (more than 360000), I set a gravity value was 1e5 times than normal value for saving time, and dump only atoms ID, type, and xyz coordinates.

Second, I used the read-data command to import the granular atoms into the 2nd simulation, and here, I set the normal value of gravity. I want to wiggle the wall to make all particles move. However, I found the granular atoms still with a 1e5 gravity value at the beginning of 2nd simulation, which made it unstable and boom.

When I tried the same way but only pour 50000 atoms, the 2nd simulation can become stable (only at the beginning, the granular atoms still with a 1e5 gravity value, they would have the normal value after importing them successfully).

Could you give me some suggestions to make the 2nd simulation stable?

Best regards,

LAMMPS data files usually store positions and velocities (and certainly not accelerations by default). So the most likely explanation is either that your particles have very unstable positions, or very high velocities.

Note that LAMMPS data files are human-readable, and you can quickly verify what sections a LAMMPS data file contains and quickly graph its contents (even in Microsoft Excel, if you do your text import correctly!).

If you are doing something rather unusual, and therefore expect that the source of the issue may be something rather more unusual, it’s best if you upload your input script here so we can take a look at it. (It’s easiest for everyone – including you! – to find a problem if you can simplify your input as much as possible to get a “minimal working example”, which will be complicated enough to show your result but as simple as possible otherwise.)

When you heavily compress some elastic object and then instantly remove the compressing force, what should happen?
You can try releasing the stress by running a minimization before continuing with an MD.

Why not use “create_atoms” and create them close to their final location?
Perhaps followed by “set” to set specific properties.

Thanks for your reply. As Axel Kohlmeyer said, I need to release the stress among atoms due to the unusual setting of gravity.
I tried creating atoms with the lattice distribution command, it works well. Thanks!

Thanks for your reply.
When I simulated a pouring process with 1e6 times normal gravity and removed the particles into the normal environment, I found all of them rebounded.
I tried releasing the stress using a minimization command, but my simulation contains more than 300000 particles, which made it difficult to keep stable even with minimization.
Thanks for your suggestion, I think create_atoms with lattice distribution is better than the pour command in my case. Thanks!