non-periodic Boundary applied over moving rough walls acts as periodic Boundary

LAMMPS LAMMPS Mailing List Mirror
1

Dear Lammps users,

I am simulating a Couette flow using moving rough walls. I applied periodic boundary condition in x and y directions and a fixed boundary at z-direction in which we have the rough walls. The Rough walls consists of grouped atoms; lower wall is stationary and the upper wall move through “fix velocity” and “fix nve” commands, and also “fix setforce” command set their forces to zero. Other Atoms interact with each other and the walls through Lennard-Jones Potential and integrated through “fix nvt” command. I saved both wrapped positions and unwrapped positions into separate dump files. The input script is provided below.

When I look at the dump file of unwrapped positions, I notice that there are atoms which has escaped the walls and moved beyond the walls and the program did not result in any error. When I take a look at the wrapped positions, you could notice any problem.

I tested several stuff like integrating all atoms motion together and applying thermostat to the flow by velocity rescaling; I also tested “fix move” for the walls (this results in escaping atom error and program stop); I also tested different non-periodic boundary (m and s) instead of fix. None of them helped me to solve the issue. I thought that the walls may be excluded from interactions with the flow atoms, but the motion of flow atoms with the moving wall contradicts this :frowning:
I hope you guys can help me to solve this issue or a way to avoid it.

Best Regards

2

When I look at the dump file of unwrapped positions, I notice that there are atoms which has escaped the walls and moved >beyond the walls and the program did not result in any error.

3

Dear Steve,

Thanks for your response. As it is noted in the input script, I set the boundaries as “p p f”. and I know that unwrapped positions exclude periodic boundary condition and gives us the real position but in the direction where we have fixed boundary and walls, you expect that no atoms leave the domain.

Best regards

4

Dear Steve,

Thanks for your response. As it is noted in the input script, I set the
boundaries as "p p f". and I know that unwrapped positions exclude periodic
boundary condition and gives us the real position but in the direction where
we have fixed boundary and walls, you expect that no atoms leave the domain.

but what if they had already left *before*?
in other words, have you checked if your data file has the correct
image flags set?

keep in mind that MD is strictly following the GI-GO principle
(garbage in -> garbage out). :wink:

axel.