excessive pressure printed in log file

LAMMPS LAMMPS Mailing List Mirror
1

Hi,

I have simulated water flow between two parallel walls. My problem is that pressure printed in log file shows a very high value. The pressure value is about 400000 atm!!! Also the total energy is in the order of 130000 kcal/mol.

It is worth mentioning that I have validated my work by 3 different works and my velocity and density profiles seem to be true. So, what is the reason of this excessive pressure?

Thanks,

Omar

2

Hi,

I have simulated water flow between two parallel walls. My problem is that
pressure printed in log file shows a very high value. The pressure value is
about 400000 atm!!!! Also the total energy is in the order of 130000
kcal/mol.

It is worth mentioning that I have validated my work by 3 different works
and my velocity and density profiles seem to be true. So, what is the reason
of this excessive pressure?

there are multiple possible reasons. it is very difficult to say
anything with so limited information, though. two things come to mind:
- your potential or simulation parameters are bad. did you run a check
on a non-flowing, regular, all-periodic bulk system?
- you didn't account for the impact of the flow on kinetic energy and pressure.

axel.

3

Dear Axel

Thanks for your answer.

I have checked my simulation and found that the problem is because of the solid walls. My system consists of two parallel walls each atom of which is fixed in its place by Spring/self command with K=1500 (if I use less K, I will get a lost atom error). I set the wall temperature at T=300K by NVT ensemble.

When I remove the walls and simulate a bulk system, I do not deal with such a problem.

I even simulated a system including just the two walls (without any fluid) and found that the printed pressure is very high yet.

So, what is the problem here? What is wrong with my simulation?

4

You are probably getting large contributions to the virial from wall-wall interactions. Use neigh_modify exclude to turn off the interaction between wall atoms. note that if the wall-fluid interaction is anything other than true pairwise additive potential, using neigh_exclude will probably change the wall-fluid interaction. Now that I think about it, you are probably also getting large contributions from the tether forces. I don’t think LAMMPS provides a user-command to remove these contributions. These contributions should have a time average of zero, but the instantaneous fluctuations could be very large. There is no simple solution to this problem, except to accept that the reported pressure is an artifact of the fake walls you have constructed. You can still measure pressure in other ways e.g.

  1. total force between fluid and wall
  2. local pressure obtained by summing over per-atom virial in a region containing only fluid
  3. measure fluid density at point farthest from walls and assume pressure is that of bulk fluid at same density

This problem goes away if you freeze the wall atoms and set their forces to zero. But that makes the fluid-wall interface less realistic.

We could add a switch to LAMMPS to exclude the contributions of certain forces to the virial, but I don’t think there is any way to exclude forces in a way that would yield a physically meaningful pressure.