Different velocities by changing thermostating way.

Dear LAMMPS users:

I am running two simulations of pressure-driven water flow (spce) inside a CNT. Both are at the same temperature 298K and both are under the same pressure gradient P1 by using fix addforce in the z-direction. Both tests are basically the same, the only thing that I changed is the way of thermostating.

In test1 I used 2 thermostats one for the CNT and another for the water; in this case I got an average velocity of 250 m/s.

In test2 I used only one thermostat for the whole system, and I got an average velocity of 400 m/s.

In both cases I used fix nvt and temp/profile for removing the z-direction degree of freedom for the temperature saw by the thermostat. I am using a 20 timesteps Tdamp. Each timestep is fs (real units).

I expected to get the same velocity in both cases, what am I misunderstanding or what am I doing wrong?

I am using Tersoff for the CNT and LJ for interactions water-CNT. I am using a May15-2015 version.

Thank you very much for your help.

Jose Cobena

Mork Family Department, USC.

Dear LAMMPS users:

I am running two simulations of pressure-driven water flow (spce) inside a
CNT. Both are at the same temperature 298K and both are under the same
pressure gradient P1 by using fix addforce in the z-direction. Both tests
are basically the same, the only thing that I changed is the way of
thermostating.

In test1 I used 2 thermostats one for the CNT and another for the water;
in this case I got an average velocity of 250 m/s.

In test2 I used only one thermostat for the whole system, and I got an
average velocity of 400 m/s.

In both cases I used fix nvt and temp/profile for removing the z-direction
degree of freedom for the temperature saw by the thermostat. I am using a
20 timesteps Tdamp. Each timestep is fs (real units).

I expected to get the same velocity in both cases, what am I
misunderstanding or what am I doing wrong?

i would expect to see *different* temperature profiles for the water in the
two cases and thus the biasing would be different and consequently you
should get different results.

axel.‚Äč