I was trying to simulate a Couette flow problem with a temperature gradient along the height of the channel. In order to do this I tried to thermostat the two walls at different temp. using a langevin thermostat. The wall atoms were fixed to their site by spring/self. This limited me from applying a set velocity to the wall (which for a simple Couette flow problem I was doing using the 'velocity' and 'fix setforce' commands). I could alternatively just simulate a temp. gradient just in the fluid but have a doubt that it would affect the fluid-wall interface hydrodynamics.
I would appreciate any help to simulate using Lammps a temp. gradient between walls of a channel and also have the wall move with a specified velocity.
Thermostats can subtract out a "bias" before being
applied, like a translational or advection velocity. See
Section 6.16 of the manual for details.
yea I was doing that. It is the first time that I am trying to implement a temperature gradient so guess I might be missing something. In order to check if thermostating was messing things up I removed the temperature gradient and just did a simple Couette flow problem with fixed spring wall moving atoms, but I was not obtaining the linear velocity profile. As for the temp. profile for the temp. gradient case that was coming as expected. What I have a doubt on is can we have net velocity in one direction to atoms that are fixed to their lattice sites using 'spring/self'. If not, what would be the way to implement a constant temp. moving wall.
Appreciate your help.
Find the portion of the input below also have attached the full input script for your ref.
yes, you can assign a velocity to wall atoms. And you can
thermostat them thermally by removing the bias of the velocity.
When you monitor the temp you have to be sure you are printing
out the correct unbiased velocity/temperature.