How do I model Couette Flow NVE system with temperature control by walls

Dear all

I am looking a way to simulate Couette Flow. My concern is to study viscous heating effects. Thus, i need to model an NVE system without any thermostat applied on fluid molecules. Im using lammps 2Nov10.

I have a current 3D NVE model with Argon confined in Silver walls of 4 layers each. Out most wall layers are only defined as a “region” (static), in order to keep all molecules to stay inside defined simulation domain. Rest of the molecules are dynamic. Fluid-fluid and wall-fluid interactions are LJ, while the wall-wall interactions are EAM. Im keeping wall molecules at constant temperature by “fix Langevin”.

What i need is so simple. I just need to assign each wall molecules same velocity and keep them at constant temperature while interacting with both each other and fluid molecules. I know that i need to “fix_modify” my Langevin with a temperature calculation by “temp/com”.

Thus my wall needs to move all together (like a slab) at a constant velocity and at the same time, independent to its center of mass velocity in one direction, it should be dynamic interacting with both each other and fluid molecules in all directions.

I already tried to use “velocity” command but it did not work. There appears shear effects between wall molecules that each layer moves at different velocity and create undesired viscous heating.

i will be glad if someone can help me.

Best Regards

Murat Barisik
PhD Student
Old Dominion University

You can use fix aveforce to allow a group of atoms (e.g. the wall)
to respond to force as a single entity, i.e. they will move together.

But if you keep them rigid (as one body), I don't see how you
can thermostat them, or use them to thermostat the fluid.
I suggest you thermostat the fluid separately. There are many
ways to do that, even if they are flowing. See Section 4.16 of
the manual for an overview.

Steve