How to simulate a poiseuille flow without adding the solid wall

LAMMPS LAMMPS Mailing List Mirror
1

Dear lammps users，

Now I’m trying to simulate a poiseuille flow by “fix addforce” command. I also use the “fix wall/reflect” command and does not add an actual solid wall at the boundary.
But the results show that the velocity distribution is uniform, not parabolic.

So What’s wrong with my treatment?
Can I simulate a poiseuille flow without adding the actual solid wall? This means the system contains only pure fluid atoms.

I’m looking forward to having your help.

Best wishes
Ruo-Yu Dong

2

You are asking several questions that have little to do with LAMMPS and more to do with classical mechanics. Nevertheless, I will answer the more obvious.

Dear lammps users，

Now I’m trying to simulate a poiseuille flow by “fix addforce” command. I also use the “fix wall/reflect” command and does not add an actual solid wall at the boundary.
But the results show that the velocity distribution is uniform, not parabolic.

A simple classical mechanics argument should tell you that you cannot do this. You are forcing particles in a direction orthogonal to a reflecting wall. Momentum of the particles in the forcing direction cannot be changed by a collision with a reflecting wall perpendicular to it(write out some vector equations an this should be obvious).

So What’s wrong with my treatment?

See above.

Can I simulate a poiseuille flow without adding the actual solid wall? This means the system contains only pure fluid atoms.

In short, no. At least not without some boundary condition to alleviate the problem above, and any conservative potential serving as your wall will give you the same problem as above.

I’m looking forward to having your help.

3

Dear lammps users，

Now I'm trying to simulate a poiseuille flow by "fix addforce" command. I
also use the "fix wall/reflect" command and does not add an actual solid
wall at the boundary.
But the results show that the velocity distribution is uniform, not
parabolic.

why should it be parabolic? think about what it is *exactly* that
slows atoms down that are close to the wall.

So What's wrong with my treatment?

that you don't consider the physics of the model.

Can I simulate a poiseuille flow without adding the actual solid wall? This
means the system contains only pure fluid atoms.

you can, but it is not a trivial as you currently think it is.

axel.