[lammps-users] Simple spinning cylinder


I'm trying to model particles in a vortex flow and need them to spin.
I've tried just about everything I can think of to no avail.
I have a cylindrical 'tank' of granular particles and I'd like to get
them all spinning as if in an airflow vortex.

Ideas that failed:

wall/gran zcylinder shear - spins the outside few layers, but never
spins up the entire group. Multiple cylinders inside the particles
can spin up most of them, but they stay in ringed zones.

fix addforce - divided the tank into a grid and pointed each one in
the right direction to spin. They keep going in that direction, but
since it works by groups instead of regions, each group of atoms keeps
going in that same direction, ultimately sticking to the walls.

set vx vy - works by region instead of group, but only sets initial
values. The particles get pushed in the right direction, but friction
quickly steals away all angular momentum and they stop.

fix rigid - make some kind of elaborate rigid mixer assembly. It's
not what I need anyway, but rigid gives me a bus error and I don't
know why.

So, unless there's a way to make the wall/gran zcylinder porous,
define fix addforce by region (seemingly best option), or make set vx
vy reassert itself regularly, I am completely out of ideas. Please
tell me there's some command for cylindrical coordinates that is going
to solve all my problems...


You could write a modified fix addforce that worked by region.
Wouldn't be hard to do.

I know fix wall/gran works to rotate a densely packed granular
system. Assuming you rotate slowly enough that there is not
large slip at the boundary. Is the issue that you are not
densely packed and the walls don't effectively interact
deep into the material?

What is your model for the flow force that should be acting
on a particle as a function of its position? It might be simplest
to just write a fix that imposes that force.