Avoid moding mesh using force fields

Hello everyone!
I'm trying to simulate a ball mill and I'm using at the moment a very simple geometry to obtain acceptable computation times since I'm learning now LIGGGHTS. While looking at the documentation page, I read about the possibility to assign variables instead of exact values to both the magnitude and the direction of hte gravity vector and I was thus wondering if there's a way to avoid heavy computation by using this trick instead of moving he mesh.

As a first attempt, I first inserted the particle into my geometry using a classical gravity vector, using:

fix grav all 9.81 vector 0.0 0.0 -1.0

until all the particle settled; after that, this gravity was unfixed and substituted with the following lines:

variable ang_vel equal 8.5 #it was originally the angular velocity of the geometry
variable actualstep equal step
variable theta equal ${ang_vel}*(step-${actualstep})*${tstep} #direction of the gravity vector

#please ignore the step-actualstep, it was just written to adjust the timestep since the simulation was divided in 2 parts using a restart;

variable gx equal -sin(v_theta)
variable gz equal -cos(v_theta)

fix gravi all gravity 9.81 vector v_gx 0.0 v_gz

Obviusly, what I obtained was completely different from the real simulation in which the mill is rotating: the particles "slide" following the internal surface of the geometry and they never fall down.

I thought that what I've tried is conceptually incorrect since from a physical point of view, the force that keeps the particles attached to the surface is the centrifugal force, the gravity force instead, makes them fall down. I think I'm able to do it using a centrifugal force field, but still, there would be no realistic relative movement between periferical particles and mesh and thus the model would be incorrect

Any idea about where I could find more information about this topic?
Thank you in advance for your support