The separation equilibrium doesn't depend on the size, only on the azimuthal speed. If you want to separate a suspended powder from the liquid, as in your other discussion, the probably the equilibrium is trivial and needs only a ridiculous azimuthal speed.
So the size and speed depend only on the desired kinetics. You need to evaluate how quickly the powder particles drift under the centrifugal field, as a consequence of the density difference, particle size and fluid viscosity. It's probably a laminar regime, for which algebraic solutions exist, if assimilating the particles to spheres.
I don't expect one solution that results from the design needs. You have to engineer something sensible and efficient, and many possibilities exist, with big variations. More azimuthal speed separates the particles faster, but above Mach 0.7 aerodynamic losses increase more steeply. Probably not necessary for a suspension. You also may want to decide early if you want liquid-tight joints where one end rotates and the other doesn't (difficult and it restricts the possible speeds) or if you can disconnect all hoses before starting the rotation, which is easier with human intervention.
One non-trivial estimate is how imbalanced the rotor will be, especially if a powder is to settle at the wall. This decides much the rotation speed, the bearings, the wall resistance.
So, what is the input mix, especially densities and grain size? Throughput?