Still on the back of my mind I thought possibly or just maybe the Brazil Nut Effect could be modeled and simulated in CFDEM I refer to this paper:
http://guava.physics.uiuc.edu/~nigel/courses/563/Essays_2005/PDF/missel.pdf
Anyone in the group have some insight on this as starting point? Let start the discussion. My motivation was brought about trying to create a "software sieve" for sand particles to arrive at distribution of sizes of fine sands along a study area on sand bar or tourist beach resorts. Back then I used entire an image processing approach and machine vision algorithms on a high definition image capture of sample sand particles. I am also thinking about the sediments in the river banks that somehow self organize with the impinging of water flow in their banks.