Anisotropy in Simulated Random Packing of Equal Spheres

WE have used an IBM 1130 computer to simulate the very slow settling of rigid equal spheres from a dilute slurry into a randomly packed bed. Settling took place in a box defined by { x,y,z ∣0≤ x ≤20, 0≤ y ≤20, z ≥0}, where the dimensions are given in sphere radii. Initial x and y coordinates of the spheres were chosen by pseudorandom numbers. Once an incoming sphere made its first contact, the process was completely deterministic: the sphere rolled and sometimes fell until it reached a stable position. Thus some spheres rolled all the way from a peak to a valley. Gross wall and end effects were avoided by varying, by means of pseudo-random numbers, the position of walls and floor for each sphere. The final packed bed consisted of 1,561 spheres, but measurements were made on interior spheres only.