Experimental and computational analysis of random cylinder packings with applications

Abstract

Random cylinder packings are prevalent in chemical engineering applications and they can serve as prototype models of fibrous materials and/or other particulate materials. In this research, comprehensive studies on cylinder packings were carried out by computer simulations and by experiments. The computational studies made use of a collective rearrangement algorithm (based on a Monte Carlo technique) to generate different packing structures. 3D random packing limits were explored, and the packing structures were quantified by their positional ordering, orientational ordering, and the particle-particle contacts. Furthermore, the void space in the packings was expressed as a pore network, which retains topological and geometrical information. The significance of this approach is that any irregular continuous porous space can be approximated as a mathematically tractable pore network, thus allowing for efficient microscale flow simulation. Single-phase flow simulations were conducted, and the results were validated by calculating permeabilities. In the experimental part of the research, a series of densification experiments were conducted on equilateral cylinders. X-ray microtomography was used to image the cylinder packs, and the particle-scale packings were reconstructed from the digital data. This numerical approach makes it possible to study detailed packing structure, packing density, the onset of ordering, and wall effects. Orthogonal ordering and layered structures were found to exist at least two characteristic diameters from the wall in cylinder packings. Important applications for cylinder packings include multiphase flow in catalytic beds, heat transfer, bulk storage and transportation, and manufacturing of fibrous composites.