Computer Simulation Of Particulate Systems

Computer models for simulating the construction and for calculating the properties of particulate solids in two- and three-dimensional systems (coverings of circles and packings of spheres, respectively) were developed. By simulating the physical process of constructing circle coverings and sphere packings, in which the circles or spheres were randomly selected from many different distributions of sizes, physically realizable loose random assemblies were simulated whose physical and geometrical properties were predictable with the computer. Many of these properties, such as the number of contacts, are extremely-difficult to measure accurately, or considerable ingenuity is required in their determination. All of these properties were easily and accurately calculated by the computer models for many different coverings and packings constructed from equal, discrete, uniform, and log-normal distributions of particles. In addition to the loose random assemblies, dense random covering and packing constructions were simulated whose properties are ideal upper limits for those of the loose random coverings and packings. The properties of the dense random coverings and packings, heretofore only obtainable by theoretical analysis, were also simply obtained. The statistical geometry of loose random coverings and packings was studied, and several new interpretations of the area-covering or space-filling requirements in these systems were made. In two dimensions, it was found that the properties of all loose random coverings are predictable by taking suitably weighted averages of the corresponding properties of just four kinds of polygons whose edges are defined by connecting the centers of mutually tangent circles in the covering. The distributions of these polygons that permit area-covering were determined, and empirical equations for predicting the density, average coordination number, and average pore area in loose and dense random coverings were formulated. The properties of loose packings were determined by performing a simplicial division of the computer-simulated assemblies to produce a unique set of minimum volume, space-filling tetrahedra whose edges connected the sphere centers in each packing. A simple empirical equation was formulated that permits the calculation of the average number of contacts in any sphere packing, random or regular. Some of the statistical geometrical properties of the simplicial tetrahedra in the computer-simulated packings were analyzed. The tetrahedron edge length distributions in all the loose random packings were nearly identical when the edges were presented in terms of sphere diameters, regardless of the sphere population in the packing. Moreover, the gap lengths in the loose packings were to an excellent approximation representable by a uniform distribution of sizes. The distributions of spheres in tetrahedra resulting from the simplicial division of the loose random packings also existed in the tetrahedra in dense random packings. These observations were employed in developing a unique method for predicting the geometrical properties of loose random packings without simulating the assembled packing. Instead, the space-filling simplicial graph could be predicted for any given density of packing of any sphere population. The cumulative packing properties of these tetrahedra were shown to be nearly identical to those calculated by all loose random packing simulations attempted in this investigation. The simplicial tetrahedra completely describe a packing and can therefore be used to predict all properties of interest.