A novel approach of random packing generation of complex-shaped 3D particles with controllable sizes and shapes

Abstract

This paper presents a novel computational-geometry-based approach to generating random packing of complex-shaped 3D particles with quantitatively controlled sizes and shapes for discrete modeling of granular materials. The proposed method consists of the following five essential steps: (1) partitioning of the packing domain into a prescribed number of random polyhedrons with desired sizes and form-scale shapes using the constrained Voronoi tessellation; (2) extraction of key points from the edges and facets of each polyhedron; (3) construction of a freeform curve network in each polyhedron based on Bézier curve fitting; (4) generation of solid particles with smooth, convex surfaces using the biharmonic-based surface interpolation of the constructed network; and (5) creation of concavity by superimposing spherical harmonic-based random noise. To ensure that the obtained shape descriptors (e.g., the elongation, flatness, roundness and convexity ratio) match the hypothesized values, an inverse Monte Carlo algorithm is employed to iteratively fine-tune the control parameters during particle generation. The ability of the proposed approach to generate granular particles with the desired geometric properties and packing is demonstrated through several examples. This study paves a viable pathway for realistic modeling of granular media pertaining to various engineering and industrial processes.