A particle packing method for non-uniform sizes and prescribed filling ratio

This work presents a particle packing method based on a distance minimization procedure with non-uniform sizes and prescribed filling ratio. In the study of engineering problems involving granular systems, like those that employ numerical tools such as the discrete element method, an important modeling task is to produce the initial geometric configuration for a given set of particles. Several methods of particle packing are currently employed for this purpose. In this work, the employed methodology uses the Levenberg–Marquardt minimization strategy and does not require a mesh to generate the initial geometric configuration for the particle set. Also, the technique aims a good approximation for the input parameters of granular media, such as grain size distribution and filling ratio. The proposed methodology is divided into four macro-steps: (a) definition of the initial particle assembly, (b) distance minimization procedure, (c) particle reallocation and d) particle removal. The input data related to the size distribution and filling ratio are used to generate the initial particle assembly, and the following steps are used to eliminate the overlapping between the circles. The packing algorithm presents good computational performance and converges to the input parameters. Granular models generated from real soil data are presented for validation of the proposed strategy.