The Discrete Element Method (DEM), developed in the 70’s, has become more a attractive with the increasing computer processing capacity and the development of high-performance computational techniques. This scenario induced growing interest of many researchers as a tool to study engineering problems. A very appealling field of study is the modeling of phenomena associated with granular materials, including the compaction of granular packages, such as metal powders in the steel industry, sand production and proppant flowback in the petroleum industry, which is motivation to this work. The application of the method in its first step requires the generation of the particles’ initial configuration or the filling of domains with them. Some studies have focused on the development of algorithms to generate dense packing of particles. This work presents an algorithm to generate random dense packing of particles that correspond to a given granulometric curve and are consistent with other parameters set for the granular package. In the present work special attention is given to generation of two-dimensional packings of disks to model the phenomenon of fractures filling in wells stimulated by hydraulic fracturing. The proppant flowbak generated by the oil pumping is a leading cause of production decline in fractured wells, besides causing damage to the production equipment. The proposed algorithm was implemented in Java language and the optimization of packings was performed according to the porosity using genetic algorithms. Applications of the algorithm to some packings of particles and the problem of filling of fractures are presented.