For mineral processing, an ore can be defined as the assemblage of useful mineral grains and gangue mineral grains. The first step of ore mineral processing involves crushing the material in order to achieve required mineral liberation. When it is achieved, in the ideal case, resulting fragments are formed: 1) only by useful minerals or 2) only by gangue minerals. This work consists to develop a numerical tool to determine the liberation mesh size of a given ore. With this aim, 1) a sample of the considered ore is modelled by a binary 3D image segmentation. Then, 2) a grinding process is modelled by a digital approximation of a Voronoi tessellation of the image, with random seeds. Each Voronoi cell represents a fragment of the ground ore sample. Fragments must meet several conditions so that they reproduce the characteristics of the crushing products in the mineral industry (e.g. the particle size distribution must follow a Rosin-Rammler law and the shape of the fragments must be close to spheres). This modelling allows obtaining the characteristics of the liberation mesh size. This is an important parameter to evaluate the feasibility of a mining project because the grinding operations are very expensive. This study would: 1) to evaluate and predict grinding costs from the bulk ore texture and 2) to provide an assessment tool downstream of the means of current observations and conventional analysis.