A geometrical texture model for multi-mineral liberation prediction

We present a novel methodology for numerically reconstructing a three-dimensional (3D) texture for simulating multi-mineral rock and estimating the particle properties and liberation characteristics of its breakage products. A stochastic geometrical model has been developed to describe the intact structure of ores at any scale, potentially containing several mineral types, each with unique grade and grain size distribution parameters. Additional controls over the intact texture include the ability to vary the proportion of grain inclusions within grains of a different mineral phase and by varying the degree of spatial clustering of grains within the texture. Repeated random 2D and 3D sectioning of the 3D texture to simulate particle production and to calculate their compositions is used to generate outputs comparable to reports from mineral liberation analysis systems; including density distributions, liberation distributions, and particle class distributions. These outputs enable texture parameters to be fitted to corresponding measured liberation data and is demonstrated for a four-mineral sulphide ore, yielding a realistic distribution of particle properties, essential for accurately modelling particle separations.