Multi-scale stochastic modelling of ore textures at the George Fisher mine, Queensland, Australia

"The link between ore texture and mineral liberation has previously been established at a microscopic scale. To construct ore texture models that can be used in mining applications requires this relationship to be retained at a significantly larger scale. This is initially dealt with by finding a basic set of microtextures that form the elements of the ore textures at the next larger mesoscopic scale, simulating these mesotextures at a fine scale, then changing the support of the mesotexture simulations to generate the expected mining block/stope mesotexture distributions. Thus, a mutually exclusive and exhaustive set of spatially locatable mesotextures can be used to fully characterize the metallurgical properties of in situ material.A method that generates expected distributions of mesotextures for any mining block size or shape is proposed, and includes: (1) characterizing the spatial correlation of the set of mesotextures with a single covariance function; (2) generating multiple simulations of mesotextures at a fine scale; and (3) using the simulations at a fine scale to generate the expected mining block/stope mesotexture distributions. A case study, using the George Fisher sediment-hosted base metal deposit in Queensland Queensland, Australia, is developed to demonstrate the proposed methodology as well as show the inadequacy of off-the-shelf approaches.Mining-scale texture models can be used to improve the in situ discrimination of ore from waste by exploiting the joint spatial variability of grades and textures to quantify the expected metal recovery of in situ material, then calculate local marginal cut-off grades. An artificial case study is presented to demonstrate the changes in ore/waste selection when considering ore texture models compared to constant metal recoveries."