Assessment of uncertainty in grade-tonnage curves of a multivariate lateritic bauxite deposit through min/max autocorrelation factor transformation

Abstract

Assessing uncertainty in grade-tonnage curves is rather crucial in resource estimation, as these curves provide information on how much ore within the deposit is above/below a given cut-off grade value. Conditional simulation algorithms can provide robust quantification of the ore production by generating multiple realisations of the ore deposits. However, when the ore deposit is characterised by multiple variables that are spatially correlated, these attributes need to be jointly simulated in order to preserve the spatial correlation. The application of co-simulation algorithms to the grade attributes requires large co-kriging matrices to be solved, making the whole process computationally inefficient and impractical. The minimum/maximum autocorrelation factor (MAF) transformation, however, offers a solution to the simulation of the multivariate deposits by transforming the original attributes into non-correlated orthogonal factors which can then be simulated independently, while preserving the spatial correlation among the attributes. This paper discusses the joint simulation of Al2O3, SiO2, Fe2O3, and TiO2 variables of a lateritic bauxite deposit through the MAF transformation and the results of the grade-tonnage curves for each realisation to express the uncertainty.