Local recoverable reserves prediction with block LU simulation

Abstract

Prediction of local recoverable reserves is an important problem in ore reserve evaluation. Wide spacing of exploration data leads to unavoidable uncertainty at unsampled locations. Deterministic methods such as kriging cannot explicitly account for this lack of perfect information and equipment selectivity. We present a methodology and software for assessing recoverable reserves at Selective Mining Unit (SMU) resolution. Reserves and their uncertainty are calculated by performing matrix (LU) simulation at a fine resolution and then scaling these simulated models to SMU size. Multiple variables are simulated using a Linear Model of Coregionalization. Measures of uncertainty in SMU reserves are reported.