Using production data to improve grade estimation in underground mining

Abstract

All available data should be used to build a geostatistical model. In underground mining, data have different support volumes: drillhole data are defined at a quasi-point support, while production data represent tonnes of ore mined during a period of time (stopes). Due to the support difference, production data are frequently ignored to update the block grade model. We propose a block kriging approach to combine these two sources of information (point and volumetric support data). A synthetic underground mining case is presented. Two estimation scenarios are evaluated: the first considers only drillhole data, while the second considers both drillhole and production data. Results show that the use of production data improves grade estimation. The improvement is more pronounced where diamond drillholes are sparsely located.