Algorithmic Optimization of an Underground Witwatersrand-Type Gold Mine Plan

Abstract

In the mining environment, mine planning is complicated by the presence of unfavorable environmental conditions, limited knowledge of the shape and size of the deposit, ore body characteristics, and volatile market conditions. In this paper, we propose a top-down algorithmic approach to strategically optimize the cutoff grade and net present value (NPV), and implement its solutions at the operation level, while simultaneously mitigating operation risks, to maximize the life of an ultra-deep gold mine from the Witwatersrand Basin, South Africa. To date, the Witwatersrand Basin has contributed about 28% of the world’s total gold supply from a series of Mesoarchaean quartz pebble conglomerate units (referred to as reefs). Through a quantitative analysis using algebraic and stochastic methods, we ranked mining variables in terms of their margin sensitivity and impact/adjustability efficacy. The results of this study showed the following. By using our proposed approach, an underground mine plan can be optimized by focusing on few key variables. Strategic mining of combinations of high-grade panels with low-grade panels and counter-balancing their risk profiles can yield optimal executable mine plan results (i.e., higher NPV, ideal profit margin, and lower risk) without sterilizing a given mineral resource for underground mining operations.