Better management of production incidents in mining using multistage stochastic optimization

Among the many sources of uncertainty in mining are production incidents: these can be strikes, environmental issues, accidents, or any kind of event that disrupts production. In this work, we present a strategic mine-planning model that takes into account these types of incidents, as well as random prices. When confronted by production difficulties, mines which have contracts to supply customers have a range of flexibility options including buying on the spot market, or taking material from a stockpile if they have one. Earlier work on this subject was limited in that the optimization could only be carried out for a few stages (up to 5 years)and in that it only analyzed the risk-neutral case. By using decomposition schemes, we are now able to solve large-scale versions of the model efficiently, with a horizon of up to 15 years. We consider decision trees with up to 6¹⁵ scenarios and implement risk aversion using Conditional Value-at-Risk, thereby detecting its effect on the optimal policy. The results provide a “roadmap” for mine management as to optimal decisions, taking future possibilities into account. We present extensive numerical results using the new sddp.jl library, written in the Julia language, and discuss policy implications of our findings.