Network-flow based algorithms for scheduling production in multi-processor open-pit mines accounting for metal uncertainty

The open-pit mine production scheduling problem (MPSP) deals with the optimization of the net present value of a mining asset and has received significant attention in recent years. Several solution methods have been proposed for its deterministic version. However, little is reported in the literature about its stochastic version, where metal uncertainty is accounted for. Moreover, most methods focus on the mining sequence and do not consider the flow of the material once mined. In this paper, a new MPSP formulation accounting for metal uncertainty and considering multiple destinations for the mined material, including stockpiles, is introduced. In addition, four different heuristics for the problem are compared; namely, a tabu search heuristic incorporating a diversification strategy (TS), a variable neighborhood descent heuristic (VND), a very large neighborhood search heuristic based on network flow techniques (NF), and a diversified local search (DLS) that combines VND and NF. The first two heuristics are extensions of existing methods recently proposed in the literature, while the last two are novel approaches. Numerical tests indicate that the proposed solution methods are effective, able to solve in a few minutes up to a few hours instances that standard commercial solvers fail to solve. They also indicate that NF and DLS are in general more efficient and more robust than TS and VND.