Underground mine stope boundary optimisation using a hybrid network flow-particle swarm algorithm

This study introduces a new hybrid algorithm designed to optimise stope boundaries in sublevel-stoping mines. Stopes were delineated based on the raises’ locations and relevant constraints, including minimum and maximum stope dimensions, stope overlaps, and geotechnical considerations. Particle swarm optimisation was used to determine the optimal locations of raises, while a network-flow algorithm applied constraints and generated stopes surrounding each raise, defining their size and shape. The hybrid algorithm was implemented on a copper deposit in Chile at a depth of 1,000 meters below the surface. A comparative analysis against the network-flow and maximum-value neighborhood algorithms showed that the economic value of stopes produced by the hybrid algorithm exceeds that of the network-flow algorithm by about 4% and exceeds that of the maximum-value neighborhood algorithm by about 7.5%. Moreover, the proposed algorithm was able to produce stable stope layouts for large-scale problems within a reasonable solution time. These findings are promising, particularly given the growing prevalence of high-tonnage, low-grade deposits in the mining industry.