A genetic algorithm for temporal and spatial alignment of long- and medium-term mine production scheduling for open-pit mines

Open-pit mine production scheduling essentially comprises long-term (LT), medium-term (MT) and short-term (ST) schedules, which have typically been optimized in isolation to each other. However, by independently optimizing these schedules, temporal and spatial scheduling misalignment between consecutive schedules occurs and can lead to lower mining project net present values (NPVs). Therefore, it is important to integrate these schedules for improved scheduling alignment. Accordingly, this paper developed a mixed integer programming (MIP) model that integrates LT and MT production schedules to improve scheduling alignment between LT and MT schedules compared to separately optimizing the schedules. The model was solved using a genetic algorithm (GA), which is a stochastic algorithm. The combined MIP model and GA approach was tested on a Geovia Surpac® block model and generated a 2.20 % higher NPV than for the isolated LT schedule. Using the same input parameters on MineLib, the approach was validated by comparing its results to the best-known feasible linear programming (LP) relaxation solutions obtained using a TopoSort heuristic algorithm. For the Newman, Zuck Small and KD block models, the approach generated comparable NPVs, which were 4.90 % lower, 10.90 % higher, and 2.36 % lower, respectively. However, for the four block models, the approach achieved 100 % temporal alignment between LT and MT production schedules, while the isolated schedules had temporal misalignment ranging between 86.22 % and 105.47 %. Therefore, this paper's contribution is on incorporating temporal and spatial alignment between LT and MT production schedules to achieve LT objectives at the MT horizon.