This article presents a novel algorithm for solving a short-term open-pit production-scheduling problem in which several objectives, of varying priority, characterize the quality of each solution. A popular approach employs receding horizon control, dividing the horizon into N period-aggregates of increasing size (number of periods or span). An N-period mixed integer program (MIP) is solved for each period in the original horizon to incrementally construct a production schedule one period at a time. This article presents a new algorithm that, in contrast, decomposes the horizon into N period-aggregates of equal size. Given a schedule for these N periods, obtained by solving an N-period MIP, the first of these aggregates is itself decomposed into an N-period scheduling problem with guidance provided on what regions of the mine should be extracted. The performance of this hierarchical decomposition-based approach is compared with that of receding horizon control on a suite of data sets generated from an operating mine producing millions of tons of ore annually. As the number of objectives being optimized increases, the hierarchical decomposition-based algorithm outperforms receding horizon control, in a majority of instances.
- hierarchical decomposition
- mixed-integer programming
- Short-term open-pit production scheduling