Long-term scheduling in underground coal mining using mixed binary programming

Abstract

Recognizing the complexity of coal mining management, e.g., the scarcity of financial resources and the high level of uncertainty, a mixed binary programming model has been developed as an aid for generating production schedules which maximize the associated net present value. Defining the mine layout as a precedence network, with the nodes representing mining blocks, a solution procedure is developed, based on Benders' partitioning scheme. That is, the procedure iterates between two problems, namely, the master (primal) problem, solved by a combination of heuristic and exact methods, and the subproblem (dual problem), solved partly by inspection and partly as a minimal cost network flow problem. The heuristic methods are based on improvements of existing algorithms for scheduling precedence-related jobs on m processors. Computational experiences are presented and the procedure is demonstrated on a mining case.