Scheduling trucks in a time-space network of open-pit mines

This study investigates a truck scheduling problem in open-pit mines, where trucks transport raw coal and rock from electric shovels to unloading stations. The raw coal is used to produce commercial coal for sale, which requires a consistent calorific value between the mined raw coal and the blended commercial coal. Truck congestion significantly impacts work efficiency, so proper scheduling is necessary to prevent congestion and improve efficiency. We model the problem as a mixed-integer linear programming model using a time-space network to minimize the total operation time of all trucks. We design a column generation-based algorithm to solve the model, integrating state-reduction-based dynamic programming and machine learning to enhance efficiency. Effective inequalities are also incorporated to accelerate the solution process and improve computational performance. Experimental results show that, for small-scale instances, the proposed algorithm reduces solving time by 24% compared to CPLEX while maintaining solution quality. For large-scale instances, CPLEX fails to find an optimal solution within 1800 seconds, but the proposed algorithm consistently produces better solutions in a shorter time. Sensitivity analyses based on real data from an open-pit mine show that the variable speed mode improves truck transportation efficiency and reduces congestion compared to the constant speed mode. Our results also suggest that mine operators should carefully choose truck speed modes, and truck size combinations, as well as the distribution of unloading stations and shovels. Utilizing speed modes with more choices, higher maximum speeds, and suitable gradients, along with incorporating larger trucks into fleets, can reduce the total operation time of all trucks.