New heuristics for the stochastic tactical railway maintenance problem

Efficient methods have been proposed in the literature for the management of a set of railway maintenance operations. However, these methods consider maintenance operations as deterministic and known a priori. In the stochastic tactical railway maintenance problem (STRMP), maintenance operations are not known in advance. In fact, since future track conditions can only be predicted, maintenance operations become stochastic. STRMP is based on a rolling horizon. For each month of the rolling horizon, an adaptive plan must be addressed. Each adaptive plan becomes deterministic, since it consists of a particular subproblem of the whole STRMP. Nevertheless, an exact resolution of each plan along the rolling horizon would be too time-consuming. Therefore, a heuristic approach that can provide efficient solutions within a reasonable computational time is required. Although STRMP has already been introduced in the literature, little work has been done in terms of solution methods and computational results. The main contributions of this paper include new methodology developments, a linear model for the deterministic subproblem, three efficient heuristics for the fast and effective resolution of each deterministic subproblem, and extensive computational results.