A Matheuristic Approach for a Special Case of Train Unloading Planning Problem

This paper addresses the Train Unloading Planning Problem with fixed zones, a structured variant of the Train Unloading Planning Problem (TUPP) introduced in [4], arising in non-automated maritime container terminals. The distinguishing assumption is that each reach stacker is assigned to a designated zone of the export yard. This fixed-zone constraint removes the reach stacker allocation degree of freedom from the optimization. A matheuristic approach is proposed to solve real size instances of this problem; it is derived from the Iterated Local Search proposed in [4]. The derived approach extends and modifies the Iterated Local Search by introducing four complementary seed strategies and a new Greedy Bay-location Re-routing subroutine (GBR) that, embedded within every neighbourhood evaluation, ensures each candidate sequence is assessed under bay-location-optimal conditions. Computational experiments show that no single seed dominates across all instance sizes, confirming the value of the multi-start design. Averaged across instances with up to 70 containers, the matheuristic achieves a mean improvement of 25.5% over time-limited MILP solutions with computation times below 65 s.