Abstract

Background: Rail operations form a critical interface between maritime container terminals and their hinterlands, yet their operational planning remains underexplored; this paper introduces the Train Unloading Planning Problem (TUPP), which jointly determines the container unloading sequence, gantry crane movements, reach stacker routing, and the assignment of containers to compatible export yard bay-locations. Methods: We propose a mixed-integer linear programming (MILP) formulation that captures the tightly coupled interactions among these decisions, strengthened by valid inequalities that reduce CPU time; because exact optimization becomes intractable at operationally realistic sizes, we develop an Iterated Local Search (ILS) heuristic that refines an MILP warm-start using two complementary neighborhood operators, each embedding a greedy reach stacker re-routing procedure, together with a controlled perturbation mechanism. Results: On 16 groups of benchmark instances reflecting Italian terminal operations, the MILP proves optimality only for the smallest instances, with optimality gaps high on medium sized instances. Averaged across all groups, the ILS improves solution quality by 16.62% over the time-limited MILP while requiring about 360 s of total solution time, roughly a ten-fold reduction, with gains reaching 36.56% on the largest instances. Conclusions: The framework provides an effective decision support tool for operational planning in non-automated maritime container terminals.

Keywords