Battery-swapping-considered scheduling for automated guided vehicles in container terminals: a two-phase decision framework

This study addresses an integrated scheduling problem for battery-powered automated guided vehicles (AGVs) at automated container terminals, optimizing AGV scheduling for efficient container transport and simultaneous battery swapping. Considering practical requirements—such as battery capacity, station capacity, battery availability—and scalable problem sizes (combinations of stations, AGVs, and tasks), the inherent complexity of the problem necessitates the development of a two-phase decision framework. The first phase generates feasible routes, which account for multiple transport tasks and associated empty travels, while constrained by battery capacity. The second phase schedules AGVs to serve these routes, considering station capacity and battery availability. The proposed framework integrates a multi-label setting algorithm (for the first phase) with an enhanced logic-based Benders decomposition (LBBD) method (for the second phase). The LBBD method’s enhancements include an improved lower bound, strengthening cut, and an efficient subproblem algorithm. Numerical experiments, benchmarked against MILP and a genetic algorithm, demonstrate the superior performance of the framework: it reduces average computational time by over 60% for small-scale instances and improves optimality gaps by 23% for large-scale instances when compared to time-limited MILP. Furthermore, it significantly outperforms the genetic algorithm with an 18.79% lower makespan and 26% faster computation for large-scale instances under 300-second time limits.