Optimizing manpower-dependent berth allocation problem in automotive terminals

This paper studies a manpower-dependent berth allocation problem arising in automotive terminals. Here, manpower refers to groups of drivers responsible for transporting cars between vessels and storage yards. We extend the classical berth allocation problem by jointly deciding berth allocations and manpower assignments, where vessel loading and unloading are constrained by limited baseline manpower ability, congestion-induced manpower ability loss, vessel-specific manpower capacity, and tidal time windows. To solve this problem, we formulate a mixed-integer linear programming model and propose a Hybrid Adaptive Large Neighborhood Search (HALNS) framework consisting of two phases: initialization using a Two-stage Parallel Genetic Algorithm (TPGA) and an ALNS-based improvement phase. Structurally, this hybrid framework integrates the global exploration capacity of TPGA with the local improvement of ALNS. Computationally, to address the complexity introduced by manpower-related features, we develop two dedicated algorithms, one for each phase, to generate high-quality solutions efficiently. Extensive numerical experiments on realistic instances demonstrate the effectiveness of the proposed methods and validate the contributions of algorithmic operators. To benchmark solution quality, we further develop a column generation algorithm that provides tighter lower bounds than the 0-1 relaxation of the MILP. Finally, sensitivity results show that a 5% increase in baseline manpower ability reduces average turnaround time by 3.81%, and improved car-flow congestion management in the terminal can reduce it by up to 10.53%.