Adaptive Metaheuristics for Multi-Objective Berth Allocation and Scheduling

Abstract

The Berth Allocation and Scheduling Problem (BASP) is a challenging combinatorial optimisation problem arising in container terminal operations, where decisions must balance conflicting goals including operational efficiency, economic cost, environmental impact, and infrastructure utilisation. This paper investigates the multi-objective discrete dynamic BASP with time windows and introduces a four-objective mixed-integer linear programming formulation that simultaneously captures vessel service efficiency, CO₂ emissions, operating cost, and berth workload balance.
To solve this computationally challenging problem, we develop and systematically evaluate adaptive multi-objective metaheuristic approaches. First, we adapt four evolutionary multi-objective algorithms through problem-specific variation operators and an adaptive operator selection mechanism. Second, we extend Adaptive Large Neighbourhood Search (ALNS) to the multi-objective BASP setting and propose two variants: SMOALNS, based on an external Pareto archive, and NSALNS, which integrates ALNS with NSGAII environmental selection.
Extensive computational experiments demonstrate that adaptive operator selection consistently improves evolutionary baselines and that the proposed multi-objective ALNS approaches achieve the best overall performance across benchmark instances. The results indicate that adaptive destroy-repair search mechanisms are particularly effective for highly constrained berth scheduling problems, providing high-quality approximations of the Pareto front for realistic port management scenarios.