A predict-then-optimize framework for port state control officer routing

Port state control (PSC) is a critical mechanism for ensuring maritime safety, yet the mismatch between limited inspection resources and the vast number of visiting ships challenges the efficient identification of high-risk ships. While predict-then-optimize (PO) frameworks address this, existing studies often rely on oversimplified routing assumptions, such as fixed inspection times and negligible fuel consumption, and lack efficient exact algorithms for large-scale problems. To overcome these limitations, this study proposes a novel inspection routing model under the PO framework that explicitly incorporates cruising speed selection and fuel consumption of launch boats to maximize economic benefits while promoting port decarbonization. To solve this model, an efficient branch-and-price exact algorithm tailored to this problem is designed. Computational experiments on real-world data from the Port of Singapore show that the proposed framework can improve the net economic profit of daily inspection plans by over 62.83% compared to the traditional risk-based approach. Further analysis confirms that enabling flexible cruising speeds is crucial for balancing inspection rewards against fuel costs. This flexibility provides the model with operational resilience against fuel price fluctuations, effectively mitigating cost increases and reducing physical emissions through dynamic speed adjustments. Additionally, the analysis of the selection patterns reveals that the model preferentially targets older and smaller ships, as well as specific ship types like oil tankers and bulk carriers with poor historical records. This study contributes to enhancing the efficiency and reward of PSC inspection while taking port decarbonization into consideration, shedding light on ship inspection management and port sustainable operations.