Risk assessment of water pollution from port ship accidents: an improved Bayesian network model

Abstract With the rapid development of the marine shipping industry and green ports, the risk of water pollution caused by vessel traffic accidents has attracted widespread concern. Port water pollution risk assessment can provide guidance for green port construction and the optimal allocation of emergency resources. This study develops a structured risk assessment framework that integrates a Bayesian network (BN), the Analytic Hierarchy Process (AHP), and a fuzzy risk matrix. The BN is used to estimate accident probability under uncertainty, the AHP is used to quantify multidimensional accident consequences, and the fuzzy risk matrix incorporates uncertainty and decision-makers’ risk preferences into the final risk evaluation. Furthermore, the improved fuzzy risk matrix addresses uncertainty in port water pollution risk assessment. A case study was conducted at Zhanjiang Port to validate the capability of the model. The results indicate that the probability of a ship accident is approximately 0.0528 and that the accident consequence index is 3.304. The comprehensive risk indices are 3.27 (Easy), 3.72 (Standard), and 3.99 (Hard) for the different risk matrices. These results fall within the Tolerable Unacceptable or Unacceptable risk levels. The comprehensive risk indices are obtained using three fuzzy risk matrices with different risk preferences and costs. The proposed framework enhances the transparency and practical applicability of port water pollution risk assessment and provides structured decision support for marine environmental management.