Investigation of the severity of maritime accidents considering the interaction between human factors and operating conditions: A case study on collision accidents in China

Maritime accidents are characterized by low frequency but high severity, with significant consequences often influenced by the interplay between human factors (HFs) and operating conditions (OCs). This study examines the individual and combined effects of HFs and OCs on the severity levels of maritime accidents using an improved Human Factors Analysis and Classification System (HFACS) framework, association rule mining (ARM), and a data-driven Bayesian Network (BN) model. Initially, a novel database is constructed by categorizing HFs into four hierarchical levels under the HFACS framework and OCs into four categories: vessel characteristics, navigational conditions, weather conditions, and temporal factors. Subsequently, ARM and the Tree-Augmented Naïve Bayes (TAN) algorithm are utilized to identify causal relationships among HFs and their associations with OCs, which served as the foundation for constructing the BN model. Finally, forward and backward reasoning, along with sensitivity analysis, are applied to explore the individual and joint contributions of HFs and OCs to different accident severity levels. The case results indicate that unsafe acts, particularly improper collision avoidance operations and steering errors, are the most critical HFs across all severity levels of collision accidents. Their combined effects with adverse OCs, such as adverse weather and navigation conditions, high traffic density, and large vessel size, significantly exacerbate accident severity. These findings are expected to guide policymakers and maritime stakeholders in implementing targeted interventions to mitigate accident severity and enhance maritime transportation safety.