Bulk carrier accident severity analysis in Australian waters using a data-driven Bayesian network

Maritime accident consequences often entail substantial property losses, environmental contamination, and even loss of life. To investigate the consequences of maritime accidents in Australian waters, this paper develops a data-driven Bayesian network (BN) model with new features derived from a new accident database. The localised vital risk factors influencing bulk carrier safety in Australian waters are generated, which can help develop new measures for consequence mitigation. Compared to the previous relevant research, this article makes new contributions in that 1) manual analysis of each ATSB maritime accident report to formulate a new comprehensive database containing the key influential factors (IFs) influencing bulk carrier accident consequences. Maritime risk analysis and safety management suffer from insufficient accident databases and hence, this development will address the research gap and stimulate data-driven maritime risk analysis in future; 2) the development of a new data-driven BN model to investigate the accident consequences in Australian waters which attracts little attention compared to the other regions of high maritime traffic. The results aid in formulating a new baseline to benchmark Australian maritime accident consequences research; 3) the raw data is trained to configure and quantify the interdependence and dynamics of all the IFs. Given the country's crucial role in international dry bulk trade, this paper contributes to ensuring maritime safety from Australian national and global bulk carrier perspectives. The results reveal that the critical IFs are accident type, emergency handling, navigational condition, ship speed, visibility, safe act, time of the day, loaded or ballast condition, and lookout. Furthermore, the new BN can realise the real-time analysis of a ship's consequence severity in Australian waters and provide valuable insights for transport authorities to mitigate the consequences of accidents. © 2024 The Authors