Risk management of ship independent navigation in ice-covered Arctic waters: an early warning model

Journal article, 2026

With the retreat of Arctic sea ice, independent ship navigation in ice-covered waters is increasing, especially in areas with lower ice concentration and thinner ice, reducing reliance on icebreaker escort. However, this trend raises risks of ship-ice (or iceberg) collisions and ship besetting in ice. This study proposes a risk early warning model for such scenarios by integrating multi-state fault tree analysis (MS-FTA) and Bayesian network (BN). First, risk influencing factors (RIFs) related to environmental, technical, human, and organizational aspects are identified from maritime accident reports and set as model nodes. Then, MS-FTA is used to construct model structures for the two accident scenarios, with if-then rules formulated as early warning conditions. These rules are embedded into the BN framework to compute accident probabilities and identify critical RIFs. Based on the results, targeted risk control options are recommended. The proposed model provides a systematic tool for assessing and managing the risk of ship independent navigation in Arctic and other ice-covered waters.