The paper presents an approach to estimating the effects of safety culture on maritime accident risk expressed as the number of ship collisions in a sea area. The problem is approached probabilistically with a Bayesian belief network technique that combines qualitative and quantitative modeling. According to the preliminary results of the model, having poor safety culture onboard would increase the number of collisions by approximately 20 % compared with ships with an excellent safety culture. Although the model needs improvement and validation, it shows potential to serve as an aid for decision makers within maritime risk management.

Maritime safety; safety culture; Bayesian networks; accident causation