ABSTRACT
Only in the last decades the International Maritime Organisation (IMO) has formally recognised the importance of adopting risk assessment procedures in its decision process. It defined Formal Safety Assessment (FSA) as a structured and systematic methodology aimed at enhancing maritime safety, including protection of life, health, the maritime environment and property by using risk and cost-benefit assessments (IMO, 1997). The integration of techniques, both established and novel, to assess risks is a current goal within many maritime organizations. In fact the application of probabilistic methods in order to model some of these high risks has been a current practice due to their potential to help in the process of decision making which would allow to propose regulatory changes (Guedes Soares and Teixeira, 2001). Recently IMO investigated the use and effectiveness of BBN modelling as part of the FSA, in relation to the quantification of Risk Control Options (IMO, 2006). In this document, an evaluation of the effect of ECDIS (Electronic Chart Display and Information System), ENC (Electronic Navigational Charts) and Track control was performed. Nevertheless the number of published works with the application of BBN modelling in the maritime field is still very scarce when compared with other industrial fields. These applications can, almost, be summarized in the works performed by Or and Kahraman (2002), Pedrali et al
(2004), Leva et al. (2006), Eleye-Datubo et al. (2006), Trucco et al. (2006) and Norrington et al. (2008). However, most of these studies are based on elicitation through experts’ judgments and with low number of states per node. Mainly binary states or canonical Noise-OR or Noise-AND are typically used for both reduction in the computational demand and aid on the elicitation process. These procedures, although, most of the times necessary due to the lack of consistent and reliable data, have the drawback of not allowing an accurate estimate of the inferences. This, obviously, can affect the process of decision making. Therefore, the present study aims at applying real datawithout any experts’ judgements elicitation process, which therefore reduces the uncertainty typically associated with this type of studies.
