Proceedings of the

The 33rd European Safety and Reliability Conference (ESREL 2023)
3 – 8 September 2023, Southampton, UK

A Bayesian Inference and Metaheuristics Model for Estimating Maritime Accidents: The Case of Fernando de Noronha

Paulo Gabriel Siqueiraa, Thais Campos Lucasb, Victor Hugo Resende Limac, Maria Luisa Barreto de Goisd, Isis Didier Linse, Marcio Chagas Mouraf and Heitor Duarteg

Center for Risk Analysis, Reliability Engineering and Environmental Modeling (CEERMA), Industrial Engineering Department, Universidade Federal de Pernambuco, Brazil.


Toxic spills that arise from maritime accidents can lead to catastrophic environmental damage to animals. The numerous oil tankers that travel the planet raise the risk of potential oil spills that can affect sensitive ecosystems such as oceanic islands. To evaluate those risks, frequency estimation is an essential step. However, dealing with events with low frequency and high consequence poses a challenge, since classical statistical approaches strongly relies on data, which are scarce in this case. To overcome this shortcoming, a Bayesian population variability-based method is proposed to assess the accident rates considering accident data from various databases combined with the expertise of professionals such as academics, captains, pilots, and chief officers. As a real case application, we used this framework to estimate the frequency of accidents near Fernando de Noronha Archipelago. The results can support decision-making regarding measures to prevent accidents or reduce risks.

Keywords: Bayesian analysis, Maritime accidents, Frequency assessment, Quantitative risk assessment, Oil spills.

Download PDF