Determination of a safety criterion via risk assessment of marine accidents based on a Markov model with five states and MCMC simulation and on three risk factors

Abstract

Risk modeling of marine accidents involves many technical and environmental safety factors which cannot be predicted in advance. The aim of this paper is to determine a safety criterion via risk assessment of marine accidents for a certain sea area. The risk is assessed with a combination of two approaches: one is based on a Markov model and MCMC simulation, and the other on three risk factors. On the one hand, the state space of the Markov model consists of five states considering major accident types and its transition rates are obtained by MCMC simulation. On the other hand, accident occurrence probabilities are modeled based on the accident frequencies according to three risk factors—sea state, traffic density and ship's length, respectively. The risk value is estimated as a weighted average of results from the two approaches. The safety criterion for navigation is determined based on a set of risk values obtained from the historical accident data for the sea area. The proposed method is illustrated with a numerical example, performing a sensitivity analysis and comparing the safety criterion with risk values for the accidents. The comparing results are in good agreement.