Abstract: In order to solve the problem of evaluating the risk of accidents from oil and gas release in FPSO system under the condition of scarce data,a new methodology based on HBA was proposed for risk analysis of oil and gas release in FPSO.Firstly,a sequence model was built for FPSO leakage accidents based on the analysis of safety barrier and event tree.Secondly,on the basis of data on accident precursor and Bayesian inference,a new layer of estimation was added to the basic distribution parameters of the safety barrier and probabilities of FPSO oil and gas release accidents were obtained.Finally,the barrier failure probability in real time was updated to realize dynamic quantitative risk analysis for FPSO accidents.The results show that HBA can make full use of scarce data and add information priors from relevant data to obtain the posterior probability density distribution of each parameter,and that with the increase of time,the posterior probabilities tend to be stable and high,and the operation safety should be paid enough attention at the barriers' action which cause disasters and simultaneously prevent release accidents.

Key words: event tree, hierarchical Bayesian analysis (HBA), scarcity data, floating production storage and offloading (FPSO), quantitative risk analysis