Risk assessment of general FPSO supply system based on hybrid fuzzy fault tree and Bayesian network

One of the major risks to the operation of a FPSO (Floating, Production, Storage and Offloading) is collision of supply vessels. Experiences have shown that collisons of supply vessel take place frequently, although the consequences are limited. The offshore oil&gas industry has established relevant design standards and operational procedures, which are proven to be effective in reducing this risk. As the risk assessment becomes increasingly applied, there is a need to quantify the likelihood of supply vessel collisions especially duing the early design/project stage when historical incident data does not exist. This paper proposees a method for estimating the likelihood of supply vessel collision based on fault tree analysis (FTA), Bayesian network (BN), hybrid theory and fuzzy theory. In liew of using historicval data, this method combines the fuzzy logic with expert heuristics to determine the basic events(BEs). The triangular fuzzy membership functions and a hybrid theory were adopted to define factors that lack data or had no data in the events. To account for the imprecise information of human error, a scheme is devised to use the fuzzy fault probability(FFP) based on the hybrid theory, or a hybrid fuzzy fault tree analysis (HFFT). The algorithm transforms the FT into a BN with a conditional probability table (CPT). This new method was demonstrated in an example of a FPSO, and the results were has been validated through case studies and accident reports. A sensitivity analysis was also conducted to illustrate key parameters. •Applying fuzzy fault tree and Bayesian network methods to assess marine platform risks.•Using a hybrid fuzzy fault tree approach to address the limitations of missing or limited data for risk factors in the risk assessment process.•Using Bayesian network fault diagnosis to adjust root node prior probability and calculate posterior probability.