Quantitative Risk Analysis of an LNG terminal using two-step Bayesian analysis

The QRA of relatively recent industries such as the Liquefied Natural Gas (LNG) industry presents a difficulty in the frequency analysis step, because of the very sparse or even inexistent failure data. The use of Bayesian Analysis can overcome this difficulty-it allows failure frequency calculation from more statistically reliable data from similar and older industries and updating it using specific information of the plant. It is important to consider the non-homogeneity of the data obtained from the industries databases and experts' opinions. The Two-Steps Bayesian Analysis considers this non-homogeneity, performing a non-homogeneous calculation in the first step, with the use of generic data, and updating the result with specific data from the plant through homogeneous calculation in the second step. This paper proposes the use of the Two-Steps Bayesian Analysis in the calculation of failure frequency for the LNG industry and applies it to an offshore LNG terminal.