Bayesian estimation of occurrence probability of cracks in offshore platforms

To ensure safety of offshore platforms, non-destructive inspections are often conducted after the installation of offshore platforms. It is often impractical to inspect 100 percent of tubular joints of offshore platforms due to high costs and possible delays. Accordingly, the presence of cracks may not be fully detected as only a limited number of tubular joints can be inspected. This paper proposes a procedure to predict the occurrence probability of crack in tubular joints of offshore platforms based on ultrasonic inspections supplemented with engineers' experiences. This procedure is developed in the Bayesian framework. To conduct the updating and estimating of the occurrence probability of cracks, formulae for calculating the updated probability distribution and Bayesian estimator of occurrence probability are derived. Then, the occurrence probability of crack is updated based on results from ultrasonic inspections. Results from numerical examples indicate that a reasonable estimation of the occurrence probability of crack can be obtained using the proposed procedure. Both the inspection results from the ultrasonic inspection and the crack size have a significant influence on the estimated occurrence probability of crack. The use of the available information from engineers' experiences or similar projects can lead to a more accurate estimation of the occurrence probability of crack. The quality of offshore platforms will be overestimated if the uncertainty in crack size is not accounted for.