Predicting maximum depth of corrosion using extreme value analysis and Bayesian inference

In the present study, overall thickness profile data for the backside corrosion of the bottom floors of the same oil storage tank were collected in 2006 and 2011. A custom-developed program was used to create virtual discrete thickness data from the overall thickness profile data. Using the created discrete thickness data, the scale parameter of Gumbel plots was examined by comparison with the data from 2006 and 2011. Using the experimental findings, a method combining extreme value analysis and Bayesian inference was proposed. With the proposed method, the actual maximum depths of corrosion were accurately predicted using the detected maximum depth of corrosion.