Statistical Characterization of Pitting Corrosion—Part 2: Probabilistic Modeling for Maximum Pit Depth

It is proposed that for immersion corrosion conditions extreme pit depths are those resulting from immediate pit initiation and pit growth without significant metastable activity. This is termed “super-stable” pit growth. As a result of this and the very similar material and environmental influences, there is likely to be a high degree of dependence among the depths of extreme (deepest) pits. Also, the statistical population describing such pits is likely to be different from that of the remaining pits. These observations call into question the conventional use of extreme value distributions (such as the Gumbel distribution) for modeling the uncertainty in maximum pit depth, since such distributions are based on the assumption of independent (or asymptotically) statistical events. In contrast, empirical observations suggest that extreme pit depths appear to be representable by a normal distribution. This provides the basis for a review of probability theory that may be used for dealing with systems of highly dependent events. It is then applied to the probabilistic modeling of maximum pit depths. The implications for extrapolation of maximum pit depth are illustrated. Significantly lower probabilities of the occurrence of extreme depth pits are predicted. This may have significant practical implications.