Representation of uncertainty in maximum depth of marine corrosion pits

Pitting can be a critical form of corrosion for mild and high strength low alloy structural steels subject to saline environments. For pipelines, tanks and nuclear and other containers perforation is the main design criterion. The present paper outlines a recently developed model for the progression of 'maximum' observed pit depths, based on new field trials using conventional coupon testing. The variability of the maximum pit depth and its changing behaviour with time of exposure is an important aspect for structural reliability studies. The most common approach uses the Gumbel distribution. However, recent data for mild steel reveals inconsistencies. These are described and have been noted also for pitting in some other metals. It is proposed that proper account has not been taken of sample inhomogeneity and, more importantly, that extreme pits are unlikely to be independent events, even asymptotically, as required by extreme value theory. A much simpler approach based on structural systems reliability theory is proposed. This has important practical implications for extrapolation to larger areas of steel plate or for long-term exposures, such as for nuclear containers.