Statistical modelling for pitting corrosion of cast iron pipelines

External corrosion is the main reason of the deterioration of cast iron water main in the Australia Water Industry. Prediction of the life time and the maximum pit depth which causes failure can help to reduce the cost of water system maintenance. Due to the complexities of the mechanism of pitting as well as existence of numerous uncertainties, none of the conventional electrochemical approaches is capable of accurate prediction of pipe lifetime. This study, therefore, aims at investigating the maximum pit depth based on the probabilistic theory in order to assist with the prediction of the lifetime of pipelines. A probabilistic model based on the extreme value theory is suggested for this study. Real field data of pitting corrosion of a series of old underground pipelines have been gathered from some sites in Australia. Although Gumbel distribution has been widely employed for extrapolation of maximum pit depth of pipelines, analysis of these real data indicates that the variability of maximum pit depth in longterm exposure fits the Frechet extreme value distribution. Moreover, results of this study show that for cases whose ages have been 30-50 years the maximum pit depth tends to follow one Frechet distribution, while two subsequent Frechet curves have been observed in the probabilistic distributions of data of cases which are older than 50-60 years.