Stochastic modelling of corrosion damage propagation in active sites from field inspection data

A stochastic model for prediction of corrosion damage evolution in active sites, applicable under professional practice conditions is presented here. The damage of a material and its evolution are determined from the damage state at a given time instant and the rate of damage occurrence. To this end, probability density function of the corrosion damage depths of the system is estimated and four models to calculate corrosion damage velocities at localized defects are shown. Their application depends on the amount of inspection reports available. This work takes into account two settings: the first considers that the system has only one inspection report and the second assumes that there are two inspection reports; this latter setting has two variations, the first, when the same defects can be identified at both inspections, and the second, when they are not identifiable. Furthermore, the work introduces a Bayesian model that allows updating corrosion damage velocity on the basis of new measurements found in successive inspection reports. The stochastic model is exemplified by inspection data from a real pipeline system. Its analysis takes into account technical specifications of the system, measured depths of corrosion defects and the number of defects. Additionally, it considers measurement errors during inspection and the variability of corrosion phenomenon under field conditions. Model robustness lies in the fact that corrosion damage estimates are based on measurements reported during inspections. It implicitly considers multiple factors, such as aggressive chemical environment, microstructure composition, operating conditions (temperature, fluid velocity, etc) intervening in the corrosion process, as well as their correlations and variability.