Prediction of Aluminum Pitting in Natural Waters via Artificial Neural Network Analysis

One mission of the U.S. Department of Energy's Savannah River Site (SRS) is to store spent nuclear fuel (SNF) and other waste products while permanent storage facilities for such materials are prepared. This extended storage increases the probability of pitting corrosion for Al-based SNF stored in natural (fresh) waters. Such pitting can be slowed, or even eliminated, through careful regulation of water chemistry. Although no universal predictive models linking water chemistry with pitting rate in Al have been developed, empirical models specific to particular alloy and storage environment combinations can be created using previously collected pitting data. Archival pitting data for Al alloy AA1100 (UNS A91100) was used with the back propagation of error method to train and test a feed-forward artificial neural network model. The trained model was found capable of predicting expected pit depth as a function of water pH; concentrations of carbonate $(CO3−2)$⁠, copper (Cu⁺²), and chloride (Cl^-); and storage time.