Localized Corrosion in Seawater: A Bayesian Network-Based Review

Abstract

This paper reviews the localized corrosion of passive Ni-Fe-Cr-Mo-N alloys immersed in seawater using a Bayesian network (BN) method. Making alloy performance decisions using data from the literature on seawater is challenging because a large body of data is generated using various methods in various natural conditions. There is a significant scatter in the data and cross-comparison of data from different techniques is difficult. The BN approach serves to integrate diverse sources of knowledge and data in this area and evaluate the data in a probabilistic manner. The paper shows that the predicted probability of localized corrosion agrees reasonably well with field data. The challenges and opportunities to improve the BN model are discussed.