Modeling the Evolution of Corrosion: A feature-based interacting particle model for growth prediction

The Government Accounting Office reports that the Department of Defense spends approximately $20 Billion per year on prevention and repair of corrosion. The resources required to combat corrosion problems are desperately needed for equipment and personnel in today's high operations tempo environment. Despite its significance, modeling for the spatial-temporal prediction of corrosion evolution has been largely ignored in the literature. The purpose of this research is to develop a feature-based interacting particle model for corrosion evolution incorporating features available in time sequenced images of growth. A posterior probability of growth function is incorporated into a generic interacting particle model definition to influence growth direction based on the feature values at each location. Images of filiform growth on samples of AA2024-T3 reveal heterogeneities such as intermetallic particles and surface holes which are used as features in supervised learning methods to define the posterior probability of growth. This information is used by a simple interacting particle model to simulate the growth of corrosion. The accuracy of the predicted growth is compared to results from a random model without the feature information. Results for multiple filaments from different samples validate the use of the proposed model and suggest the inclusion of the heterogeneity features in the model to improve the predictive ability.