Predictive Modeling of Corrosion Evolution in Automotive Multifunctional Coatings Using Electrochemical Testing and Artificial Neural Networks

Abstract The natural and induced harsh conditions such as temperature, humidity, rain, and de-icing salt, influence the lifetime performance of different parts of a vehicle. The coating/metallic substrate interface experiences various degradation stages during exposure to corrosion conditions. The corrosion performance of three different coating systems (Physical barrier (polymer-based), sacrificial (Nickel zinc), and the combination of both (hybrid)) used for automotive applications was studied. In this work, we used electrochemical and surface experimental data with mechanistic analysis to develop a mathematical framework. The understanding of the degradation mechanisms under steady-state conditions served to train a mathematical model that was later used as a predictive tool. The electrochemical impedance spectroscopy (EIS) technique was utilized to characterize the coating and substrate performance during steady-state, and to train an Artificial Neural Network (ANN) as an arrangement of Time Series Prediction (TSP). After a few days of testing, the ANN successfully predicted the coating performance over multiple years, where experimental results validated these estimates. Moving forward, accelerated methods and field data will be utilized to further test and validate the mathematical framework developed under laboratory conditions, allowing this approach as an effective predictive tool.