Session: RIP: Predictive Modeling and Characterization of Corrosion Processes in Complex Environments (In Honor of Professor Digby Macdonald) (Part III of IV)
Artificial Intelligence-Enabled Pipeline Corrosion Prediction and Uncertainty Analysis (RIP2026-00002)
This study demonstrates the use of artifical intelligence (AI) for pipeline corrosion prediction and uncertainty analysis. The proposed Bayesian Neural Network (BNN) model demonstrates superior prediction accuracy as compared to different empirical and data-driven models, while capturing uncertainty information. Shapley Additive Explanation (SHAP) analysis is used to compare the importance of input factors. Key factors influencing corrosion include chloride content, pH, pipe-to-soil potential, and pipeline age. Specifically, conditions such as a pipe-to-soil potential exceeding -0.85 V, a pipeline age over 20 years, and pH values below 7 are found to significantly accelerate external corrosion depth. The model also presents the dynamic changes in prediction uncertainty over time, especially under higher chloride content, lower pH values, and higher pipe-to-soil potentials. This knowledge of corrosion uncertainty is crucial for quantifying probability of failure for pipeline and making informed decisions on pipeline inspection and maintenance strategies.
