Methodology for Estimating the Corrosion-Induced Degradation Trajectory of 304L Stainless Steel in Nitric Acid

Abstract

Corrosion-induced degradation is a major challenge for the integrity management of industrial equipment operating in aggressive nitric acid environments, particularly in the nuclear industry. Austenitic stainless steel 304L is widely used in such conditions due to its corrosion resistance. However, predicting its long-term degradation remains difficult because corrosion kinetics depend on multiple operating parameters (e.g., temperature, nitric acid concentration, exposure history, and oxidizing species), while available experimental data remain sparse, heterogeneous, and mostly limited to laboratory studies.

This work proposes a methodology for estimating the corrosion-induced degradation trajectory of 304L stainless steel using data extracted exclusively from the scientific literature. First, a structured database of gravimetric corrosion tests was built from published studies and standardized by converting mass losses into equivalent thickness losses under the assumption of uniform corrosion. The collected data were categorized according to exposure conditions, and only renewed nitric acid environments were retained as representative of industrial operating conditions.

Based on these data, a power-law degradation model was identified to describe thickness loss as a function of time. The parameters of this model were then estimated using machine learning approaches based on decision-tree regression, allowing the prediction of degradation parameters as functions of operating conditions such as temperature, nitric acid concentration, and environmental descriptors. Model performance was evaluated using a leave-one-out cross-validation strategy adapted to the limited dataset size.

Finally, the predicted degradation parameters were combined with operating-condition sequences in order to reconstruct cumulative degradation trajectories under variable conditions. The proposed approach provides a reproducible and physics-guided framework for estimating degradation trajectories despite limited data availability and constitutes a promising basis for prognostics and remaining useful life (RUL) assessment of equipment exposed to nitric acid environments.