Data-driven design and application of corrosion-resistant low-alloy steels
Corrosion processes are affected by a variety of materials and environmental factors and can hardly be effectively interpreted and predicted using conventional ‘trial-and-error’ and experience-based approaches. Recently, the concept of ‘corrosion big data’ was proposed as a general framework to elucidate complex corrosion problems and to accelerate the design of corrosion-resistant materials, by integrating i) high-through corrosion data accumulation, ii) data standardization and organization, iii) data mining and modeling and iv) data sharing and applications. This talk will present a demonstration of this data-driven approach in the design and application of corrosion-resistant low-alloy steels. The corrosion processes of low-alloy steels with varying contents of Cr, Mo and Sn elements, as well as different microstructures and grain sizes are scrutinized based on corrosion-monitoring data obtained in field atmospheric environments. Furthermore, machining learning models are constructed to process the data for better understanding of the roles of different materials and environmental factors and for predicting atmospheric corrosion kinetics.