For automotive exhaust valve applications, future vehicles will need affordable, durable materials capable of operating at higher temperatures with predictable response to severe oxidizing environments. Both commercial and model Ni-based alloys were tested in 1-h cycles at 800-950°C in wet air, and the oxide scales formed on wrought Ni-(14-25)wt%Cr binary alloys were characterized to gain a better understanding of the behavior of chromia-forming alloys under these conditions. The mass change curves were used to quantify the behavior of the tested alloys and fit growth and spallation rates using the kp-p model. We systematically analyzed the correlation between elemental alloy compositions and the manually fitted kp and p values to select high-ranking features to be included in a machine learning analysis. The machine learning models for the rate, kp, could be trained with a surprisingly high accuracy even with limited data, while only modest fitting was obtained for p, the spallation parameter. A preliminary theoretical framework that can predict kp and p of hypothetical alloys was established, however, improving the accuracy of surrogate models is needed to assist in alloy development for this transportation application.
Key words: Ni alloy, chromia scales, cyclic oxidation, exhaust valve, mass change modeling, correlation analysis, machine learning