| About this Abstract |
| Meeting |
MS&T24: Materials Science & Technology
|
| Symposium
|
Computation Assisted Materials Development for Improved Corrosion Resistance
|
| Presentation Title |
Predicting Oxidation Behavior of Ni-Based Superalloys with Physics-Informed Machine Learning |
| Author(s) |
William Trehern, Aditya Sundar, Leebyn Chong, Richard Oleksak, Madison Wenzlick, Kyle Rozman, Martin Detrois, Paul Jablonski, Michael Gao |
| On-Site Speaker (Planned) |
William Trehern |
| Abstract Scope |
Oxidation continues to be a challenge for high-temperature in energy applications. Currently, there is no method to effectively approximate a materials susceptibility to oxidation at high temperatures making designing new oxidation-resistant alloys challenging. In this work, a large, high-quality database of 50,000+ data entries for alloy oxidation behavior has been created and used in a physics-informed machine learning workflow. Manual literature data collection was performed and compiled with internal, high-fidelity experimental data. A custom physics-informed descriptor library was created to generate physically meaningful features by transforming the composition, processing, and test parameters. Using the oxidation rate (kp) and mass change as the target parameters, we identify key features that impact oxidation in Ni-based superalloys. Multiple regression models are evaluated and the best is selected for use in a multi-objective optimization schema to design new Ni-based superalloys with superior high-temperature oxidation performance. |