About this Abstract |
Meeting |
2020 TMS Annual Meeting & Exhibition
|
Symposium
|
Materials Design Approaches and Experiences V
|
Presentation Title |
Accelerated Design of High-temperature Alloys with Data Analytics and Supercomputing |
Author(s) |
Jian Peng, Andrew Williams, Sangkeun Lee, Yukinori Yamamoto, J. Allen Haynes, Dongwon Shin |
On-Site Speaker (Planned) |
Jian Peng |
Abstract Scope |
We present a modern data analytics framework that has the potential to accelerate the design of high-temperature alloys significantly. Our starting point is a series of machine learning models that can predict creep properties of alumina-forming austenitic (AFA) alloys, trained with a decade of ORNL’s consistent experimental data augmented with key thermodynamic features (e.g., volume fraction and degree of supersaturation) calculated from a state-of-the-art computational thermodynamic database. We use a design of experiments (DOE) approach to populate tens of thousands of hypothetical AFA alloys as inputs for machine learning models. The thermodynamic features of each virtual alloy have been computed with ORNL’s world-class supercomputer to generate a substantial volume of synthetic data in a high-throughput manner. We have identified a number of promising AFA alloys with improved creep, for future experimental validation. This research was sponsored by the Department of Energy, Vehicle Technologies Office, Propulsion Materials Program. |
Proceedings Inclusion? |
Planned: Supplemental Proceedings volume |