About this Abstract |
Meeting |
2021 TMS Annual Meeting & Exhibition
|
Symposium
|
Materials for High Temperature Applications: Next Generation Superalloys and Beyond
|
Presentation Title |
Rapid Screening, Machine Learning, and Multi-objective Optimization for Refractory Alloy Development |
Author(s) |
Andrew Detor, Meinolf Sellmann, Scott Oppenheimer, Emily Cheng, James Ruud |
On-Site Speaker (Planned) |
Andrew Detor |
Abstract Scope |
Alloy development suffers from the curse of dimensionality, particularly in high entropy or complex concentrated systems. In a field that is already regarded as slow and expensive, a new approach is needed to efficiently develop custom alloys for demanding applications. In this work we review a 15 month effort to experimentally screen 400+ new refractory alloys for performance across a range of metrics including strength, ductility, oxidation resistance, high temperature phase stability, and cost. Active machine learning is used to supplement a more traditional metallurgist intuition-driven approach. Multi-objective optimization routines are also employed to direct experiments and develop alloys with the best balance of properties for the intended use. The methods, tools, and approach detailed in this talk demonstrate the practical and accessible benefits machine learning and multi-objective optimization bring to today’s material development challenges. |
Proceedings Inclusion? |
Planned: |