| About this Abstract |
| Meeting |
MS&T24: Materials Science & Technology
|
| Symposium
|
Understanding High Entropy Materials via Data Science and Computational Approaches
|
| Presentation Title |
Machine Learning Design of Additively Manufacturable Tungsten-Based Refractory Multi Principle Element Alloys with Enhanced Strength at Extreme Temperatures |
| Author(s) |
Zhiyang An, Bo Ni, Benjamin Glaser, Amaranth Karra, Bryan Webler, S. Mohadeseh Taheri-Mousavi |
| On-Site Speaker (Planned) |
Zhiyang An |
| Abstract Scope |
Designing additively manufacturable tungsten-based refractory multi principle element alloys (R-MPEA) with high strength at temperatures beyond 2000°C is extremely challenging. These alloys will be used in our next-generation structural components in reactors or jet engines. Here, we explore the high-dimensional compositional and processing space using hybrid machine learning (ML)/CALPHAD-based ICME techniques. We combine our simulations with experimental data and train various ML models to guide the design and validate in experiments. We show that while linear models provide first-order valuable insights, they have limitations in capturing nonlinear interactions. We will discuss how much our nonlinear models can improve the predictions of strength at room temperature and 2000°C. We will compare the experimental results with our predictions and discuss how the models can be modified for the next round of enhanced iterations. Our design framework and concepts can be used to discover the high strengths of various HEAs at extreme temperatures. |