About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Additive Manufacturing: Materials Design and Alloy Development VII – Design With Multi-Modal and Field Data by Integrating Uncertainty
|
| Presentation Title |
Multi-Terminal Compositionally Graded Alloy Design for High-Throughput Materials Exploration |
| Author(s) |
James Hanagan, Sk Md Ahnaf Akif Alvi, Mrinalini Mulukutla, Raymundo Arróyave |
| On-Site Speaker (Planned) |
James Hanagan |
| Abstract Scope |
Compositionally graded alloys (CGAs) present a unique opportunity for fabricating and characterizing libraries of compositions for exploring large alloy design spaces, allowing for rapid evaluation of microstructure, processability, and indentation-based mechanical properties as a function of position and composition. This idea is demonstrated in this presentation with a 4-terminal CGA designed leveraging a graph-based representation of the CoCrFeNi space while considering material constraints from CALculation of PHAse Diagrams (CALPHAD). The CGA is represented in chemical space by a tree spanning a region of compositions characterized by high uncertainty in Vickers hardness predictions from a multi-task deep Gaussian process regressor. The regressor is trained on a sparse dataset of hardness values from additively manufactured samples and augmented with physics-based yield strength model predictions. The compositions from the CGA are then mapped to a test sample geometry, the results from which can enable further enhancement of model predictions through iterative learning. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Additive Manufacturing, Computational Materials Science & Engineering, Machine Learning |