| About this Abstract |
| Meeting |
2024 TMS Annual Meeting & Exhibition
|
| Symposium
|
Additive Manufacturing: Process-induced Microstructures and Defects
|
| Presentation Title |
Batched Additive Manufacturing Meets Parallel Bayesian Optimization – Highway for Materials Design |
| Author(s) |
Jonathan W. Pegues, Anh Tran, Hannah Sims, John Emery |
| On-Site Speaker (Planned) |
Jonathan W. Pegues |
| Abstract Scope |
The advantages of additive manufacturing (AM) technologies have been well established; however, adoption has been slow due to relative uncertainty in material reliability. To accelerate AM adoption, new agile approaches to simulate advanced processing strategies and associated material responses are necessary. In this talk, we employ Bayesian optimization to optimize high-dimensional problems and efficiently navigate design of experiments for AM. Results are discussed in the context of linking key process variables (laser power and velocity) to material structure (defect population and grain size) and then to mechanical performance (strength and ductility). Experiments are accelerated using a high-throughput materials characterization coupon to accommodate x-ray computed tomography, electron microscopy, and hardness testing of several distinct metallurgical samples. Paired with high-throughput torsion testing, process-structure-property characterization can be accelerated by >10X compared to conventional characterization techniques.
Sandia National Laboratories is managed and operated by NTESS under DOE NNSA contract DE-NA0003525 |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Additive Manufacturing, Modeling and Simulation, Machine Learning |