| About this Abstract |
| Meeting |
2024 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2024)
|
| Symposium
|
2024 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2024)
|
| Presentation Title |
GAN-Based High-Throughput Single Track Characterization to Reduce Variability and Experimental Costs in Laser Powder Bed Fusion |
| Author(s) |
Jiahui Ye, John Coleman, Gerald L. Knapp, Amra Peles, Chase Joslin, Alex Plotkowski, Alaa Elwany |
| On-Site Speaker (Planned) |
Jiahui Ye |
| Abstract Scope |
Most process optimization methods in laser powder bed fusion additive manufacturing rely on single-track experiments, an efficient screening step to conduct preliminary exploration of large processing parameter space and guide subsequent manufacturing of coupon, mechanical test samples, and ultimately full-scale parts. However, most approaches overlook variability in single-track experiments due to melt pool instabilities. Replicates are thus needed to account for these variabilities. We propose a new high-throughput single-track analysis method based on Generative Adversarial Network (GAN) for the rapid investigation of heteroscedastic variance in tracks while minimizing labor-intensive efforts. Next, we couple this method with physics-based defect criteria to develop optimal process maps with uncertainty bands. We further show how this approach is readily generalizable to different processing conditions with minimal computational and experimental burden. This research was sponsored by NSF, the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, Advanced Materials and Manufacturing Technology Office. |
| Proceedings Inclusion? |
Definite: Post-meeting proceedings |