About this Abstract |
| Meeting |
2027 TMS Annual Meeting & Exhibition
|
| Symposium
|
AI-Enabled Materials Processing: Integrating Accelerated Experimental Workflows and Processing-Aware Machine Learning
|
| Presentation Title |
From process observation to parameter selection in Directed Energy Deposition: An Expertise-Guided, Bead-to-Build Methodology |
| Author(s) |
Christos Papaioannou, Michail S. Koutsokeras, Evangelos Georgousis, Harry Bikas, Konstantinos Salonitis, Panagiotis Stavropoulos |
| On-Site Speaker (Planned) |
Panagiotis Stavropoulos |
| Abstract Scope |
Directed Energy Deposition (DED) is increasingly used for complex, high-value parts. However, selecting process parameters often relies on trial-and-error experimental methods. This work presents a new approach for choosing process parameters following a bead-to-build hierarchy. Machine Learning is used to connect process parameters to the geometry of the deposited bead. It also captures how parameters interact and supports parameter selection based on a target geometry. This method incorporates process knowledge in defining workable parameter ranges and the design of experiments structure. This way, fewer deposition trials can produce useful, learnable data. The method is tested on DED-Arc, with a modeling structure designed to generalize across materials and conditions. It can connect to in-process monitoring, which feeds new data back into the models. Operators can also add new information over time, such as additional experiments, materials, or operating conditions. As a result, predictions improve continuously with ongoing use of the system. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Additive Manufacturing, Machine Learning, Other |