| About this Abstract |
| Meeting |
MS&T24: Materials Science & Technology
|
| Symposium
|
Additive Manufacturing Modeling, Simulation, and Machine Learning: Microstructure, Mechanics, and Process
|
| Presentation Title |
Machine Learning Surrogate Model of Spatter Transport in a Laser Powder Bed Fusion Machine |
| Author(s) |
Nicholas Obrien, Satbir Singh, Jack Beuth |
| On-Site Speaker (Planned) |
Nicholas Obrien |
| Abstract Scope |
Laser Powder Bed Fusion (LPBF) still suffers from random defects due to spatter particles, which prevent the process from being adopted for fatigue-life-critical applications. The inert gas flow in LPBF machines is not capable of removing all spatter particles, especially ones larger than the virgin powder. In the future, machine users will need to choose process parameters and place parts to avoid spatter contamination. To facilitate this process, we have developed a quick and reliable surrogate model of spatter transport in the EOS M290. The model is trained on computational fluid dynamics (CFD) simulations of spatter trajectories from our prior work, which was shown to agree with real-world flow and spatter measurements. This talk discusses the model selection process, the validation of the final model, and how this type of model empowers machine users to tune their build parameters to reduce spatter contamination and rogue defects. |