About this Abstract |
Meeting |
2025 TMS Annual Meeting & Exhibition
|
Symposium
|
Additive Manufacturing Modeling, Simulation and Machine Learning
|
Presentation Title |
Using part-scale in-situ defect formation monitoring to assist post-build qualification and predict fatigue performance in LPBF |
Author(s) |
Ziheng Wu, Justin Patridge, Gabe Guss, Nicholas Calta |
On-Site Speaker (Planned) |
Ziheng Wu |
Abstract Scope |
Porosity remains to be a major obstacle for the wider adoption of laser powder bed fusion. The evaluation of the porosity population in the AM components mostly relies on the time-consuming post-build characterization that is often limited by tool availability. The photodiode signal generated during printing carries information of pore formation. Many state-of-art in-situ monitoring efforts focus on single tracks which may not represent the actual fabrication conditions due to the over-simplified thermal history. Here, we account for interlayer interaction and utilize machine learning approaches to connect the pore formation events with the photodiode signals on the part-scale level where the subsequent and adjacent remelting can interact with the existing pores. Each pore is assigned a criticality index based on its size, shape, location, etc. The criticality index is later validated by fatigue testing results from the corresponding AM printed coupons. |
Proceedings Inclusion? |
Planned: |
Keywords |
Additive Manufacturing, Machine Learning, Characterization |