About this Abstract |
Meeting |
2024 TMS Annual Meeting & Exhibition
|
Symposium
|
Additive Manufacturing: Advanced Characterization with Synchrotron, Neutron, and In Situ Laboratory-scale Techniques III
|
Presentation Title |
F-27: Machine Learning-aided Laser Absorptance Prediction and Keyhole Feature Simulation in Laser Powder Bed Fusion |
Author(s) |
Jiahui Zhang, Yu Zou |
On-Site Speaker (Planned) |
Jiahui Zhang |
Abstract Scope |
Numerical simulation is an intrinsic, robust, and cheap approach to understanding the complex physical mechanism in the laser powder bed fusion printing process. However, the existing heat source model falls short in accurately replicating real-world conditions due to the unknown laser absorption rate. Consequently, the simulation model fails to provide quantitative predictions for keyhole features. In this study, we utilized experimental laser absorbance values and corresponding X-ray images to develop and validate a precise thermal-mechanical-fluid coupled model using the finite volume method. This model allows us to effectively simulate the keyhole features in Ti64 welding, Ti64 LPBF, and Al welding processes. Furthermore, we present a novel self-iterative framework that combines our established simulation model with a well-trained machine learning model. This framework enables automatic iteration of the laser absorptivity for various power-scan speed combinations until the simulated keyhole features closely match the X-ray images. |
Proceedings Inclusion? |
Planned: |
Keywords |
Additive Manufacturing, Computational Materials Science & Engineering, Machine Learning |