About this Abstract |
| Meeting |
TMS Specialty Congress 2026
|
| Symposium
|
World Congress on Reproducibility, Qualification, and Standards Development of Additive Manufacturing and Beyond (RQSD 2026)
|
| Presentation Title |
A Gaussian Process-Assisted Phase-Field Framework for Physics-Informed Prediction of Grain Structure Evolution in Laser Powder Bed Fusion |
| Author(s) |
Xinxin Yao, James Hanagan, Mohsen Taheri Andani, Veera Sundararaghavan, Raymundo Arroyave, Lei Chen |
| On-Site Speaker (Planned) |
Xinxin Yao |
| Abstract Scope |
Grain size and morphology critically influence the mechanical performance of laser powder bed fusion (LPBF) parts. These features are governed by nucleation within the melt pool, which is sensitive to local thermal conditions and alloy composition, making accurate prediction challenging. We develop an integrated Gaussian Process Regression (GPR)-Phase Field (PF) framework for physics-informed prediction of grain structure evolution. Using Thermo-Calc columnar-to-equiaxed transition thermodynamic analysis, the equiaxed grain fraction under specific temperature gradient and solidification rates is obtained. The GPR model reconstructs a continuous distribution of equiaxed fractions, which is converted into a local nucleation probability through a derived relation. Coupling this nucleation probability into a phase-field model enables quantitative prediction of grain nucleation and growth under varying processing conditions. The framework is validated with experimentally EBSD- measured grain structures. This framework directly links process parameters to nucleation kinetics and grain evolution, providing a data-informed route for microstructure control in LPBF. |
| Proceedings Inclusion? |
Undecided |