|About this Abstract
||MS&T22: Materials Science & Technology
||Additive Manufacturing Modeling, Simulation, and Machine Learning: Microstructure, Mechanics, and Process
||Computational Fluid Dynamics Data-driven Heat Source Model for Finite Element Process Simulation in Laser Powder Bed Fusion Additive Manufacturing
||Seth T. Strayer, Florian Dugast, Albert To
|On-Site Speaker (Planned)
||Seth T. Strayer
Thermal field prediction of the laser powder bed fusion (L-PBF) process via the finite element (FE) method can help optimize the process while avoiding the cost of experimental techniques. However, FE models require the abstraction of critical physics into an analytical heat source model, which is not accurate for simulating moderate to high energy melting regimes. This work attempts to mitigate these issues via a data-driven heat source model. In this approach, the thermal fields from a higher-fidelity computational fluid dynamics (CFD) simulation obtained via deep learning are imposed onto the FE solution and entirely replace any analytical heat source model. The resulting thermal fields and melt pool sizes are within 10% error regarding the CFD simulation and experiment, respectively, while the computational expense is significantly reduced compared to the CFD simulations. Hence, this model provides a path for improving the accuracy and potential of thermal FE modeling for L-PBF.