About this Abstract |
| Meeting |
MS&T25: Materials Science & Technology
|
| Symposium
|
Additive Manufacturing Modeling, Simulation, and Machine Learning: Microstructure, Mechanics, and Process
|
| Presentation Title |
Multi-Fidelity Framework to Predict the Melt Pool Characteristics for Laser Powder Bed Fusion of Inconel 718 |
| Author(s) |
Abdul Khalad |
| On-Site Speaker (Planned) |
Abdul Khalad |
| Abstract Scope |
Melt pool characteristics in laser powder bed fusion (L-PBF) are crucial for building defect-free parts. Prediction of melt pool characteristics is quite challenging due to the complex interactions between laser parameters, powder properties, and process conditions. In the present work, the low-fidelity model generates the data points using the Eager-Tsai model with little effort but needs calibration of the heat source model. The high-fidelity model contains single-track, single-layer experiments of Inconel 718 on L-PBF. This paper presents a multi-fidelity framework by combining low-fidelity and high-fidelity models using Gaussian Process Regression (GPR) to predict melt pool characteristics. By incorporating data from both models, the predictions become more reliable, resulting in better part quality and fewer defects. Hyperparameter tuning of GPR yielded the lowest mean absolute error compared to traditional regression models. This current work focuses on the L-PBF of Inconel 718, but it can be applied to any other AM process. |