Digital Engineering-Digital Twin-based Machine Control for Adaptive Additive Manufacturing Processing of Metallic Aerospace Components

Due to the physical complexity and intrinsic randomness in metal additive manufacturing (AM), the process control is intractable, the part quality is not repeatable, and the existing models fail to give consistent results to guide process design. To address this challenge, Global Engineering and Materials, Inc. (GEM) and its team members propose to develop a digital twin (DT) for AM process to enable intelligent, self-adaptive control and guarantee part quality. In this DT framework, physics-based models, data-driven models, and machine learning reduced-order models are combined to maximize predictive capability and computational efficiency. The observation data from in-situ monitoring is assimilated by the DT in real time, and the DT self-evolves to cope with the physical complexity and randomness. Based on the anomaly detected from the real-time data, the DT autonomously provides corrective measures (e.g., dynamically adjusts laser power) to heal defects and tailor microstructure. A probabilistic graphical network is established to quantify and propagate the uncertainty. In virtual experiments, we demonstrate the DT-enabled adaptive closed-loop control when a laser powder bed fusion process is subject to material variation and machine degradation.