Multi-Scale Model Predictive Control for Laser Powder Bed Fusion Additive Manufacturing

Abstract

Abstract Additive manufacturing (AM) process stability is critical for ensuring part quality. Model Predictive Control (MPC) has been widely recognized as a robust technology for controlling manufacturing processes across various industries. Despite its widespread use, there has been limited exploration into the application of real-time MPC for controlling the laser powder bed fusion (LPBF) AM process through in-situ process monitoring. This paper presents a new framework designed to develop MPC strategies for real-time LPBF control. The framework accommodates various multi-scale approaches, including pointwise, trackwise, layerwise, and partwise methods. This framework considers the diverse needs for material state representation when formulating predictive models, constraints, and objective functions while allowing for predictive control implementation at different scales and frequencies. The utility of this framework is demonstrated through three trackwise MPC case studies, all employing high-speed co-axial melt pool imaging. Simulation results indicate that LPBF systems enhanced with MPC achieve superior performance compared to those governed by open-loop control systems. Additionally, we find that MPC implementations that utilize feedback control at finer scales provide improved process stability, albeit at the expense of increased computational demands. This framework serves as a guide for industrial practitioners, outlining how the implementation of MPC in AM process control can be optimized based on available in-situ sensing capabilities and data acquisition techniques.