Spatiotemporal monitoring and control for foam additive manufacturing processes of thermoplastics

Abstract

Foam Additive Manufacturing (Foam-AM) offers a novel approach to fabricating architected structures with tunable density by leveraging in-situ foaming. However, precise control of the foaming process remains challenging due to the complex interplay of multiple process parameters. This complexity has limited the broader adoption of Foam-AM in applications such as packaging, protective gear, automotive components, and beyond. To address these challenges, this study proposes a novel spatial-temporal monitoring and control method using a multi-sensor platform to optimize the performance and geometry of Foam-AM. The platform integrates a thermal camera and positioning encoders to monitor the temperature field and speed distribution, identifying bead interference and speed variations as the primary causes of foaming defects. Additionally, a line laser profiler is used to measure sample’s spatial information, complemented by extruder encoder data for in-situ density estimation. This in-situ approach facilitates high-throughput data acquisition, forming the basis for a process-performance model developed using Invertible Neural Networks (INN). Leveraging the INN model, two major advancements are achieved: (1) off-line control strategies effectively minimize bead interference, ensuring consistent density and geometric precision; and (2) localized foaming defects, especially in regions with rapid speed changes, are accurately identified and addressed through real-time parameter adjustments, significantly improving overall print quality.