Development of in-situ optical sensor-based monitoring methodologies for Additive Manufacturing processes

Process monitoring has proven to be essential in manufacturing fields, where quality and precision are paramount. The introduction of AM techniques has enabled the production of complex parts but has also increased the complexity of manufacturing systems. This has led to the demand for monitoring systems able to provide significant information about the process while it is running as well as give real-time insights to detect and prevent, when possible, process anomalies. Reliable monitoring systems provide a window into the process dynamics, allowing for process parameters adjustment that could maintain product quality and reduce waste of materials and energy. Despite the numerous advantages introduced by monitoring systems, widespread traditional solutions remain limited, prompting significant effort from researchers and industries in this direction. This thesis addresses the current critical gaps in terms of monitoring purposes by proposing innovative and cost-effective in-situ optical-based monitoring methodologies for three key AM processes: Material Extrusion Additive Manufacturing, Laser Powder Bed Fusion and Direct Energy Deposition. Material Extrusion showed to be lacking adequate metrics able to describe the process, together with the absence of solutions exploiting profilometers as monitoring sensors. The monitoring activities on L-PBF were driven by the lack of robust monitoring systems able to characterize lattice structures. Finally, the monitoring activities on L-DED were led by the current absence of real-time monitoring system capable of monitoring more than one direction, since the majority of the system focused on single direction thin wall monitoring. First, two layerwise monitoring approaches for MEX, based on a high-resolution blue laser line profilometer embedded within a consumer grade MEX printer, were presented to characterize the surface quality, to detect defect occurrence layer-by-layer and to expand the metrology field related to the MEX process, which is currently lacking, by proposing new metrics. The three proposed quality indexes (ADLH, RAD and slope s) proven to be representative of the layer height accuracy, the occurrence and distribution of surface defects, such as over/under-fill, and, also, the process stability respectively. In the second approach, functional analysis tools were successfully used to detect, localize, and characterize the topology of common surface defects caused by over-extrusion and under-extrusion conditions. An optical-based monitoring procedure was also developed and applied for layerwise in-situ monitoring of complex geometries produced by L-PBF through a tailor-made image processing algorithm. Based on High-Resolution Optical Tomography (HR-OT), this procedure effectively detected geometric distortions and, at the end of the process, provided a 3D reconstruction of the lightweight structure suitable for post-process quality assessment. Finally, a preliminary study on a cost-effective off-axis dual-camera real-time 3D monitoring method for L-DED process was conducted. The aim was to enable a reliable measure of the melt pool height regardless of the laser head scanning direction. Results showed that the proposed methodology was able to provide acceptable melt pool height measures for all the scanning direction tests. By introducing robust process signatures and defect metrics, this work significantly advances the metrology of AM processes. The comprehensive monitoring methodologies developed during the research activities not only improve in-process quality evaluation and process stability but also pave the way for real-time, closed-loop corrective actions.