Adhesion dynamics under time-varying deposition: A study on robotic assisted extrusion

Abstract

Recent advances in robotic assisted-additive manufacturing (RA-AM) have enabled rapid material extrusion-based processing with comprehensive data collection. The following study investigates the adhesion dynamics of the initial printed layer across parameters such as surface energies, stand-off heights, and extrusion speeds of up to 100 mm/s, using an applied in-situ thermal analysis technique. Observations indicate that the characteristic length parameter, $L_c$ < 0.05 mm, is adequate in anchoring the thermal melt, which adheres to the substrate when the nozzle proximity to the surface increases. Up to 100% molten area is contacting the surface prior to translation, and a final eccentricity over 0.85 has been observed. Through an analysis of variance, operational parameters of lower nozzle heights, printing speeds, and higher surface energy were statistically significant. The resultant in-situ characterization-driven data, was used to train a convolutional neural network (CNN). The model tested at an accuracy of 90.9%, and was able to distinguish between failed prints and initially adhered structures.