Modeling and Sensing for Control of 3D Printing: Examining the Roles of Buckling and Heat Transfer in Fused Filament Fabrication (FFF)

Abstract

Fused filament fabrication (FFF) offers a low-cost additive manufacturing method for prototyping and small-scale production of thermoplastic polymer, polymer composite, biological, ceramic, and even metal structures with internal geometries infeasible before the advent of additive manufacturing. However, inferior quality, production speed, and repeatability of parts produced via FFF is a hindrance to its wider adoption in mass production. This work aims to understand the extrusion dynamics that produce these extrusion errors which limit quality, speed, and repeatability. This work is organized into five research itemstasks tasks to improve the modeling and control of extrusion dynamics. The first work, task wasincluded was to collect the modifyingication of an existing FFF 3D printer to study theidentify time constant trends inof the extrusion response on a modified FFF 3D printergiven various step inputs in extrusion conditionsvelocity. The experimental results from the experiments show the extrusion system as a nonlinear dynamical system, with a varying a parameter-varying “time constant” parameter . This result explains the failure of using linear model control techniques onf the extrusion system when the across a range of extrusion setpoints are changedspeeds. Therefore, a deeper study of the extrusion response and associated physics was required. The This study was addressed in the second workresearch task,, which involved building a highly instrumented FFF system using metrology- grade stages and sensors, placed at the extruder, including a melt pressure sensor, a loadcell to measure extrusion force, a thermocouple with direct polymer melt contact, an extrusion servo drive with a current feedback signal, a thermal camera, and a laser line profilometer. The sensing platform was used to collect large experimental datasets to build complete process maps of the extrusion process at different temperatures and layer heights. The first application of the experimental data was shown in tThe third workresearch task, which involved using experimental data collected on the system built in the second work to creatinge a data driven model of extrusion dynamics using a NARX-Net approach. The trained NARX-Net model provided decent performanceproved capable of predicting the extrusion response with a realistic sensor delay. However, to probe the sources cause of the dynamics and understand the causes of the of the now-identified “time constant” variability, an investigation of the underlying dynamics within the extrusion system was necessary. To that end, tThe fourth workresearch task, involves involved conducting experiments to explore the dynamics of testing compression of the filament in response to an extrusion event and the impact of compression on its impact on flow rate. Compression experiments using a customized experimental apparatus clearly showed clear buckling behavior in the filament path. The filament compression in response to force was modeled, and a compression model provided accurate prediction of the actual extrusion velocity from the force signal. Another source of dynamics is the temperature response of the inner wall of the liquefier to extrusion. The temperaturee dynamics were studied in the fifth workresearch task, which includeds deriving an analytical model forof the thermal system at the extrusion nozzle in response to melt flow and the PID temperature- controlled heaterer. The experimentally validated model showed a large variation between the feedback temperature of the thermistor and the modeled temperature at the inner wall, highlighting the importance of state estimation for temperature control of the polymer melt. In summary, this work advances the knowledge of extrusion dynamics in fused filament fabrication by providing new experimental data to the larger scientific community, data driven models, and simple analytical models for some of the core contributors to extrusion dynamics.