Analytical and Experimental Models for Real-time Feedback Control of Robotic Fused Filament Fabrication System
Restricted (Penn State Only)
- Author:
- Badarinath, Rakshith
- Graduate Program:
- Industrial Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- March 17, 2022
- Committee Members:
- Timothy Simpson, Major Field Member
Ed De Meter, Major Field Member
Vittaldas Prabhu, Chair & Dissertation Advisor
Chris Rahn, Outside Unit & Field Member
Steven Landry, Program Head/Chair - Keywords:
- Additive Manufacturing
Fused Filament Fabrication
Robotics
Sensing
Control
Computer Vision
Closed-Loop
Synchronization
in-situ inspection
System Identification
Process Model
Modeling - Abstract:
- Fused filament fabrication (FFF) continues to be among the most widespread additive manufacturing process for making polymeric functional prototypes, and in several cases end-use parts. Robots are well established in manufacturing and their flexible nature is well suited to serve the dynamic demands of manufacturing. There is a large installed base of serial-link industrial robots, some of which could be potentially retrofitted with an extruder head as an end-effector to serve as FFF systems. Additionally, there is a dearth of engineered efforts and process models toward process monitoring and closed-loop control of the FFF process in general. Part one of this dissertation identifies and proposes solutions to key engineering challenges that arise in retrofitting such robotic FFF systems in terms of integrating robot motion controller with extruder controller and evaluating the quality of the fabricated parts. Specifically, an approach for integration and real-time synchronization of controllers is proposed that ensures the extrusion velocity and deposition velocity match closely by building upon an analytical model for predicting road geometry as a function of process parameters. Compared to gantry mechanisms, this is challenging in serial-link industrial robots because of significantly larger and space-variant inertias. A fully functional Robotic FFF research testbed has been developed in which integration and real-time synchronization of controllers is achieved by (1) communicating space-variant process parameters in real-time using TCP/IP sockets, and (2) analog and digital I/O interfacing. Experimental testing shows excellent (𝑅2 = 0.9983) agreement between requested and actual volumetric flow rates and less than 5% errors in extrusion widths and heights in test samples fabricated across the range of physical limits of FFF process parameters. This work can serve as a basis for further engineering innovations toward cost-effectively harnessing the capacity of industrial robots to manufacture geometrically accurate parts using FFF. Part two of this dissertation is an effort to further the engineering science and to gain a better understanding of the physics of the FFF process by engineering an instrumentation system for sensing and signal processing for real-time estimation of two main process variables in the FFF process: (i) temperature of the polymer melt exiting the nozzle using a thermocouple; and (ii) polymer flowrate using extrusion width measurements in real-time, in-situ, using a microscope camera. A design of experiments approach is used to develop response surface models for two materials that enable accurate estimation of the polymer exit temperature as a function of input polymer flowrate and liquefier temperature with a fit of 𝑅2 = 99.96% and 99.39%. The live video stream of the deposition process is used to compute the flowrate based on a road geometry model. Specifically, a robust extrusion width recognizer (REXR) algorithm was developed to identify edges of the deposited road and for real-time computation of extrusion width, which was found to be robust to filament colors and materials. The extrusion width measurement was found to be within 0.08 mm of caliper measurements with an 𝑅2 value of 99.91% and was found to closely track the requested flowrate from the slicer. This opens new avenues for advancing the engineering science for process monitoring and control of FFF. Part three of this dissertation uses a computer-aided control system design (CACSD) approach for developing mathematical plant models that describe the temperature and flowrate dynamics in the FFF process and the development of control strategies. Using the process sensing capabilities developed in part two, a data-driven system identification approach is taken to develop process models for temperature and flowrate dynamics. An analytical process model for polymer flowrate dynamics is developed based on first-principles modeling and is found to be in good agreement with its equivalent empirical model developed using system identification. Finally, control strategies such as feedback only and feedback + feedforward are designed in MATLAB, and simulations of these closed-loop systems are performed to evaluate performance improvements in setpoint tracking and disturbance rejection in the FFF process. The feedback + feedforward control architecture works best to maintain the temperature inside the nozzle even during changes in the requested polymer flowrate. Extrapolating based on results in existing literature, there is a potential for a 10% improvement in the flexural strength of printed parts. This work helps to advance the fundamental process science for describing FFF dynamics and can serve as a basis for future innovations in model-based control of the FFF process to realize adaptive systems with feedback-based process correction.