Temperature control in friction stir welding (FSW) is of interest because of the potential to improve the mechanical and microstructure characteristics of a weld. Two types of active temperature control have been previously implemented for steady-state friction stir welding conditions: PID Feedback Control and Model Predictive Control. The start-up portion of a weld is an obstacle for these types of active control.To date, only minimal exploratory research has been done to develop an active temperature controller for the start-up portion of the weld. The FSW temperature controller presented in this thesis, a Position-Velocity-Acceleration (PVA) controller implemented with gain-scheduling, is capable of active control during the start-up portion of a weld. The objectives of the controller are (1) to facilitate fully-automated active temperature control during the entire welding process, (2) to minimize the rise time, the settling time, the percentage maximum post-rise error (overshoot calculated as a percentage of the settling band half-width), and the post-settled root-mean-square (RMS) of the temperature error, and (3) to maintain the steady state performance of previous control methods.For welds performed in 6.35 mm plates of 7075-T651 Aluminum with controller gains identified through a manual tuning process, the mean controller performance is a rise time of 10.82 seconds, a settling time of 11.35 seconds, a percentage maximum post-rise error of 69.86% (as a percentage of the 3◦C settling band half-width), and a post-settled RMS error of 0.92◦C.Tuning of the start-up controller for operator-specified behavior can be guided through construction of regression models of the weld settling time, rise time, percent maximum post-rise error, and post-settled RMS error. Characterization of the tuning design space is performed through regression modeling. The effects of the primary controller tuning parameters and their interactions are included. With the exception of the post-settled RMS error model, these models are inadequate to provide useful guidance of the controller tuning, as significant curvature is present in the design space. Exploration of higher-order models is performed and suggests that regression models including quadratic terms can adequately characterize the design space to guide controller tuning for operator-specified behavior.
Identifer | oai:union.ndltd.org:BGMYU2/oai:scholarsarchive.byu.edu:etd-8461 |
Date | 01 July 2018 |
Creators | Erickson, Jonathan David |
Publisher | BYU ScholarsArchive |
Source Sets | Brigham Young University |
Detected Language | English |
Type | text |
Format | application/pdf |
Source | Theses and Dissertations |
Rights | http://lib.byu.edu/about/copyright/ |
Page generated in 0.0021 seconds