K. Krishnamurthy, Y. Yu, R. L. Landers, University of Missouri–Rolla, Rolla, MO; D. R. Bolser, R. Talwar, Advanced Manufacturing R&D, Boeing – Phantom Works, St. Louis, MO; D. L. Ballard, Air Force Research Laboratory, Wright-Patterson AFB, OH

Summary:

Friction stir welding is becoming increasingly important as a joining process for many aerospace applications. Many studies have been conducted to better understand the process. These studies typically consist of either a detailed analysis of the work zone, using tools such as finite element analysis, or empirical investigations of the effect of process parameters (i.e., tool linear speed, tool rotational speed, and plunge depth). While these studies have led to a more fundamental understanding of the friction stir welding process, these models only provide process plans having constant process parameters. The friction stir welding process has many disturbances such as gaps between plates, plates whose thicknesses are mismatched, tool wear, and variations in material properties. Constant process parameter plans, even those developed from detailed models, cannot account for these disturbances. Automatic process control has the potential to adjust process parameters to account for these disturbances. To date, dynamic, control–oriented models of the friction stir welding process have not been developed and real–time control algorithms that regulate the friction stir welding process by automatically adjusting the process parameters have not been implemented. The biggest obstacle to implementing process control is the development of hardware testbeds with open control systems.

This paper details the development of two hardware testbeds that are being used to develop dynamic, control–oriented models and perform real–time process control of friction stir welding processes. The first testbed is a commercial two–axis friction stir welding machine. The axis position, velocity, and force and spindle rotational speed and torque signals are taken from the machine and fed back to a data acquisition system to measure these variables at a high frequency. The system is also augmented with two surface temperature measurement devices, whose signals are also fed back to the data acquisition system. This testbed will be used to develop dynamic, control–oriented process models. The second testbed is a robotic friction stir welding machine with an open architecture controller. This controller allows for the implementation of real–time controllers. The authors gratefully acknowledge the support of the Air Force Research Laboratory through contract no. FA8650-04-C-704 (Dr. Jaimie S. Tiley, Program Manager).