Many companies are evaluating various welding technologies as an alternative to riveting and the penalties that go with it in terms of aircraft weight and costs.
Friction stir welding (FSW) allows the welding of all-aluminum alloys of interest with good seam properties and with thicknesses from 0.3 to 50 mm or even more, according to a recent research project. The use of FSW instead of automated riveting can lead to a 20% reduction in process time and costs, researchers involved in the project say.
In FSW, a rotation symmetric tool with a shoulder and a contoured tool pin is used. The rotating tool plunges into the joint line until the shoulder comes in hard contact with the join partners. Due to the friction of the tool shoulder and the pin with the joint partners, heat that is generated plasticises the material below its melting temperature. After a short dwell, the welding feed starts and the tool pin stirs the materials together while the tool shoulder continues to generate frictional heat by its rotation under high downforces. In the case of aluminum alloys, the welding temperature in the welding zone typically is between 450 and 550°C—significantly lower than those of other fusion-welding techniques. Finally, the tool is retracted at the end of the weld and the negative shape of the tool remains.
FSW is not a new technology, but improvements are being made continuously. Airbus, for example, qualified the FSW process in 2006.
The advantages of FSW with robots were observed in the project called “RoboFSW” involving Airbus Deutschland GmbH; EADS Innovation Works Germany; Kuka Roboter GmbH, whose KR500 high-payload robot was used; and the Institute for Machine Tools and Industrial Management at the Technische Universität München.
FSW machines have very high stiffness, but in the majority of cases these machines are restricted for welding only linear joints. Aircraft fuselage sections usually consist of different and complex curvatures, which require a more flexible handling unit for the FSW tool. Industrial robots possess the needed flexibility, and their investment costs are considerably lower than those of a special machine, an NC milling machine, or a parallel kinematic robot. In addition, they are able to provide sufficient forces to weld thin aluminium sections.
In the RoboFSW project, the Kuka KR500 robot was heavily modified to apply higher process forces than the standard version. The welding spindle is hydraulically driven. Between the end effector and the spindle, a force-measurement system is integrated. The system allows, in combination with the robot control, a force-controlled welding mode. Force control is needed to obtain high seam qualities and requires no additional hardware except the force-measurement units and the signal I/O interface. The system works without any bobbin tool technology (double-sided shoulder) and requires no C-frame to keep the process forces in the tool. All required welding forces are generated by the six axes of the robot.
The system is able to apply process forces of up to 10 kN in every basic end effector direction at a medium range of the welding tool to the robot base (~1.8 m). This capability enables the robot to weld aluminum sections with a penetration depth of 5 mm and more, depending on the alloy welded. Thus, it is possible to apply the same process parameters that are also used on special FSW machines. The maximum process forces are limited by the maximum power of the robot engines and the maximum torques of the gears, respectively.
Industrial robots feature a maximum flexibility, which means a high capability for 3-D seam geometries. Usually, the FSW tool is moved over the joint line under a tilt angle of 2 to 5°. To apply this angle, an axis handling machine of at least five axes is necessary to weld simple 2-D joint lines like a circle. It could be shown that the robot is able to produce welds with equal properties in 12 overall working positions/directions, four in each Cartesian end-effector position. These experiments have been performed with a programmed process force of 9 kN on the AW-6056-T6 material. The measured force values revealed an overall scatter of less than 7% (comparison of the average values).
To evaluate the robot’s flexibility to weld on 3-D shapes, a convex-concave device was built up with radii of 500 mm. This curvature is of interest for aircraft welding applications and provides a basis for preliminary tests to weld real parts later in the project. The system was able to generate a constant seam quality over the whole length of the weld even in different welding directions. The sample showed a bead-on-plate welded aluminium sheet of the alloy AW-6061-T6 with a penetration depth of 1.8 mm and a welding feed rate of 700 mm/min. The sample was welded three times in different welding directions and with identical process parameters. The angular distortion was below 1.5° after welding. Subsequent to the welding, 16 specimens for the metallographic evaluation were prepared out of each sample; except for some increased flash formation due to non-optimal clamping, micrographs showed nearly identical properties without defects. The hardness at each welding sample was measured at one welding position. The nearly identical distribution over all three measurements confirms the robustness of the robot driven welding process.
This article is based on SAE technical paper 2007-01-5811 by G. Voellner, Technische Universität München; M. Zaeh, also of Technische Universität München; J. Silvanus of EADS Innovation Works Germany; and O. Kellenberger of Kuka Roboter GmbH.