Subscribe to our newsletter to receive the latest news and events from TWI:

Subscribe >
Skip to content

Robotic Friction Stir Welding

Joining Technologies Banner
Innerpage Title Banner

The use of robots for friction stir welding (FSW) poses a number of challenges compared to conventional, rigid FSW machines. However recent advancements in robotics, together with new FSW techniques, have increased the potential of robotic FSW substantially.

A major difference between FSW and fusion welding processes is that FSW is an in-contact machine operation, which means that there are forces acting between the machine and the welded component. As the tool is pressed into the plasticised material, a forging-like operation is initiated where the plasticised material is forced to flow around the rotating tool. This requires high forces from the machine in order to keep the tool down into the material. For FSW of aluminium sheets, the axial force is typically several kilonewtons, depending on the alloy grade and thickness. For example, a 3mm thick 6000-series aluminium alloy requires around 5kN of axial force.

The majority of industrial robots are articulated arm robots, where the tool is actuated by a serial chain of links and typically six rotary joints, driven by electric servo motors. Due to the high force required for FSW, the demands on the machine are high. Consequently FSW was initially not suitable for implementation on such industrial robots.

A breakthrough came with the development of heavy-duty industrial robots in the late 1990s. While mainly developed for lifting car bodies in the automotive industry, the robots were also proven to be suitable actuators for implementation of FSW. In a race for the title of most powerful robot across machine manufacturers, payloads increased from a maximum of 500kg around the year 2000 to the 1500kg-plus that current robots are capable of lifting, with corresponding improvements in accuracy and stiffness. These developments have made industrial robots more and more suitable for FSW.

Robots have the great advantage of being flexible to reprogram and reconfigure, and being capable of following three-dimensional joint lines. Furthermore, industrial robots are usually less expensive than comparable five-axis machines, very reliable and well known in many industry segments. This makes them well suited for high-volume industrial FSW applications.

But challenges remain when using industrial robots for FSW. One of the main complications is the limited stiffness of robots, which can cause them to deflect under high loads. Because positioning of the tool, relative to the joint line, is important to guarantee a consistent weld quality, some additional control actions are needed:

  • Force control: Implementation of force control allows the FSW tool to ‘float’ on the surface of the material while welding with a constant contact force, independent of its axial position. The most accurate approach is to use a force sensor connected to the robot controller. Alternatively, a kinematic model, relating robot stiffness to axial force, can be used to make an indirect force controller.
  • Path deviation compensation: Deviations of the FSW tool from the weld path due to machine deflection can cause, for example, lack-of-penetration defects. This can be resolved by implementation of seam tracking, either sensor-based such as laser scanners and camera machine vision, or without sensors, using a robot deflection model.

Increased complexity of the weld geometry, especially along multi-dimensional joint lines, can cause significant changes in heat dissipation along the weld. This means that, with a constant heat input, the weld temperature and consequently the weld quality will vary. In some cases, this leads to over-softening of the material, causing the force-controlled tool to sink down into the workpiece. Implementation of a feedback temperature controller which adjusts factors such as the rotational speed during welding can significantly reduce this risk.

TWI has developed new variants of the conventional FSW process, which reduce some of the complexity of robotic FSW. One such example is stationary shoulder FSW. With this technique, unlike conventional FSW, only the tool probe is rotating, while the non-rotating shoulder is pressed against the material surface. Stationary shoulder FSW provides improved surface finish, with lower heat input and hence lower distortion. Because the shoulder is moving over relatively cold material, the risk of sinking down in over-softened material is greatly reduced. The stationary shoulder FSW technique has been expanded to non-cylindrical shoulders such as corner tools, allowing corner FSW of T-shaped components.

Programming a FSW robot to weld along complex joints is not an easy task. The position of the tool must be accurate, and the orientation relative to the surface is critical. Typically, a tilt angle towards the trailing edge of the tool should be applied; small changes in tilt angle have a significant influence on the joint quality. By using offline programming software tools, CAD models of the components can be imported into a virtual environment and robot paths can be generated automatically. This improves accuracy of the path, while reducing programming time. Furthermore, programming of a future product can already be done while the present is still in production. This reduces overall down-time of the system

Several applications across various industries have adopted robotic FSW as a production method. A notable example is the Apple iMac computer, but automotive chassis parts and heat exchangers are also welded using FSW robots.

Using current robot systems, applications are mainly restricted to aluminium, magnesium and copper in thicknesses of up to 7mm and speeds up to 3m/min, but continuous developments in FSW and robotics are constantly pushing this limitation further.

For more information, please see Joining Technologies or please email: