Prediction of Weld Defects during Friction Stir Welding in Real Time (30-Nov-2011)

Problem Definition

Friction Stir Welding (FSW) is a solid state joining technique that uses a rotating (spinning) tool to stir metal together to form a joint. Depending on the robustness of FSW process parameters, welds may contain various types of defects such as undercut, wormholes, lack of fusion, inadequate joint penetration and material flash etc. Several factors that influence these defects are tool dimensions and design, workpiece thickness and welding process parameters such as rotation speed, travel speed and the plunge force etc. These welding process parameters affect the material flow which is the dictated by the combined effects of tool pin and shoulder on the material surrounding them. These defects can be categorized as either flow or process related. The geometric lack of penetration/fusion defect occurs due to improper pin tool penetration depth.

During FSW, various variables such as forces on tool, machine torque and tool temperature are measured to monitor the weld quality. These variables are recorded to provide information on time-domain variation of these variables during welding as shown in Fig. 1. Any anomaly present in the recorded data indicates the presence of a defect in the weld. In general, machine data is a noisy waveform that conveys little information to the human eye. Currently, tool temperature measurement is generally used to control the weld quality. This information only provides an indication of deviations in the actual process parameters. Reduction in tool temperature is generally considered as disengagement of the tool shoulder from the workpiece surface. Similarly, an increase in tool temperature is generally considered as excessive penetration of the tool in the workpiece. However, this approach does not provide any specific details on the type and location of the defect. Similarly, any fluctuations in the other measurements such as forces on the tool and machine torque indicate deviation in the process parameters but can not predict the defect.

Figure 1: Plot showing variation of forces on tool and rpm recording during actual welding operation. X-load, Y-load and Z-load on tool represent the loads on tool in welding direction, transverse direction and plunge direction, respectively.

The presence of any defect in the welds reduces the welding productivity and project economics due to increased repair rate. Furthermore, any missed defect during weld inspection will impact structural integrity. Also, any wrong indication of defect which does not correspond to an actual defect (called as false alarm) will lead to unnecessary repair and increase in welding cost. Therefore, there are both economic and safety incentives to accurately detect defects during welding and to take appropriate remedial measures. To date, there is no framework to predict the optimal window of process parameters or to predict in real time weld defects, which may h...