Detection of kissing bond defects in the friction stir welding using linear and nonlinear ultrasonic techniques

The development of NDT techniques for the detection of kissing bond defects in Friction Stir Welding (FSW) is still a challenging and ongoing research topic. Kissing bond defects are root flaws that arise during the solid-state welding procedure and may have an adverse impact on the fatigue resistance of the welded components. Conventional off-the-shelf ultrasonic weld inspection methods are not capable of reliably identifying kissing bond defects due to the tight and tiny characteristics of the flaws. Therefore, two potential ultrasonic examination techniques are proposed in this presentation: an immersion linear ultrasonic technique and a contact nonlinear ultrasonic inspection method. The former technique takes advantage of an empirical focusing procedure at oblique incidence and a timing gate in backscattering mode using a 3.5 MHz transducer, whereas the latter method exploits the pulse-inversion technique for isolation and enhancing the nonlinear scattering induced by the kissing bond defect in a pitch-catch mode using a combination of a 3.5 MHz transmitter and a 7 MHz receiver. Figure 1 (left) displays angled C-scan results obtained by the linear ultrasonic backscattering inspection method. The horizontal and vertical axes represent the distance along and from the weld centerline, respectively. Defects can be identified in the zone between the two black horizontal lines along the weld centerline in the backscatter C-scan image which correspond to the Thermo- Mechanically-Affected Zones (TMAZ). A pulse inversion based B-scan spectral image obtained by the nonlinear technique is displayed in the right hand side subfigure of Figure 1. The horizontal axis represents the frequency and the vertical axis is the distance along the weld. The emitting transmitter was excited with a chirp signal ranging from 2 to 5 MHz. Two out of phased back-wall reflections (A-scans) were received in each position along the weld centerline. Next, a pulse inversion spectrum of each couple of A-scan signals was normalized by the maximum amplitude squared of the corresponding positive polarity signal. Ultimately, the normalized spectra are stacked up to obtain a 2D heat map (B-scan spectral image), representing frequency components on the horizontal axis and the position along the weld centerline on the vertical axis. Using pulse inversion, the fundamental frequency components are completely canceled out, while the second harmonics are amplified. As such, defect zones can be easily identified and localized along the weld path and their size can be measured.status: accepte