Developments in NDT for Detecting Imperfections in Friction Stir Welds in Aluminium Alloys

Abstract

Friction stir welding (FSW) has dramatically changed how aluminium alloys can be welded. The quality of FS welds is usually excellent, but some imperfections periodically occur. The geometry, location, and microstructural nature of these imperfections bear no resemblance to the imperfections typically found in aluminium fusion welds. Consequently, it has been difficult to identify FS weld imperfections with common non-destructive testing (NDT) techniques. Therefore, further development of NDT techniques must be done to enable the detection of FS weld imperfections. This paper presents an integrated, on-line, NDT inspection system for FS welds, which employs a data fusion algorithm with fuzzy logic and fuzzy inference functions. It works by analyzing complementary and redundant data acquired from several NDT techniques (ultrasonic, Time of Flight Diffraction (ToFD), and eddy currents) to generate a synergistic effect that is used by the software to improve the confidence of detecting imperfections. The system was tested on friction stir welded AA5083-H111 specimens. The results indicate that by combining the output from various NDT processes, an improvement in finding imperfections can be obtained compared to using each NDT process individually. The methodology implemented in the QNDT_FSW system has given good results and improved reliability in the NDT of friction stir welds.