Spectral Correlation Based Defect Recognition in Friction Stir Welding
By Yuri Hovanski (Brigham Young University), John Hunt (Brigham Young University), Carl Sorensen (Brigham Young University), Brian Mazzeo (Brigham Young University)

As FSW transitions from a welding novelty to a production process, greater emphasis is required to qualify and quantify the welding process. While external inspection technologies, destructive and non-destructive, are being adapted to accommodate FSW, these fail to take advantage of the inherent benefits within the process controls; namely that all process forces can be continuously monitored while welding. Initial investigations starting in 2008 demonstrated that spectral analysis of Z-force could yield the ability to detect the presence of a defect at a particular position along the weld length; however, no attempt was made to characterize the position of the defect within the weld nor the magnitude or type of defect present at that location. We show how spectral evaluation of process forces can be used to both recognize the presence of a defect, and derive specific attributes related to such presence recognition.