Fatigue of Friction Stir Spot Welding in Magnesium alloys
In order to realize the goal of developing a microstructure-sensitive model for predicting fatigue damage in welded joints made using friction stir technology, relationships of microstructural and geometrical features to fatigue performance were investigated using AZ31 magnesium alloy sheets joined by friction stir spot welding (FSSW). Two sets of lap-shear coupons were spot welded using different welding conditions. Optical microscopy of the initial state of the microstructure of each set of spot welds revealed differences in the hook formation, sheet thickness in the weld zone, nugget diameter, and microstructure. Both sets of welds were fatigue tested in load control until failure at various load ratios. Optical microscopy of the failed coupons revealed differences in the fracture mode between the two sets of coupons. Fractography analysis conducted in this study suggested that the effective top sheet thickness largely determined the failure mode, which in turn influenced the final number of cycles to failure. While the height of the interfacial hook was greater in the process with better fatigue performance, it was the larger effective top sheet thickness that promoted crack propagation modes more favorable to greater fatigue resistance. To further aid in determining the cause and effect relationships and to elucidate the mechanisms behind fatigue damage in (FSSW), a linear elastic fracture mechanics model  was used to correlate the fatigue life in the two processes. The fatigue model, which is a function of hook size, sheet thickness, and nugget diameter, was used in a sensitivity study to evaluate the effect of sheet thickness in the weld zone, hook height, and nugget diameter.
Figures 20 and 21 show the comparison of the experimental to model results for the number of cycles for through-crack initiation versus maximum applied load for R=0.3 and R=0.7, respectively. As the graphs demonstrate, model predictions differ from process #1 and #2 results only in terms of sheet thickness, nugget diameter, and hook height. The model showed good correlation in capturing the differences in fatigue lifetimes for the two processes. However, the model was slightly non-conservative compared to process #1 for R=0.7 and slightly conservative for process #2 at R=0.3.