Multivariate optimization of mechanical and microstructural properties of welded joints by FSW method

Abstract

This paper investigates the microstructural and mechanical properties in the joining of aluminum sheets using the friction stir welding (FSW) method. First, the microstructural properties are investigated, and then the effect of process parameters on the grain size, defect formation, microhardness, and fracture energy is discussed. The grain size in the stir zone is remarkably refined compared to the base metal, while the microhardness is reduced due to the annealing effects of the FSW. According to the findings, the material integrity in the stir zone is the most critical aspect in determining the value of absorbed energy in the fracture test. The existence of a cavity in the stir zone severely reduces the fracture energy, while the negative effects of the weld line remnant defect on the fracture energy are negligible. For each pin profile and rotational speed, the safe window of traverse speed for defect-free joint production is widened from the conical cylinder pin to the square pin and then to the threaded pin. The ratio of rotational speed to traverse speed (w/v) is the most important parameter in determining almost all characteristics of a joint produced by a specific tool. For each type of pin profile, there is a minimum amount for this ratio, below which the defect formation is very likely to the occurrence. The lowest w/v ratio required to generate a defect-free weld with the conical cylinder pin tool is nearly 2. While that is desirably lowered to 1.6 and 1.25 for square and threaded pin tools, respectively. For all three pin shapes, the hardness value is increased by increasing the traverse speed and decreasing the rotational speed. The grain size and Charpy impact fracture energy values are also reduced by decreasing the rotational speed and increasing the traverse speed. While, a power regression curve (that is very similar to the Hall-Petch relationship) is fitted well to describe the relationship between hardness and grain size, a second-order equation is fitted to the results of fracture energy vs. hardness and grain size. Furthermore, based on the obtained results, it can be concluded that the fracture energy value is more sensitive to the pin profile compared to other outputs of this research (i.e., hardness and grain size). Using the neural network, a relationship was obtained between the input parameters of the process, such as traverse speed, rotational speed, and tool shapes, with the mechanical and microstructural properties of the joints. This relationship was then used in a hybrid multiobjective optimization to find the optimal process parameters. As the best compromise option, hybrid multiobjective optimization proposes a traverse speed of 309 mm/min, a rotational speed of 495 rev/min, and a threaded cylinder pin shape.