Comparison of the Optimized Process Parameters of Double-Sided Friction Stir Welded Aluminium Alloy Joints Using Statistical and Evolutionary Techniques

Abstract

One of the most innovative solid state welding techniques used in the aerospace, automotive, defence, rail and marine industries is Friction Stir Welding (FSW) process, as it is used for joining aluminium, copper and magnesium alloys. The weld quality is decided by the FSW process parameters such as rotational speed, welding speed and pin profile. A regression model was developed relating the welding input parameters (tool rotational speed, welding speed and pin profile) and the output response parameters (tensile strength, hardness and toughness) based on the experiments carried out with the help of Response Surface Methodology. The obtained regression equations were used in determining the optimal welding process parameters. A new method, Elitist Non-dominated Sorting Genetic Algorithm (NSGA-II) based on evolutionary algorithm has been used in the optimisation. The optimum results gathered from the desirability approach through Response Surface Methodology (RSM) were compared with those obtained through the evolutionary algorithm. The results show that the proposed evolutionary method is much effective, faster than the desirability approach discussed in the work.