Microstructural and Strength Comparison Friction Stir Processing of Aluminium Alloys and Multi-Response Optimization by Taguchi GRA-PCA based Hybrid Methods

Abstract

Friction stir welding input parameters play a significant role in determining the weld joint quality. The quality of weld joint can be assessed in terms of microstructure, weld nugget geometry and mechanical properties. The present study takes up the multi-response optimization of friction stir welding (FSW) of aluminium alloy, which is suitable for liquid rocket metal tanks of launching vehicles. This paper has reported a review of friction stir welding key parameters and their influence on weld quality. Experiment work has been conducted on 6mm thickness AA 2050 aluminium alloy to observe the influence of process parameters on the microstructural change and features of mechanical properties variation. A search for an appropriate set of process parameters capable of creating high-quality friction stir weldment has been made. The optimal process environment is made up of a number of process control characteristics known as factors. This thesis looked at two process variables: rotating speed and welding speed. Taguchi's L16 Orthogonal Array (OA) was used for testing parameters such as tensile strength, yield strength, ductility, hardness, bending strength and heat affected zone width. These various objective functions were added together to generate an overall quality indicator, which was then optimized. The goal was to optimize a multi-objective optimization problem by converting it to a single objective function. The Taguchi method, paired with grey connection notion, was used as a hybrid technique. RSM-integrated desirability approach has been used for the parametric optimization. To overcome the limitations of the desirability approach, an integrated method, RSM-GRA-PCA is used to solve the multi-response optimization problem. It can be inferred that RSM-GRA-PCA method has an edge over Taguchi-GRA-PCA method by comparing the overall weld qualities calculated by the two methods at their respective optimum process parameters. Optimal results have existed in the experimental work and no additional verification is required. And finally, the validation of the optimal results has been done by Analysis of Variance (ANOVA).