Manufacturing Technology 2022, 22(2):156-167 | DOI: 10.21062/mft.2022.016

Multi-Response Optimization of Friction Stir Welding of AA2050 Using Response Surface Methodology Coupled with Grey Relational Analysis and Principal Component Analysis

Raju Kamminana ORCID..., Venkata Subbaiah Kambagowni ORCID...

Department of Mechanical Engineering, Andhra University, Visakhapatnam, Andhra Pradesh, India

The third generation aluminium-lithium alloy AA2050 finds wide applications in defence and aircraft industries by virtue of its high strength-to-weight ratio and excellent corrosion resistance. Friction stir welding (FSW), relatively novel technique, is more suitable to join this alloy compared to other conventional fusion welding techniques. In this work, the overall quality of the weld joint was decided from the higher values of tensile strength, yield strength, percentage elongation, hardness of weld zone, hardness of heat affected zone, bending load and lower value of width of heat affected zone. The optimal (combined) design was used to design the experiments with four numeric factors (traverse speed, rotational speed, tilt angle and shoulder diameter) and a categoric factor (tool pin profile). The multi-response optimization problem was reduced into a single-response optimization problem using grey relational analysis (GRA); principal component analysis (PCA) was used to assign optimal weighting values for the responses in the process of dimensionality reduction. Mathematical model for the reduced single response, which can be perceived as overall weld quality, was developed by the response surface methodology (RSM) and the optimization of process parameters was also carried by the RSM. Analysis of variance (ANOVA) was carried to evaluate the significance of each parameter on the overall weld quality and the adequacy of the developed model. The confirmation tests conducted at optimum levels of parameters proved the effectiveness and robustness of the method.

Keywords: Multi-response optimization, grey relational analysis, principal component analysis, response surface methodology, analysis of variance

Received: April 23, 2021; Revised: February 16, 2022; Accepted: March 2, 2022; Prepublished online: March 31, 2022; Published: May 15, 2022  Show citation