Indexed on: 15 Aug '10Published on: 15 Aug '10Published in: The International Journal of Advanced Manufacturing Technology
In this study, rolled plates of AA 2198 T3 aluminium alloy are friction-stir welded in butt configuration varying two fundamental process parameters: rotational and welding speeds. Two sets of empirical models based on regression analysis are developed. The first one predicts the stationary values of the in-plane and downwards forging welding forces in dependence of the process parameters under investigation. The second one predicts the mechanical strength, in particular yield and tensile strength, of the friction-stir welded joints as function of the same parameters. For the development of the empirical models, two 32 full factorial designs are used: one having the stationary values of the welding forces and the other having the yield and tensile strength as observed responses, respectively. Statistical tools such as analysis of variance, F tests, Mallows’ CP, coefficient of determination etc. are used to build and to validate the developed models. By using the desirability function approach, the optimum process parameters to simultaneously obtain maximum possible yield and tensile strength are found within the investigated range. The developed models can be effectively used to predict the stationary forces and the mechanical proprieties of the joints at 95% confidence level.