This paper reports a regression modeling and optimization procedure for activated tungsten inert gas (A-TIG) welding process of AISI316L austenitic stainless steel parts using Taguchi design of experiments (DOE), regression modeling, analysis of variance (ANOVA) and simulated annealing (SA) algorithm. Welding current (I), torch speed (S) and welding gap (G) are the most significant process input parameters has been considered in this study. Depth of penetration (DOP) and weld bead width (WBW) have been taken in to account as the most essential quality measures of the process under study. SiO2 surface activating flux powder (in scale of nano-particles) have been used to enhance TIG welding process characteristics. To gather required data for modeling and statistical analysis purposes, DOE based on orthogonal array (OR) Taguchi method has been used. Then, modeling has been performed using different regression equations (including linear, curvilinear and logarithmic models). Statistical analysis based on analysis of variance (ANOVA) has been performed and the most fitted models were selected as an authentic representative of the process characteristics. Next, these models were used for optimization of process parameters in such a way that DOP is maximized and WBW minimized using simulated annealing (SA) algorithm. Finally, a set of experimental tests has been carried out through which the proposed method has been validated. Furthermore, results of SA optimization method have been compared with ones gained using signal to noise (S/N) analysis based on Taguchi procedure. Results shown the proposed procedure is quite efficient in modeling and optimization of the A-TIG process.

Keywords