In the present study, a technique has been addressed in order to model and optimize gas tungsten arc welding (GTAW) process which is one of the mostly used welding processes based on the high quality fabrication acquired. The effects of GTAW process variables on the joi
More
In the present study, a technique has been addressed in order to model and optimize gas tungsten arc welding (GTAW) process which is one of the mostly used welding processes based on the high quality fabrication acquired. The effects of GTAW process variables on the joint quality of AISI304 stainless steel thin sheets (0.5 mm) have been investigated. The required data for modeling and optimization purposes has been gathered using Taguchi design of experiments (DOE) technique. Next, based on the acquired data, the modeling procedure has been performed using regression functions for two outputs; namely, heat affected zone (HAZ) width and ultimate tensile stress (UTS). Then, analysis of variance (ANOVA) has been performed in order to select the most fitted proposed models for single-objective and multi-criteria optimization of the process in such a way that UTS is maximized and HAZ width minimized using simulated annealing (SA) algorithm. Frequency, welding speed, base current and welding current are the most influential variables affecting the UTS at 22%, 21%, 20% and 17% respectively. Similarly, base current, welding current, frequency and welding speed affect the HAZ at 28%, 20%, 16%, and 15% respectively. Based on the results considering the lowest values for current results in the smallest amount of HAZ. By the same token in order to acquire the largest amount of UTSs the highest values of current must be considered. Setting welding and base current, frequency, speed, and debi at 42 and 5 apms, 46 Hz, 0.4495 m/min, and 5 lit/min respectively resulted the optimized HAZ and UTS simultaneously. The proper performance of the proposed optimization method has been proved through comparison between computational results and experimental data with less than 6% error.
Manuscript profile
