A New Approach for Predicting andOptimizing Weld Bead Geometry in GMAW

Abstract

Generally, the quality of a weld joint is directly<br><br>influenced by the welding input parameter settings. In this study, the<br><br>regression modeling is used in order to establish the relationships<br><br>between input and output parameters for Gas Metal Arc Welding<br><br>(GMAW) process. To gather the required data for modeling, actual<br><br>tests were carried out based on the proposed Taguchi experimental<br><br>matrix design. The process variables considered here include voltage<br><br>(V); wire feed rate (F); torch Angle (A); welding speed (S) and<br><br>nozzle-to-plate distance (D). The process output characteristics<br><br>include weld bead height, width and penetration. To develop<br><br>mathematical models, various regression functions have been fitted<br><br>on the experimental data. The adequacies of the models are then<br><br>evaluated using analysis of variance (ANOVA) technique. The best<br><br>and most fitted model is then selected based on the ANOVA results<br><br>and other statistical analysis. The ANOVA results recommend that<br><br>the curvilinear model is the best fit in this case. In the next stage, the<br><br>selected model is implanted into a Simulated Annealing (SA)<br><br>optimization algorithm. This optimization procedure has been<br><br>developed in order to determine the best set of process variables<br><br>levels for any desired weld bead geometry characteristics.<br><br>Computational results show very good compatibility with<br><br>experimental data and demonstrate the effectiveness of the proposed<br><br>modeling and optimization approach.