Quadratic Response Surface Modeling for Prediction of Bead Geometry in Submerged Arc Welding

Abstract

The effect of various process control parameters on bead geometry of submerged arc w/eldment has been aimed to Investigate using quadratic response surface methodology. Each of the features of weld bead geometry has been assumed and expressed as a function of selected process parameters so that the function takes into account the linear, quadratic as well as interaction effects of the predictors. Experiments are conducted with different levels of process parameters like voltage, welding current, wirefeed rare and traverse speed to obtain beadon- plate weld on mild steel plates. Based on multiple linear regression, mathematical models have been developed for prediction of bead geometry, for different settings of factor level. Using statistical software package MINITAB, reduced models are built up with significant factors and coefficients. MINITAB's Backward elimination option in stepwise regression is used to eliminate insignificant factors from the models and to recalculate the coefficients of the significant factors. Developed mathematical model for penetration has been optimized (maximized), considering (minimizing) the reinforcement, depth of HAZ and bead width as constraints. Sensitivity analysis was also carried out to study the change in value of the objective function (bead penetration) due to change of limit of the constraints from the optimum value. The direct and interactive effect of various process control parameters on the features of bead geometry and HAZ have been measured quantitatively and represented graphically. The methodology, proposed in the study can be used to obtain superior quality weld bead and also to achieve high productivity.