A Semi-Analytical Nonlinear Regression Approach for Weld Profile Prediction: A Case of Alternating Current Square Waveform Submerged Arc Welding of Heat Resistant Steel

Abstract

The complexity in weld profile caused by abrupt change in polarity in square waveform welding is investigated through the development of a model capable to accurately predict weld profile. A semi-analytical model is conceived wherein characteristic attributes of a composite parabolic–elliptic function, which represent the weld profile, are obtained through nonlinear regression (NLR). The proposed model is demonstrated for its efficacy in the prediction of weld profile over a wide range of welding parameters, vis-à-vis, welding current, frequency, electrode negative (EN) ratio, and welding velocity. The investigation suggests that the center and outer cores of welding arc remains more active during positive and negative polarity, respectively, that leads to distinct macroscopic zones in weld cross section and thus, necessitates a composite profile for representation of weld profile. The intersection of the zones forms a metallurgical notch which the investigation offers a method to estimate and thus control. Unlike the convention continuous arc welding, the waveform arc welding caters welding at higher velocity without compromising the weld penetration and almost abolishing the metallurgical notch as well.