An analysis of curvilinear regression equations for modeling the submerged-arc welding process

Abstract Curvilinear regression equations have traditionally been used for modeling the weld bead geometrical features of the submerged-arc welding process, but, however, only the correlation coefficients have been reported. This investigation was carried out to study the relationship, if any, between the correlation coefficient and the standard deviation of the deviations between the predicted and measured values, with respect to melting rate, total fusion area, penetration, deposit area, bead height and bead width. It is found that there is no strong relationship between the correlation coefficient of a correlation equation and the standard deviation of the deviations between the predicted and measured values. A low correlation coefficient does not necessarily mean that there are standard-deviation consequences for weld-bead features. The statistical analysis has provided other useful information also. It is found that the mean deposit area obtained directly from the melting rate is higher, since it is assumed that all the melted metal goes into the weld deposit without any losses. It is estimated that the metal loss through vaporization is about 4% and 8% for positive electrode polarity and negative electrode polarity, respectively.