Feasibility study on sensing and prediction of backside weld geometry in cold metal transfer welding of X65 pipeline in the vertical-up position

Good root-pass welding is the basic guarantee for the formation of high-performance welded joints. In the (vertical) position, the difference in the directions of gravity and arc force induces the issues of uneven weld shaping and discontinuous weld penetration more serious than those in the 1G (flat) position. In order to obtain good weld penetration state for cold metal transfer (CMT) root-pass welding in the vertical-up position, characteristic signals that could characterize the backside weld geometry were explored, and a prediction model for the backside weld width Wb was established. In this study, the topside weld pool images and the electrical signals were collected in real time. It was found that the two-dimensional topside weld pool geometrical rameters, such as weld pool area A, cannot accurately characterize the backside weld geometry. Due to difference in arc force, the backside weld geometry with large differences can be obtained with a similar A. After the analysis of the electrical signals, the characteristic signals, namely welding heat input HI and the peak current time ratio PTR, which can represent the heat and force in the CMT welding process, respectively, were extracted, and the two were combined to establish a Wb prediction model, which had a good accuracy and laid foundation for the weld penetration control after that.

Automated summary

Good root-pass welding is crucial for high-performance welded joints. Welding in the vertical position presents challenges of uneven shaping and discontinuous penetration. This study explores characteristic signals to predict the backside weld width (Wb) for cold metal transfer (CMT) root-pass welding. By analyzing topside weld pool images and electrical signals, the welding heat input (HI) and peak current time ratio (PTR) were identified as representative signals for heat and force in the CMT welding process. Combining these signals, a precise Wb prediction model was established, laying the foundation for weld penetration control.