Abstract
Lack of sidewall fusion (LOSWF) is a critical defect in arc welding that compromises structural integrity, especially in multi-pass welds where buried discontinuities require highly advanced volumetric imaging techniques for detection. Traditional non-destructive testing (NDT) methods are often unable to identify such defects until fabrication is complete, increasing rework rates and overall build time. This study presents a novel approach, combining in-process ultrasonic imaging with controlled experimentation to enable LOSWF detection capability during welding. An experimental setup is introduced in which a static phased array probe is positioned ahead of the welding torch, allowing B-scan acquisition in real-time, during welding. Characteristic signal loss is observed prior to sidewall fusion, followed by echo recovery upon solidification-providing a dynamic indicator of fusion status, with a distinct amplitude drop from 60 to 0%, highlighting the binary nature of the monitoring. To benchmark detection limits, artificial LOSWF flaws were introduced into single-layer welds and evaluated using a roller probe configuration. In addition, experiments were performed to analyze signal degradation and recovery due to thermal disturbance, captured through C-scan sidewall echo analysis. The results demonstrate that ultrasonic imaging deployed during welding can offer both predictive and confirmatory information about fusion quality. This integrated approach provides a foundation for automated, embedded weld inspection systems that can identify fusion defects earlier in the process chain.