On the nondestructive testing and monitoring of cracks in resistance spot welds: recent gained experience

The current paper addresses the challenge of steel and modern automotive industries in the detection of liquid metal embrittlement (LME) crack when occurring during resistance spot welding of advanced high strength steel. While progress has been made to understand and prevent LME crack occurrence, its detection and characterization are still the main concern. Unfortunately, these aspects have received scarce attention in the literature. In this study, various non-destructive testing solutions (NDT) for crack detection and characterization (in particular depth estimation) were developed using novel algorithms and hardware. Despite the complex features of LME crack, its detection and characterization are promising: acoustic emission testing (AT) was found to be efficient for crack detection using a novel criterion based on Shannon entropy. Unprecedented AT tests have even been carried out in serial automotive production environments to assess deployment feasibility. The machine vision algorithm was implemented for magnetic inspection to assess crack detection automation. For crack depth estimation, three solutions are discussed using pulsed thermography, Eddy current array, and industrial large scale tomography (RoboTom). This paper focuses on gained experience and recommendations that would help the research community to increase the technology readiness of LME crack detection and characterization.