Signal-based detection of arc phenomena in underwater electrothermal cutting with CAMG for the decommissioning of nuclear facilities

Abstract Contact Arc Metal Grinding (CAMG) is a fast, robust underwater cutting process. This makes it a potentially valuable solution for nuclear decommissioning. Until now, the measurement of tool wear has been based on parameters such as diameter, volume, or weight of the electrode. All of these variables require direct manipulation and contact with the electrode in order to be determined. In contaminated environments, contactless condition monitoring is crucial. The study proposes and evaluates a signal-based, contactless indicator for tool electrode wear, derived from electrical process data. Transient current signals were transformed into histograms, and the integral over the high-current region was used as a wear descriptor. CAMG trials with various additively manufactured tool electrodes with functional areas (different monolithic metallic alloys and CuSi3Mn1-based composites with fused WC) were conducted under water to evaluate this feature. A dedicated model test setup with different electrodes and high-speed imaging for the isolated observation of individual arc events was designed to facilitate the interpretation of process signals. For monolithic metallic functional areas and a reference steel electrode, the high-current integral decreased monotonically with wear and showed strong, statistically significant negative linear correlations. Despite initial disparities in levels, most notably for the thinner reference disc, all electrodes converged to similar values at the end of service life, once the functional area had been consumed. In contrast, CuSi3Mn1-based composites with fused WC particles exhibited weak or non-significant correlations, consistent with wear-driven changes in surface composition that alter local contact and arc attachment conditions. High-speed imaging was employed to confirm the occurrence of discrete arc events and to demonstrate that discontinuous voltage transients are consistent with rapid current interruptions and inductive voltage spikes. This enabled a correlation previously assumed in the literature to be confirmed experimentally for the first time.