RMS-based b-value Analysis of Nuclear Piping Cracks under Realistic Noise Environments

Abstract

Current Acoustic Leak Monitoring Systems (ALMS) in nuclear power plants (NPPs) are primarily designed for leak detection, often failing to identify precursor crack growth due to the masking effects of high-level mechanical background noise. This study proposes a robust RMS-based b-value analysis to evaluate crack progression under realistic high-noise conditions. To simulate an operational environment, actual background noise recorded from nuclear pumps was digitally synthesized with Acoustic Emission (AE) signals obtained from tensile tests of SM45C steel specimens. While conventional hit-based methods failed to distinguish valid signals from noise, rendering the Kaiser effect indiscernible, the proposed RMS-based approach effectively smoothed stochastic fluctuations and clearly identified the Felicity effect during irreversible damage accumulation. Furthermore, the RMS-based b-value exhibited a stable, monotonic decrease corresponding to crack propagation, in contrast to the scattered trends observed in amplitude-based analysis. These results demonstrate that the proposed algorithm significantly enhances the Signal-to-Noise Ratio (SNR) and offers a practical solution for the early warning of structural degradation in high-risk piping systems.