Anomaly detection in BWR fuel cell using neutron noise analysis techniques. Slow control rod detection

Abstract

A two-step methodology based on noise analysis techniques was applied to detect and monitor a resonance to do with a control rod-related anomaly, notified as SLOW in a BWR-6 core. The methodology utilizes LPRM (Local Power Range Monitor) neutron flux signals. The first step involves applying traditional noise analysis to all available neutron detector signals. Both NAPSDs (Normalized Auto Power Spectral Density) and CPSDs (coherences and phases) were estimated. The second step uses EMD (Empirical Mode Decomposition) and VMD (Variational Mode Decomposition) in a sequence of successive decompositions which allow both the resonance under study to be isolated in only one IMF (Intrinsic Mode Functions) and the main resonance peak parameters (frequency, amplitude, and damping) to be estimated. On the one hand, it was possible to delimit the zone where perturbation takes place, though the phenomenon seemed to be quite local and not all detectors could provide information on it. On the other hand, intrinsic mode functions make it possible to observe traces of potential instrumentation tube impacts in the Hilbert Spectrum. The methodology proved to be very effective to monitor resonances even when these were not visible in the spectrum. The changes observed coincide with the corrective actions carried out by the plant.