Fault Detection Using Acoustic Methods for the Sodium-Water Reaction Monitoring in a Liquid Metal Reactor

Abstract

This study was aimed at developing a new method for detecting the sodium-water reaction as a result of a water leakage into the liquid sodium boundary for the liquid metal reactor. In the case of a passive acoustic method, to develop the leak detection algorithm, several signal processing methods have been evaluated. When the acoustic emission signal has a relatively high signal-to-noise ratio for the acoustic noises the spectral estimation method could be used to detect the sodium water reaction. In the case of a low signal-to-noise ratio within -5 dB to -15 dB, the system modeling and the identification methods using an autoregressive and an adaptive algorithm could be used to detect the sodium-water reaction. Regarding the active acoustic method, the basic attenuation factors of the ultrasonic beam were evaluated in the case of normal and abnormal plant conditions.