Intercomparison of Auto- and Cross-Power Spectral Density Surveillance Systems for Sodium Boiling Detection in Fast Reactors

Theoretical and experimental investigations on detection systems for small narrow-band components in noise signals were conducted. These detection systems are based on the continuous surveillance of the power spectral density for characteristic peaks. Detection sensitivity for auto- and cross-correlation measurements was computed for signals with normally distributed amplitudes in dependence of signal coherence. The derived detection criteria allowed us to compare auto- and cross-power spectral density surveillance. Theoretical results were confirmed in a number of experimental parameter studies. Special theoretical investigations were done for the optimal detection of local sodium boiling in liquid-metal fast breeder reactors.