Cofrentes nuclear power plant instability analysis using ensemble empirical mode decomposition (EEMD)

Abstract

Several instability events have occurred in the world in various commercial BWRs since the 80s. The dynamics of these events has been studied for many years. Basically, the power in normal operating conditions is a noisy signal with no significant frequencies. When instability takes place, oscillations around 0.5Hz and 1Hz become more relevant in the reactor dynamics behaviour due to the void fraction feedback. In the last years, time frequency analysis has been used by several authors to study instability events so as to isolate certain harmonics. Among them, we can cite wavelets, Short Time Fourier Transform (STFT), Hurst exponent, Hilbert Huang Transform (HHT), etc. The latter consists of decomposing the original signal into a subseries (Empirical Mode Decomposition, EMD). The Hilbert transform is then applied to each mode to obtain the instantaneous amplitude and frequency. However, when a frequency component within the signal comes into existence or disappears from it entirely at a particular time scale, the EMD does not work properly and one mode can present more than one frequency component (mode mixing problem). In this work, a modification of the HHT methodology has been applied to Local Power Range Monitors (LPRM) signals during an instability event. A different signal decomposition method is used, the Ensemble Empirical Mode Decomposition (EEMD), and compared with the previous EMD. The EEMD produce Intrinsic Mode Functions (IMF) with no frequency mixing problem. The frequencies and modes extracted this way describe the instability dynamics more accurately.