Cepstral Subtraction for Ultrasonic Signal Processing in Liquid-film Thickness Measurement under High-temperature and High-pressure Conditions of Boiling Water Reactors

Abstract

We have been developing the ultrasonic liquid film sensor to measure the liquid film thickness under high-temperature and high-pressure conditions of Boiling Water Reactors (BWRs), focusing on thin film thickness in pre-/post-dryout. The liquid film sensor utilizes Time-of-Flight (ToF) detection based on the ultrasonic pulse-echo method. To detect the ToF from output ultrasonic signal of the liquid film sensor, the waveform subtraction in the time domain has been used as conventional method. However, subtracting signals in the time domain has some concerns for the ToF detection under actual steam-water conditions; e.g. a calibration test is forced to be carefully conducted in the single-phase flow. In the present study, we improved signal-processing performance for the liquid film measurement using cepstral subtraction method. In this method, the ToF was directly detected from convoluted waveforms by calculating cepstrum from Fourier transform and logarithm spectrum. Moreover, avoiding noise effect on the cepstrum was achieved by subtracting mean cepstrum in the qufrency domain; because multiple waveforms were acquired from repetition pulsing of ultrasonic signals at the same experiment, the noise effect could be removed due to unsteady characteristics of liquid film flow. In this study, the basic principle of the cepstral subtraction method was confirmed by a calculation test using simulated ultrasonic signals. Assuming the additive white Gaussian noise with average noise power of -90dB, the maximum error between input thickness and output thickness was 21% in the range of liquid film thickness of 0.008-0.300mm. Furthermore, the time-series liquid-film thickness was measured from actual signals acquired under BWRs’ operating conditions at temperature of 286℃ and pressure of 7MPa. Comparing to the thickness processed from the conventional time-domain subtraction method, the processed results of cepstral subtraction method showed good agreement. Consequently, we confirmed appropriateness of the cepstral subtraction method for the liquid film sensor under high-temperature and high-pressure conditions.