Detection of an acoustic source inside a pipe using optimized vibroacoustic beamforming

Abstract : In an intent to improve the monitoring of steam generators, a technique based on vibration measurements is developed for the detection of a water leak into sodium. Background noise can mask the leak-induced vibrations. In order to increase the signal-to-noise ratio (SNR), a beamforming technique may be considered. In the purpose of studying the feasibility and the efficiency of this technique for the present configuration, experimental investigations have been performed on a mock-up composed by a straight cylindrical pipe coupled to a hydraulic circuit through two flanges. A sound emitter introduced in the pipe simulates the source to detect, whereas a varying flow speed controls the background noise vibrations. Beamforming is applied on the signals measured by an array of accelerometers externally mounted on the pipe. Two different kinds of beamforming are considered the conventional (Bartlett) one and a statistically optimized one based on SNR maximization. After a brief presentation of the mock-up's vibroacoustic characteristics, we study the efficiency of the two beamforming treatments for narrowband and broadband analysis.