Automated detection of compressed air leaks using a scanning ultrasonic sensor system

Compressed air is a costly but useful utility in production facilities. However, losses up to 50 percent are reported due to leakage. In this paper a system for detecting compressed air leaks using its air-coupled acoustic emissions is presented. An ultrasonic microphone is combined with a pan-tilt unit to automatically scan an area. The directivity of the microphone is enhanced using a parabolic mirror or an acoustic horn. The acquired data is used to detect a leak. The influence of the two different microphone setups and several parameters, such as distance, leak size, and obstacles on the success of leak detection are investigated. A detection threshold based on the sound pressure level and the signal-to-noise ratio is determined. The experiments show the feasibility of this method.