Leak Detection Algorithm for Pipelines in Noisy Environments

Many methods have been proposed for leak detection in pipelines. The method based on the acoustic signals created by the leakage is a non-destructive method and has presented reliable results. However, applying such a method in noisy environments is still a challenge. In this paper, a leak detection method applied in pipes which are located in a highly noise environment of the process area of an oil refinery is introduced. The proposed detection algorithm takes into account the frequency contents of the noise and the signals of the leaks as well as some statistical characteristics of the noise. Processing of measurements in time windows of 1s are found enough for decision making, keeping the response time of leakage detection very low. Results have shown reliable detection of leaks in sufficient distances without false alarms while there is the ability for the discrimination of some accidental or periodic events from defining them as leaks. The limited processing and storage requirements of the algorithm make it appropriate for efficient implementation.