Hydroinformatics based technique for leak identification purpose – An experimental analysis

Abstract

Analysis of single point pressure transient signal has the potential of providing information on the pipeline system. The technique uses the frequency response and experiments are satisfactory for low-frequency bandwidth. The concept behind this approach is that reflections are detected by leak-induced disturbances. Such reflections are often very difficult to identify, particularly because of excessive noise. The common approach to boost the analysis is to use a data denoising approach. The wavelet transforms for leak detection in water pipelines are proposed in this article. The wavelet filter is optimally calibrated using maximal kurtosis values and is chosen for leakage detection. The suggested approach is tested using high-noise laboratory test signals. The findings show that the method is superior to current signal processing methods for the conditions used. The system permits not only the detection of leakage but also the location and evaluation of its magnitude with errors of less than 7%.