Analysis of acoustic signals for leak detection in water distribution networks




Shrivastava, Runal

Journal Title

Journal ISSN

Volume Title



Acoustic leak detection methods have proven effective in identifying leaks before they can cause interruptions in water supply and financial loss. However, most studies in this area have relied on laboratory experiments, and there is a need to expand the use of acoustic-based leak detection methods for real-time monitoring in the field. This study focuses on using hydrophones for leak detection in four different parts of a water network. The aim is to identify features that can differentiate between simulations of leaks and normal flow conditions and to assess the impact of network and leak characteristics on these features. Data was obtained from leaks simulated in the field, and the signals were analyzed using continuous wavelet transform and power spectral density. The results showed that acoustic signals from sites with cast iron pipes exhibit a higher power value in the frequency range of 200 to 400 Hz during most leak tests. Factors such as the distance of leaks from the sensors and network topology affect the magnitude of power for this frequency range, thus making detection more challenging. This frequency band can be used to establish a historical baseline and differentiate normal and abnormal conditions, thus, facilitating leak detection. The study concludes that acoustic-based leak detection methods have the potential to detect leaks in cast iron pipe networks. However, further research is necessary to tackle challenges present in real water distribution networks, including background noise, changes in pipe properties, and complicated network topology.


LCSH Subject Headings