Autocorrelation Analysis of Vibro-Acoustic Signals Measured in a Test Field for Water Leak Detection

Abstract

Reducing losses in water distribution networks is a worldwide challenge and all utilities are developing proper strategies for the active control of leaks. Temporary or permanent grids of units for the continuous monitoring of pipelines through vibro-acoustic measurements are probably the most commonly adopted leak detection systems. Such systems generally rely on the definition of proper thresholds to detect increments in the vibration levels associated with leaks. Since the thresholds are strongly dependent on the local boundary conditions of the monitored network, the initial setup is costly and time consuming, and the risk of undetected leaks or false alarms increases. This work aims to investigate leak detection methods based on the inherent properties of the measured signals instead of their relative amplitude. In particular, the possibility of detecting water leaks in small-diameter plastic pipes by analyzing the autocorrelation function of vibro-acoustic signals is assessed. An experimental campaign is conducted in a full-scale test facility that simulates the actual network. The measurements concerning artificially generated leaks are attained by two accelerometers and one hydrophone. The experimental results confirm the effectiveness of the proposed approach, which is therefore proven as a promising tool for leak detection.