Abstract Summary
Leaks in water distribution systems are a major problem as a large amount of water is wasted between treatment plants and consumers. Water supply companies use different strategies to identify and locate leaks in their pipe networks. Among these are vibro-acoustic methods and devices (i.e. listening sticks, geophones, and leak noise correlators). Although listening devices are cheap and simple to use, they rely on the operator's hearing ability to locate leaks from single point measurements. Moreover, this may lead to a search based on trial and error as only vibration intensity is used, so no information is obtained regarding the direction of the leak from the measurement position. Leak noise correlators, however, use two sensors positioned on the pipe or at access points, and the leak can be located by estimating the time of flight between the two measured signals and the speed at which the leak noise propagates. This paper concerns an investigation of a new technique where leak noise signals measured on the ground surface using a camera (non-contact measurements) are pre-processed using computer vision techniques to extract meaningful information concerning the location of the leak. A preliminary study is reported in which measurements are made on a test bench that simulates the vibration response due to a buried pipe at two positions. Tests are conducted to evaluate the camera sensitivity for leak noise measurements regarding the distance from the vibration source, lighting, contrast intensity, and the relation between image resolution and leak detection are performed. The obtained results are compared to classic contact measurements using accelerometers showing that the results from the proposed technique are promising.
