Ordinarily, city officials only know about leaks in municipal water pipes once those leaks have become quite large and troublesome. A new artificial intelligence-based system, however, could catch such leaks much earlier – by listening for them.

Currently under development at Canada's University of Waterloo, the system incorporates inexpensive hydrophone-equipped sensors which are installed within existing fire hydrants. This is done without taking the hydrants out of service, requiring no excavation nor affecting their functionality.

The hydrophones extend down into the city's water-supply pipes, where they monitor the sound of the flowing water. Utilizing machine learning algorithms, the sensors are able to detect the telltale acoustic signature that's created by even very small amounts of water escaping from those pipes.

In a lab setup consisting of PVC pipes and a fire hydrant, the technology has successfully detected leaks as small as 17 liters (4.5 US gal) per minute. Presently, leaks of this size can remain undetected for years, resulting in huge cumulative losses of water. Additionally, they may allow contaminants to enter the water supply, plus the leaked water can damage the foundations of buildings and other structures.

The research team is now field-testing the technology, and is working on methods of determining the exact location of leaks.

"This would allow cities to use their resources for maintenance and repairs much more effectively," says lead researcher Roya Cody, a civil engineering PhD candidate. "They could be more proactive as opposed to reactive."

A paper on the research was recently published in the Urban Water Journal.