In many of the world’s cities, a quarter or more of the treated water supply never reaches a tap: it leaks into the ground from ageing pipes, unbilled and unseen. Finding those leaks used to mean walking streets with listening sticks. It now involves satellites, acoustic sensors and machine learning, and the economics are compelling enough to embarrass every other supply option.
The scale of the loss
Non-revenue water, the industry’s term for water produced but never paid for, combines leakage with theft and metering failures, and in poorly maintained systems it can exceed half the supply. Even well-run networks in rich countries commonly lose a fifth. Every lost litre was abstracted, treated and pumped at real cost in energy and chemicals.
The leaks also damage the network itself: escaping water erodes pipe bedding and road foundations, and the pressure losses make contamination more likely. A leak is never just a leak.
Listening from space and street
Satellites can now detect treated water underground by sensing the soil moisture signature around a leak, scanning whole cities and flagging suspect zones for ground crews. Within those zones, acoustic loggers clamped to pipes listen at night, when the network is quiet, and triangulate the hiss of escaping water with a precision that surprises the engineers who once did it by ear.
Machine learning binds the system together: models trained on a network’s history predict which pipes are most likely to fail next, so replacement budgets target the worst risks rather than the loudest complaints or the oldest streets.
Why utilities are slow anyway
If the returns are so clear, the lag is organisational. Leakage reduction competes for capital with visible projects; its benefits arrive as absence, water not lost, bills not raised, which politicians find hard to inaugurate. Regulatory regimes that reward new assets over maintenance deepen the bias.
Data quality is the other brake. A utility that does not know where its pipes are, a common situation with century-old networks, must map before it can model. The mapping, by ground-penetrating radar and patient archive work, is unglamorous and indispensable.
For more coverage of the stories shaping the day, visit our homepage and the latest reporting in News.
The repairs themselves are changing
Finding leaks faster only helps if fixing them gets faster too. Trenchless repair techniques, lining pipes from inside, inserting new pipes through old ones, have cut the cost and disruption of renewal dramatically, and robotic inspection crawlers now grade pipe interiors continuously rather than at excavation.
Some utilities have restructured around the new economics entirely, creating leakage teams whose budgets are justified like supply projects: cost per litre saved against cost per litre from any alternative source. On that comparison, leakage work wins so consistently that its historical neglect looks like an accounting illusion.
What customers notice
For households the benefits arrive as resilience: fewer burst-main disruptions, steadier pressure, and bills that rise more slowly than they would if losses were met by building new reservoirs. Drought orders also bite less hard on systems that are not haemorrhaging their supply into the soil.
The deepest change may be cultural. Water has been priced and managed as if it were abundant in places where it no longer is. The leak hunters, with their satellites and algorithms, are in the end making an argument about value: that the water already found, treated and pumped deserves more respect than the next dam.
Further reading: BBC News and The Guardian.
Every era of water engineering has its emblematic project. Ours may turn out to be invisible: a network that keeps what it carries. The cheapest, cleanest litre of water is the one that never needed producing, and an industry is forming to deliver it by subtraction.
