Artificial intelligence used to detect underground pipe leaks in Buzău

The regional water utility in Buzău County has reduced network water loss using AI-supported noise sensors. The system monitors the sounds and vibrations produced by water escaping from pressurized pipes, then uses the data to identify areas where a hidden pipe break or leak is likely located.

It is worth clarifying the 13 percent figure mentioned in the headline. Water loss did not decrease by 13 percent; rather, it dropped from the 22–23 percent level measured in autumn 2025 to about 13 percent by August 2026. This represents a decrease of 9–10 percentage points, or roughly a 40 percent proportional improvement. The company originally set a target of bringing losses below 20 percent by 2030, meaning that, according to its own data, it has already significantly exceeded that goal. Water escaping from pipe breaks produces characteristic vibrations that travel along the pipeline. So-called noise loggers installed at designated points continuously record these signals, while the software filters out the acoustic patterns that may indicate a leak. This means the technical team no longer has to inspect the entire network street by street and connection by connection; instead, they can target the suspect section, pinpoint the location of the fault, and begin repairs.

The sensors are being installed primarily on aging pipe sections that fail more frequently. The program currently covers the city of Buzău, Vernești, the Slănic Valley, the town of Pătârlagele, and Râmnicu Sărat. The utility plans to install more than 3,500 noise loggers across the county network; 150 sensors each have already been installed in the Slănic Valley and in Râmnicu Sărat, while the first ten devices are operating in Pătârlagele. Reducing water loss matters not only for conserving resources. Every cubic meter of water that leaks out had first to be extracted, treated, pumped, and delivered into the network, meaning that leaks consume not only water but also energy and money. Faster fault detection can therefore simultaneously reduce the utility's costs, ease the strain on the network, and lower the risk of supply disruptions.

Source: G4Media