Water Utilities Shift to AI for Leak Detection and Efficiency Gains

Water utilities have traditionally used technicians with listening sticks to detect leaks by prodding the ground and listening for sounds from underground pipes. Now, companies are turning to artificial intelligence to identify leaks, optimize energy use, and reduce sewage outflows.

Countries such as China, Japan, and Singapore lead in this area, according to Asit Biswas, a water management authority who has advised 23 countries. In Singapore, municipal authorities collect several terabytes of data daily using AI to anticipate and fix leaks.

Singapore and Japan Singapore’s National Water

Agency appointed a chief information officer more than a decade ago to standardize data collection across its network. This approach allows identification of weak sections in water, sewerage, and drainage pipes for preventive maintenance, with 2 percent of pipes replaced annually, Biswas stated.

In Japan, AI has contributed to low leakage rates, with Tokyo and Osaka losing about 3.8 percent of treated water. Singapore reports around 5 percent loss, and AI has enabled utilities in these areas to reduce staff numbers, according to Biswas. Leakage rates in these countries are 75 percent lower than in England and Wales, where about 20 percent of treated water is lost, leading to hosepipe bans in 2025 and concerns over shortages.

Utilities in England and Wales are starting to adopt AI and related technologies. Northumbrian Water, serving areas including Newcastle and Durham, uses technology from Origin Tech, a start-up founded in 2019 by John Marsden and Phil Surtees, former oil and gas engineers.

Origin Tech employs satellite radar signals to detect leaks, noting differences in signal bounce from wet versus dry ground, along with subsidence and overgrowth indicators. AI combines this data for 80 percent accuracy in leak detection, and repairs involve injecting a non-toxic sealing substance without excavation, the company stated.

Gregg Herrin, vice-president of water at Bentley Systems, an infrastructure software firm, said AI allows utilities to plan maintenance proactively rather than reacting to failures. He added that it provides continuous automated monitoring beyond manual processes.

Water uses AI with weather forecasting and 750 in-pipe sensors to predict sewer capacity and divert waste to holding tanks, potentially preventing 80 percent of storm overflows. Southern Water, serving south-east England, has installed 34,000 radar sensors under manholes to monitor inflows and alert staff to prevent flooding.

This system has prevented 4,000 sewage flooding incidents in homes over the past year, Southern Water reported. Yorkshire Water is collaborating with Socotec and UnifAI Technology to develop AI models for predicting water quality at 20 inland bathing sites based on samples taken every four weeks.

Faye Cossins, coastal delivery and engagement manager at Yorkshire Water, stated the project aims to reduce the time and cost of bacteria monitoring. Several utilities, including South East Water, which experienced outages affecting tens of thousands of customers in recent months, still use listening sticks, and South East Water declined to comment.