Listening for Leaks: How AI and Acoustic Sensors Save Water
Image Credit: Jacky Lee
A new industry briefing published on 16 January 2026 highlights a shift in smart infrastructure: water utilities are increasingly pairing acoustic sensors with AI analytics to find hidden leaks earlier and plan maintenance before failures disrupt streets and supply. The approach is being framed less as experimentation and more as an operational response to ageing networks, rising water losses, and tighter maintenance budgets.
Why Leaks Matter in Smart Infrastructure
In water networks, “non revenue water” refers to water put into the system that does not generate revenue, including real losses from leaks as well as metering issues and other losses. Cutting these losses matters for costs, reliability, and resilience, particularly during dry periods when utilities are under pressure to stretch existing supplies. Sydney Water’s public reporting shows non revenue water at 63,460 ML in 2023 to 24, with network leakage (real losses) reported at 48,446 ML for the same rolling water balance table. The report also notes uncertainty in leakage estimates because leakage is calculated by deduction from system inputs and measured usage.
How Acoustic Sensing and AI Fit Together
Acoustic monitoring aims to pick up the sound signatures created when pressurised water escapes through cracks or faulty fittings. In practice, utilities still need analytics to separate leak noise from everyday background sound and to prioritise which alerts justify field crews. Sydney Water’s 2022 to 23 water conservation reporting describes “lift and shift” and semi permanent acoustic sensors used in high density areas, noting that when combined with machine learning, a leak’s soundwave signature can be learned over time so sensors can raise alarms rather than relying only on manual field surveys. The same report says trials indicated acoustic sensors could detect leaks as low as 0.01 L/s.
Sydney Water and UTS: What the Public Record Shows
Sydney Water’s work with the University of Technology Sydney has been widely referenced as an Australian example of acoustic sensing plus analytics. A 2022 report on the CBD trial described 600 acoustic sensors deployed across 13 km of CBD water mains, reporting 9,000 ML saved over two years, valued at $20 million, and linking the rollout to a purpose built portal consolidating acoustic data from multiple sensor manufacturers. That same reporting attributed to Sydney Water the discovery of 160 hidden leaks, including some estimated to have been active for years.
UTS separately reports that acoustic sensors have saved 10,000 ML and $23 million in supply and repair costs, and describes a long running collaboration with Sydney Water dating back to 2009. UTS also reports an AI tool that predicts where a water main is likely to fail with 80 per cent accuracy, positioning it as part of a broader asset management approach.
From Alerts to Prediction: the Operational Change
The main operational promise is moving from “find and fix” to “predict and plan”. Work published through the Australian Water Association’s Water e journal describes how acoustic sensor deployment was tied to prioritised zones and follow up workflows, including review of sensor outputs and issuing work orders for investigation and repair. That paper also describes the need to consolidate multi vendor sensor data, combine it with other network information, and feed it into machine learning systems so confidence improves as the dataset grows.
What Can Slow Adoption
The same sources also point to practical barriers that can blunt the benefits if not managed well. Acoustic accuracy can be affected by external noise, sensor differences, and the need for careful calibration. Governance is also a recurring theme: utilities need clear processes for validating alerts, retraining models, and ensuring staff trust the system enough to change how maintenance is scheduled. Sydney Water’s own reporting on leakage also underscores that even measuring leakage at system level carries uncertainty, which can complicate performance tracking and business cases if not explained well.
How This Compares with Other Leak Detection Paths
Acoustic sensing is not the only analytics led approach in play. Sydney Water has also been profiled using central event management style analytics that aggregate multiple data streams to detect anomalies in near real time, then trigger targeted investigations using traditional tools like acoustic listening equipment. Separately, SA Water has publicly reported that sensors implemented in its Adelaide CBD helped detect around half of water main leaks and breaks, supporting proactive repair. In regional NSW, the Rous County Council has described acoustic leak detection delivered under the NSW Government’s Water Efficiency and Regional Leakage Reduction Program, reporting cumulative statewide survey and repair figures published by the council.
What to Watch Next
The near term direction is integration and scale: more utilities are trying to combine sensor feeds, network telemetry, and asset data so leak detection becomes part of routine operations rather than a standalone pilot. Sydney Water’s current five year plan explicitly flags expansion of acoustic sensors and other IoT approaches as part of improving proactive leak detection. The credibility test for the sector in 2026 will be whether these systems keep improving in real deployments, with clear validation, transparent reporting, and measurable reductions in avoidable losses.
TheDayAfterAI News
We are a leading AI-focused digital news platform, combining AI-generated reporting with human editorial oversight. By aggregating and synthesizing the latest developments in AI — spanning innovation, technology, ethics, policy and business — we deliver timely, accurate and thought-provoking content.