Water Leak Detection: Finding and Fixing Non-Revenue Water Losses
Somewhere under nearly every Indian city, treated water is escaping into the ground before it ever reaches a tap. Utilities call this non-revenue water, or NRW, and by most estimates, Indian cities lose somewhere between 35 and 40 percent of the water they treat and pump. Some systems run considerably higher: Delhi's losses have been estimated at close to 50 percent, while Hyderabad and Bengaluru typically report figures closer to 35 to 40 percent. For context, the Central Public Health and Environmental Engineering Organisation (CPHEEO) recommends that urban utilities keep NRW under 15 percent of the water they supply, which gives a sense of how much ground the sector still has to cover.
Leak detection is where closing that gap actually begins. Because most water infrastructure runs underground, a pipeline can leak for months, sometimes years, before anyone notices, usually only once the ground above it starts to sink or a road begins to buckle. This piece walks through how leak detection actually works: the acoustic methods utilities have relied on for decades, the pressure-based techniques that narrow down where to look, the sensor and AI-driven tools reshaping the field, and where non-revenue water management fits into the wider picture of water security.
What Counts as Non-Revenue Water
Water engineers generally sort NRW into three categories, a framework known internationally as the water balance methodology developed by the International Water Association (IWA). Physical losses are water that literally escapes through cracked pipes, failed joints, and worn-out fittings before it reaches a customer. Apparent losses come from meters that under-register consumption, unauthorised connections, or billing errors that make real usage disappear from a utility's books. Unbilled authorised consumption covers water a city legitimately supplies but doesn't charge for, such as firefighting, public standposts, or its own parks and municipal buildings.
The distinction matters because each category needs a different fix. Physical losses call for leak detection and pipe rehabilitation. Apparent losses call for meter upgrades and better data systems. Unbilled consumption is more a policy and accounting question than an engineering one.
Physical losses in particular have a way of compounding. A leak that goes unrepaired for a year doesn't just waste one year's worth of water; it keeps wasting water every single day until someone finds and fixes it, which is why even a modest, slow leak can add up to an enormous cumulative loss. There's a public health dimension too: wherever water can escape a pipe, contaminants can just as easily be drawn back in, particularly during the low-pressure periods common in intermittent supply systems. That's one reason leak detection isn't purely a cost-recovery exercise; it's also a water-quality safeguard.
How Leak Detection Works: Acoustic Methods
Acoustic detection remains the backbone of most leak-finding programmes. The logic is simple: pressurised water escaping through a crack or joint makes a distinct hissing or rushing sound, and that sound doesn't stay put. It travels along the pipe wall and through the water inside it, often carrying much farther than it would through the loose backfill soil surrounding the pipe, which is a comparatively poor conductor of sound. In a rough sense, a pipeline behaves the way a stretched string does on an instrument: its length, diameter, and material all shape how far and how clearly a sound travels along it.
How far exactly depends on conditions, and this is worth being precise about rather than quoting a single fixed number. Sound generally carries farther through metal pipes under good pressure, and considerably less far through plastic pipes such as PVC, where field surveys often work in ranges of a few dozen to a few hundred metres rather than the longer distances possible on well-pressurised metal mains. Soil type, pipe age, and background noise from traffic or nearby equipment all affect the result, so engineers treat these figures as planning ranges rather than constants.
The equipment used to listen for that sound comes in a few forms. Mobile ground microphones are carried or wheeled along a pipe route by a technician, with the signal growing louder as the operator nears the leak's location. Permanently mounted sensors, by contrast, sit at fixed points along the network and continuously monitor for leak sounds without needing a person to walk the route. A trained crew using ground microphones can typically pinpoint a leak's location to within about a metre, which is precise enough to guide excavation without digging up more road than necessary. The main limitation is environmental: thick pavement, deep burial, or heavy traffic noise can all muffle the signal at the surface, which is why acoustic surveys often work best at night or in quieter stretches of the network.
Narrowing the Search: Pressure and Flow-Based Methods
Acoustic surveys work best once a search area has already been narrowed down, and that's where pressure and flow monitoring come in. Utilities place flow meters and pressure gauges at strategic points along a pipeline. If the flow rate measured at the start of a segment is noticeably higher than what's measured at its end, or if pressure has dropped compared to the system's normal operating baseline, that's a strong signal that a leak sits somewhere between the two measurement points.
Used together, the two methods complement each other well. Pressure and flow data bracket a rough stretch of pipe where a leak is likely, and acoustic surveys then pinpoint the exact spot within that stretch. Doing it this way saves considerable time compared to walking an acoustic survey along an entire pipeline network with no starting point.
Beyond the direct cost of lost water, leaking infrastructure carries costs that are easy to underestimate. Every litre lost after treatment has already consumed energy to pump and, often, to treat, so leaks translate directly into wasted electricity as well as wasted water. Aging, leak-prone infrastructure also tends to bring the surrounding infrastructure down with it over time: saturated soil shifts, pavement above buried pipes buckles, and nearby structural foundations can be affected as the ground beneath them settles unevenly.
Emerging Technologies for Network-Level Monitoring
The acoustic and pressure-based methods above have been the industry standard for decades, but the last several years have brought genuinely new tools into use at the network level, particularly for utilities managing large, sprawling distribution systems.
AI and machine learning on SCADA data. Utilities that already collect continuous flow and pressure readings through SCADA systems can run that data through anomaly-detection models trained to flag the kind of pattern that typically precedes or accompanies a leak, often catching problems earlier than a scheduled acoustic survey would.
Smart meters and AMI. Advanced metering infrastructure doesn't just automate billing. Many modern smart meters carry built-in acoustic or pressure sensors and can flag unusual overnight consumption patterns, a classic sign of a leak on the customer side of the connection, well before a bill would ever reveal it.
Also read: How AMI Technology Is Evolving for Modern Water Utilities in 2026
Fibre-optic distributed acoustic sensing. Where a fibre-optic cable already runs alongside a pipeline, it can double as a continuous acoustic sensor along its entire length, picking up leak signatures anywhere on the route without needing separate point sensors installed at intervals.
Satellite and remote-sensing methods. Techniques that detect soil moisture or ground movement anomalies from above are being piloted in some markets to flag likely leak zones before a crew is ever sent out. In India, this remains a more niche and costly option than the methods above, and it generally works best as a first-pass screening tool over long transmission mains rather than a replacement for ground-level acoustic surveys.
None of these tools eliminate the need for a trained crew with acoustic equipment to do the final pinpointing. What they do is help utilities decide where to send that crew first, which matters a great deal when a network runs into hundreds of kilometres of ageing pipe.
Facility-Level Leak Detection
Everything covered so far deals with leaks in the public distribution network, the pipes a utility owns and operates. Leak detection inside a customer's own facility, a factory, a large housing complex, a hospital, is a related but genuinely different problem, closer to a matter of continuous vigilance than of specialised field survey work.
The traditional tool here is the spot detector, a sensor placed at a single point, such as a low spot beneath a piece of equipment, that triggers an alert once water reaches it. Spot detectors are inexpensive and simple to install, but their coverage is limited by design: water that never physically touches the sensor's probes goes undetected. Non-conductive sensing cable improves on this by running continuously along a route rather than sitting at one fixed point, which means it can catch a leak anywhere along its length rather than only where a puddle happens to form, and because it's non-conductive, it won't short out if it brushes against metal surfaces or building structures.
When evaluating any new facility-level sensing technology, a few questions are worth asking before adopting it: how many different situations can it actually be used in, can its sensitivity be adjusted for different liquid volumes, how quickly can it be reset after a trigger, how easy is it to install, will it scale as the facility grows, and how easily does it integrate with existing building management or control systems. Technology that struggles on more than one or two of these fronts tends to see limited adoption regardless of how well it performs on paper.
How SPML Approaches Non-Revenue Water Management
Reducing NRW at scale takes more than any single tool. SPML Infra's water business covers the full cycle, from treatment and transmission through to water loss and NRW management, and its NRW work leans on the same layered approach described above: pressure and flow monitoring through SCADA and instrumentation systems, trenchless rehabilitation and automated valves to fix what's found, and remote sensing to keep watching a network after repairs are complete.
That approach extends into how SPML manages metering and billing as well, through its SPMLAQUA platform for utility asset and network management. One example is the Delhi Jal Board's South Zone smart metering pilot in the Mehrauli to Vasant Vihar area, a 12-year programme covering roughly 90,000 smart water meters, built to support 24x7 supply with minimal losses across the zone. Projects like this reflect the same idea that runs through this article: leak detection only pays off when it's paired with the metering, monitoring, and rehabilitation work needed to act on what it finds.
The Bottom Line
Water leak detection isn't a single technology so much as a layered process: pressure and flow data to narrow the search, acoustic surveys to pinpoint the exact spot, and increasingly, AI-driven monitoring and smart metering to catch problems earlier and prioritise where crews go first. For a country where NRW routinely runs above 35 percent in urban systems, getting this process right isn't just a technical exercise. It's one of the more direct ways a city can stretch its existing water supply further without building a single new source.
Frequently Asked Questions
What is non-revenue water (NRW)?
NRW is treated water that a utility supplies but doesn't collect revenue for, whether because it physically leaked out of the network, was lost to meter errors and unauthorised connections, or was legitimately used without being billed, such as for firefighting.
How much water do Indian cities typically lose to NRW?
Estimates commonly put the national average around 35 to 40 percent, with some cities running higher. CPHEEO recommends utilities aim to keep NRW under 15 percent.
How accurate is acoustic leak detection?
A trained crew using ground microphones can typically pinpoint a leak's location to within about a metre, though the achievable accuracy depends on pipe material, depth, and surrounding noise conditions.
What are the warning signs of an underground water leak?
Unexplained wet patches or ponding at the surface, a drop in system pressure, unusually high flow readings with no matching increase in billed consumption, and localised pavement or ground settlement are all common indicators.
How often should a utility run a leak detection survey?
This depends on pipe age, material, and NRW history, but many utilities schedule acoustic surveys on a rolling basis across their network annually, supplemented by continuous SCADA or AMI monitoring where it's available, rather than relying on a single yearly sweep.
Can smart water meters detect leaks on their own?
Many modern smart meters can flag unusual overnight consumption patterns that suggest a leak, which is useful for catching customer-side leaks early, though they generally work best alongside, not instead of, acoustic and pressure-based methods for the main distribution network.