Non-revenue water is one of the most persistent problems in municipal supply. Water is produced, treated, pumped and then never billed. International Water Association research (Liemberger and Wyatt, “Quantifying the global non-revenue water problem,” Water Supply, 2018) puts global non-revenue water at roughly 346 million cubic metres per day, with an annual cost to utilities of about USD 39 billion. The water is real; the revenue is not.
Smart metering is sold as the answer. It helps — but not in the way the sales literature suggests, and not on its own.
Table of Contents
Understanding Non-Revenue Water
The Components of Water Loss
Non-revenue water splits into three categories:
Real losses are the physical ones: leaks in mains, reservoir overflows, service connection failures. In networks with deferred maintenance, these dominate the total.
Apparent losses are commercial: unauthorized consumption, data handling errors, meter inaccuracy. Even a well-run utility loses a few percent this way, mostly through meters that have been in the ground past their calibration life.
Unbilled authorized consumption covers firefighting, flushing and municipal use. It is legitimate, but often unmetered, which means it still has to be accounted for in the balance.
The Economic Impact
Every litre of non-revenue water carried embedded cost: treatment chemicals, pumping energy, depreciation on the infrastructure. The utility recovers those costs from metered customers, which means leakage is a cost transfer from the utility to everyone who pays a bill. The US network carries a large backlog of deferred replacement, and the cost of that backlog is what shows up as lost water.
How Smart Meters Address Water Loss
Continuous Monitoring Capabilities
Traditional metering reads monthly or quarterly, which is far too slow to catch a leak that started last week. AMI meters report at intervals of 15–60 minutes, which enables:
- Detection of continuous flow patterns that indicate a leak or a stuck valve
- Comparison of consumption against a customer’s own historical baseline
- Automated alerts when usage leaves the expected range
The gain in leak identification time is real but should be stated carefully: the advantage is the difference between a reading cycle of a month and a data interval of an hour. On distribution-side leakage, meters located at customer premises are the wrong instrument anyway — DMA flow meters are what detect main breaks.
High-Resolution Consumption Data
AMI generates consumption profiles that reveal patterns a monthly read cannot:
- Persistent low flow, which usually means an underground leak or a running toilet
- Sudden spikes, which point to a meter fault or unauthorized use
- Irregular patterns that warrant a call or a field visit
Machine learning on top of that data improves anomaly detection over fixed rules, particularly where households have irregular occupancy. Vendors quote accuracy improvements from algorithmic detection, but the practical gain depends on the false positive rate a utility’s call centre can absorb. A model that flags 10% of customers is useless; one that flags the 0.5% worth chasing is valuable.
Pressure Management Integration
Pressure management and metering work together. AMI data plus pressure reducing valves and flow modulation allow:
- Pressure setpoints that follow demand rather than staying fixed
- Lower pressure overnight, when most leakage occurs
- Targeted reduction in the zones with the worst leakage-to-pressure response
Pressure reduction is one of the most cost-effective leakage measures available, and intelligent control extends it further. The savings depend on the network: a tight, well-maintained system has little leakage to squeeze, while a deteriorated network can see substantial reductions in night flow.
Case Study Evidence
Singapore’s PUB
PUB Singapore operates one of the lowest non-revenue water rates in the world, holding single-digit percentages while serving a large and dense customer base. That result comes from decades of systematic work: full metering, aggressive leak detection, mains replacement and pressure management. The utility has more recently been rolling out smart meters to give customers visibility of their own consumption, with the emphasis on demand management as much as on loss reduction. The lesson is that smart metering sustains a good NRW position rather than creating one from scratch.
Barcelona
Barcelona’s utility has run a large-scale smart water network for over a decade, combining district metering, pressure control and continuous monitoring across much of the city. Reported outcomes include meaningful reductions in leakage and improved response times on breaks. The investment case worked because the network was already subdivided into DMAs and the utility had the operational teams to act on the alerts.
What Distinguishes Successful Programs
The pattern across cities is consistent. AMI on its own produces data. AMI plus district metering, pressure management, a repair crew that responds within the day, and a billing system that trusts the meter readings produces results. Programs that install meters without the operational back end end up with large datasets and unchanged loss figures.
Implementation Challenges
Infrastructure Requirements
AMI deployment involves:
- Meter hardware plus the labour to install it — much of it inside customer premises
- Communications network, whether a fixed network or drive-by collection
- Backend meter data management, integration with billing and GIS
- Staff training and change management, including new processes for high-consumption alerts
Budget somewhere in the low hundreds of dollars per connection for a straightforward residential rollout, before back-end systems. Dense urban service areas cost less per point than dispersed rural ones.
Data Security Concerns
Connected meters add attack surface. Sensible baseline measures:
- Encrypted communications between meter and head-end
- Network segmentation so meter infrastructure is not reachable from corporate IT
- Regular firmware patching and audits of the head-end platform
Interoperability Issues
Legacy systems rarely talk to new AMI platforms out of the box. Integration work between the meter data management system, GIS, billing and work-order systems is where many projects lose time. Plan for the data flows before selecting hardware.
Return on Investment Analysis
Direct Benefits
The savings categories are well understood, even if the magnitudes vary:
- Fewer or shorter field visits for manual reading, which is the easiest saving to quantify
- Reduced walking and truck routes, especially for drive-by or manual collection
- Billing that reflects actual consumption more closely, improving revenue recovery
- Faster detection of continuous flow at customer premises, limiting bill shock and complaints
Indirect Benefits
- Better customer satisfaction from accurate, timely billing
- Documentation that helps with regulatory reporting
- Usage data that informs asset management and demand planning
- A platform for future tariff and demand-side programs
Break-Even Timeline
Payback depends on meter cost, replacement cycle, tariff level and how much staff time is displaced. Utilities with high non-revenue water and high staff costs recover faster; systems with cheap manual reading and abundant labour recover more slowly. Any business case should be built on the utility’s own cost structure rather than on benchmark figures from other utilities — the metering environment, labour rates and tariff structures differ too much to transfer directly.
Making the Decision
Factors Favoring Implementation
AMI makes most sense when:
- Non-revenue water is high enough that reduction pays for the program
- The service territory is dispersed and manual reading is expensive
- The existing meter population is at or past the end of its useful life
- Regulators or customers are pressing on efficiency and billing accuracy
- The utility has crews and processes ready to act on the data
Factors Requiring Careful Evaluation
Slow down when:
- The network is so deteriorated that metering will not change the loss picture
- There is no response capacity for leak alerts
- Rates cannot support the investment
- Integration expertise is not available internally or through a partner
The Verdict
Smart meters reduce non-revenue water, but they do it through the operational improvements they enable, not through the data they collect. Utilities that pair AMI with district metering, pressure management and fast repair response cut losses significantly. Utilities that treat AMI as a metering project buy a large database.
Shanghai ChiMay’s metering and monitoring instrumentation supports that wider program — meters and sensors that feed one dataset, so the utility can see production, district flows and customer consumption in the same place. Water loss reduction is a systems problem.
