What you will learn
- How acoustic sensors, pressure loggers, and district metering each find leaks
- How night-flow analysis separates leaks from legitimate use
- What a leak alert needs to send a crew to the right street
- How detection data feeds repair dispatch and records
- When leak detection pays and when it does not
What the project is
A distribution network loses water through leaks at joints, valves, and service connections, most of them invisible from the surface. Detection uses three overlapping methods. Acoustic sensors listen for the sound of water escaping under pressure. Pressure and flow loggers watch for patterns that indicate a break. District metered areas divide the network into zones, so the water entering a zone can be compared with the water billed in it, and a growing gap points to a leak in that zone.
The software combines those signals into prioritised alerts. A strong acoustic signal on a main road, a zone whose night flow doubled, and a pressure drop in the same area together make a confident repair job. Any one signal alone is a guess. The product's value is in combining signals, ranking leaks by estimated loss, and getting a crew to the right location with the right equipment.
Core features
Ranking is the feature. A utility with fifty possible leaks and five crews needs to know which five to fix first, and why.
- Sensor data ingestion from acoustic loggers, pressure sensors, and flow meters, with health monitoring for dead or drifting sensors.
- District metered area balances: inflow versus billed consumption per zone, tracked over time.
- Night-flow analysis that separates background leakage from bursts, using the hours when legitimate use is lowest.
- Leak ranking by estimated loss rate, location confidence, and risk to roads and property.
- Work order creation from alerts, with location, suspected pipe, and equipment notes for the repair crew.
- Repair records closing the loop: leak found, pipe repaired, and the zone balance re-measured afterwards.
- Maps showing sensors, zones, active leaks, and repair history, so patterns across the network become visible.
How it is built
Pilot in one district before buying sensors network-wide. Sensor behaviour on real pipes, with real background noise, is what the datasheet cannot tell you.
- Network study: map zones, pipe materials and ages, and existing meter coverage, and pick a pilot district with known loss.
- Sensor deployment: install acoustic and pressure logging in the pilot district, and establish the baseline night flow.
- Detection platform: ingest sensor data, build zone balances and night-flow analysis, and rank the first alerts.
- Repair loop: send crews to ranked alerts, record outcomes, and measure how many alerts were real leaks.
- Expansion: roll out district by district, using the pilot's hit rate and water saved to justify each next district.
Tech considerations
Signal processing and honest data handling are the technical core. The dashboard is straightforward once the signals are trustworthy.
- Sensor ingestion with time-series storage, because leak analysis is about patterns over weeks, not single readings.
- Acoustic analysis that distinguishes leak noise from traffic, pumps, and other network sounds, with human review for borderline cases.
- GIS integration so sensors, pipes, zones, and leaks share one map and one location language.
- Work order integration with the utility's maintenance system, so an alert becomes a job without retyping.
- Alert tuning controls, because thresholds that cry wolf will be ignored, and thresholds set too high miss real leaks.
Challenges to plan for
Leak detection fails on false alarms and on repairs that never happen. Detection without a repair loop is just an expensive map of problems.
- Background noise: traffic, pumps, and industrial use create acoustic signals that look like leaks to naive analysis.
- Sensor maintenance: loggers in chambers flood, batteries die, and a dead sensor looks like a quiet zone.
- Repair backlog: detecting fifty leaks when crews can fix five a week creates a list that demoralises rather than motivates.
- Private-side leaks: a leak past the meter is the customer's, not the utility's, and the system must separate the two or it creates disputes.
Who it is for, and honest limits
This fits utilities with high non-revenue water, ageing pipe networks, and districts where losses are concentrated. It fits less well for a new network with plastic pipes and low losses, where the same money buys more by replacing the oldest mains directly.
It does not repair pipes. It finds leaks and ranks them. If the utility cannot fund the repairs the system finds, detection will document a growing problem without fixing it, which is still useful for a funding case, but not a fix in itself.
Frequently asked questions
How accurate is acoustic leak detection?
On metal pipes with good sensor placement, acoustic methods locate leaks well. On plastic pipes, sound travels less far, so sensor spacing must be tighter and confidence lower. Always confirm with a crew on site before digging.
What is a district metered area?
A zone of the network with a metered inflow, small enough that the gap between inflow and billed use points to a specific area. Zones are usually sized so a crew can search them in a day or two.
How is night flow used?
Between roughly 2am and 4am, legitimate use is at its lowest, so flow in those hours is mostly leakage. A rising night flow in a zone is one of the strongest leak indicators available.
Do sensors work on all pipe materials?
No. Acoustic detection works best on metal pipes and less well on plastic, where sound attenuates quickly. Pressure and zone-balance methods work regardless of material, which is why combining methods matters.
How is success measured?
By water saved: the zone balance before and after repairs, and the volume of confirmed leaks repaired. Sensor counts and alert counts are activity, not outcomes, and should never be reported as savings.