Municipal water main leak detection. The network leaks. Here's how it gets found.

A pressurised leak emits continuous broadband noise that propagates along the pipe wall and water column. Sensors on two accessible contact points — hydrants, valves — both receive it; the correlator measures the time difference between arrivals. With pipe material, diameter, and sensor spacing known, the leak's position falls out of the velocity equation. On metallic mains, accuracy within half a metre is routine; the method is why modern repair crews dig one hole.

Plastic and asbestos-cement attenuate the higher frequencies that carry best on metal, so leak noise travels far shorter distances and arrives quieter and lower-pitched. The countermeasures: tighter sensor spacing, low-frequency sensors and hydrophone approaches, more ground-microphone legwork, and survey timing at maximum pressure and minimum ambient noise. BC's postwar AC stock and modern PVC mean a crew's PVC technique matters as much as its equipment.

It's the flow into a metered zone during the lowest-demand hours, typically 2–4 a.m. Subtract a small allowance for legitimate night use and what remains approximates real losses in that zone. It matters because it's continuous, cheap, and ruthless: a new leak shows as a step change within days, zones rank themselves for survey budget, and post-repair drops prove recovery in the utility's own data. If a system has zone metering and isn't watching night flows, it's leaving its best instrument switched off.

Real losses are physical leakage — mains, services, appurtenances, reservoir overflows. Apparent losses are accounting losses — meters under-registering, billing gaps, unauthorised use: the water was delivered, just never paid for. The split (an AWWA M36 water balance produces it) decides where money goes: detection crews fix real losses; meter replacement and billing work fix apparent ones. Programs that skip the split routinely spend detection budget hunting losses that live in a spreadsheet.

A crew covers several kilometres of distribution main per day depending on contact-point density and how many points of interest demand correlation — so a small municipality's whole system fits comfortably in a survey season, and larger systems run rolling programs (commonly a third of the network per year) or logger-led targeting. Aerial corridor screening multiplies coverage where alignments suit it.

They aim them. Loggers answer 'where is there persistent leak-like noise?' across a whole zone for weeks of nights — far more listening than any crew pass. But a logger POI is a neighbourhood, not a dig location: correlation and ground mic work still produce the paint mark. The economics win because crew hours concentrate on ranked POIs instead of walking quiet pipe.

Whenever the leak won't talk: deep mains, soft backfill swallowing noise, PVC attenuation, or loss spread across many small weeps that never correlate individually. Step testing is pure hydraulics — close valves in sequence at night, watch the zone meter, and the inflow drop at each step quantifies leakage inside the isolated segment in litres per hour. It finds the guilty block; acoustics then finds the guilty metre.

Yes — pressure is what makes leaks audible, so live mains are the working condition. Large-diameter work changes the physics (low-frequency noise, sparse contact points, greater depth) and calls for specialist sensors and patient spans, often preceded by aerial thermal screening of the alignment to rank where intensive time gets spent. Shutdowns belong to internal inspection, a different discipline our findings often help justify — and target.

Escaping water follows the trench, and pipe bedding gravel is an excellent conduit — flow travels along it, sometimes tens of metres, until it finds a weak exit: a utility cut, a joint in the asphalt, a low shoulder. Digging at the puddle finds wet gravel and intact pipe. Correlating the noise finds the break. The two-hole, three-hole excavation stories every works crew tells are the cost of skipping that step.

Surveys price per kilometre or per zone; programs are sequenced so early phases (appurtenances, loggers) recover cheaply found water that funds the harder phases. The benchmark that matters: marginal production cost per cubic metre times the volume a located leak was wasting. A modest 0.5 L/s find wastes ~15.8 ML/year — against that arithmetic, a survey week pays for itself with one good find, and programs are designed to find dozens.

Whatever exists of: system length and material breakdown, production and billing volumes (any recent year), zone metering and night-flow data if metered, break history, and the trouble list operations already suspects. None are prerequisites — programs start from incomplete data constantly — but each one sharpens the first scope. The free consult turns whatever you have into a ranked starting plan: 604-239-9934.

Because they fail differently. Aerial thermal reads surface expression across kilometres per flight but can't hear a deep leak that hasn't expressed; acoustics hears what's invisible but covers ground at crew pace. Sequenced — fly to rank, listen to pinpoint — each covers the other's blind spot, and Leak.ca runs both under one roof, which keeps the evidence chain (and the accountability) unbroken from flight line to paint mark.