Tracer gas leak detection, and what it needs to work
The water pipe is drained and isolated, a mixture of hydrogen and nitrogen is put into it in place of the water, and a detector is used at the surface to find where the hydrogen comes up. Hydrogen is used because it is the smallest molecule there is, so it escapes through openings too small to produce a useful noise and rises rather than spreading sideways. It has nothing whatever to do with a fuel supply. The cost of the method is that the pipe has to be emptied first, which is not always practical, and that a sealed surface above the pipe can send the gas somewhere other than straight up.
Last reviewed 14 August 2026
The short answer
Empty the pipe, fill it with a harmless hydrogen and nitrogen mixture, and find where the hydrogen comes up through the ground.
It is the method for the cases acoustic detection cannot handle, because it does not depend on pressure, on noise, or on what the pipe is made of. What it does depend on is being able to drain the pipe, and on the gas having a path to the surface.
How this works in practice
The method swaps the thing in the pipe for something easier to detect outside it. Four steps, and the first is the one that decides whether it can be done at all.
- The section of pipe is isolated and drained, so the water is out of the way and the gas can reach the whole run.
- The mixture is introduced at a convenient point, usually where the pipe is already accessible, and allowed to fill the pipe.
- It escapes at the leak, exactly where the water was escaping, and rises through soil, sand or paving towards the surface.
- A detector is passed over the ground above the run and picks up where the hydrogen is emerging.
The reason it reaches what acoustic cannot is that none of its requirements are the ones acoustic needs. There is no noise involved, so a slow loss is no harder than a fast one. There is no pressure needed, so a drained or unpressurised system is testable. And the pipe material barely matters, because the gas is inside the pipe and comes out at the hole rather than travelling along the wall.
That last point is worth holding on to, because it is the whole reason both methods exist. Acoustic depends on the pipe carrying something. Tracer gas depends on the ground carrying something. A property that defeats one may be perfectly suited to the other, and they fail for unrelated reasons.
What changes the answer
Three things, and the first is a practical obstacle rather than a technical one.
| Condition | Why it matters | What it means for you |
|---|---|---|
| The pipe cannot be isolated or drained | The gas has to occupy the space the water is in. If the section cannot be emptied on its own, it cannot be filled on its own. | Find out where your stop taps are and what each one isolates, before booking. On some layouts the answer is inconvenient. |
| A sealed surface above the pipe | A concrete slab, a membrane or a tarmac drive can stop the gas surfacing above the leak and route it sideways to a joint or an edge. | It may still be found, but where it emerges is then not where the leak is, so the result is a lead rather than a position. |
| Being off the water while it happens | The supply is drained for the duration of the test. | A practical consideration rather than a technical one, but worth planning for rather than discovering on the day. |
The middle row is the failure mode most worth understanding, because it does not look like a failure. The detector finds hydrogen, somebody marks the spot, and the mark is real. What it marks is where the gas got out of the ground, which under a sealed surface can be a long way from where it got out of the pipe. A result under a slab is worth asking about specifically.
Where the water in the pipe is hot rather than cold, there is a third approach that needs neither draining nor listening. what a thermal camera can and cannot show sets out what it actually detects.
Assumptions and sources
Drip Calculator: how much water does a leaking faucet waste?
United States Geological Survey, Water Science School · primary source
- Faucet drip volume, adopted: 0.25 ml
- Faucet drip volume, measured range: 0.2 to 0.33 ml
- Bath tap drip volume: 0.5 ml
- Drips per US gallon: 15140 drips
- Drips per litre: 4000 drips
The pharmacopoeial metric drop, 20 drops to 1 millilitre
United States Pharmacopeia, British Pharmacopoeia and European Pharmacopoeia convention · primary-standard source
- Standard drop volume: 0.05 ml
Tate's law: the weight of a falling drop
Thomas Tate, Philosophical Magazine, 1864, with the later drop-weight literature · primary source
- Drop weight: W = 2 * pi * r * gamma equation
Plumbers shine the spotlight on dripping taps during Water Saving Week
WaterSafe, with Waterwise · untraceable source
- Claimed annual waste from a dripping tap: 5500 litres per year
International Organization of Legal Metrology · primary-standard source
- MPE, accuracy class 2, lower zone Q1 to Q2: 5 %
- MPE, accuracy class 2, upper zone Q2 to Q4: 2 %
- MPE, accuracy class 1, lower zone: 3 %
- MPE, accuracy class 1, upper zone: 1 %
- In-service MPE multiplier: 2 ratio
- Q2 / Q1: 1.6 ratio
- Q4 / Q3: 1.25 ratio
Aquadis+ DN15 and DN20 specification sheet
Itron · manufacturer source
- Q1 minimum flow rate, DN15, Q3 2.5, R160: 15.6 l/h
- Q2 transitional flow rate, DN15, Q3 2.5, R160: 25 l/h
- Q3 permanent flow rate, DN15: 2500 l/h
- Typical starting flow rate, DN15: 0.4 l/h
- Typical starting flow rate, DN20: 2 l/h
- Q1 minimum flow rate, DN20, Q3 4.0, R160: 25 l/h
- Flow rate at which accuracy is within +/- 5%, DN15: 3 l/h
- Flow rate at which accuracy is within +/- 2%, DN15: 5 l/h
- Minimum scale interval, DN15 and DN20: 0.02 l
Water resources 2024 to 2025: analysis of the water industry's annual water resources performance
Environment Agency · primary source
- Household per capita consumption: 136.5 l/person/day
- Household per capita consumption, prior year: 137 l/person/day
- Household per capita consumption, dry-year adjusted: 140.3 l/person/day
- National leakage: 2617 Ml/day
- National leakage as share of water put into supply: 19 %
PR24 common performance commitments: per capita consumption (PCC), version 2.1
Ofwat · primary-standard source
- PCC formula: (measured household consumption + unmeasured household consumption) / total household population l/person/day
Leaky loos, why it's not as simple as faulty flush valves
ech2o, reporting research attributed to Thames Water · secondary source
- Toilets inspected: 58551 toilets
- Toilets found leaking: 4854 toilets
- Prevalence: 8.3 %
- Average residential leak rate: 400 l/day
- Average commercial leak rate: 2100 l/day
- UK-wide average across all WCs, leaking and sound: 20 l/day
- Share of leaking toilets with flush valves rather than siphons: 81 %
Waterwise · untraceable source
- Single leaking toilet: 215 to 400 l/day
- Share of toilets leaking: 5 to 8 %
- UK total from leaking toilets: 400000000 l/day
Methods for Calculation of Evaporation from Swimming Pools and Other Water Surfaces
M. M. Shah, ASHRAE Transactions SE-14-001 · primary source
- Outdoor unoccupied pool evaporation: E0 = (C1 + C2 * u) * (pw - pa) / ifg kg/(m2*h)
- C1, SI: 235 constant
- C2, SI: 206 constant
Location-specific long-term averages, Wisley, Surrey
Met Office · primary source
- Annual mean daily maximum temperature: 15.41 degrees C
- Annual mean daily minimum temperature: 6.66 degrees C
- Annual rainfall: 667.92 mm
- Annual mean wind speed at 10 m: 5.06 knots
Discover Water, the water industry transparency site run with Water UK, Ofwat, CCW, the Drinking Water Inspectorate and Defra · primary source
- Average annual household bill, combined water and sewerage: 639 GBP
- Average annual household bill, water only: 309 GBP
- Average annual household bill, sewerage only: 330 GBP
SI Brochure: The International System of Units (SI), 9th edition (2019), version 4.01, June 2026
Bureau International des Poids et Mesures · primary-standard source
- cubic metre: coherent derived unit of the SI, Table 5 m3
- litre: non-SI unit, Table 8. 1 l = 1 L = 1 dm3 = 10^-3 m3 l or L
- bar: non-SI unit, Table 8. 1 bar = 0.1 MPa = 10^5 Pa bar
Authority URLs for the named entities in content/entities.ts
Various, see per-entity notes · primary source
WC flush volume: three UK instruments, not one
UK Statutory Instruments, Northern Ireland Statutory Rules, and Scottish Water · primary-standard source
- Maximum single flush, all three jurisdictions: 6 litres
- Dual flush, lesser flush maximum: two-thirds of the largest flush volume ratio
- England and Wales transitional maximum, 1 Jul 1999 to 1 Jan 2001: 7.5 litres
Per-event central heating top-up volume, derived
Derived by Q3 from Boyle's law and typical UK domestic expansion vessel specifications · derived source
- Water held in an expansion vessel: V_water = V * (1 - P0 / P), pressures absolute equation
- Typical top-up, 8 l vessel at 1.0 bar pre-charge, 0.5 to 1.5 bar: 1.6 litres
- Typical top-up, 8 l vessel at 0.75 bar pre-charge, 0.5 to 1.5 bar: 2.4 litres
- Typical top-up, 12 l vessel at 1.0 bar pre-charge, 0.8 to 1.5 bar: 2.4 litres
- Typical top-up, 8 l vessel at 1.0 bar pre-charge, 1.2 to 1.5 bar: 0.9 litres
- Realistic range: 0.9 to 2.5 litres
Monodispersed Bubble Generation Using Hydrophobic Orifices: The Extended Tate’s Law
Bo Liu, Hao Zhang and co-authors, ACS Omega 2024, 9(17), 18854 to 18861 · primary source
No proportion is given for the mixture, and no accuracy figure or detection depth appears, because no sourced measurement of any of them was found. What is stated is why hydrogen is the molecule used and why it is mixed with nitrogen rather than used alone, which is chemistry rather than a claim about performance.
This page is careful about naming because the name misleads. Tracer gas leak detection is a water leak method: the gas is a detection medium rather than a fuel, and it goes into a drained water pipe. Nothing on this site concerns fuel supplies, appliances or their safety, and this page is the one place a reader might reasonably wonder, so it says so directly rather than avoiding the word.
What is not claimed: that tracer gas succeeds where acoustic fails, as a general rule. What is claimed is that it does not depend on the same things, which is why it is the usual alternative. Whether it works on a particular property depends on that property.
What to do next
Three things, and the first is the one that decides whether this method is available to you at all.
- Find every stop tap and work out what each one isolates. If the run you suspect cannot be shut off on its own, that is worth knowing before a survey is arranged rather than during it.
- Note what is above the suspect run. Soil and gravel let the gas up. A concrete slab, a membrane or a tarmac drive may not, and the answer changes what a result means.
- Plan for the water being off while the test runs, and say if that is difficult, because it affects when the work can sensibly be done.
What tracer gas leaves unknown is depth, and how far the gas travelled underground before it surfaced. Over open soil that distance is usually small. Under a sealed surface it can be substantial, and the honest form of the result there is a place to start digging rather than a position.
Common questions
- Is this related to the fuel coming into my house?
- No. It is a detection technique used on water pipes. A harmless mixture of hydrogen and nitrogen is put into a water pipe that has been drained and isolated, and a detector finds where it escapes. It has nothing to do with fuel, and nothing on this site concerns fuel, appliances or their safety.
- Why hydrogen?
- Because it is the smallest molecule there is, so it gets out through openings that would be too small to make a useful amount of noise, and because it is far lighter than air, so it rises through soil and paving towards the surface rather than dispersing sideways. That combination is what lets a detector at ground level find a pipe buried below it.
- Is it safe to put hydrogen in my pipework?
- The mixture used is mostly nitrogen, which is inert and is the majority of the air you are breathing now, with only a small proportion of hydrogen. It is chosen to be non-flammable at that proportion, which is the reason it is mixed rather than used pure, and it is vented out and the pipe refilled with water afterwards.
- Why does the pipe have to be drained?
- Because the gas has to occupy the space the water is in to reach the leak. That is the method's main practical cost: draining and isolating a supply is a job in itself, it puts your water off for the duration, and on some layouts the section you want to test cannot be isolated on its own.
- Will it work under a concrete floor or a driveway?
- Sometimes, and less predictably than over soil. The gas needs a path to the surface, so a sealed slab, a membrane or a tarmac drive can hold it under and let it travel sideways before it emerges. It can still be found, at an expansion joint or an edge, but where it surfaces is then not where the leak is, which is the failure mode to be aware of.