Start with what is being asked
There are two different tests that both get called a pressure test, and confusing them causes most of the arguments.
The first is a commissioning test on new work. Schedule 2 paragraph 12 of the Water Supply (Water Fittings) Regulations 1999 requires that the water system shall be capable of withstanding an internal water pressure not less than one and a half times the maximum pressure to which the installation is designed to be subjected in operation. Paragraph 5 says the same of each individual fitting. The point of that test is strength and integrity of an installation before handover. It is a proof test.
The second is a diagnostic test on an existing system that is losing water. Nobody is proving it is fit for handover; you are trying to establish whether there is a real loss, roughly how large it is, and which section it is on. Same gauge, entirely different question, and the acceptance criteria from the first test are the wrong tool for the second.
The numbers, so you can read a test certificate
Where BS EN 806 or BS EN 805 is followed, the recommended test pressure is 1.1 times the maximum design pressure, where maximum design pressure is 1.5 times maximum operating pressure. On a system designed for 5 bar working pressure, that gives 1.1 multiplied by 7.5, or 8.25 bar, against a statutory minimum of 7.5 bar. Water Regs UK note that an installation tested to BS EN 806 therefore satisfies the regulations by default.
The procedures differ by material, and this is where the interesting part is.
Rigid systems, meaning no plastic anywhere, such as copper or stainless steel. Charge with wholesome water and remove all air, ideally 24 hours before the test. Pressurise the whole system to the test value, isolate it, and leave it for one hour. The pressure should be maintained and not fall below the initial test pressure. That is a clean pass or fail with no tolerance.
Systems containing plastic get two permitted procedures, because plastic expands. In Test A the system is pumped to test pressure and held by pumping for 30 minutes, the pressure is then reduced to one third of test pressure, and the test is satisfactory if the pressure does not drop further over the following 90 minutes with no visible leakage. In Test B the system is pumped to test pressure and held by pumping for 30 minutes, and is then satisfactory if the pressure does not fall more than 0.6 bar over the next 30 minutes and then a further 0.2 bar over the following 120 minutes, with no visible sign of leakage.
The first thing a pass does not mean
Read Test B again. A plastic system can lose 0.6 bar in half an hour and another 0.2 bar over two hours and still be recorded as satisfactory. That is not a loophole, it is physics: plastic pipe stretches under pressure and the material relaxes, so a leak-free system genuinely does show a decaying curve. But it means the sentence "it held pressure" carries almost no information on its own.
When somebody tells you a test passed, the useful questions are which procedure was used, what the start and end pressures were, and over what period. A test result without numbers is an opinion.
Temperature, which is the quiet one
Water expands and contracts with temperature and so does the pipe. A sealed system that cools overnight loses pressure without losing a drop of water. This is why the water industry guidance specifies recording with a calibrated data logger rather than reading a gauge by eye, and why those loggers are calibrated to take account of temperature, which affects recorded pressure in a pipeline.
The domestic version of this mistake happens constantly. A boiler gauge read at nine in the evening with the heating just off, and again at seven the next morning in a cold house, will read lower. That is not evidence of a leak. Compare cold to cold, at the same time of day, with the system at the same temperature, or you are measuring the weather.
What it cannot tell you: location
This is the biggest misunderstanding. A pressure test proves a loss exists, and the rate of decay gives you a rough size, which is genuinely useful because it separates a pinhole from a split and tells you how urgent this is. It gives you no information at all about where. Pressure is the same everywhere in a connected body of water.
Locating is a separate exercise with separate tools, and there is a walkthrough of the sequence in how plumbers find leaks. The pressure test comes first because it stops you searching for a leak that is not there, not because it points anywhere.
What it cannot see: everything outside the tested section
A test only covers what is inside the valves you closed. Three ways that bites:
- A passing isolation valve. If the valve you used to define the section does not fully close, you are testing more than you think, and the loss you measured may be beyond it. Old gate valves and service valves that have not been turned in twenty years are the usual culprits.
- The section you capped off. Isolating the suspect leg and testing the rest is how you narrow it down, but it also means a clean result on the rest proves nothing about the leg you removed.
- Anything downstream of a check valve. A non-return valve stops test pressure propagating past it, so a whole branch can sit outside the test without anybody noticing.
The way to use this rather than be caught by it is deliberate sectioning: isolate, test, and halve the system each time, keeping a written note of what was in and out of each test. That is slower than one big test and it is the only version that produces an answer.
What it cannot separate: a leak from a vessel
On a sealed heating system there is a failure that behaves exactly like a leak and is not one. If the pressure climbs unusually high when the heating runs and then drops below normal as it cools, the expansion vessel has probably lost its air charge and the system has nowhere to put the expansion, often discharging through the pressure relief valve outside. No water is being lost to the building at all.
A static pressure test on a cold system does not distinguish these. What does is watching the gauge through a full heating cycle, and checking whether the discharge pipe outside drips after the boiler fires. There is more on the pattern in boiler pressure dropping with no visible leak, and the whole subject sits under the boiler pressure hub.
What it cannot catch: leaks that need conditions
A static test applies steady pressure to a still system. Some leaks do not appear under those conditions at all:
- A shower tray or bath that only weeps under the load of someone standing in it.
- A waste or overflow that only leaks while an appliance is discharging, because it is not under pressure otherwise.
- A heating joint that only passes while the pump is running and the circuit is at working temperature.
- A soil stack joint that leaks on a single flush and is dry the rest of the day.
A clean pressure test on a property with an obvious wet patch is therefore a result, not a dead end. It tells you to stop looking at pressurised pipework and start looking at waste, seals and rainwater.
Drains: where the standard explicitly allows loss
Worth knowing, because it is the clearest illustration of the whole point. Approved Document H sets out how gravity drains are tested. In the air test for pipes up to 300mm, the pipe is pressurised to 110mm water gauge, held about 5 minutes, and should then hold an initial 100mm with a maximum loss of head on a manometer of 25mm over 7 minutes. In the water test, the test pressure is maintained for 30 minutes and losses should not exceed 0.15 litres per square metre of surface area for lengths with only pipelines, 0.20 with pipelines and manholes, or 0.40 for manholes and inspection chambers alone.
A drain that loses a quarter of its manometer head, or a measured volume of water, passes. Passing a Building Regulations test and being watertight are not the same statement.
How we use it, and when it is not worth doing
On a detection visit the test goes first, before any gas or cameras, because it answers the only question that determines whether the rest of the visit is worth paying for: is there a real loss, and on which circuit. Then the locating methods run against a known target rather than a suspicion.
It is not worth doing when you can already see the water leaving the pipe. If a joint is visibly dripping, testing it tells you what your eyes told you. It is also not worth paying for on a system that has already emptied itself and will not hold anything, where the right first step is tracer gas on the drained circuit.
Detection is £150 per hour, fixed before we attend, with no find, no fee. The test, the trace, the repair and the making good are the same visit and the same firm, which is the part most detection-only companies leave to you.
