Why this is the first question on site
Ask a leak engineer what they want to know before they look at a wet ceiling and the honest answer is not the symptom. It is the age of the property, what has been done to it, and what the pipe is made of. Material tells you the failure mode, failure mode tells you the likely location, and likely location decides which detection method goes on first. Get that order wrong and you spend two hours confirming what a two-minute look under the sink would have told you.
A typical London house has three or four materials in it at once: a lead or MDPE supply from the boundary, a copper rising main, copper or microbore heating from a 1980s conversion, and plastic push-fit in whatever was refitted last. The leak is rarely in the middle of a nice straight run. It is at the point where one thing meets another.
Lead
Dull grey, soft enough to mark with a coin, and it scratches bright silver. It bends in sweeping curves with no fittings, and where it does have a joint the joint is a swollen, lumpy collar wiped on by hand rather than a machined fitting. Tap it and it sounds dead rather than ringing.
Before 1970 lead pipework was commonly used in domestic distribution systems, and although its use has since been banned, plenty of pre-war London stock still has it underground and sometimes inside. The Drinking Water Inspectorate is clear about who owns what: the communication pipe from the main to the boundary belongs to the water company, and the supply pipe from the boundary into the property belongs to you. The lead standard is 10 micrograms per litre, and consumers are largely protected from exposure by phosphate dosing at treatment works. In 2024 there were 53 lead failures from 13,394 samples, and water companies have collectively declared an ambition to be lead pipe free by 2050.
Where it fails. At the wiped joint, and at any point where it has been disturbed. Lead work-hardens: a run that has been dug against, leaned on by a new patio slab, or nudged when someone fitted a stopcock will crack months or years later at the point of the disturbance rather than randomly along the run. This matters on site because it means the search is not a general sweep of the garden. It is the joints, the entry point through the wall, and anywhere the ground has been opened.
What finds it. Lead carries leak noise well, so acoustic detection is usually the first and often the only method needed on a pressurised lead supply.
Copper
Copper does not usually fail as a split. It fails as a pinhole, and the pattern of pinholes is diagnostic. The National Physical Laboratory's guide to corrosion in plumbing systems sets out the main forms, and two of them turn up repeatedly in London.
The first is flux attack. Soldering fluxes contain ammonium or zinc chlorides, which are aggressive to copper. An excess of flux runs downstream of the joint and pin-holes the pipe, usually within a few months of installation and typically a few centimetres past the soldered fitting. A leak on a recent refit, just downstream of a solder joint, is very often this.
The second is Type 1 pitting, historically associated with hard borehole supplies and carbon film left in the tube bore during manufacture. BS EN 1057, introduced in 1996, set limits on those residues and the problem largely went away, but similar pitting has since been found in hard water areas with high phosphate in the deposits. London sits in that hard water region: Thames Water describe all the water in their region as hard, because it passes through chalk and limestone. Type 1 pits sit under a blue-green mound of malachite, they appear randomly along a length, and the deeper ones are nearly always near the bottom of horizontal runs.
There is a lot more to this than fits here. The full breakdown is in why copper pipes get pinholes and why there is rarely just one.
Galvanised steel
Heavy, threaded, and screwed together with tapered fittings rather than soldered. It rings when you tap it and a magnet sticks to it, which is the quickest way to tell it from copper under old paint. It turns up on pre-1960s cold services and rising mains, often with copper grafted on to it where later work was done.
BSI publishes a whole standard on the question of how likely it is to corrode, BS EN 12502-3, covering influencing factors for hot-dip galvanised ferrous materials used as steel tubes and cast iron fittings in drinking water systems in buildings. The practical version on site is that the threads are the thin part. Thread cutting removes wall thickness and removes the galvanising with it, so where a run gives way, it gives way at or just behind a fitting.
What it tells you. Galvanised is usually a re-run rather than a repair. Cutting into an old galvanised run to fit a repair section means unscrewing joints that have been made up for sixty years, and the ones either side of your repair are the next ones to go.
MDPE
The blue coil that replaced lead on supply pipes. Medium density polyethylene to BS 6572, laid at a minimum depth of 750mm and a maximum of 1350mm below finished ground level, at least 350mm clear of other services. Where the ground is contaminated with oil, petrol or creosote, standard MDPE is not acceptable because those chemicals permeate it, and a barrier pipe to BS 8588 is required instead.
The regulatory guidance says it is best practice to lay supply pipes in a way that avoids joints, and that is the whole story of how they fail. A continuous coil from the boundary box to the house does not develop holes in the middle. What fails is a coupler somebody added halfway along, a compression fitting where the insert was left out, or a point where a sharp stone or a flint was backfilled hard against the pipe.
What finds it. Not acoustic, usually. Leak noise in plastic pipe attenuates badly, which is covered in the acoustic detection guide. On a buried MDPE supply the reliable methods are a correlation across a known run where two contact points exist, or tracer gas once the pipe can be drained.
Plastic push-fit inside the house
Grey, white or barrier pipe with a printed spec along it, joined with collet fittings that click. Kitemarked to BS 7291 and entirely legitimate when installed correctly, which is the qualifier that does the work. Failures here are almost never the pipe. They are the joint, and specifically the O-ring, the insert and how far the pipe went in. There is a full post on it: how plastic push-fit plumbing leaks and why it hides.
Microbore
Soft copper in 8mm and 10mm, run as continuous coils from a manifold to each radiator. BS EN 1057 covers seamless copper tube from 6mm outside diameter upwards, so these are ordinary tube sizes, not a bodge. The design intent is that there are no joints under the floor at all: the only connections are at the manifold and at the radiator tail.
Where it fails. At the tail, where a compression nut has been over-tightened or the soft tube has been flexed once too often when a radiator was taken off for decorating. Second place is a kink, because 8mm coil kinks easily when it is pulled through a joist. A kink does not leak on the day it is made; it leaks after the pipe has expanded and contracted through a few winters.
Microbore also produces a symptom that is not a leak at all. A cold radiator on microbore is usually a blocked 8mm run full of magnetite, not a loss of water, and the pressure gauge will be fine. If your pressure holds and one radiator is cold, that is a power flush conversation, not a detection one.
Fittings, which are their own failure mode
Brass loses zinc. Dezincification leaves a porous copper skeleton the same shape as the original fitting, weaker and eventually leaking, and it comes in a layer form and a plug form. Dezincification-resistant brasses are marked CR or DRA, and Schedule 2 paragraph 7(3) of the Water Supply (Water Fittings) Regulations 1999 requires any concealed water fitting that is not a terminal fitting to be made of gunmetal or another material resistant to dezincification. A plain brass fitting buried in a wall is both a failure waiting to happen and a regulation breach.
There is a rarer one worth knowing because it looks inexplicable. NPL record stress corrosion cracking of brass fittings in drinking water installations occurring under phenolic foam lagging, where condensation reaches the brass and the fitting had been over-tightened. A cracked fitting under insulation, with no corrosion anywhere else, is usually this.
What the material changes about the search
| Material | Typical failure | What finds it first |
|---|
| Lead | Wiped joint, or a point of past disturbance | Acoustic |
| Copper | Pinhole, often just past a soldered joint | Acoustic, then thermal on hot runs |
| Galvanised steel | Thread at a fitting | Acoustic, then visual at every fitting |
| MDPE | An added joint, or a stone bearing on the pipe | Correlation, then tracer gas |
| Push-fit | O-ring at a joint, often a concealed one | Tracer gas, thermal on heating runs |
| Microbore | Radiator tail or a kink | Visual at the tails, then thermal |
When this is not a job for us
If you have identified a weeping compression joint on an accessible run, nip it up a quarter turn and watch it for a day. If a radiator tail is damp at the nut, that is a tail and an olive, which is a competent plumber's half hour, not a detection visit. And if your pressure holds steady and one radiator is cold, nothing is leaking.
Call us when the water is arriving somewhere and you cannot see where it left. Detection is charged at £150 per hour, fixed before we attend, and if we do not find the leak you do not pay for the time we spent looking. What we do that a detection-only firm does not is repair the pipe and put the floor, wall or ceiling back afterwards. See pipework repair and replacement for how that half of the job works.
