Power off at the consumer unit, not at the wall switch
An immersion heater is a three-kilowatt electric element screwed directly into a tank of water. When water escapes around that fitting it runs down the outside of a live appliance and, often, straight past the terminal cover at the top of it. Deal with that before the plumbing.
Go to the consumer unit and switch off the breaker for the immersion or water heater circuit. It is usually a 16A breaker on its own way, labelled immersion, water heater or cylinder. If nothing is labelled, switch off the main switch for the whole board.
The pull-cord or the switch on the airing cupboard wall is not good enough on its own. Some of those switches are single-pole, so they break the live and leave the neutral connected, and you cannot tell which yours is by looking at it. A switch that has sat in a warm damp cupboard for fifteen years is also exactly the sort of thing that fails closed. Isolate at the board.
Then leave it off. Do not switch it back on to see whether the leak gets worse when it heats. If a residual current device keeps tripping when you reset it, that is water reaching a conductor and the RCD doing the only job it has.
Some situations mean you should not be in the cupboard at all: water pooling on the floor around the cylinder, water dripping onto the terminal cover itself, or any scorching or burnt smell at the head of the element. Isolate the whole board from a dry position and get an electrician to it. The Health and Safety Executive is blunt about water and live equipment, and a domestic airing cupboard is not an exception.
Why the boss leaks
The boss is the threaded socket welded or brazed into the cylinder that the element screws into. On most British cylinders it is a 2¼ inch fitting, and it is sealed by a single flat fibre washer compressed between the element's hex head and the face of the boss. Three things go wrong there, and they go wrong for different reasons.
The washer has given up
Fibre washers are consumable. They start as a compressible gasket and end up as a hard ring that has taken a permanent set. Every heating cycle expands the copper and the brass at slightly different rates, and the washer absorbs that movement until it cannot. After ten or fifteen years it weeps, usually first as a faint green or white crust under the hex head rather than as a visible drip.
Over-tightening kills them faster. An installer who leaned on the spanner to stop a weep has crushed the fibre, and the seal it makes afterwards is worse, not better. This is the one failure mode that is genuinely a cheap fix, because a new washer costs pennies. The cost is entirely in getting the element out.
The threads have corroded
You have dissimilar metals in permanent contact with hot water: a copper cylinder, a brass or stainless boss, and an element head that may be brass, stainless or titanium-sheathed. That is a galvanic cell, working slowly but continuously for the whole life of the cylinder, and what it produces is a thread that is both eaten away and packed with corrosion product.
The practical effect is that the thread stops being a thread. It cannot be re-sealed by tightening, because there is no longer clean metal for the washer to bear against.
Scale, which in London is not optional
Most of London is supplied with hard water drawn from chalk and limestone catchments, and Thames Water publishes hardness figures by postcode. In practical terms it means calcium carbonate comes out of solution wherever water is heated, and the hottest surface in the cylinder is the element itself.
A scaled element is an insulated element. The heat it produces can no longer leave into the water fast enough, so the sheath runs hotter than it was designed to, expands further on every cycle, and in bad cases distorts. That movement goes straight into the boss and the washer. Scale also migrates into the thread, where it sets like grout and turns a fitting designed to be serviceable into one that is not. It is why a London cylinder can be on its third element while a soft-water equivalent is still on its first.
Is it actually the boss?
Water runs downhill and lands somewhere that tells you little about where it started. Before the element gets blamed, rule out what looks identical from the cupboard door:
- A pinhole in the cylinder above the boss. Copper cylinders pit from the inside. A hole two inches higher up drains down the curve of the tank and arrives at the element head looking exactly like a boss leak. Dry everything, wrap tissue tight around the boss, and see whether it wets from above or from the seal.
- The thermostat pocket. On some cylinders the stat sits in its own pocket, a second seal in the same area.
- The immersion above it. Plenty of vented cylinders have two elements, a short one near the top and a long one near the bottom. The upper one leaking drips onto the lower one.
- The cold feed or the vent connection. Compression joints at the top of a cylinder weep in the same place and look the same. Condensation on a poorly lagged cylinder is rarer, but it is also not a leak.
If the water is arriving in the room below rather than in the cupboard, the cylinder has probably been leaking for a while, and the ceiling is the symptom rather than the fault. A stain below an airing cupboard is one of the more common ways a slow boss weep finally gets noticed.
What to establish before anyone puts a spanner on it
An immersion heater that comes out easily is a forty-five minute job. One that does not can end with a scrapped cylinder and a house with no hot water for a week. The difference is worth knowing about in advance:
| What to check | Why it decides the job |
| Vented or unvented | An unvented cylinder is a pressure vessel. Work on it is notifiable under the Building Regulations and needs an operative holding the G3 qualification. A vented copper cylinder fed from a loft tank does not. |
| Age of the cylinder | Past roughly twenty-five years, spending money to keep it going rarely pays. The next failure is already coming. |
| Material | Copper tears. Stainless and glass-lined steel behave differently under the torque needed to shift a seized element. |
| Foam lagging | A factory-foamed cylinder cannot be strapped or supported the way a bare copper one can. There is nothing to hold while you turn. |
| Clearance above and beside | A box spanner and bar needs room, and the element comes straight out — sometimes nearly a metre of it. |
| Is a replacement available | If it is an odd size or a discontinued model, find that out before the old one is destroyed. |
Why removing a seized element often finishes the cylinder
To break a corroded and scaled 2¼ inch thread you need real torque, applied through a box spanner, to a fitting brazed into a thin copper shell. There is no way to apply that force without the whole cylinder wanting to turn with it. What tends to happen, in order: the cylinder rotates on its base, the boss starts to move relative to the shell, the copper around it deforms, and then it tears or the braze cracks. Once the boss has moved, no washer will seal it again, and you cannot weld a patch onto a domestic hot water cylinder.
So the honest framing is this: on an old cylinder with a corroded boss, the decision is not really repair or replace the element. It is whether you are prepared to risk the cylinder to find out. Sometimes the right answer is to leave the element where it is, plan a cylinder replacement on your own timetable, and not have it forced on you in the first ten minutes of a cheap repair.
Worth attempting the removal: a cylinder under about fifteen years old with no green staining on the boss, or one where the element has failed electrically anyway so it has to come out regardless. Usually not worth it: a cylinder already showing corrosion elsewhere, or any cylinder whose replacement would have to be ordered in.
What we actually do, and what it costs
We attend with the circuit already isolated, confirm the source rather than assuming it, and tell you which of the three outcomes you are looking at: a washer and refit, an element and washer, or a cylinder. If it is the cylinder we say so before the spanner goes on, not after.
The rate is £150 an hour, agreed before we attend, with no call-out fee. Where the water is arriving somewhere other than the cupboard and it is not obvious what is feeding it, that becomes a leak detection job first, under no find, no fee: if we do not find it, you do not pay for the looking. Full rates are on the pricing page.
If a slow weep has been soaking a cupboard floor and the ceiling below, the making-good is done by our own handymen and builders. A detection visit comes with a written report prepared to support a trace and access claim; what your policy covers is your insurer's decision, not ours.