Underfloor Heating Leaks: The Signs, the Tests That Prove It, and the Repair

The signs, and what each one actually indicates

Underfloor heating gives ambiguous symptoms, because the components that fail silently are buried and the components that are visible rarely fail. A symptom list is therefore only useful alongside what else produces the same symptom.

What you notice	Consistent with a leak	Also produced by
System pressure falling and needing topping up	Yes, strongly	Failed expansion vessel, passing pressure relief valve, air being purged
A persistently warm patch of floor with the heating off	Yes	A hot water pipe or flow and return buried in the same floor
One zone never reaching temperature	Possible	Air lock, stuck actuator, closed flow meter, blocked circuit
Damp at the perimeter or a skirting board	Yes	Rising damp, a failed damp proof course, a waste pipe in the floor
Floor finish lifting, cupping or discolouring	Yes	Screed not dried before laying, no expansion gaps, subfloor moisture
A faint hiss or trickle audible at the manifold	Possible	Circulation noise, a partially closed valve, entrained air
Boiler firing more often with no change in weather	Possible	Controls fault, a bypass left open, cylinder thermostat issue

The warm patch that is not a leak

A warm area of floor is the symptom people trust most and it is one of the least reliable, because an underfloor heating floor contains pipework that is meant to be warm. The diagnostic version of the test is done cold: isolate the heating circuits at the manifold, leave the floor to equalise for several hours, then check whether the warm area persists. A patch that stays warm with the heating circuits isolated is not being warmed by the heating, and the candidate then becomes a hot water pipe or a primary flow and return passing through the same floor. That case is worked through in a warm patch on the floor and what it usually means.

Conversely, a cold zone is much more often a control or commissioning fault than a leak. Actuators fail closed, flow meters get shut down during a previous visit and never reopened, and air collects at the high point of a long circuit and will not shift on a normal fill. Those are covered separately in underfloor heating cold spots, and they should be eliminated before anyone talks about the screed.

The system you have decides the investigation

Underfloor heating is not one construction. What sits between the pipe and the room changes which methods work and what a repair involves.

Construction	Where the pipe sits	Investigation implication	Repair implication
Pipe in a sand and cement screed	Encased, typically 50mm to 75mm cover	Slow thermal response, good for imaging once charged	Screed cut out locally, pipe jointed, screed reinstated
Pipe in a liquid anhydrite screed	Encased, thinner cover	Faster response, sharper thermal image	Repairs need compatible reinstatement material
Routed insulation or castellated panel, floating floor over	In a panel below the finish	Floor covering can often be lifted instead of cut	Much less destructive
Spreader plates between joists	Below the deck, in the void	Access from above or from the ceiling below	Usually straightforward once found
Pipe within the structural slab	Cast in	Deep cover, weak thermal signal, tracer gas preferred	Serious work; a re-route is often the better answer
Electric matting	In tile adhesive or screed	Not a water system at all; faults are electrical	Insulation resistance testing, not leak detection

Pipe material matters too. Most modern circuits are PE-RT or PEX with an EVOH oxygen barrier, which are flexible and largely joint free within the floor. Older installations may use polybutylene, or in some cases copper, which has joints and corrodes. A circuit that is genuinely joint free between manifold and manifold fails in only a small number of ways, and the commonest by far is mechanical damage from a nail, screw or core drill driven in after the floor was finished.

Proving there is a leak before anyone lifts a floor

The manifold is the reason underfloor heating is, in one respect, easier to investigate than ordinary pipework: every circuit can be isolated individually and tested on its own. That turns a vague suspicion into a specific circuit, and it is done before any thought is given to opening the floor.

Isolating and testing circuit by circuit

Each loop is closed at both the flow and return bars, then pressurised independently and watched over a fixed period with a gauge. A circuit that holds is eliminated. A circuit that loses pressure is the one. Where several circuits hold and one does not, the search area has just been reduced from a whole floor to one loop, and the manifold labelling or a thermal sweep will show roughly where that loop runs.

This is the same principle as the test carried out at installation. BS EN 1264-4, the installation part of the European standard for water based surface embedded heating and cooling systems, requires the circuits to be pressure tested before the screed is laid, with a minimum test pressure of 6 bar commonly cited and many installers testing higher and holding it for a full day. A system that passed that test and is failing now has been damaged or has developed a fault at a fitting, rather than being defective from new.

• Record the starting pressure, the time, and the ambient temperature, because a sealed system's pressure moves with temperature and a small drop on a cold night is not necessarily a leak.
• Test long enough. An hour is often not enough to distinguish a slow loss from thermal contraction.
• Eliminate the rest of the system first. A pressure loss shared by the radiator circuits and the boiler is not an underfloor heating fault.
• Check the expansion vessel and the pressure relief discharge before concluding anything. A vessel that has lost its charge produces exactly the symptom of a leak, and there is usually water at the relief pipe to prove it, as set out in why a boiler loses pressure.
• Where every circuit holds under test but the system still loses pressure, the fault is at the manifold, on the primaries, or elsewhere in the building.

Locating it: thermal imaging, then tracer gas

Once the circuit is identified, the job is to find the point. Two techniques do most of the work, and they are used in sequence rather than as alternatives.

Thermal imaging on a heated screed

The circuit is run hot with the others isolated, and the floor is imaged as it warms. The pipe run appears as a pattern of warm lines, which is useful in itself because it maps the loop. A leak commonly shows as a locally hotter or more diffuse area where escaping water is spreading heat into the screed, or as a disruption in the regularity of the pattern. Timing matters: the image is most informative during the warming phase, before the whole floor reaches a uniform temperature and the contrast disappears.

Tracer gas into a drained circuit

Where the thermal image is ambiguous, or where the cover is too deep to produce contrast, the circuit is drained and a hydrogen and nitrogen mixture is introduced under pressure. Hydrogen is a very small molecule and migrates up through screed, tile grout and floor joints to the surface, where a detector follows the rising concentration to its strongest point. It does not depend on the leak making a noise or on any temperature difference, which is why it succeeds on the cases thermal imaging cannot resolve. Our tracer gas leak detection page sets out how the gas is introduced and traced.

Method	System state required	What it produces
Circuit by circuit pressure test	Each loop isolated at the manifold	Which circuit is losing pressure
Thermal sweep, circuits run individually	Circuit hot, others isolated	The route of the loop, and candidate anomalies
Thermal imaging during warm up	Cold start, then heat applied	Best contrast for a spreading leak
Acoustic listening	Circuit pressurised	Occasionally useful on a substantial loss under thin cover
Tracer gas	Circuit drained and isolated	A surfacing point, marked on the floor
Moisture mapping of the finish	None	The extent of wetting, useful for the drying scope

What these methods cannot do

Both techniques have well defined limits, and knowing them is what stops an investigation running past the point of usefulness.

• Thermal imaging does not see pipes or water. It sees surface temperature. Deep cover, thick insulation under a floating floor, thick stone or a heavy rug all flatten the pattern to the point where nothing can be read from it.
• A uniformly hot floor is a useless image. Once the screed has equalised, the contrast that makes the leak visible has gone. Imaging a floor that has been running all day usually wastes the visit.
• Tracer gas needs a path to the surface. A fully sealed impermeable covering with no grout lines or perimeter gaps will resist it, and the gas may then surface some distance away at the nearest joint.
• Tracer gas requires the circuit off and drained. That means no heating in that zone during the survey, and a refill and re-vent afterwards.
• Two breaches read as one. Where a floor has been drilled more than once, the second leak is often only found after the first is repaired and the circuit still fails its test.
• Acoustic methods are weak here. Plastic pipe under screed transmits leak noise poorly, and small losses at heating system pressures are frequently silent.
• None of them tells you what the floor finish will tolerate. Whether a stone floor can be lifted and relaid is a question for whoever laid it, and it often decides the repair strategy more than the location does.

Repair options, from one joint to a re-route

Once there is a marked point, the choice is between repairing in place and bypassing the damaged section. The right answer depends on the construction, the finish and the age of the system.

Option	When it fits	What it costs you
Cut out locally and joint the pipe	Single point of mechanical damage, screed construction	One opening, a buried joint, screed and finish reinstatement
Lift the floating floor and repair	Panel or spreader plate systems	Little or no destructive work
Bypass the damaged section with a new surface run	Damage near the perimeter, or where cutting is undesirable	Visible pipework, or a new chase
Abandon the circuit and re-route a replacement loop	Multiple failures, or a circuit that cannot be located reliably	Significant work, but a joint free result
Cap the circuit and rebalance the rest	A small zone whose heat output can be met another way	Loss of output in that zone
Replace the floor build up	Old failing system with widespread damage	The largest option, usually only justified alongside other work

A buried joint deserves a word. A properly made compression or press joint in a heating circuit under screed will normally outlast the surrounding building, but it is a joint in a system that previously had none, and its position should be recorded on a plan and photographed before the screed closes over it. Nobody remembers where it is in eight years, and the next investigation will start from that plan.

The failures that are not in the floor

A meaningful proportion of underfloor heating leaks are at the manifold, in the cupboard, entirely visible, and missed because everyone assumed the problem was buried.

Compression fittings at the flow and return bars loosen with thermal cycling. Flow meter cartridges and their O rings weep. Actuator heads fail and mask a weeping valve body underneath. Automatic air vents pass. Isolating valves seep at the spindle. Blending valve bodies fail. Almost all of these produce a very small loss that evaporates from a warm cupboard floor without leaving a pool, which is exactly why they go unnoticed while the pressure gauge keeps falling.

The check is tedious rather than difficult: dry every fitting thoroughly, put clean paper towel under and around each one, run the system up to temperature and come back after a few hours. Staining on the towel identifies the fitting. Doing this first costs an hour and regularly saves a floor. The same principle applies further out in the building, where a pressure loss attributed to underfloor heating turns out to be a leak in the pipework under a solid floor that has nothing to do with the heating circuits.

After the repair: drying, refilling and recommissioning

The repair is not the end of the job, and the steps after it are the ones that get skipped when a room is wanted back.

The wet screed and any wet insulation need drying to measured readings before the floor finish goes back. A screed that has been saturated holds moisture for a long time, and closing an impermeable finish over it produces adhesive failure, cupping in timber and mould at the perimeter. Readings, not appearance, decide when it is ready, and a dated drying log is what shows that.

The circuit is then refilled, vented properly at the manifold and, on a system that has been open, treated. A system that has been drained and refilled has fresh oxygenated water in it and the inhibitor has been diluted or lost. Underfloor heating circuits are particularly vulnerable to the consequences, because the pipe runs are long, the flow velocities are low and debris settles rather than circulating. Finally the circuits are rebalanced at the flow meters, because isolating and refilling a single loop disturbs the balance of the whole manifold. A full account of how the system is commissioned and tested is on our underfloor heating leak detection page.