Hidden Water Leaks Behind Walls & Under Floors in Dubai

Most apartment, villa and townhouse supply pipework here is concealed by design. Cold and hot runs are chased into hollow block walls and plastered over. The branches that feed a bathroom or a kitchen are laid on the slab and buried in screed before the tiles go down. The risers climb the building inside a shaft the resident never sees. The pipe itself is usually PPR, joined by heat fusion, and a fusion joint that was under-heated or pushed home crooked on the day of the fit-out can hold for years before it opens. A weep that shows during snagging, before handover is signed, costs a re-made joint; the same joint found after the tiles have been down for two years costs a slab opening.

Three things then make the leak hard to see. Water follows the screed and the slab soffit rather than rising to the surface, so it travels along the floor, or along the ceiling below, until it meets a joint, a light fitting or a change of level. The finishes hide the early signs: porcelain tile does not stain, and a plastered wall only bubbles once the plaster has been wet for weeks. And the cavity is warm. A cold-water supply running through a wall cavity or a ceiling void in August sits below the dew point of the air around it and sweats, which produces a wet wall with no leak at all, and it has to be ruled out before anyone cuts.

Wet rooms add the membrane. Under the tiles and the screed of a bathroom there should be a continuous waterproofing layer. When it is torn at a floor drain or stops short at the shower kerb, the water from every shower goes into the screed and appears in the ceiling below exactly as a pipe leak would. Podiums and common areas add a third source: a planter, a pool deck or a landscaped podium above a car park drains into the structure and surfaces in the units and corridors beneath as a leak that nobody in the flat can explain.

Stacking makes the riser and slab-above cases common. Studios, 1 BHK, 2 BHK and 3 BHK units are stacked floor over floor with the same layout, so the wet room of the flat above sits directly over yours, and a failed joint or membrane upstairs arrives as a stain on your hall ceiling beside the shaft wall.

The meter answers the first. Every outlet in the property is closed, the roof tank is isolated so its float valve cannot refill and imitate a leak, and the DEWA meter is watched. Movement means a supply-side loss somewhere after the meter. No movement means the wet patch is fed by a drain, a membrane or a condensate line, and the visit changes course: dye down each fixture, a push-rod camera along the drain, a dammed flood test on the wet-room floor and a moisture grid on the ceiling of the flat below.

The pressure test answers the second. Hot and cold are shut off from each other and from the water heater, the line is charged and the air is bled out, because trapped air compresses and produces a gauge drop that looks like a leak. Each circuit is held at test pressure and read at the start and at the end. The one that fails to hold is the one that will be listened to. In a villa with a booster pump the pump is stopped for the test. In an apartment on a tank-fed gravity supply the static head is often so low that a leak barely makes a sound, so the circuit is pressurised from the test rig to give the microphone something to hear.

A hot-water branch leaking under the screed warms the tiles above it, and a warm plume on a marble floor with no underfloor heating beneath is one of the clearest thermal signatures there is. A cold-water or chilled-water leak, and water evaporating from a wet wall, both show cold. The image is taken once the room has been at a steady temperature for a while; a floor that had sun on it an hour earlier is still carrying that heat and prints a pattern of its own.

The moisture grid then puts numbers on the picture. A pinless meter is swept across the thermal footprint at fixed spacing, and a pin meter confirms the wettest readings and a dry reference point in the same material. Concrete and screed carry reinforcement mesh, and a pinless meter reads metal as wet, so a high reading over a slab is accepted only once the pin meter agrees. The finished map shows the extent and the direction of travel: readings that fall away from a high point say which way the water came from, and on a screed that is usually uphill of the wettest tile.

An electronic sensor pressed to the surface hears it. The technician begins where the pipe can be touched: the isolation valves, the meter, the tap tails behind the basin, the exposed length in the shaft. A contact microphone at each one says which section of the run is carrying leak noise, so the surface search covers one branch rather than the whole floor.

Then the surface search itself. A ground microphone is set on the tile directly over the traced route and moved along it in steps, and the gain is set once and left alone, because a reading that is louder only because the volume was turned up is worthless. On a wall chase or a ceiling void the same sensor is pressed to the plaster over the route and walked the same way. At each point the technician reads the level on the meter and listens through headphones. The point of strongest response is marked on the floor. Higher frequencies fade fastest, so the pitch rises as the sensor approaches the leak and drops again past it, which is a second confirmation of the peak. Bandpass filters strip the pump, the traffic and the compressor out of the signal so that the leak's own band is what is being read.

Concrete helps here. A hard slab resonates and carries the vibration to the surface, whereas grass, sand and loose fill absorb it, which is why a supply main running through a floor slab is one of the best acoustic cases in a building and one of the worst in a villa garden.

One sensor is fixed to the pipe upstream of the suspect length and one downstream, wherever it can be touched: a valve, the meter, an exposed length in the shaft. Each records the leak noise arriving along the pipe wall. The instrument cross-correlates the two recordings, finds the time offset at which they line up best, and converts that offset into a position: the leak is nearer to whichever sensor heard it first, by an amount fixed by the speed of sound in that pipe. For the calculation the technician enters the pipe material, its diameter and the true route length between the sensors. On a straight run these are known. On a slab run that doglegs around a column they have to be traced first, because a wrong length or a wrong material puts the mark in the wrong place.

What timing adds over listening. Loudness falls with distance and with everything the sound has to pass through, so a microphone reads a leak under thick screed as quieter than the same leak under thin screed, and the technician allows for that by ear. Arrival time is indifferent to how loud the leak is. Neither sensor has to be near the leak; both can sit at the far ends of a run, one in the shaft and one in a kitchen cupboard, with the whole floor between them, and on a metal main the two can be hundreds of metres apart. Background noise from a booster pump or a chiller is narrowed out by frequency band before the two recordings are matched, and because the answer is a computed offset rather than a level heard through headphones, the operator's ear stops being the judge. The result is a figure, so many metres from sensor A, that a second technician can reproduce. When two joints on the same run are both leaking, the correlator can separate them by frequency band and report both, where the ear locks onto the louder one and never hears the second.

Sensor choice follows the pipe. On copper, galvanised steel and cast iron, accelerometers clamped to the pipe wall pick up the structure-borne signal cleanly and the spacing can be generous. On plastic, where the wall damps the vibration within a few metres, the sensor is a hydrophone screwed into a tapping so that it listens to the pressure wave inside the water column instead, which travels much further than the wall vibration does. On a PPR branch inside a 2 BHK the practical version is two accelerometers a room apart, one on the isolation valve and one on the basin tap tail, with the route between them traced by the wall scanner first so that the length entered is the length the pipe actually runs.

The speed of sound is the weak input. It depends on the material, the diameter and the wall thickness, and published sound-velocity tables for a polyethylene pipe of the same size range from roughly 260 to 380 metres per second depending on whose table is used. The calculation also assumes that the noise reaching both sensors is the same noise, steady for long enough to be matched, so a leak that surges as a pump cycles is recorded across several cycles. For both reasons the correlator's answer is cross-checked against the ground-microphone peak before a tile is lifted. When the two agree, the mark stands. When the two marks sit further apart than any water could have tracked, the technician re-walks the route with the scanner, corrects the length, and listens again.

Where it earns its place. A long branch under a slab with fitted joinery over it. A riser section between two floors, correlated from the isolation valves on each landing. A run where two joints are noisy and the ear cannot separate the peaks. A plastic main in a villa garden, from two fittings, where the ground microphone hears nothing through the sand. On a short run with clear tile above the pipe, the microphone alone is faster and just as accurate, and the correlator stays in the van.

Plastic damps it. PPR, PEX, uPVC and HDPE absorb the higher frequencies, leave a low-pitched signal and carry it only a short distance. Since PPR is the usual supply pipe on a Dubai fit-out, that is the case on most apartment jobs: the technician listens closer to the run, in shorter steps, with a low-pass filter, and reaches for the correlator or for tracer gas sooner. The joint failures on PPR are point sources at a fusion joint, which suits correlation well when there is access at both ends.

Reported frequency bands put metal-pipe leak noise between roughly 500 and 1,500 hertz and plastic-pipe leak noise between roughly 70 and 850 hertz, and the same difference drives the correlator's velocity input, which is why plastic runs are cross-checked against the microphone and metal runs seldom need to be.

Low pressure. Below about one bar there is little turbulence at the orifice and almost nothing to hear. That describes the tank-fed apartment on a low floor, where the head above the tap is a few metres of pipe, and the villa whose booster pump has been switched off for the test. The fix is simple. The circuit is pressurised from the test rig and listened to again.

A small weep. A hairline at a fusion joint can lose enough water to soak a screed over a month and still be silent. This is the tracer-gas case. The branch is emptied and capped, filled with a hydrogen and nitrogen mixture (5 percent hydrogen, 95 percent nitrogen, non-flammable at that ratio) to its normal working pressure, and the detector is walked over the tiles: the gas leaves through the same hairline as the water and climbs through screed and grout to the surface.

Depth and cover. Surface pinpointing works to roughly one and a half to two and a half metres of cover for a leak losing a few litres a minute, with hard concrete at the top of that range and soft or waterlogged ground at the bottom. A deep run in a villa garden is correlated from two fittings or traced with gas.

Plant noise. Chillers, booster pumps, lifts and traffic sit in the same band as the leak. Filters suppress the steady part of it, and the correlator, which is a calculation on two channels rather than an ear on one, is far less affected than the microphone.

No pressure at all. Drains, soil stacks, floor traps, shower outlets and AC condensate lines run by gravity and are silent. The tools for those are the dye test, the drain camera, the flood test and the moisture map, and the meter test at the start of the visit is what sends the job down that path. If the drip follows the air conditioning, the cause is on the AC leaking water page rather than on this one.

Before any cut, the route is scanned with our wall and post-tension scanner. The scan shows the pipe, the reinforcement and any conduit sharing the chase, so the opening lands on the joint and misses the cable beside it. Tower slabs add the tendons. In a post-tensioned floor the steel tendons drape through the slab depth and change level across every bay, a cut tendon is a structural event, and the scan locates them before the mark is confirmed. Most tower management entities require an NOC and a work permit before a slab is opened, and the debris from even a two-tile opening leaves by the service lift, so the service-lift slot is booked with the permit and the scan and leak report go in with the application. The opening itself is a tile or two over the mark, or a strip of plaster the width of the chase. The first hole is small enough for a borescope, and the drip is seen on its screen before a second tile is touched.

After the repair the same instruments run in the other direction. The pressure test is repeated on the repaired circuit and must hold. A repeat thermal and moisture pass a few days later shows the readings falling, which is the evidence an insurer or a management entity accepts that the source was found. Making good and reinstatement of the finish follow: a matched tile, a re-plastered chase, a repainted patch. The repair itself is a separate scope, described on the leak repair page.

The riser in the shaft behind your maid's room wall, the slab over your hall and the chilled-water main serving the floor are the building's, maintained by the management entity from the service charge and covered by the liability policy Article 41 requires it to carry. Water from the fixtures of the flat above belongs to the owner of that flat, or to their tenant where neglect can be shown.

The order of the visit is built for this. Your own circuits are pressure-tested first. If they hold, your unit is cleared and the report says so in writing, with the meter reading, the pressures held and the moisture map showing water arriving from the soffit or the shaft wall. That report, with a dated notice, is what goes to the management entity, or to the Owners Association through it, and it is the document that moves the repair from your account to the building's. Article 42 sends disputes under the law to the Rental Disputes Centre, which will want the same evidence. Membrane failures on the floor above or on a podium are a waterproofing scope once they have been proved.

Tenants have the same protection one level down. Under Law 26 of 2007 the landlord carries the repair of anything that stops the tenant using the premises (Article 16) and stays liable for defects the tenant did not cause (Article 17). Article 21 is the end-of-tenancy clause: the premises go back in the condition they were received in, ordinary wear and tear excepted. A wall soaked by the building's riser is neither wear nor the tenant's doing, and a dated report is what keeps it off the security deposit at the move-out inspection.