How to Find a Flat Roof Leak: Flood Testing and Thermal Surveys

The single most useful fact about flat roofs is that the stain on the ceiling is almost never underneath the hole. Water entering a flat roof travels along the top of the deck, along the underside of the waterproofing, down a joist, sideways across an insulation layer, and enters the room below wherever it finds a fastening, a joint or a service penetration. The distance between the breach and the drip is regularly several metres, and occasionally it is on the other side of the building.

That single fact decides how the search has to be conducted. This guide sets out how water actually moves through a flat roof build-up, what to inspect before anybody tests anything, how staged flood testing works, when a thermal survey will and will not help, and what an investigation should leave you holding at the end.

Why the Stain Is Not Under the Breach

A flat roof is a layered assembly, and each layer gives water a different route. On a warm roof the waterproofing sits over insulation over a vapour control layer over the deck. On an inverted roof the insulation sits above the waterproofing under ballast. On a cold roof, which BS 6229, the code of practice for flat roofs with continuously supported flexible waterproof coverings, does not recommend, the insulation sits at ceiling level with a ventilated void above it. Each of those gives water a different amount of room to travel before it appears.

Keep a rainfall diary before anything else

Note the date and time each time water appears, and what the weather was doing. Three patterns separate three different problems. Water appearing during and shortly after rain is ingress. Water appearing hours after rain, or in wind-driven rain only, points at a detail such as an upstand, a parapet or a poorly formed outlet rather than a hole in the field of the roof. Water that appears in cold weather with no rain at all is condensation, which is covered by BS 5250, the code of practice for management of moisture in buildings, and is a ventilation and airtightness problem rather than a waterproofing one. Our guide to telling damp from a leak works through that distinction.

What to Inspect Before Any Testing

Most flat roof failures are at details rather than in the middle of the field, because details are where the membrane has been cut, turned, bonded to something else or penetrated.

• Outlets and rainwater gullies: blocked, undersized, or with a failed clamping ring.
• Upstands and skirtings: insufficient height above the finished surface, or bonded down and lifting at the top edge.
• Parapet copings and their joints, which shed water into the wall when a joint fails.
• Junctions with vertical walls, and the flashing or chase detail at that junction.
• Penetrations: soil vents, aerials, handrail fixings, air conditioning feet, cable entries.
• Laps and seams in the membrane, especially where they cross a fall.
• Ponding water, which is a symptom of insufficient fall and accelerates failure at any weak point it sits over.
• Mechanical damage: dropped tools, foot traffic to plant, and anything that has been screwed through the covering.

Look inside as well as on top

Where a roof void is accessible, a torch from inside tells you more than a walk across the surface. Water staining on the underside of a deck, rusted fixings, dark tide marks on joists and daylight at a junction all mark the route water is taking, and that route can be followed back uphill. On a warm roof with no void, this is unavailable, which is one reason warm roofs are harder to diagnose from below.

Flood Testing, and How It Is Staged

Flood testing means holding water on a defined area of the roof and watching the inside for ingress. Done properly it is the most conclusive test available on a flat roof, because it reproduces the failure rather than inferring it. Done carelessly it soaks a building and tells you nothing about where the water entered.

The delay that catches people out

Because water has to travel from the breach to wherever it can drop, ingress inside can lag the flooding of a zone by a considerable time. An observer who gives each zone ten minutes and moves on will clear a zone that is actually the guilty one. Allowing a realistic dwell time per zone is what makes the difference between a flood test and a wet afternoon.

Where flood testing is the wrong tool

It cannot be used on a roof with a fall so steep that water will not stand, on a deck whose condition is questionable, on a green or ballasted roof without stripping, or where the leak is driven by wind rather than by standing water. A defect at the top of an upstand that only admits water in horizontal rain will not be found by ponding water on the deck below it. In those cases a directed spray test, working upwards in sections with a hose, reproduces the conditions better.

Thermal Surveys and Why Timing Decides the Result

A thermal camera does not see water. It reads surface temperature, and it finds wet insulation because saturated material holds heat differently from dry material around it. That distinction governs everything about when a thermal survey is worth booking.

The window that actually works

The classic survey is carried out after a clear, sunny day, beginning after sunset. During the day the roof absorbs solar energy. After sunset the dry areas give that heat up quickly while wet areas, having far greater thermal mass, cool more slowly and show as warm anomalies against the cooled surround. The requirements are specific: enough solar gain during the day, a clear sky rather than cloud cover overnight, a dry surface at the time of the survey, and little wind. The infrared method for detecting thermal irregularities in building envelopes was set out in BS EN 13187, now replaced by BS EN ISO 6781-1, and the discipline both impose is that the conditions under which an image was taken are recorded alongside the image. A thermal picture with no stated conditions is not evidence of anything.

What a thermal survey cannot see

It cannot see through ballast, gravel or a paved terrace, because the camera reads the top surface and the anomaly is buried beneath it. It reads poorly on an inverted roof, where the insulation is above the waterproofing. It will not distinguish wet insulation from a warm services route or a poorly insulated section. And it locates saturated material, which is not the same as locating the breach: on a roof that has been wet for years, the saturated area can be extensive and the entry point a small defect at one edge of it. Our thermal imaging leak detection page sets out what a survey covers.

Other Methods and What Each One Finds

What None of These Methods Can Do

Flat roof investigation has harder limits than internal plumbing work, and most disappointment comes from expecting a method to reach past its own boundary.

Nothing finds an active leak on a dry roof in dry weather

Ingress needs water. If the roof is dry and the weather is settled, the choice is to wait for rain, to reproduce it with a controlled test, or to look for the consequences of past wetting rather than the leak itself. An inspection on a dry day that reports no leak found has established very little.

Saturation is not the same as the breach

Every moisture-based method, including thermal, maps where water has ended up. On a build-up that has been wet for a long time, that area can be large and its edges arbitrary. The breach is a small defect somewhere at or above it, and identifying it usually still needs a physical test or an opening.

One negative test rarely clears a roof

A zone that passes a flood test has been cleared for standing water at that depth on that day. It has not been cleared for driven rain, for thermal movement at a detail in cold weather, or for a defect above the flood line. Multiple failures on the same roof are common, particularly on older coverings, which is why a repair that fixes one entry point can be followed by a second stain a month later.

And no method dates the failure

The condition of the membrane, the extent of deck deterioration and the state of staining support a view on whether a failure is recent or long standing. None of it produces a date, which matters whenever the timing of the damage is in question.

What a Flat Roof Investigation Should Produce

The deliverable is not an opinion that the roof needs replacing. It is a document somebody can act on and somebody else can check.

• A roof plan with zones marked, and the result recorded for each zone including the negatives.
• The method used in each zone, with the conditions: date, weather, dwell time, water depth.
• Photographs of each defect found, referenced to the plan, plus internal photographs of the ingress point.
• A named defect and a described failure mode, not a general observation that the covering is at the end of its life.
• The distance and route between the defect and the internal damage, stated explicitly, because that is the question everyone asks.
• A scope of works separating the immediate repair from any wider recovering, and the internal making good from both.

Where the damage inside has to be recorded to the same standard as the roof itself, our guide to what a leak detection survey involves sets out what a written deliverable contains. If the roof in question is over a conservatory or a single storey extension, which fail in their own characteristic ways, our guide to a leak in a conservatory or extension roof covers those details, and if water is already coming through, start with what to do when water comes through the ceiling.