Telling a failed unit from a failed installation

Three places a problem can live, and only three

Draw a line round the glass and a second line 300mm outside the frame. Everything that goes wrong with a rooflight sits inside the glass, between those two lines, or in the room below.

Inside the glass means the sealed unit: two or three panes with a spacer and two seals holding the sandwich together. When that fails you get cloudiness, streaks or droplets in a place you cannot wipe, because it is on the cavity faces of the panes. Nothing outside the unit has gone wrong at all, and the roof is still weathertight.

Between the lines means the perimeter: the underlay, the flashings, the fixing pattern, the upstand height on a flat roof, and where the covering terminates. That is where water from outside gets in, and it is the part of the job that is buried by the time anyone sees it.

In the room means moisture the household generated, landing on cold glass or on a cold strip of plaster at the reveal. No water crossed the roof at all. Getting this third one wrong is expensive, because people replace glass that was working perfectly and the wet mornings continue.

Timing is the strongest single clue you have

Before anything else, write down when the water appears. Not roughly. Actually note it, over ten days or so, with what the weather was doing.

Water that appears during rain and for an hour or two afterwards, and never otherwise, came from outside. Water that appears on cold clear mornings with no rain at all, and is gone by lunchtime, is condensation. Cloudiness that is there every day regardless of weather, though more obvious when it is cold outside and the sun is low, is inside the glass. Those three patterns almost never overlap, and ten days of notes will separate them more reliably than a survey conducted on one dry afternoon.

The awkward case is a February week when it rains at night and the mornings are cold, because all three can present at once. That is why the log matters. Two dry cold mornings in the middle of it will tell you whether the water arrives without rain.

Where the water actually sits, read position by position

Position is the second discriminator, and it is almost as good as timing.

Droplets across the whole room-side face of the pane, evenly, are condensation on cold glass. A band of water following the frame all the way round, 30 to 80mm wide with the middle of the pane clear, is the edge zone of the glazing going below dew point, which is a spacer and frame effect described in full on warm edge spacers and where condensation starts. Cloud or streaking that stays put when you wipe the inside and the outside is between the panes.

Water running down the inside face of the frame, on the timber or the aluminium rather than on the glass, has come past the gasket or in over the head of the unit. Water appearing at the junction between the frame and the plaster, with the frame itself dry, has come round the outside of the frame rather than through it. And a wet mark in the plaster 100mm or more away from the frame, with everything nearer the glass dry, is usually not a rooflight problem at all: it is water running down a rafter from somewhere higher up the slope and arriving at the first thing that interrupts it.

Reading the shape of a mark on a ceiling

Marks on plaster carry more information than people expect, because plaster records history and glass does not.

A mark with a hard brown tide line at its edge, dry to the touch and unchanged for months, is old. Water arrived, spread to that boundary, and stopped. It records an event, not a condition. A mark with a soft diffuse edge that darkens after rain and lightens in a dry spell is live. A ring inside a ring, two or three tide lines nested, is a recurring event with a wet season behind it, which usually points at wind-driven rain from one direction rather than a continuous defect.

Black spotting rather than brown staining is a different mechanism entirely. Mould needs a surface that stays above about 80 per cent relative humidity for weeks; it does not need liquid water. Black spotting in the top corners of a reveal, appearing every February and fading by May, is a cold bridge where insulation stopped short of the frame, and it is covered on thermal bridging at the rooflight perimeter.

Wind direction, and the fault that only shows on a south westerly

Gloucestershire’s weather arrives from the south west, up the Severn and over the Forest of Dean before it reaches the vale. If a rooflight only ever wets during a south westerly, that is a strong finding.

Driving rain against a slope does not run straight down the covering. It travels sideways across it and it climbs laps that would be entirely adequate in still air. A side flashing with a lap that is 25mm short, or an apron that stops one tile too early, will hold water back for years of ordinary weather and let it past three or four times a winter when the wind gets behind it. The same is true of an upstand on a flat roof that is 100mm rather than 150mm: standing water plus wind pressure gets over it, and calm rain does not.

Note the wind as well as the rain in the log. A defect that only appears when the wind is from one quadrant is a lap or a height problem on that side of the unit, and it tells the surveyor where to open the covering first.

Cold mornings without rain, and the moisture the house made

A household of four puts something in the order of ten litres of water into the air every day from cooking, showering, washing and breathing, and more again if clothes are dried indoors. Warm air carries it, warm air rises, and the coldest surface in a loft room is the glass over your head.

So a roof window running with water on a clear January morning while every wall in the room is dry is physics working correctly rather than a defect. The test is simple. If the outside of the glass has frost on it and the inside has water on it, and it has not rained for two days, the water came from the room. Ventilation, the vent flap in the head of the unit and the shape of the reveal are what change it, and they are set out on condensation, ventilation and rooflights.

The dew point arithmetic that settles the argument

One calculation removes most of the doubt. A room at 20 degrees with 55 per cent relative humidity has a dew point of about 10.7 degrees, which means any surface in that room colder than 10.7 degrees will collect water. Push the humidity to 65 per cent and the dew point rises to about 13.2 degrees, and a great deal more of the room qualifies.

A cheap hygrometer left on the windowsill for a week is the most useful instrument a homeowner can own for this question. If it reads 65 to 75 per cent through a cold week, the glass was always going to be wet and no unit on the market would have stayed dry. If it reads 45 per cent and the glass is still streaming, the surface is far colder than it should be, which points at the glazing specification or at an air path bringing cold air to the back of the frame.

Cloudiness between the panes, and how to be certain of it

Misting inside the cavity is the easiest of the three to confirm and the one people most often doubt. Two checks settle it.

First, wipe both faces you can reach. The room side is easy. The outer face can be checked with a hosepipe or simply after rain, when the outside is wet and the cloudiness is still visible in the same pattern. If a haze survives a clean inner face and a wet outer face, it is on a surface you cannot touch, which leaves only the two cavity faces.

Second, look at the pattern in raking light. A failed unit fogs in a characteristic way: clear in the middle on a mild day, hazy at the edges, and fully clouded when the outer pane is cold and the sun is on it. Over years it leaves a permanent white deposit that does not clear even in warm weather, which is mineral bloom etched into the glass. The mechanism, and how long a unit lasts before it happens, is set out in full on misted sealed units and why they go.

Symptoms that read as water and are actually mechanical

A fair number of enquiries described as water turn out to be the unit itself wearing out, and the two feel similar from the room.

A sash that has dropped out of square no longer compresses its gaskets evenly, so it draughts along one edge and drives fine rain past the seal in a gale. A perished gasket, hard and shiny with no spring left in it, does the same thing more evenly. A blocked drainage channel in the outer frame turns the channel from a drain into a gutter, and it overflows inward at the low corner. A hinge with worn bearings lets the sash lift a millimetre at one corner under wind load, which is enough. And a top hung unit whose stay has bent will not latch fully, which the occupant reads as a stiff catch rather than as an open window.

The test is to close the unit on a strip of paper and pull. If the paper slides out easily at one point on the perimeter and grips everywhere else, the compression is uneven, and that corner is where the draught and the water are.

Flat roof units, where the whole picture changes

On a flat roof, water does not have to run anywhere, and that changes every one of the signals above.

Water at the low corner of a flush glazed unit an hour after heavy rain, and nowhere else, is ponding standing against the kerb rather than draining to the outlet. Water along the whole low edge is often a blocked weep hole in the frame rebate, so the channel that should empty onto the glass is holding a shallow tank against the edge seal. Water at the high corner is more likely to be the waterproofing termination over the top of the kerb, because that is the corner that stays wettest longest when the fall is shallow. And a dark stain in the ceiling plasterboard that spreads slowly over days rather than appearing during rain is water inside the deck build-up, travelling along the insulation before it finds a joint.

Falls, ponding and outlet positions are on flat rooflight falls, drainage and ponding, and the height question is on how high should a rooflight kerb be.

When both have failed on the same roof, which is common

The two are not mutually exclusive and on a thirty-year-old installation they usually are not.

A unit glazed in 1993 has a sealed unit at the end of its service life, gaskets that have gone hard, lead that was face-nailed down both sides and has cracked along the nail line, underlay slit rather than dressed, and insulation compressed flat for 100mm all round the frame. Every one of those is at the same stage because they were all installed on the same morning. So the honest finding is often that the misting is what the homeowner noticed and the perimeter is what actually needs the attention.

That is why a diagnosis is written down as a list of findings with a photograph against each rather than as a single sentence. It also decides whether new glass into the existing frame is sensible, which is a real option on a sound frame under about fifteen years old, or whether the whole unit and its perimeter come out together. What tends to be waiting under an old frame is set out on what is found when the old unit comes out.

Which finding turns up on which Gloucestershire roof

Thirty miles around Gloucester covers enough different construction that the likely answer changes with the postcode.

On the trussed estates at Quedgeley, Abbeydale, Bishops Cleeve and Ashchurch, the units went in with the build, between trusses at 600 centres, with a trimmer already there. The perimeter is usually sound and the finding is almost always the glass, arriving across a whole elevation within two or three years because it was all glazed in the same week. On the interwar hipped semis of Longlevens, Hucclecote and Warden Hill, concrete interlocking tiles are long and a cut one at the side of an opening carries a lot of leverage in a gale, so a loose side tile over a slit underlay is the classic wind-direction finding.

Across the Severn Vale terraces of Gloucester and Tewkesbury, the rear addition roof is shallow and short, and the recurring finding is a head flashing with nothing much to lap under because the opening was set too high in a 2.1 metre run of slope. In the Forest of Dean, around Coleford and Cinderford, the weather does the work: steeper, wetter, more exposed, and laps that were at the manufacturer’s minimum start letting water past after fifteen winters. And on Cotswold stone slate through Painswick, Bibury and Northleach, the finding is very often a cement fillet run round the frame instead of soakers, cracked along its length, with water tracking down behind it into the batten ends.

Central Cheltenham is its own case. Behind the Lansdown, Montpellier and Pittville parapets the roof is a shallow lead or felt flat, invisible from the street, and the unit on it is usually a dome on a low kerb. There the finding is height rather than workmanship: 75 to 100mm of upstand where standing water and wind-driven rain need 150.

What each answer means for the work, and how it gets priced

The three diagnoses lead to three different pieces of work, and the difference between them is large enough that it is worth being sure before anything is ordered.

A failed sealed unit in a sound frame under about fifteen years old can take new glass into the existing frame, which is the smallest version of this job. A failed sealed unit in a frame of the same age as the glass is a complete unit, because the gaskets, the sash and the flashing kit are all at the same point in their lives. A perimeter finding means the covering comes back, the underlay is renewed locally and dressed properly, the flashing kit is renewed and lapped in the direction of flow, and the insulation and vapour control layer are reinstated tight to the frame, which is why that work takes as long as a new installation. A room air finding involves no roof work at all: a vent flap that gets used, a horizontal lining under the cill so warm air washes the glass, and continuous insulation round the reveal.

Two figures follow, in this order. An indicative range by email from photographs, the data plate and a description of when the water appears. Then a fixed price after a survey, itemised so it is clear what is unit, what is labour, what is scaffold and what is Building Control. The workmanship carries a ten-year guarantee, and the glass and frame carry the manufacturer’s own warranty as a separate document with its own terms. Send the photographs and the fortnight of notes to info@severnskylightsgloucester.co.uk, and the process is set out on the costs page.