What causes condensation on a rooflight?
Condensation on a rooflight is warm indoor air meeting cold glass. Three conditions have to line up: moisture in the air, a surface below the temperature at which that air gives up its water, and not enough air movement to carry the moisture away before it lands. Overhead glass satisfies the second condition better than anything else in the house, which is why the rooflight is where the water shows up first even though the rooflight is not what caused it. Almost none of this is a fault in the unit. The strategy for fixing it, including air change rates and what a house actually generates, is on condensation, ventilation and rooflights. This page is the short diagnosis: what is happening, and which of several different problems you are looking at.
Three conditions, all of which have to be present
Air holds water as invisible vapour, and how much it can hold depends entirely on its temperature. Air at 20 degrees and 55 per cent relative humidity holds around 9.5 grams of water per kilogram. Cool that same air and its capacity falls, and at some point the capacity equals what is already there. That temperature is the dew point, and for those conditions it is about 10.5 degrees.
Any surface in the room below 10.5 degrees will collect water out of that air. Raise the temperature of the surface above the dew point, or lower the dew point by removing moisture, and nothing forms. Those are the only two levers, and everything anyone does about condensation is one or the other.
The glass is colder than every other surface in the room
A plastered wall on a cavity construction might sit at 18 degrees on a January night. A modern sealed unit in a roof runs at 14 to 16 degrees at the centre of the pane and colder around the edge. An older double-glazed unit at 2.8 W/m2K can drop to 9 or 10 degrees, which puts it straight below the dew point of an ordinary living room.
Two things make roof glass colder than a window in a wall. Heat leaving a horizontal or shallow surface travels upward, which is the direction convection wants to go anyway, so the insulating layer of still air on the room side is thinner and less effective. And the glass faces the open sky rather than the house opposite, so on a clear night it radiates to an effective sky temperature that can be 10 to 15 degrees below the air temperature. That is the same effect that puts frost on a car roof and not on its doors, and it is why the worst mornings for condensation are the clear still ones rather than the coldest ones.
Which side of the glass the water is on tells you which problem it is
This is the first question worth answering, because it separates three unrelated things.
| Where the water is | What it is | What changes it |
|---|---|---|
| Room side of the inner pane, wipes off | Ordinary surface condensation | Ventilation, warmer glass, reveal shape |
| Outer face of the outer pane, early morning | Dew on a well-insulated unit radiating to a clear sky | Nothing. It clears by mid-morning |
| Between the panes, never wipes off | The sealed unit has failed | New glass, or a new complete unit |
| In the plaster around the frame, worst in winter | A cold bridge at the perimeter | Insulation and vapour layer reinstated to the frame |
Dew on the outside surprises people and is worth explaining rather than worrying about. It happens on good units, not bad ones: the better the glass, the less heat leaks through it to keep the outer pane warm, so the outer pane cools to the sky and collects dew exactly as grass does. It is a sign the specification is working.
Water between the panes is a different fault entirely
A sealed unit is two or three panes separated by a spacer bar, with a desiccant inside the spacer and two seals around the edge. When the primary seal fails, humid air migrates into the cavity, the desiccant saturates, and from then on the cavity fogs whenever the outer pane goes cold. There is no way to dry it out again.
Roof units reach that point sooner than vertical ones because the edge seal cycles through a bigger daily temperature range, spends more of its life wet, and takes more ultraviolet. Fifteen to twenty-five years is a normal service life. The answer is a new sealed unit into a sound frame or a new complete unit, covered on replacing a failed or misted rooflight. The part of the unit where this begins is the glass edge, described on warm edge spacers and where condensation starts.
Damp in the plaster around the frame is a third thing again
A soft grey or black mark in the top corners of the plastered reveal, appearing in November and fading in April, is not the glass. It is a cold bridge: insulation stopping 20mm short of the frame, or squashed flat behind the plasterboard, or a vapour control layer trimmed to the edge of the board rather than taped onto the frame. That leaves a strip of plaster running several degrees colder than the rest of the room, and mould grows on plaster at around 80 per cent surface humidity, well before liquid water appears.
It looks like a damp problem and it is a construction detail. The junction is set out on thermal bridging at the rooflight perimeter, and the sealing of the layer itself on vapour control layers sealed to the frame.
Where the moisture came from in the first place
A household of four puts something in the order of ten to fifteen litres of water into the air every day. Breathing and perspiration account for a surprising share of it. Showering, cooking, boiling a kettle, washing up and drying clothes indoors do the rest, and a single load of washing dried on an airer releases around two litres over a day.
Warm moist air rises, so it collects at the highest point of the room. In a loft conversion the highest point is the roof window. In a kitchen with a lantern it is the ridge of the lantern. That is why the water lands there before it lands anywhere else, and why an unvented tumble dryer or a clothes airer directly under a rooflight produces a spectacular result on the first cold morning.
Nights when it will happen whatever you specify
A clear, still, cold night after a wet day is the combination. The sky is open so the glass radiates freely, there is no wind to disturb the still air layer against it, and the outside air is already close to saturation. In the Severn Vale, where river mist sits in the low ground through Longford, Maisemore and the villages along the river, those conditions arrive repeatedly between October and March.
On those mornings a light film of condensation on the lower part of the pane, gone by ten o’clock, is normal behaviour from correctly specified glass. What is not normal is water running down the frame and pooling on the cill every morning through the winter, or condensation that is still there in the afternoon. That points at a genuine moisture excess in the house, an old unit, or both.
What reduces it, in the order worth trying
Deal with the moisture first, because it is free. Extract at source in the kitchen and bathroom, with the fan running on its overrun rather than switched off at the door. Dry washing outside or in a vented machine. Open the background vent in the head of the roof window and leave it open, which is the single most commonly undone control in a house.
Then the surface temperature. Warmer glass condenses less, so an old unit at 2.8 W/m2K replaced with one at 1.2 moves the inner pane up several degrees and can take it clear of the dew point on all but the worst nights. A warm edge spacer moves the coldest strip along the edge of the pane up by a further degree or two.
Then the geometry. A horizontal lining below the window lets warm air from a radiator rise straight up the face of the glass. A splayed cill traps a pocket of still cold air against it and guarantees a wet edge. That detail is on rooflight reveals and why splaying them matters, and the whole-unit thermal question on what U-value should a rooflight have.
Severn Vale winters, and the rooms that show it first
Gloucester’s Victorian terraces have narrow rear additions with the kitchen underneath, often with one small window and a fan that vents into a duct nobody has looked at since it was fitted. Add a rooflight to bring daylight into that room and the glass becomes the coldest surface in the wettest room in the house. It is a good place for a rooflight and it is a room where the ventilation has to be sorted at the same time rather than afterwards.
Loft conversions are the other case, because converting a loft removes the ventilated cold space that used to buffer the whole roof and replaces it with a heated room whose ceiling is 100mm from the outside. Bathrooms with no external wall, common in Gloucester and Cheltenham terraces, are the third. The survey records what the room is used for, what extraction exists, and what the reveal will look like, because the glass specification alone will not settle it. The full picture is on the energy efficient rooflights hub.
Next on this subject.
- Question What U-value should a rooflight have? Ask for a whole-unit figure of 1.2 to 1.4 W/m2K on a manufactured roof window, and 1.0 to 1.3 on…
- Question Will a rooflight make my room too hot? It might, and whether it does comes down to four things: which way the slope faces, how steeply it is…
- Question Are rooflights bad for insulation? Per square metre, yes. A current rooflight loses somewhere between eight and ten times as much heat as the insulated…
- Question Is triple glazing worth it in a rooflight? Sometimes, and the deciding factor is usually not the U-value. A triple unit takes a good rooflight from about 1.2…