Energy Efficient Rooflights

What energy efficiency means once glass sits in a roof plane

In a wall, thermal performance is mostly a heat loss question. Put the same glass in a roof and it becomes two questions pulling against each other, because the roof plane is the surface with the clearest view of both the cold night sky and the high summer sun.

So the honest definition has three parts. How much heat leaves through the unit and its perimeter in winter. How much solar energy comes through it in July. And how much artificial lighting the daylight displaces across the year. A specification that wins on the first and loses badly on the second has not made the house more efficient, it has moved the running cost from the boiler to a fan.

The U-value on the brochure, and the one that matters

Glass manufacturers quote a centre pane U-value, which describes the sealed unit away from its edges. Whole unit U-values include the frame, the spacer and the edge of the glass, and they are always worse, because those are the parts that conduct.

The gap between the two is not small. A double glazed unit with a soft coat and argon can hit around 1.0 W/m²K at the centre of the pane and still land near 1.4 to 1.6 W/m²K as a complete rooflight, depending on frame depth and how much of the opening the frame occupies. On a small unit the frame is a large share of the area, so the whole unit figure drifts further from the glass figure the smaller the rooflight gets.

The plane a rooflight was measured in

Here is the subtlety that catches out even careful buyers. The same rooflight can be declared with two different U-values, and both are honest, because the test standards assess glazing at different angles and heat moves differently through a cavity that is horizontal rather than upright.

Convection inside a near flat sealed unit is more vigorous than in a vertical one, so the horizontal figure is the less flattering of the two. A unit quoted at 1.2 W/m²K in the vertical plane might be nearer 1.6 in the horizontal. Neither number is wrong. Comparing one supplier’s vertical figure with another’s horizontal figure is, and it happens constantly. Ask which plane the declaration refers to and compare like with like.

What Building Regulations set as the ceiling

Approved Document L sets a limiting whole unit U-value for rooflights, and in an existing dwelling that threshold is the one a replacement has to satisfy. Current modern units from the mainstream manufacturers clear it without effort, which is why the regulation rarely constrains a domestic job.

What it does do is create a paper trail. Fitting or replacing a rooflight is notifiable work, so the performance figure has to be evidenced against a certificate. Sunspire handles the notification as part of the installation, which means the glazing specification and the declared U-value are recorded at the point the work is signed off rather than reconstructed years later when a buyer’s solicitor asks. The wider process is set out across the rooflight and skylight services pages.

Why a roof plane loses heat faster than a wall

Three things stack up against a horizontal or shallow surface, and they are all physics rather than product quality.

Warm air rises, so the air in contact with a rooflight is the warmest air in the room, and the temperature difference driving heat through it is larger than at a window near floor level. Inside the sealed unit, a near horizontal cavity convects more freely than an upright one, which slightly undermines the gas fill. And the glass faces the open sky, so on a clear night it radiates to an effective sky temperature well below the air temperature, which is why frost forms on a rooflight before it forms on the tiles beside it.

None of this makes rooflights a poor idea. It means the specification has to work harder for the same result than the equivalent window in a wall.

Frame material, and the heat the frame itself carries

Frames conduct at wildly different rates, and on a rooflight the frame is proportionally larger than on a big window, so the choice moves the whole unit figure more than people expect.

The line worth noticing is the last but one. Aluminium without a proper thermal break is a heat path and a condensation site, and it is still sold. Where a slim aluminium sightline is the design intent, the depth and continuity of the break is the question to ask, not the powder coat finish.

Continuity is the operative word. A polyamide break twenty four millimetres deep does very little if a stainless fixing runs from the outer profile to the inner one at every corner, because the metal shorts across the barrier. Good profiles keep fixings on one side of the break or use an isolating sleeve where they cross it. It is not something you can inspect once the unit is assembled, which is why it belongs on the order rather than in a conversation on site.

Where triple glazing earns its place overhead

Triple glazing improves a rooflight’s U-value, typically taking a whole unit figure from the mid ones into the region of 0.9 to 1.1 W/m²K. It also adds weight, adds cost, cuts light transmission by roughly another tenth, and on a large unit it can push you into a heavier frame or a bigger opening structure.

It earns its keep in three situations. Where the room below is a bedroom or a home office occupied through the winter. Where the rest of the fabric is already good, so the rooflight is the weakest element and worth lifting. And where rain noise matters, because the extra pane damps as well as insulates.

On a lightly used utility space, or in a house where the loft insulation is thin and the walls are solid, the money returns more heat somewhere else first.

Triple glazing also has a side effect worth knowing about in advance. Because the outer pane is now insulated from the room behind it, it runs colder overnight, and external condensation on the outside of the glass on clear autumn mornings becomes common. It clears by mid morning and it is a sign the unit is working, but on a rooflight you are looking straight up at it, and nobody enjoys being surprised by it in the first week.

Gas fills, and the gap that suits a unit lying near flat

Argon in the cavity slows conduction and convection, and it is standard rather than an upgrade on any decent unit. Krypton performs better in a narrow cavity and costs considerably more, which is why it appears mostly in slim heritage style units where the sightline caps the gap width.

Cavity width has an optimum rather than a more is better curve. Around 16mm is the usual sweet spot for argon in a vertical unit, and for a near horizontal rooflight a slightly narrower cavity often performs better, because a wide horizontal gap lets convection circulate freely between the panes. A supplier who has thought about this will specify the gap for the installed angle rather than reusing the window figure.

The kerb, and the cold bridge built into it

Everything above concerns the unit. This section and the next two concern the roof, and in practice they decide as much.

A flat rooflight sits on a kerb, an upstand built off the roof structure and standing typically 150mm above the finished surface. That upstand is a projection of timber or metal sticking up out of an insulated roof, and unless it is deliberately insulated it forms a continuous cold bridge running right around the perimeter of the opening. The heat loss through it can rival the loss through the glass it carries, and it never appears on a quote, because it is not part of the product.

You can find it with a thermal camera on a cold morning. It reads as a bright rectangle outlining the rooflight, and it is entirely avoidable.

Insulating an upstand properly

The principle is continuity. The insulation line in the roof deck has to be carried up the outside of the kerb, or built into it, so that there is no uninterrupted timber or metal path from the warm inside face to the cold outer skin.

In practice that means an insulated kerb, either a proprietary unit with rigid insulation bonded into the construction or a site built upstand with insulation dressed up its outer face beneath the covering. The internal face is then lined and sealed to the ceiling. Where the roof build-up is warm deck, the detail is straightforward. Where it is a cold deck with insulation at ceiling level, the kerb sits outside the insulated envelope entirely and the detail needs thinking through before the opening is cut.

Air leakage around the opening

Conducted heat loss is calculable. Air leakage is not, and in an older Essex house it is often the larger number.

Every rooflight opening creates a junction between the ceiling membrane, the roof structure and the frame, and if that junction is left as a nominal fit with insulation stuffed into the gap, warm moist air moves through it into the roof build-up. That costs heat and it carries moisture to a cold surface, which is a durability problem as well as an energy one. The fix is cheap and it happens once: an airtightness tape or membrane taken from the frame onto the ceiling line, continuous around all four sides, before the reveal is boarded.

It takes an hour. It cannot be done afterwards without taking the lining off.

There is a test for it if you want one. A blower door pressurises the house and a smoke pencil held at the rooflight perimeter shows any path immediately. That is normal practice on a new build and rare on an extension, but if the house is being tested anyway, the rooflight junction is one of the two or three places worth walking the pencil around.

The reveal lining, and its surface temperature

A deep roof build-up gives a deep reveal, and that lining is a surface with a room on one side and a cold roof void on the other. Insulated, it stays near room temperature. Left as plasterboard on battens over the timber, its inner face runs cold, drags radiant comfort down and collects condensation at the corners.

Insulated plasterboard, or rigid board behind a skim, closes it. There is a comfort argument as well as an energy one. The surface temperature of a large area near your head influences how warm a room feels more than the air temperature does, which is why a room with cold reveals gets the thermostat turned up.

Solar control coatings, and the daylight they cost

Solar control is a coating on the sealed unit that reflects part of the near infrared while passing as much visible light as the chemistry allows. It is described by the g-value, the fraction of solar energy that gets through, where a neutral double glazed unit sits around 0.5 to 0.6 and a solar control unit around 0.25 to 0.4.

Where the coating sits matters as much as what it is. A sealed unit has four glass surfaces, numbered from the outside in, and solar control belongs on surface two, the inner face of the outer pane, so the reflected energy is rejected before it reaches the cavity. The low emissivity coating belongs on surface three, facing the cavity from the inner pane, where it returns long wave heat to the room. Put either on the wrong surface and the unit looks identical while performing considerably worse.

The trade is light transmission and a slight tint. Modern coatings are far more selective than the bronze tinted glass of thirty years ago, but selective is not free: expect somewhere near sixty to seventy per cent light transmission on a solar control double glazed unit against seventy five to eighty on a neutral one. In a north lit room that reduction is noticeable. On a south facing extension roof in July it is the difference between using the room and closing the door on it.

Shading from outside, and why it does more

An internal blind stops light after the solar energy has already passed the glass and entered the room. The heat is inside; the blind simply decides where it lands. An external awning or shutter stops it before it gets in, and the difference in effect is large.

For rooflights the practical options are an external fabric awning in a cassette above the glass, which cuts a substantial share of the gain and still lets diffuse daylight through, or a solar powered external shutter on units designed to take one. Both are ordered with the rooflight, because the mounting provision sits in the frame. Internal blinds remain worth having for glare and for blackout, and integrated cassette blinds are covered in more detail with the units themselves.

Purging heat at night without running anything

An opening rooflight at the highest point in a room is a heat exhaust, and it costs nothing to run. Warm air stratifies at ceiling level, so opening the rooflight and a window lower down sets up a stack that moves air continuously while there is any temperature difference at all.

Used at night in summer, this pulls the accumulated heat out of the structure and leaves the floor slab and walls cooler at breakfast, which flattens the following afternoon’s peak. It works best in rooms with exposed mass, a tiled floor over a slab or a plastered masonry wall, and it works barely at all in a lightweight room with a carpet and a stud partition. Where the opening is out of reach or the room needs to be left secure overnight, an electric opener with a timer does the same job.

Two details make it work rather than merely sound good. The opening at ceiling level wants to be the higher of the two by a clear margin, because the stack is driven by the height difference, and the low inlet wants to be on a shaded or sheltered elevation so the air being drawn in is the coolest available. A rooflight and a window in the same wall, at similar heights, move very little.

Overheating, and what Part O changed

Approved Document O brought overheating into Building Regulations for new dwellings in England, and it did so by limiting glazed area on the sun facing elevations and by requiring a demonstrable route for removing heat.

Extensions to existing houses are not assessed under it. The reason to know about it anyway is that it codified what the trade already knew: a large area of unshaded glass facing between south east and west, in a room without a way of purging air, produces temperatures that make the space unusable in a heatwave. Applying the same logic voluntarily to an extension is sensible, and it turns into three decisions. Cap the glazed area, specify solar control on the exposed planes, and make at least one unit open.

Winter gain, the useful side of the ledger

Solar gain is treated as a problem for eleven of these sections, so it is worth stating the other half. Between October and March, sunlight through a rooflight is free heat arriving exactly where you want it, and because the roof plane sees the whole sky it collects gain even when the low sun is behind a fence or a neighbouring roof.

This is why blanket solar control across every rooflight in a house is a blunt answer. On a north facing slope there is almost no summer gain to control and a modest amount of useful winter light to lose, so neutral glass is usually right. The strategy is per plane, not per house, which is one of the questions a survey settles.

Daylight against the lighting bill

The lighting saving from a rooflight is real but it is the smallest of the three effects, and it is worth being honest about the scale. Lighting is a modest fraction of a home’s electricity now that filament lamps have gone, so the saving is measured in tens of pounds a year rather than hundreds.

The larger effect is behavioural. A room that needs a light on at two in the afternoon in November gets used less, and a room lit by daylight from above holds usable light much later into a dull day, because a rooflight collects from the whole dome of the sky rather than the strip visible through a wall opening. That is a comfort and utilisation gain that never shows up on a bill.

Glazed area, and the point where more glass starts costing you

Glazing area of roughly fifteen to twenty per cent of the floor area below is the usual daylight starting point. Thermally, more glass is always worse than the roof it replaces, because even an excellent rooflight at 1.2 W/m²K sits against an insulated roof at around 0.15.

So the area decision is a balance rather than a maximum. Past about a quarter of the floor area, the extra daylight is subject to diminishing returns while the winter heat loss and the summer gain keep climbing in a straight line. A well placed three square metres almost always outperforms a poorly placed five, and where a wide expanse is wanted the better answer is often two smaller units spread down the room rather than one large one over the middle.

Orientation and pitch, the two free variables

Before any glass is specified, two geometric facts set the ceiling on what the specification has to achieve, and neither of them costs anything to get right at design stage.

Orientation decides how much summer sun arrives. Pitch decides the angle it arrives at, and a shallow or flat unit receives the July midday sun almost square on, which is the worst case for gain. A rooflight in a 40 degree north facing slope and a flush unit in a flat roof are different thermal problems even with identical glass in them. Where the room and the roof allow a choice of plane, choosing the cooler one is the cheapest solar control available.

Three Essex starting points, three different answers

What an efficient rooflight is worth depends entirely on the fabric it lands in, and the county gives three common cases.

• Victorian brick terrace, Chelmsford or Colchester, solid walls and a rear addition. The fabric loses heat everywhere, so an excellent rooflight is a small share of the total. Specify a good double glazed unit, spend the attention on the kerb detail and the airtightness, and put the saved money into the walls.
• Interwar semi, hipped concrete tile roof, commuter towns along the Great Eastern line. Usually cavity filled and loft insulated by now, so the rooflight becomes a visible weak point. This is where triple glazing starts to pay, and where a cold reveal is felt immediately.
• Trussed roof estate house, Great Notley, South Woodham Ferrers or Beaulieu Park. Tight fabric and low air leakage already, so the airtightness detail around the opening matters more here than anywhere. The unit is constrained by truss spacing rather than by thermal choice, and two narrow roof windows between trusses is the common answer.

Swapping an old dome or polycarbonate unit

The largest single efficiency step available on an Essex roof is taking out a twin skin polycarbonate dome from the nineteen seventies or eighties and putting a modern sealed unit in its place. The old dome performs somewhere around 3.0 W/m²K when new and worse once the skins have crazed and the perimeter has moved.

Going from that to a modern unit at 1.4 cuts the heat loss through that opening by well over half, and it fixes the draught around the frame at the same time, which is often the bigger felt improvement. Where the existing kerb is sound and square, a like for like replacement is frequently a single day. It is also the moment to insulate the upstand, because the covering is already open at the perimeter and the access will never be cheaper.

How a rooflight shows up in an EPC

Energy Performance Certificate assessments treat rooflights as part of the glazing element, and the assessor records the glazing type, the frame and the approximate area rather than the manufacturer’s declared figure. A modern double glazed rooflight is logged as double glazing installed during or after a given period, with default performance assumed for it.

Two consequences follow. Adding glass to a roof can move an EPC score down slightly, because the software sees more glazed area against an insulated roof. And keeping the paperwork matters: where a genuinely high performing unit has been fitted, a documented U-value lets an assessor enter the real figure rather than a conservative default. That documentation comes out of the Building Control notification, which is another reason it is worth having filed properly.

Condensation risk, and why a warmer frame stays drier

Surface condensation is a temperature problem before it is a ventilation problem. Moist air condenses on whatever surface in the room falls below its dew point, and in a room with a rooflight the coldest surfaces are almost always the frame corners and the bottom edge of the glass.

Improving the unit therefore reduces condensation directly. A warm edge spacer lifts the temperature at the perimeter of the pane by a couple of degrees against an old aluminium spacer bar, and a properly broken frame does the same at the sightline. In a kitchen or a bathroom, where moisture output is high and the air is warm, those couple of degrees decide whether water runs down the glass in January. Ventilation still does the rest of the work, which is why a trickle vent or an opening unit belongs in the specification of any rooflight over a wet room.

What to put on the order so performance is contractual

A performance specification only exists if it is written down before anything is made. A rooflight order that can be checked on delivery names each of the following.

• Whole unit U-value in W/m²K, and the plane it is declared in.
• The glass make-up, pane by pane, including which surface carries the low emissivity coating and which the solar control.
• The g-value and the light transmission, as two separate figures.
• Cavity width and gas fill.
• Spacer type, warm edge or otherwise.
• Frame material and, for metal frames, confirmation that the profile is thermally broken.
• The kerb: whether it is insulated, and by whom it is being built.

That last line is the one most often missing, and it is the one that decides whether the glass figure ever gets realised in practice. Indicative costs for the different build-ups are set out on the costs page.

Checking afterwards whether it worked

Two checks are worth doing in the first winter, and both are cheap. On the first genuinely cold morning, put your hand on the internal frame and on the reveal lining. Both should feel close to room temperature. A frame that reads noticeably cold is either unbroken metal or a break that has been bridged by a fixing.

The second is to look at where condensation appears, if it appears at all. Even misting across the whole inner pane on a cold night is normal physics in a humid room and clears with ventilation. Water pooling only at one corner, or a damp mark on the ceiling at one edge of the reveal, points at a thermal bridge or an air path at that point rather than at the glass. Both are worth photographing and raising, because both are fixable while the job is fresh, and both are far easier to trace in the first year than in the fifth.