Reusing an existing kerb, and when not to

Why the upstand carries the whole detail

A flat rooflight does not keep water out by being watertight at the glass. It keeps water out by sitting above the water.

The roof covering is dressed up the outside face of the upstand and terminated near the top, so the membrane, not the unit, is what holds the wet side. The unit lands on that and is sealed and fixed down. Everything about that arrangement depends on the box underneath being tall enough, sound enough and flat enough to do its part. The glass could be perfect and the frame new, and a soft upstand at the deck junction will still put water on a ceiling.

What was built here, decade by decade

Opening up around an old rooflight in Essex produces one of about five things, and each behaves differently under assessment.

Nineteen sixties and seventies work is often a proprietary plastic or aluminium kerb frame supplied with the dome, seventy five to a hundred millimetres tall, bedded on mastic and screwed through the covering. Seventies and eighties timber work is usually untreated softwood, nailed together, with felt dressed up it and turned over the top. Nineties work introduced treated timber and better heights. Post-2000 work on warm roof build-ups usually has a properly insulated box off the deck. And a proportion of everything is a kerb built by a competent joiner to no particular standard at all, which can be excellent or can be three lengths of two by one.

Probing the base, which is where it always fails

Old timber upstands rot from the bottom, not the top, and the failure is invisible from above.

Water that has found a way past a tired detail runs down the inside face of the box and collects at the junction with the deck, where end grain sits against a cold surface and nothing dries. The top edge can look and feel perfectly sound while the bottom fifty millimetres has gone soft enough to take a screwdriver to the handle. That matters more than it sounds, because a new frame is fixed through its flange into the top of the kerb, so a soft base does not announce itself at the moment of fixing. It announces itself later, when wind lifts a frame whose fixings have nothing left to hold.

The test is mechanical rather than visual. A bradawl or a slim screwdriver pushed firmly into the timber at the deck junction on all four sides tells you in ten seconds what a photograph never will.

Level, square and out of wind

A flush unit lands on a continuous seal along its flange, and that seal compresses evenly or it does not seal.

Three separate checks matter here. Level, taken along each side, since a kerb built off a deck that has deflected will follow the deflection. Square, taken on the diagonals, since a box built by eye in 1979 is very rarely square and a difference of more than about eight millimetres across the diagonals means the unit sits on three corners. And out of wind, which is the twist test: sighting across the top from two ends, or laying two straight edges across opposite sides and looking along them. A twisted top is the one that catches people out, because it reads as flat with a spirit level and still puts a frame under permanent stress.

Height, measured from the right surface

The figure quoted across the trade is one hundred and fifty millimetres from the finished roof surface to the top of the upstand, and both ends of that measurement need care.

It is measured from the top of the covering, not from the deck, so an upstand that reads as adequate in bare timber loses whatever the insulation and membrane build-up adds. Where a roof has been over-boarded at some point, which is common on flat roofs that have been re-covered, the effective height has already been reduced by the thickness of the additional layers. The number itself is derived from the depth water can pond on a level roof combined with how far wind driven rain travels up a vertical face. Where the roof drains cleanly and the position is sheltered, some manufacturers will warrant less. Where the roof is genuinely level, has a history of standing water, or sits exposed on the marshes or out towards the coast, a hundred and fifty is the floor rather than the target.

What the covering around it says about the answer

The upstand and the membrane dressed up it are a single detail, and one cannot be kept if the other is finished.

Felt that has gone brittle, has lost its mineral surface or shows crazing at the fold up the face of the box will not take being worked around, and a new unit set onto it is glass on a covering with two winters left. Single ply that is sound will accept a cut back and a hot air welded insert without difficulty. Where the covering across the whole roof is at the end of its life, the sensible sequence is to sort the roof first and set the glass into something that will still be there. That is a longer conversation but it is a cheaper one than doing the rooflight twice.

Cold bridging, and the box that was never insulated

A timber upstand of untreated softwood, uninsulated, is a continuous cold ring around a rooflight, and on a modern warm roof it is thermally the worst part of the assembly.

What that produces is condensation at the perimeter of the glass and dark marks along the inside edge of the reveal that get blamed on the unit. Fitting a new unit with a 1.2 U-value onto an uninsulated 1979 box wastes a good deal of what was bought, because the perimeter goes on losing heat and going on condensing. Insulating an existing kerb is possible where there is room on the outside face to add board before the covering is dressed back up it. Where there is not, that alone can be the argument for rebuilding.

Fixings into timber that has been damp

Even where an upstand passes the probe test, the quality of the fixing it will offer is worth a separate thought.

Softwood that has been through repeated wetting and drying cycles for forty years loses fibre strength around any existing screw or nail holes, and the top of an old box is usually peppered with them from the previous frame. A new frame fixed into the same holes has a fraction of the pull-out resistance the fixing schedule assumes. Where the timber is otherwise sound, the answer is fixing to fresh positions clear of the old ones, into full timber, at the centres the manufacturer specifies. Where there is not enough clear material left to do that, the top of the box needs renewing whatever the probe says.

Raising an upstand rather than rebuilding it

Where the timber is sound and dry and the only failing is height, adding to the existing box is a legitimate route and it is a good deal quicker than starting again.

A new treated timber section is fixed to the top of the existing one, screwed through into full material and glued, then the whole assembly is insulated on the outside face and the covering is dressed up the new full height and terminated. Done properly the result is indistinguishable from a new box. The two conditions are that the existing timber is genuinely sound all the way to the deck, and that the covering can be taken back far enough to be dressed up the greater height without stretching it. If either fails, the saving disappears and the rebuild is the honest answer.

Dropping a preformed upstand over what is there

Proprietary insulated upstands are supplied as a factory made box, usually rigid insulation with a facing, sized to a standard unit and delivered ready to set down.

They are quick, thermally excellent and dimensionally accurate, and on a rebuild they are frequently the right choice. Where they do not work is as a cover over a suspect existing box, because the failure underneath is still there and is now considerably harder to inspect. A preformed upstand goes onto a prepared deck, not over old timber that somebody hopes is fine. Used properly they take a rebuild from a full day to a morning, which changes the arithmetic on the whole decision.

Metal and plastic frames from the dome era

The proprietary kerb frames supplied with domes in the seventies and eighties are a separate category and they are almost always removed.

They are shallow by current standards, they were designed to carry a light plastic shell rather than a glazed unit, and their fixing arrangement was through the covering into the deck rather than into a structural box. Aluminium versions are also a straight cold bridge with nothing to insulate. Where one of these comes off, a timber or preformed upstand is built in its place and the covering is made good around it, which is precisely why a dome swap is usually a day and a half rather than a day.

Deck condition, which is found at the same moment

The upstand cannot be assessed without looking at what it is standing on, and the two failures travel together.

Water that rotted the base of a box has been sitting on the deck around it for the same period. On a timber deck that means soft boarding for a few hundred millimetres out from the opening, which is probed at the same time and cut back to sound material if found. On a chipboard deck, which appears on a lot of nineteen eighties work, the degradation is faster and the affected area is usually larger. On a concrete or screeded deck the upstand was formed differently, often in blockwork or as an upstand in the slab itself, and the assessment is about the surface and the fall rather than about rot.

Getting the plan size to suit something purchasable

An upstand that passes every physical test can still be the wrong one to keep, and the reason is arithmetic rather than condition.

Flat rooflights are supplied to fit a stated kerb outside size, and the ranges are built around their own grids. An upstand built in 1985 to suit a dome that nobody sells now may sit fifty millimetres off every standard size available. At that point the choice is a made to measure unit against rebuilding the box to a standard size, and rebuilding is often cheaper and faster than the bespoke lead time. That comparison is set out properly on matching a replacement to an obsolete size.

What the survey writes down

The point of all of this is that the quotation is fixed rather than provisional, so the survey records the findings rather than the conclusion.

The probe results at all four base corners. Level, diagonals and twist, with the figures. Measured height from the finished surface, stated on each side. The covering type and its condition at the upstand. The insulation arrangement. And the measured kerb outside size, taken at both ends and in the middle. From that the quotation says either that the upstand is retained, or that it is rebuilt, and prices accordingly. A quotation that says “subject to condition of existing kerb” has not had the survey done, and the number in it is a conversation waiting to happen on the day.