Stopping Condensation on a Curtain Wall

Dry Glass, Wet Aluminium

First hard frost after handover, seventh floor, north-west corner. Facilities has folded a paper towel onto the window board beneath the corner mullion because water has been tracking down the aluminium since about six in the morning, and by nine there is a stain on the carpet tile. The glass on both sides of that mullion is perfectly clear. Somebody has already taken a photograph, because glass is what you photograph when you believe you have a condensation problem, and the photograph shows nothing worth looking at.

The complaint will nonetheless be logged as sweating windows, and the first document anybody pulls will be the glass certificate. That is the wrong half of the wall. The vision glass and the frame holding it are two separate surfaces facing the same room, exposed to the same outdoor temperature, and separated by a factor of two or more in thermal transmittance. Each has its own room-side surface temperature, and moisture leaves the air at whichever of them is colder. On a framed facade that is the frame: at the mullion face, at the transom, along the perimeter and hardest of all in a corner, where two cold members meet and the air moves least.

So a first-rate insulating unit in a second-rate frame is a bay that runs with water while every certificate in the file passes. Sorting it out means treating the frame as a component with its own number, finding where the room's dew point actually sits, and knowing which of several quite different waters is on the wall — because some of them are not room-side condensation at all.

The Dew Point Belongs to the Occupants

Everything above the line in that calculation is fixed by the facade and the weather. The line itself is fixed by the people inside, and it moves without anybody telling the facade team. Dew point is a function of the air's moisture content alone: raise the temperature of a room and the dew point does not follow, but add moisture — from occupants, from a humidifier commissioned to protect a timber floor or a server room, from a kitchen, from a pool, from a wet trade still drying out on the floor above — and it climbs immediately. The psychrometrics chapter of the ASHRAE Handbook — Fundamentals is where the relationship is set out properly, and CIBSE Guide A carries the equivalent design conditions in British practice.

The table below runs the arithmetic in the direction that is actually useful on a facade job. Start with the humidity somebody wants to hold in the space, read across to the dew point that implies, and read across again to the highest frame U-factor that keeps a room-side face above it on a minus-ten morning. It is an uncomfortable table for anyone who has been told that a humidified building and a standard commercial framing system can share an elevation. The two right-hand columns are the whole argument between the facade consultant and the mechanical engineer, written out before it happens.

Blinds, Diffusers and a Coefficient Nobody Checks

The interior film conductance is the input in that equation that looks like a constant and behaves like a variable. The conventional 8 W/m²K represents a vertical surface with room air circulating freely against it. Close a blackout blind against the glass and the strip of air between blind and facade stops being room air: it cools, it stratifies, and its dew point stays where the room's is because moisture diffuses through the gaps at the head and sill perfectly well even when heat does not. Push a run of storage units along the perimeter, hang a heavy curtain, install a bench seat that boxes in the trench heating, and the same thing happens over a longer stretch of wall.

Halve that conductance and watch the arithmetic. The frame at 3.5, twenty-one inside and minus ten outside, sat at 7.4 degrees with the film conductance at 8. At 4 it falls to about minus six — four degrees off the outdoor air, and a surface that collects water at anything above roughly sixteen per cent relative humidity, which is to say at every humidity a building ever holds. A good frame at 2.4 W/m²K is not exempt either: comfortable at 11.7 degrees in open air, it drops to about 2.4 degrees behind the same blind, which is the dew point of a room held at thirty per cent, and thirty per cent is a dry office in a cold snap rather than a design target. This is why a facade that was dry for two winters starts weeping the season after a fit-out, and why the bays that fail are so often the ones a tenant has furnished hardest.

The counter-measure is the oldest in the trade and it is a mechanical decision rather than a facade one: wash the interior surface with supply air. A perimeter slot or a sill unit throwing warm air up the inside face keeps the boundary layer moving and the effective film conductance high, which is why the ASHRAE Handbook treats fenestration surface temperature as a reason perimeter air distribution is arranged the way it is. Balancing that zone is therefore a condensation measure and not only a comfort one — and a diffuser taped over by an occupant who finds it draughty is a facade defect waiting to be reported as one.

The Wettest Hour Follows the Setback

Buildings that run a night setback have a specific and reliably misdiagnosed failure. Drop the interior air from twenty-one to sixteen overnight and the room's moisture content does not drop with it — the vapour is still there, so the dew point is unchanged. The surface temperature, however, falls: differentiate the check and each degree lost from the interior air costs the surface (1 − U ÷ hi) of a degree. On that 3.5 frame the factor is 0.56, so five degrees off the room takes nearly three off the frame. The mullion that sat at 7.4 in the evening is at about 4.6 at five in the morning, against an unmoved dew point.

So the wall is at its wettest in the hour before anybody arrives, and by half past nine, with the heating recovered and the sun on the elevation, the evidence has evaporated. What survives is a tide mark on a window board, a soft patch of plasterboard at a jamb, a spot of corrosion on a fixing, and an occupant nobody believes. A weekend shutdown does the same thing over sixty hours instead of eight, which is why so many facade condensation reports are dated on a Monday.

Air Carries the Water; Diffusion Only Trickles

Vapour diffusion through a sealed spandrel build-up moves a small and calculable quantity of moisture. Air leakage through a gap in the same build-up moves an amount that is not in the same league, because the air is carrying its moisture with it rather than letting it find its own way through a material. This is why the sequence of failures on a wet facade so often runs the other way round from the investigation: the wall is not failing because the vapour control layer has the wrong permeability, it is failing because there is a route for humid room air to reach a cold surface, and that route is usually at a junction rather than in a panel.

The floor edge is where the routes are: a back pan seal stopped short at a bracket, a smoke seal installed as a fire measure and assumed to be an air seal, a perimeter joint sealed to the slab in one bay and to the mullion in the next, a service penetration cut after sign-off. Stack effect then supplies the pressure difference for free — in winter the upper floors of a tall building push warm humid air outward through every gap in the envelope, all day, which is exactly the direction that lands room moisture on cold facade components. The lower floors pull the other way and stay dry, and the report arrives describing a problem confined to the top three levels.

Testing tells the two apart. ASTM E283 measures air leakage through the assembly in the laboratory and ASTM E783 does the same on a wall already standing; ASTM E1105 is the water penetration test that the same wall usually passes, because water penetration and air leakage are different failures and a wall can be watertight and leaky at once. Field air leakage testing on a mock-up is the cheapest hour anybody spends on a facade, and on a humidified building it is not optional — AAMA CW-DG-1 and the CWCT Standard for Systemised Building Envelopes both set out what that programme covers.

Three Waters, Told Apart Before Anyone Is Blamed

Almost every facade moisture complaint resolves into one of a small number of conditions, and they are distinguishable on a single site visit if you know what you are looking at. The distinguishing evidence is where the water sits, whether it wipes away, and what the weather was doing in the hours beforehand. Getting this right early matters because the three have different owners: one belongs to the mechanical brief, one to the facade contractor's detailing, and one to the glass processor's warranty.

What Is Left to Change Once the Wall Is Standing

The levers are unequal in cost and in effect, and they come in an order. Interior humidity is first because it is the only one that moves the dew point rather than the surface temperature, and where humidification was specified for a reason — an archive, a data hall, a pool — that conversation belongs to the mechanical engineer and the client rather than the facade contractor alone. Ten percentage points off the set point buys three or four degrees of dew point, more than most remedial work to a standing facade will ever buy.

Air distribution is second, and often the cheapest genuine fix: recommissioning a perimeter zone, reopening a diffuser somebody taped over, changing a throw so the supply reaches the glass line, cutting a blind pocket back so the blind stops sealing the boundary layer against the frame. Third comes the interior surface itself — trench heating, a warmed sill, or in extreme cases a secondary internal glazing line putting a new, warmer surface in front of the cold one. Fourth, and by a distance the most expensive, is the frame, because retrofitting thermal isolators into a standing wall is not generally available and this becomes a replacement conversation.

On a new project the whole list collapses into one early decision: the frame's thermal performance is chosen against the interior humidity the building will actually hold, verified for the real profile by simulation to ISO 15099 or BS EN ISO 10077-2 rather than taken from a system's headline figure. The polyamide isolators that do the work in a thermally broken extrusion — Ensinger's insulbar and Technoform's profiles are the ones most often named in system literature — are tested structurally as well as thermally, under AAMA TIR-A8 and AAMA 505, and the depth and geometry of the isolator moves the frame U-factor far more than the product family name suggests. Two profiles from one range can be a full watt per square metre kelvin apart.

Log It Rather Than Argue About It

The whole dispute turns on two readings taken at the same moment, and neither is usually available when it is needed. Put a combined temperature and humidity logger in the occupied zone at the perimeter — not on a column mid-floorplate and not above the ceiling, because the air at the facade is colder and stratified and it is the only air touching the wall — and let it record through a cold spell at fifteen-minute intervals. Pair it with a surface probe taped to the mullion face and a note of the outdoor temperature from the nearest station.

What the three traces show is a dew point line and a surface line crossing somewhere around dawn, and the gap between them is the answer: how many degrees short the frame is, and therefore how much humidity reduction, air movement or better component would close it. Hand that record on with the operation and maintenance information. The alternative, in year three, is a specialist at an occupied elevation with a cover plate and a screwdriver, reconstructing from a stain what a cheap logger would have recorded at the time.