Windows · Moisture

Window condensation

A window is not one surface, it is three, and they run at three different temperatures. The middle of the pane is the warmest, the frame is colder, and the inch of glass around the spacer bar is colder still — which is exactly why the black mould appears in a neat line around the edge of the glass while the middle of the window stays perfectly clear. This page works out all three temperatures and tells you the highest indoor humidity you can run before each of them goes wrong.

The room and the weather

Inside °C

Inside RH %

Outside °C

A living room in winter runs 40–55 % RH, a bedroom overnight 55–65 %, a kitchen or bathroom in use 70 % and above. Condensation on windows is a symptom of that number meeting a cold surface — not of the window being faulty.

The glazing

U_g centre of glass W/(m²·K)

R_si m²K/W

Don't know U_g? Work it out from the build-up with the glazing U-value calculator. Single glazing is about 5.7, old double 2.8, modern low-E argon 1.1, triple 0.6.

The spacer and the frame

Spacer bar at the glazing edge

Frame

f_Rsi is the coldest internal surface temperature written as a fraction of the inside-to-outside difference, and it comes from a thermal model of the actual section — the numbers here are typical figures for orienting yourself. Note the pattern: the spacer is almost always the coldest thing in the window.

—

Centre of glass

— °C

Glazing edge

— °C

Frame

— °C

Room dew point

— °C

How the three temperatures come out

centre of glass — from the U-value, because the glass is a plain surface θ_glass = θ_i − U_g · R_si · ( θ_i − θ_e ) edge and frame — from the temperature factor of the section θ_si = θ_e + f_Rsi · ( θ_i − θ_e ) then the two lines every surface is judged against dew point p_sat(θ_dew) = φ_i · p_sat(θ_i) // below this, water mould limit p_sat(θ_mould) = φ_i · p_sat(θ_i) / 0.80 // below this, mould and the number people actually want max room RH before water = 100 · p_sat(θ_si) / p_sat(θ_i) max room RH before mould = 80 · p_sat(θ_si) / p_sat(θ_i)

Two consequences fall out of that and they explain almost every complaint about double glazing.

First, the edge is not the glass. The centre of a modern pane is warm because the cavity insulates. Around the spacer bar the two panes are bridged by a strip of metal roughly 15 mm wide, and that strip short-circuits the whole cavity. Replace the aluminium with a warm-edge spacer and the glazing edge gains three degrees for a couple of euros per metre — far more temperature per euro than anything else you can do to a window.

Second, mould comes before water. A surface has to fall all the way to the dew point before anything visibly wets. Mould only needs 80 % humidity at the surface, which happens three or four degrees earlier. So the honest failure sequence is: nothing looks wrong, then the rebate goes black, then — much later — the glass runs.

Typical temperature factors

Component	f_Rsi	Note
Aluminium spacer	0.55 – 0.62	the coldest point of most windows
Warm-edge spacer, stainless	0.68 – 0.72	worth about 3 K at the edge
Warm-edge spacer, composite	0.72 – 0.77
Timber frame	0.70 – 0.75
PVC-U frame	0.70 – 0.75
Aluminium, thermally broken	0.60 – 0.70	depends entirely on the break
Aluminium, no thermal break	0.30 – 0.45	will run with water most winters

Worked example you can check by hand

The window this page opens with: 20 °C and 50 % RH inside, 0 °C outside, U_g 1.1, aluminium spacer, PVC frame.

p_i = 0.50 × 2337 = 1168 Pa dew point = 9.3 °C mould line: p_sat = 1168/0.80 = 1461 Pa = 12.6 °C centre of glass 20 − 1.1 × 0.13 × 20 = 17.1 °C → 60 % at the surface, fine frame 0 + 0.70 × 20 = 14.0 °C → 73 %, marginal glazing edge 0 + 0.58 × 20 = 11.6 °C → 86 %, mould

The edge is 2.3 K above the dew point, so there is not a drop of water anywhere on the window — and it is a degree below the mould line, so the rebate will go black. Swap the aluminium spacer for a warm-edge one at f 0.70 and the edge rises to 14.0 °C, clear of the line. Or hold the room below 47 % RH, which nobody does.

What this page does not do

It does not calculate f_Rsi. That needs a two-dimensional model of the actual frame and glazing rebate to EN ISO 10077-2. The values offered here are typical for the component type and are for orientation, not for a report.
Steady state, one moment in time. Real condensation is a night-time, intermittent, thermal-mass affair: the glass cools fastest, fogs at four in the morning and clears by nine. This page tells you whether the conditions allow it, not how long it lasts.
Curtains and blinds are only half modelled. Anything that stops room air washing over the glass makes the surface colder, which is why the worst condensation is always behind a closed blind — but the higher R_si options here only cool the glass. The published f_Rsi of a frame or spacer already has a surface resistance baked into it, so a blind makes the edge and the frame colder too and this page will not show it. Read the edge and frame figures as the optimistic case.
No radiator under the window, no draughts, no solar gain. All three move the surface temperature by degrees, and all three are why one window in a house behaves differently from the identical one next to it.
Condensation between the panes is a different problem entirely — a failed seal, not a humidity question, and no calculation will help.

Read this. Use this to understand a window and to compare options, not to settle a dispute. A condensation claim needs measured indoor conditions over time and the manufacturer's own f_Rsi for the actual section.