Glass · Thermal
U-value calculator for insulating glass
Almost all of the heat crossing a sealed unit crosses the cavity, and most of that is radiation, not conduction. That is why a low-E coating matters far more than the gas: adding one to a 4-16-4 unit takes it from about 2.7 to about 1.4 W/(m²·K), and swapping the air for argon then takes it to about 1.1. This page follows EN 673 and shows the radiation and gas parts of every cavity separately, so you can see where the heat is going.
This is the centre-pane value, which is what glass is sold on. The whole window is always worse, because the spacer at the edge and the frame both leak more than the middle of the glass — put this figure into the window U-value calculator to see how much worse. Weight of the same unit is in the weight calculator.
Typical U-values of double and triple glazing
Ug of the units that are actually ordered, worked out by the calculator above to EN 673: 4 mm panes, gas at 90 % fill, one low-E coating per cavity. Two things in it surprise people. A wider cavity is not always better — with argon, 20 mm (1.13) does slightly worse than 16 mm (1.10), because past about 16 mm the gas starts to circulate. And the coating matters far more than the gas: going from no coating to a soft coat halves the U-value, while swapping air for argon takes off another fifth.
| Unit | Gas | Low-E coating | Ug, W/(m²·K) |
|---|---|---|---|
| 4-16-4 | Air | None | 2.74 |
| 4-16-4 | Air | Hard coat, ε 0.16 | 1.69 |
| 4-16-4 | Argon 90 % | Hard coat, ε 0.16 | 1.49 |
| 4-16-4 | Air | Soft coat, ε 0.03 | 1.35 |
| 4-12-4 | Argon 90 % | Soft coat, ε 0.03 | 1.26 |
| 4-16-4 | Argon 90 % | Soft coat, ε 0.03 | 1.10 |
| 4-20-4 | Argon 90 % | Soft coat, ε 0.03 | 1.13 |
| 4-12-4 | Krypton 90 % | Soft coat, ε 0.03 | 1.05 |
| 4-12-4-12-4 | Argon 90 % | Two soft coats, ε 0.03 | 0.71 |
| 4-16-4-16-4 | Argon 90 % | Two soft coats, ε 0.03 | 0.61 |
| 4-18-4-18-4 | Argon 90 % | Two soft coats, ε 0.03 | 0.62 |
| 4-12-4-12-4 | Krypton 90 % | Two soft coats, ε 0.03 | 0.58 |
Declared values on a data sheet are rounded to one decimal, and the coating's real emissivity comes from the manufacturer — put it in above. For the whole window, frame and spacer included, the window U-value calculator takes this Ug as its input.
How EN 673 does it
Two things fall out of those equations, and they are the whole story of insulating glass:
- Radiation dominates an uncoated cavity. With plain glass on both sides, hr is about 3.7 W/m²K against roughly 1.6 for the gas. A low-E coating drops hr to around 0.15 — a factor of twenty-five. Nothing else you can do to a sealed unit comes close.
- There is a best cavity width, and wider is not better. Below it the gas conducts too well; above it the gas starts to circulate — the Nusselt number climbs above 1 — and convection undoes the gain. For argon that peak sits near 16 mm, for air a little wider, for krypton nearer 10 mm. That is why 16 mm is the number you see everywhere.
Gas properties at 283 K
| Gas | ρ kg/m³ | µ ×10⁻⁵ kg/(m·s) | λ W/(m·K) | c J/(kg·K) |
|---|---|---|---|---|
| Air | 1.232 | 1.761 | 0.02496 | 1008 |
| Argon | 1.699 | 2.164 | 0.01684 | 519 |
| Krypton | 3.560 | 2.339 | 0.00900 | 245 |
| Xenon | 5.689 | 2.226 | 0.00529 | 161 |
A "90 %" fill is treated as 90 % of that gas and 10 % air, with the properties averaged by volume, which is what EN 673 prescribes for mixtures.
Values you can check us against
| Make-up | Ug W/(m²·K) |
|---|---|
| Single 4 mm | ≈ 5.8 |
| 4-16-4 air, no coating | ≈ 2.7 |
| 4-16-4 air, one low-E ε 0.03 | ≈ 1.4 |
| 4-16-4 argon 90 %, one low-E ε 0.03 | ≈ 1.1 |
| 4-16-4-16-4 argon, two low-E ε 0.03 | ≈ 0.6 |
Type those into the calculator. If it disagrees with this table, the calculator is wrong and we want to know.
What this page does not do
- Centre-pane only. No spacer, no edge effect, no frame. The window's Uw is always worse than this — often by 0.2 to 0.5 W/m²K once a real frame and a metal spacer are in.
- Vertical glazing. Sloped and horizontal units convect differently and use different A and n.
- No solar gain. U-value is heat loss only; g-value and light transmission are separate properties and can pull in the opposite direction.
- Nominal gas fill. Real units lose gas slowly, and a unit that has leaked performs like an air-filled one.
- Standardised boundary conditions, not your weather: he = 23, hi = 8, Tm = 283 K, ΔT = 15 K. That is how U-values are declared, and it is why the number is comparable between suppliers rather than realistic on any given day.