Envelope · Thermal

Thermal bridges

Insulation is sold by the square metre and lost by the metre run. Every place two insulated surfaces meet — a lintel, a sill, a corner, the line where the floor meets the wall — leaks extra heat that no U-value accounts for. Added up over a whole house those junctions routinely give back a fifth to a third of what the insulation gained, and they are also where the surface goes cold enough to grow mould. This page does both sums: what the junctions cost you in watts, and whether they will go black in the corner.

The building

Total envelope area m²

Design ΔT K

Envelope area is every exposed surface added together — walls, roof, floor, windows, doors. It is the same total the whole-house calculator uses, and it is what the y-value gets divided by.

How well are the junctions detailed?

This sets every ψ below to a typical figure for that level of care. Override any of them with the real number from your own detail — that is what the box is for.

The junctions

Junctionmetresψ W/(m·K)

Each row is a junction: the type, then its length in metres and its ψ.

Measure each junction once, along its length. A ground-floor-to-wall junction runs the whole perimeter; a lintel is only as long as the opening it sits over. Negative ψ is real — an inverted corner has more inside surface than outside, and genuinely loses less than the flat wall it replaces.

—

Bridging heat loss

— W/K

Equivalent y-value

— W/(m²·K)

Saved vs the 0.15 default

— W/K

Worst junction

— W/K

Will it go mouldy?

A junction fails on temperature long before it fails on heat. Mould needs 80 % humidity at the surface, not in the room, and that happens several degrees above the dew point.

—

Surface temperature

— °C

Humidity at that surface

— %

f_Rsi you actually need

—

Dew point of the room

— °C

The two calculations

heat H_tb = Σ ( l · ψ ) // watts per kelvin y = H_tb / ΣA_envelope // the number regulations ask for surface θ_si = θ_e + f_Rsi · ( θ_i − θ_e ) φ_si = p_i / p_sat(θ_si) // humidity AT the cold surface mould once φ_si > 80 %, i.e. once f_Rsi < ( θ_si,min − θ_e ) / ( θ_i − θ_e ) where p_sat(θ_si,min) = p_i / 0.80

Two things fall out of that second block that catch people every time. First, the surface humidity is what matters, not the room humidity: air at 20 °C and 60 % arriving at a surface at 15 °C is already at 82 %, because the air next to a cold surface is cold air. Second, mould arrives well before water does. The dew point of that same room is 12 °C, so the corner stays visibly dry — and grows mould anyway, three degrees earlier.

Typical ψ values

Indicative only, for getting your bearings and for sanity-checking a figure someone has handed you. The real ψ for a junction comes from a two-dimensional heat-flow model of that detail, or from a published accredited detail built exactly as drawn.

Junction	No attention	Good practice	Calculated
Lintel over window or door	0.60	0.30	0.075
Window or door jamb	0.15	0.05	0.02
Window or door sill	0.12	0.04	0.02
External corner	0.18	0.09	0.04
Inverted (internal) corner	−0.10	−0.10	−0.06
Ground floor to wall	0.32	0.16	0.04
Intermediate floor into wall	0.14	0.07	0.02
Eaves	0.12	0.06	0.02
Gable or verge	0.24	0.12	0.04
Flat roof to wall	0.20	0.10	0.04
Party wall to external wall	0.12	0.06	0.02
Balcony or canopy through the wall	0.90	0.45	0.10

Why the default is a penalty

If you do not calculate the junctions, most compliance routes make you assume y = 0.15 W/(m²·K). That is deliberately pessimistic — it is meant to be worse than anything you would build on purpose, so that calculating always pays. On an ordinary house it is worth a third of a kilowatt, which is the difference between two heat pump sizes. The odd corollary is worth knowing: if your calculated y comes out above 0.15, you are better off not calculating it, and rather better off redesigning the junctions.

Worked example you can check by hand

The set of junctions this page opens with, on a house with 222 m² of envelope.

lintels 12 m × 0.30 = 3.60 W/K jambs 30 m × 0.05 = 1.50 sills 12 m × 0.04 = 0.48 corners 20 m × 0.09 = 1.80 floor/wall 36 m × 0.16 = 5.76 mid floor 36 m × 0.07 = 2.52 eaves 24 m × 0.06 = 1.44 gables 16 m × 0.12 = 1.92 H_tb = 19.02 W/K y = 19.02 / 222 = 0.0857 W/(m²·K) default 0.15 × 222 = 33.30, so calculating saves 14.28 W/K = 0.33 kW at ΔT 23 K

And the mould check at 20 °C / 60 % inside, 0 °C outside, with a junction at the UK limit of f_Rsi = 0.75:

p_i = 0.60 × 2337 = 1402 Pa θ_si = 0 + 0.75 × 20 = 15.0 °C p_sat(15.0) = 1704 Pa φ_si = 1402 / 1704 = 82 % → over 80, it grows needed: p_sat(θ_min) = 1402/0.80 = 1753 Pa → θ_min = 15.43 °C f_Rsi,min = 15.43 / 20 = 0.772

So a junction built exactly to the regulation minimum fails, in an ordinary room, at an ordinary humidity. The 0.75 figure was calibrated for a drier house than most people live in.

What this page does not do

It does not calculate ψ. That needs a two-dimensional finite-element model of the actual junction, with the actual materials and the actual geometry. This page does the accounting once you have the numbers, and gives you typical figures so you can see whether an answer is plausible.
Point bridges are ignored. Wall ties, fixings, brackets and balcony connectors have a χ value in W/K each rather than a ψ per metre. A few hundred stainless ties are usually small; a dozen steel brackets are usually not.
Lengths are easy to get wrong. Double-counting a junction where two elements meet, or measuring a perimeter internally when the ψ was derived for external dimensions, moves the answer more than the ψ itself does.
The mould check is a single steady-state snapshot. The proper EN ISO 13788 assessment runs monthly and uses the running mean humidity, and real mould also depends on how long the surface stays wet, on the substrate and on whether anyone opens a window.
f_Rsi assumes a standard internal surface resistance of 0.25 m²K/W for the corner case — higher than the 0.13 used for U-values, because corners are sheltered from room air movement. Behind a wardrobe it is worse again, and that is exactly where mould appears.

Read this. The ψ values on this page are typical figures published for common construction types. They are not a substitute for a calculation of your own detail, and they must not be used for a compliance submission. Where the answer matters, get the junction modelled.