How the junction Ψ is worked out
A junction Ψ can't come from adding up layer resistances the way a U-value does, because the heat turns a corner. It is found the way BS EN ISO 10211 and BRE's BR 497 conventions set out: by modelling the section in two dimensions and seeing how much more heat flows through it than the U-values alone would say.
It is worked out for each outside wall on the ground floor in turn, with that wall's own build-up meeting the floor build-up of its room, so a brick extension on a cavity-wall house, or a room with a different floor, gets its own Ψ. Each room's junction loss is then the sum over its outside walls of Ψ × the wall's length.
1. Building the section
- The outside wall's build-up, layer by layer, running 1.5 m up from the finished floor (BR 497 asks for at least 1 m).
- The floor build-up of the room it bounds, running in from the wall for half the floor's B′, with its top 150 mm above the ground outside.
- The ground: 20 m beside the house and 20 m down, at 2.0 W/mK.
- A foundation under the wall, 100 mm proud of it each side, of the material and depth you set (none if the depth is 0).
- The edge insulation where you put it (horizontal or vertical, inside or outside), and the upstand if there is one.
2. The boundaries
The room is held at 1 K above the outside air. The wall's inner face sees the room through Rsi = 0.13 m²K/W and the floor's top through 0.17 (heat going down); the wall's outer face and the ground outside see the air through Rse = 0.04. The cut edges of the model (the top of the wall, the far end of the floor, the far and deep ground) are adiabatic: no heat crosses them.
3. Solving it
The section is cut into a grid of rectangles, about 1 cm across near the junction and growing to 1.5 m far out in the ground, with a grid line on every material edge. Each cell's heat balance (flow in from its neighbours equals flow out) gives one equation; together they are solved by conjugate gradients. It takes about a quarter of a second.
4. From heat flow to Ψ
The heat flowing in from the room, per metre of wall for 1 K, is L2D. The wall and floor U-values already count part of it, over the same internal lengths:
What is left is the junction's own extra loss. It can come out negative where the floor's U-value already overstates the edge, as with a thick stone wall on an uninsulated floor.
5. Checking it
| Check | Expected | This model |
|---|---|---|
| ISO 10211 case 2: heat flow | 9.5 W/m | 9.49 W/m |
| Case 2: four surface temperatures | Within 0.1 K | All four within 0.05 K |
| Plain 2-D floor against the ISO 13370 U-value | Close | Within 1 to 4 % |
| Grid fineness | Settled | Within 0.005 W/mK of a grid four times finer |
Test reference case 2 is the standard's own check for 2-D software: a 500 × 47.5 mm section of concrete, insulation, an aluminium L-section and a timber batten. Its surface temperatures come out at 7.06, 0.76, 16.77 and 18.33 °C against the standard's 7.1, 0.8, 16.8 and 18.3, matching commercial tools (COMSOL, SimScale, QuickField) to 0.01 K.
What it leaves out
- The edge insulation, upstand and foundation are the same all round the house.
- Corners of the house (a 3-D effect) aren't modelled separately.
- Steady state only: no daily or seasonal swing in the ground.
- The foundation is a simple block of one material, and details like cavity closers, damp-proof courses and wall ties are left out.
- Suspended floors aren't modelled; they take SAP's default Ψ.
- The soil is uniform at 2.0 W/mK; wet clay conducts more, dry sand less.
- Its floor-to-ISO 13370 gap of a few per cent carries into Ψ as roughly 0.01 W/mK, as it does in BR 497's own method.