Research
The energy code does not set the wall thickness. Statics and moisture margin do.
Swedish timber walls are commonly built with 220 mm of insulation "because the energy code requires it". We built a U-value engine to test that claim, and for our house it is false: the requirement is met with wide margin already at 170 mm. The wall is 220 mm for other, better reasons, and this paper names them.
Finding
For Kult's reference house (4.2 × 4.2 m, 14° monopitch roof), the area-weighted average heat-transfer coefficient is Um = 0.202 W/m²·K against the Swedish ceiling of 0.30, a 33 % margin. Sweeping the wall insulation down to 170 mm still passes, at Um = 0.224. The last 50 mm of wool (170 → 220 mm) buy only about 0.022 W/m²·K.
The 220 mm thickness therefore has three owners, none of which is the U-value: it reuses the same 45×220 lumber dimension that carries the roof (one dimension across the whole timber list); it gives a warmer wind barrier and a larger condensation margin; and it leaves headroom for the primary-energy calculation still ahead. An unexamined industry default became a deliberate, named choice.
Method
We implemented the calculation method of SS-EN ISO 6946 as a software engine over the same envelope model that feeds the structural checks: thermal resistance per layer (R = d/λ), surface resistances, ventilated air gaps excluded per the standard, repeating thermal bridges (studs through insulation) by the λ-value method with conservative timber fractions, and non-repeating bridges (corners, junctions, window reveals) as a documented per-part surcharge at ISO 14683 level.
The engine reports U per building part and the area-weighted Um, and re-runs on every design change. For this study the only parameter varied was wall insulation thickness (170 / 195 / 220 / 245 mm), and each row is a full recalculation rather than an interpolation.
Numbers
| Part | U [W/m²·K] |
|---|---|
| Exterior wall (220 + 45 mm mineral wool) | 0.207 — incl. stud and junction surcharges |
| Roof (335 mm wool) | 0.136 |
| Floor (300 mm wool, ventilated crawl space) | 0.183 |
| Window (product value) | 1.10 |
| Exterior door (product value) | 1.00 |
| Area-weighted Um | 0.202 ≤ 0.30 (BFS 2024:14) — passes with ~33 % margin |
| Wall insulation | Um [W/m²·K] |
|---|---|
| 170 mm | 0.224 — passes, 0.076 margin |
| 195 mm | 0.212 — passes |
| 220 mm (current design) | 0.202 — passes, 0.098 margin |
| 245 mm | 0.194 — passes |
The slope is roughly 0.0009 Um per millimetre of wall wool. In a window-realistic variant of the house (7.2 × 6.0 m, ~13 % glazed wall area), windows become the single largest U·A item and Um rises to 0.241, which still passes. The lever on Um is glass area and window U-value rather than the last centimetres of wall.
Assumptions: mineral wool λ = 0.036, structural timber λ = 0.14, gypsum λ = 0.25 W/m·K (Swedish design values, ISO 10456); the 45 mm service layer is insulated; the floor build-up is a documented default.
What this does not show
- Um is not the whole energy code. Sweden's primary-energy number also weighs ventilation losses, heating system and climate zone; that calculation is a separate, scheduled step. A comfortable Um margin does not mean the energy requirement as a whole is settled.
- Junction thermal bridges are a documented per-part surcharge, not yet a ψ·L calculation per junction. The surcharge is chosen conservatively (it errs toward the requirement) and a junction library will replace it.
- Window and door U-values are product placeholders (1.10 / 1.00 W/m²·K) rather than per-product data.
- These are calculated values on a design, verified against the standard's method; no built wall has been measured yet. Measured data comes with the instrumented room module of the prototype programme.
Sources
- SS-EN ISO 6946 — building components: thermal resistance and transmittance (the calculation method).
- ISO 10456 — design thermal-conductivity values; ISO 14683 — thermal bridges, simplified defaults.
- BFS 2024:14 (Boverket) — Um ≤ 0.30 W/m²·K for small dwellings: boverket.se
- Kult internal research, "The U-value engine as requirement judge" (2026-07-10) — full tables and assumptions; engine: cult.docs.u_value.
Read next
One storm board cannot replace the racking layer
The same wall judged by the structural gate: what the sheathing verification found.
Most of the money in a turnkey house goes to labour
The cost stack of a small Swedish build, including which slice self-build moves.
All papers
The research hub collects every published paper, and the note on why the work is open.