Embodied-carbon proxy
A material-quantity model: element areas measured from the plan, multiplied by per-element coefficients, reported per square metre of floor. Product stage (A1–A3) of the modelled elements only. It is not a life-cycle assessment and none of its numbers should be quoted as one.
What it counts
External walls net of glazing, glazing, party walls, internal partitions, intermediate floors, the ground slab and the roof — the elements a floor plan actually determines.
What it leaves out
Services, finishes and fit-out, transport and construction (A4–A5), replacement during use (B), end of life (C), biogenic carbon, and foundations beyond a nominal slab.
Where the coefficients come from
They are illustrative values for light-frame timber assemblies, chosen so that whole-building intensity for plausible geometry falls inside the published residential band — the LETI Embodied Carbon Primer's 2030 target of about 300 kgCO₂e/m² GIA (A1–A5), against a typical current baseline near 500. A test in the repository fails if a feasible design leaves the 250–700 range. Claims here are comparative: a common rescaling cannot reverse them, and each coefficient is perturbed by ±50% below.
Coefficients, kgCO₂e per m² of element
- opaque external wall
- 95
- window
- 120
- party wall
- 75
- internal partition
- 28
- intermediate floor
- 65
- ground slab
- 210
- roof
- 110
Score band: 620 kgCO₂e/m² scores 0, 260 scores 1.
Density against carbon
Within a brief, denser designs carry less modelled material per square metre: mean Spearman ρ between the density score and the carbon score is 0.88 (0.83–0.92) over 58 briefs. A fixed ground slab and roof spread over more floor area is most of the reason.
By method
| Method | p10 | median | p90 | designs |
|---|---|---|---|---|
| Heuristic | 297 | 332 | 352 | 2,159 |
| Random sketch + repair | 311 | 336 | 360 | 7,952 |
| NSGA-II | 306 | 341 | 362 | 10,758 |
| Simulated annealing | 326 | 342 | 362 | 5,108 |
| Random search | 334 | 345 | 357 | 178 |
| Heuristic-seeded NSGA-II | 319 | 345 | 366 | 11,121 |
| CP-SAT | 335 | 357 | 382 | 1,684 |
kgCO₂e/m² across each method's feasible designs. Under every one of the 14 single-coefficient perturbations (±50%) the method with the lowest mean best intensity stays the same in 100% of cases.