A project carried through

The brief

A six-storey office over 8 000 m² of floor. The architect proposes a 12 m column-free grid for letting flexibility. The client has committed publicly to an embodied-carbon target. The floor zone is agreed at 1 100 mm including services. Advise on the structural scheme.

Everything below is fictional and is worked with this course's own libraries.

Steps 1 to 4 — before any model

1. Problem definition. Which structural scheme meets the brief at the lowest embodied carbon, within the agreed floor zone, and what does the grid choice cost?

Note the second clause. The brief asked about the scheme; the question asks about the grid too, because Module 20 established that the grid is a decision of comparable size and it was about to be settled without a number.

2. Concept options. Four genuinely different families: composite steel, concrete flat slab, concrete band beam, CLT and glulam hybrid.

3. Parametric variables. Grid span 6 to 12 m (varying); scheme (varying); floor area and storeys (fixed by the brief); carbon factors (uncertain, to be tested).

4. Digital workflow. Rate-based comparison at this stage — no model. A model is not built until step 9 and only for the scheme that survives.

Steps 5 to 8 — deciding the model

5. Model strategy. None yet. The scheme question is answered on rates, and a model would be false precision. This is a legitimate answer to step 5 and it is worth saying explicitly rather than by omission.

6, 7. Elements and solver. Deferred with the model.

8. Prediction. Written before anything is computed:

• Timber will have the lowest carbon and the greatest depth.
• The flat slab will be shallowest and highest-carbon.
• 12 m will cost roughly 80 % more carbon than 6 m.
• Steel will be lighter than the concrete options and not lower-carbon.

Steps 12 to 13 — the exploration

At 12 m, the depths are: flat slab 414 mm, band beam 482 mm, CLT 600 mm, composite steel 600 mm.

Against a 1 100 mm zone including services, all four fit — but the timber and steel options leave only 500 mm for services across the whole floor, which is tight for a 12 m span. Flagged, not resolved: it needs the services engineer.

Carbon at 12 m tells the story the prediction expected, and one thing it did not. At a 9 m grid the four schemes come out at CLT 447, band beam 666, composite steel 782 and flat slab 866 tCO₂e — and the steel frame is lighter than the band beam and carries more carbon, which was predicted, and is worth having predicted.

The grid comparison: doubling the span from 6 m to 12 m costs the steel scheme 83 % more carbon (562 → 1 028 tCO₂e). The prediction said roughly 80 %.

Robustness. Perturbing every carbon factor by ±30 %, one at a time, does not change the winner. The ranking is worth acting on despite the factors being unverified.

Step 14 — AI-use assessment

Three candidate uses, assessed against Module 18's rule:

• Predicting scheme carbon from past projects — no. A rate-based deterministic calculation exists, is transparent, and is what was used.
• Suggesting a starting grid from comparable buildings — defensible as a recommendation to a human, with the deterministic comparison run afterwards. Not used here; the four-scheme sweep was cheaper.
• Screening the completed model for unusual warnings — yes, later, at step 9, with a human reviewing the output.

Steps 15 to 16 — record and recommendation

Recommendation. Adopt the CLT and glulam hybrid at a 9 m grid rather than the proposed 12 m.

On what evidence. Lowest carbon of the four schemes at every span examined, by a margin of 49 % over the next scheme — which survives a ±30 % error in any single carbon factor. The grid reduction from 12 m to 9 m saves roughly a third of the frame carbon on its own.

What it depends on. That 9 m is acceptable for letting; that the services zone works at 500 mm; that a timber supply chain can be secured early; and that the carbon accounting excludes sequestration, which is the conservative choice and is stated.

What would change it. Verified factors showing the timber figure materially higher; a letting requirement that genuinely needs 12 m; or a fire or acoustic constraint that rules out exposed timber.

What is excluded. Foundations and substructure — which would widen the timber advantage, since it is the lightest scheme by a factor of two. Cost. Programme.