Concrete remembers

Three buildings, all being taken apart rather than knocked down, all in the same eighteen months. Demolition is the only phase of the life cycle nobody draws, and standing inside one is the fastest way to find out what a building was actually made of, as opposed to what the specification said.

The first thing you notice is how little of it can come back. Not for technical reasons — the frames in two of the three were sound enough to re-erect — but because nobody can certify a beam whose history is a scan of a 1974 drawing and a surveyor's opinion.

The arithmetic that follows is unforgiving and widely ignored. A new frame of the same span carries roughly four times the embodied carbon of the one being removed, before anything is clad. Reuse closes that gap entirely and opens a different one, in programme and in risk, which is where the conversation always stops.

To put a reclaimed steel section back into a building you need to know what it is. Not approximately — you need a grade, and a grade is a claim about chemistry and about how the metal behaves after it yields. A section rolled before the mid-1970s was made to a standard that no longer maps cleanly onto the one the engineer is designing to, and the mill certificate, if it ever existed, went with the contractor who went into administration in 1991.

So you test. Testing means cutting coupons out of the member, which means the member is shorter, which means either the grid changes or the coupon comes from the part you were going to cut off anyway, which is the part that tells you least. Three coupons per batch is the usual ask. On the second of the three buildings that came to eleven thousand pounds of testing to release about sixty tonnes of steel, and the number that made it work was not the carbon. It was that the replacement sections had a fourteen-week lead time and the testing took three.

Concrete is worse, because concrete does not come apart. A reinforced frame is a single object that happens to have been poured in stages. You can cut it, and people do, but what you recover is a slab fragment with unknown reinforcement, unknown cover and a cut face that was never meant to be an edge. The second building gave up eleven hundred square metres of sound slab. All of it was crushed.

Every structural engineer we spoke to during the three projects gave a version of the same answer. They are not against reuse. They are against signing for a member whose provenance is a photograph. The liability for a frame sits with one named individual for the life of the building, and that individual has no mechanism for sharing it with the 1974 fabricator.

Embodied carbon has no such owner. It is counted in a report, compared against a target, and reported to a client who has already paid for the building. Nobody is personally liable for it, nobody is sued over it, and no professional indemnity policy has ever been triggered by it. Put an unowned cost next to an owned risk and the outcome is not in doubt — and it is not, in the strict sense, irrational. It is what the incentives say.

Which is why the interesting work on reuse is not technical. The engineering was solved on the second building in about a fortnight. What took nine months was arranging for the testing regime to be agreed with the approving authority before demolition began rather than after, so that the steel was characterised while it was still standing and still had drawings attached to it.

That is the whole intervention: move the paperwork earlier. It saved sixty tonnes. On the first building, where it was attempted after the frame came down and the members were in a yard with their connections cut off, it saved nothing at all, and the carbon case was identical.