Skip to content

This takes no carbon figures and produces none. It tests whether a comparison between refurbishing and rebuilding is counted symmetrically enough to support a conclusion, and names which option each failure favours. Nothing is stored and no account is needed.

CarbonSpec / Retrofit or rebuild

Retrofit or rebuild — is the comparison fair?

Whole-life carbon comparisons between refurbishing and rebuilding routinely reach opposite conclusions on similar buildings. The reason is rarely the arithmetic. It is that the two options are counted differently — over different modules, different study periods, sometimes different floor areas — and in a headline figure a difference in method looks exactly like a difference in the buildings.

This takes no carbon figures and produces none. It asks how a comparison was counted, and tells you which of six symmetry conditions it fails — and which option each failure favours. You can run it on a report somebody else wrote.

The refurbishment

Modules counted

The new build

Modules counted

This comparison cannot decide it

1 of the six symmetry conditions fails, so this comparison cannot support a conclusion either way. An asymmetric comparison does not produce a less certain answer; it produces an answer about the method. Every failure favours the new build, so the stated result is the one this method was always going to produce.

  • Are both options counted over the same EN 15978 modules?

    Holds

    Both cover A1-A3, A4, A5w, A5a.

  • Is the same reference study period used for both?

    Holds

    Both are assessed over 60 years.

  • Is energy in use modelled for both options, or for neither?

    Holds

    Neither claims an operational figure, so the comparison is embodied-only on both sides.

  • Is demolition of the existing building charged to the option that demolishes it?

    Fails · Favours the new build

    The new build does not count demolition of the existing building. Clearing the site is a cost of rebuilding that the refurbishment does not incur, and leaving it out removes the one certain difference between the options.

    What fixes it: Count C1-C4 for the existing fabric against the new build. It is not optional and it is not small.

  • Is the carbon already embodied in retained fabric excluded from both options?

    Holds

    Retained fabric carries no figure on either side, which is the only treatment that does not decide the comparison in advance.

  • Do both options deliver the same floor area?

    Not stated

    At least one option states no floor area, so whether they are alternatives cannot be established.

On retained fabric

Carbon already embodied in retained fabric is excluded from both options. It was emitted when the building was built, it is spent whichever option is chosen, and neither demolishing the building nor keeping it changes it. Credited to the refurbishment it is usually the largest figure on the page and makes retrofit unbeatable by arithmetic; charged to the refurbishment it bills a design for emissions it did not cause.

On floor area

Areas differing by more than 5% are not two ways of delivering the same thing. That is not a tolerance to widen — past it, neither comparison basis is neutral, and the question stops being well posed rather than becoming less certain.

What passing does not mean

A failed condition proves the comparison cannot support its conclusion. All six holding proves only that none of these six failures is present — the quantities, the replacement scenarios, the end-of-life routes and the condition survey behind both options are untouched by any of this, and any one of them can be wrong on its own.

The asymmetry runs one way on purpose. A comparison can be refused on its stated terms and cannot be accepted on them, which is the same rule that governs every verdict CarbonSpec issues.