GGBS at 50%: what actually happened

Replacing half the Portland cement in a mix with ground granulated blast-furnace slag cuts the embodied carbon of that concrete by roughly forty per cent. It also changes the set time, the colour and the shape of your programme. Everyone quotes the first number. The other three are what you actually manage on site.

Why we tried it

Concrete is the largest single line in the embodied carbon of almost everything we build, and cement is nearly all of that line. Slag replacement is the least exotic lever available: no new technology, no unfamiliar supply chain, no performance argument to win with a structural engineer. It is also, in our region, technically favourable for reasons that have nothing to do with carbon — lower heat of hydration in thick pours, and materially better resistance to chloride and sulfate attack, which matters a great deal in saline coastal ground.

The set time

Slag gains strength more slowly than Portland cement, particularly early and particularly when it is cool. Through most of the year here that is a non-issue or an advantage. In January, with night temperatures in the low teens, it is a real one. Our 28-day strengths were comfortably met on every pour. Our three-day strengths, which are what actually govern striking and therefore the cycle, were not.

The practical consequence is that striking times extended by around fourteen hours on winter pours. That is not a disaster if you planned for it. It is a lost day per cycle if you did not.

The colour

Slag concrete is paler than Portland concrete — noticeably so once cured, and stranger than that when fresh. Break a young slag-rich pour open and the core is a distinct blue-green that fades over weeks as it oxidises. On buried structure nobody cares. On exposed concrete it matters enormously, and the risk is not the colour itself but the variation: two pours at different replacement levels, or the same level in different weather, will not match.

We now treat exposed slag concrete the way we treat stone. Single source, single mix design, sequential pours, and a sample panel that everybody signs before the first structural pour happens.

The programme

This is where a good idea meets a real site. The extra fourteen hours only costs you if your cycle assumed the old figure. Ours did, for the first two pours, and we lost two days learning it. After that we planned the cycle around the mix rather than planning the mix around the cycle, and the cost went to zero.

That reversal is the entire lesson. Low-carbon concrete is not a substitution you make at procurement and forget. It is a programme input, and it needs to be in the programme before the programme is agreed with the client.

The carbon

A 42% reduction against the same mix at CEM I, on the pours where we used it. We are deliberately not extrapolating that to a headline figure for the whole building, because the honest number depends on how much of the structure the mix was used in, and the answer was not all of it.

Where we will not use it

• Fast-cycle slabs in January. If the programme genuinely cannot absorb the striking time, the right answer is a lower replacement level, not a heroic one.
• Exposed concrete that has to match existing work. You will not match Portland-grey with slag-pale, and trying is expensive.
• Anywhere the supply is intermittent. A mix you can only get sometimes produces exactly the variation you were trying to avoid.

What we specify now

50% replacement as the default for substructure, rafts, cores and any mass pour, where the lower heat of hydration is a benefit in its own right. 30% for fast-cycle suspended slabs in winter. A single mix design, sample panel and single source for anything exposed. And the striking times written into the programme at tender rather than discovered on the second pour.