Epoxy mortar
Three parts sand to one of resin, and the only thing on this page poured below the floor
Concrete fails locally long before it fails as a slab. A machine foot rusts and jacks a cone of concrete up around its bolt; a die gets dropped; a threshold spalls where the rain has been running off a roller door for fifteen years. What you are left with is a hole, and there is no film thin enough to bridge a hole.
The instinct is to fill it with the same resin that is going on the floor, because it is already open and already mixed. That instinct is the reason this material exists. Neat resin poured six millimetres deep shrinks measurably as it cures, and it sets a fraction below the concrete around it, so the repair you could not see on handover day is a dull saucer in the gloss by the end of the first year.
Why it is three parts sand
Gauging the resin with kiln-dried silica at roughly one to three by volume changes what the material does as it cures. Sand does not shrink. The more of the volume that is sand, the less of it is free to move while the resin goes off, which is the whole trick: the repair stays where it was trowelled instead of sinking away from the edges.
It also ends up harder in compression than the slab it is patching, and considerably cheaper than filling the same void with neat resin at two litres a square metre. The sand has to be kiln-dried rather than merely dry: a trace of moisture in the aggregate will haze an epoxy cure from the inside, and builder’s sand off a Nairobi yard in April is never as dry as it looks.
Six millimetres, or we cut it deeper
A mortar repair needs depth to hold. Feather-edged out to nothing it has no body at the perimeter, and the thin margin debonds under the first wheel that crosses it. So six millimetres is a floor, not a target: where a spall is only three deep, we cut it out until it is six.
This is why a repair sometimes looks worse before it looks better, and it is worth saying out loud on the day: a client who watches a grinder make the damage visibly bigger deserves to have been told first that this is the repair working, not the repair going wrong.
Where we have got this wrong
Early on we treated the twelve-hour overcoat as advisory on a hot day, and put a base coat over mortar that had been down since the morning. It looked perfect. Four months later that patch read as a matt rectangle through a gloss floor, because the mortar was still giving off cure underneath a film that had already closed over it.
It cost us a bay and a weekend to put right, and it is the reason the overcoat figure on this page is printed as a rule rather than a guide. Twelve hours, and the mortar is walked on and checked before anything goes over it.
How to tellSix months in, a mortared repair is invisible. A repair filled with neat resin instead of mortar shows as a slightly different gloss in raking light, because a deep pour of unfilled resin shrinks as it cures and sinks a fraction below the floor around it.
What to ask
Ask what actually goes into a spall, and in what ratio. "Epoxy" is not an answer: neat resin and a gauged mortar are both epoxy, and only one of them is still flush with your floor in five years.