Every calculation on this site has fcu in it. The beam's moment capacity, the column's axial capacity, the shear a section can take without links — all of them assume a grade of concrete that somebody promised. Cubes are the only evidence that the promise was kept.
This page is about reading the certificate: what the number means, what it is allowed to be, and what happens next when it is not.
Load to strength, in one division
fcu = load (N) / 22 500
= load (kN) / 22.5
Example 750 kN / 22.5 = 33.3 N/mm²
100 mm cube area = 10 000 mm² -> load (kN) / 10
A 100 mm cube reads a little higher than a 150 mm cube of the same concrete — a smaller specimen has fewer flaws to fail at. If your specification is written for 150 mm cubes, test 150 mm cubes. Mixing sizes and comparing the numbers directly is not a small approximation, it is a bias in the direction that flatters the concrete.
Weigh the cube before you crush it
Normal-weight concrete is about 2400 kg/m³, and a 150 mm cube is 0.003375 m³, so a sound cube weighs roughly 8.1 kg. A cube coming in at 7.5 kg has air in it that should not be there, and it will crush low for a reason that has nothing to do with the mix. The mass is on the certificate; read it.
What "grade 25" actually asks for
Grade 25 is a characteristic strength: the value below which not more than 5% of results should fall. It is a statement about a distribution, not about one cube. Concrete varies from batch to batch, so meeting a characteristic value means aiming above it by a margin that covers the variation:
Good control, sd = 5 25 + 8.2 = 33 N/mm²
Ordinary site, sd = 8 25 + 13.1 = 38 N/mm²
So a set of cubes averaging 25 on a grade 25 job is not good news. It means the mix was targeted at the grade rather than above it, and roughly half of everything poured is below the number the design used. The looser the control on site, the higher the mean has to be — which is the real argument for batching by weight rather than by headpan.
The compliance rules
BS 8110 does not judge compliance itself; it points at the specification standard, and which one applies depends on what the contract names. The two you will meet:
| Rule set | Mean of consecutive results | Any individual result |
|---|---|---|
| BS 5328, as referred to by BS 8110 | ≥ fcu + 3 over any 4 consecutive | ≥ fcu − 3 |
| BS EN 206 / BS 8500, initial production | ≥ fck + 4 over the first 3 | ≥ fck − 4 |
Both have the same shape: a mean comfortably above the grade, and a floor a little below it that no single result may cross. Note that the Eurocode family states strength as a cylinder value — fck 20 goes with fcu 25, which is what C20/25 means. See BS 8110 vs Eurocode 2.
Three sets, read against grade 25
| 28-day results, N/mm² | Mean | Verdict on BS 5328's rule |
|---|---|---|
| 34.2, 31.8, 33.5 | 33.2 | COMPLIES — mean well over 28, lowest well over 22 |
| 26.2, 24.4, 25.1 | 25.2 | FAILS — every cube "looks like" grade 25, and the mean is 2.8 short |
| 30.1, 29.6, 20.8 | 26.8 | FAILS — the individual minimum, 22, is breached by one result |
The middle row is the one worth remembering. Nothing about it looks wrong to somebody reading down a column of numbers next to the word "25", and it is the commonest way a job ends up with concrete a whole grade below its design without anybody noticing.
The 7-day cube is an alarm, not a verdict
Reading 22 at 7 days -> 22 / 0.67 = 33 expected at 28
Reading 15 at 7 days -> 15 / 0.67 = 22 expected at 28 act now
The point of the 7-day set is that it arrives while you can still do something: stop pouring the same mix, find out what changed, and keep the suspect pour identified. It cannot accept or reject anything — the specification is written at 28 days, and blended cements gain strength on a different curve entirely.
The cube tests the cube
A cube result describes concrete that was compacted properly, stored in water at a controlled temperature, and crushed on a calibrated machine at a controlled rate. Break any of those and the number describes the sample, not the structure:
- Curing. Cubes left beside the trench in the sun dry out and crush low. Cubes cured in a tank when the structure itself was never watered crush high. Either way you have measured something other than what is standing there.
- Compaction. Filled in the wrong number of layers, or rodded a token amount, and the cube is honeycombed. The mass gives it away.
- Identification. A result you cannot tie to a specific pour is a result you cannot act on. Which column, which floor, which day.
In-situ concrete is never as strong as its cubes, even when everything is done right — it is compacted and cured under site conditions, not laboratory ones. Cores cut from the finished member read lower than cubes made from the same load, and the relationship between them comes from the testing standard, not from a rule of thumb. Do not compare a core result with a cube result as though they are the same measurement.
What a shortfall actually costs the design
Not everything in a calculation is equally sensitive to fcu, which is worth knowing before anybody starts talking about demolition. Take grade 20 delivered where grade 25 was designed:
| Quantity | Effect of 25 → 20 |
|---|---|
| Beam flexural steel | Barely changes — the lever arm moves a little, and steel does the work |
| Shear capacity vc | Down about 7%, following the cube root |
| Bond, and every lap length | Down about 11%, so every lap should have been about 12% longer — see lap lengths |
| Column axial capacity | Down about 13% — this is the one that bites |
The 225 square column carrying 590 kN with 4T16, designed to equation 39, makes the point:
At fcu 20 N = 0.35(20)(49 821) + 0.67(804)(460) = 597 kN PASS, by 1%
At fcu 18 N = 0.35(18)(49 821) + 0.67(804)(460) = 562 kN FAIL
Read the middle line rather than the last one. A cube at 20 still "passes" — and the entire margin the design was carrying has gone, spent on concrete that was specified and paid for at 25. The third line is only 2 N/mm² further down, which is inside the variation of a well-run site, let alone a badly run one. Columns are mostly concrete carrying load in direct compression, so their capacity tracks fcu almost proportionally: the cubes from a column pour matter more than the cubes from a slab pour, and "the beams are fine" is not an answer.
A low result: the order to do things in
- Read the certificate properly. Age at test, curing history, dimensions, mass, machine calibration date. A surprising number of "failures" are failures of the sample.
- Identify the concrete affected. Which elements came off that batch. If nobody wrote it down, the answer is everything poured that day, which is why records are cheap insurance.
- Check the design against the delivered strength. Re-run the affected members at the strength actually achieved. Slabs and beams often still pass; columns often do not. This step settles most cases without anybody cutting anything.
- Core the element if it still matters, to the relevant testing standard, and interpret the cores as cores.
- Load test, to BS 8110 Part 2, as a last resort.
- Strengthen or remove if it comes to that. It rarely does, and it never does quietly, so steps 1 to 3 deserve to be done carefully.
Nothing on this list is a rebound hammer reading. Surface hardness correlates loosely with strength, needs calibration against cores to mean anything, and reads the outer 30 mm of a member. It is useful for comparing one part of a floor with another. It is not evidence.
Re-run the design at the strength you got
Step 3 above is the one that resolves most disputes, and it is exactly what a design tool is for: change fcu, re-run, and read the PASS or FAIL on every check with the arithmetic beside it. Single members — slab, beam, column, footing, stair, lintel — are free to run, as many as you like.
Open Structura in your browser
On Android, get it on Google Play; on iPhone and iPad, get it on the App Store — same account as the browser.