What a Concrete Cube Test Actually Tells You

A cube result lands on the desk reading 34.5 MPa for C30/37 concrete. Is that a pass? The answer is that the question is not well formed, and understanding why is the difference between a sensible site conversation and an expensive argument.

The grade is a fractile, not a minimum

In EN 1992-1-1, $f_{ck}$ is the characteristic cylinder strength: the value below which 5% of results are expected to fall. It is not a floor that no specimen may cross. By construction, one result in twenty is supposed to be below it.

C30/37 pairs a cylinder strength of 30 MPa with a cube strength of 37 MPa. The two numbers describe the same concrete tested on different specimen shapes. A cube is squatter, the platen restrains it more, and it reads roughly 20 to 25% higher than a cylinder of the same material. Which one you compare against depends on which one the lab cast, and mixing them up is the most common way to misread a report.

The design strength is lower again:

$$f_{cd} = \alpha_{cc}\,\frac{f_{ck}}{\gamma_C}$$

With $\gamma_C = 1.5$ and $\alpha_{cc} = 1.0$, C30/37 gives $f_{cd} = 20$ MPa. So between the 37 MPa printed on the delivery ticket and the 20 MPa used in the calculation there is almost a factor of two, and every part of that gap is doing a job.

What the acceptance rules actually require

EN 206 assesses conformity on groups of results, not one at a time. For initial production, with 3 results in the group, the criteria are:

• mean of the group $\ge f_{ck} + 4$ MPa

• every individual result $\ge f_{ck} - 4$ MPa

For continuous production, with 15 or more results, the mean criterion becomes $f_{cm} \ge f_{ck} + 1.48\,\sigma$, where $\sigma$ is the established standard deviation of the production, while the individual criterion stays at $f_{ck} - 4$ MPa.

Two things fall out of this. First, a single low result is not a failure by itself: it has to be below $f_{ck} - 4$ before it fails the individual criterion at all. Second, a producer with tight control, a small $\sigma$, is allowed to target a lower mean. That is why a good supplier's results cluster close to the grade while a poor one has to over-design the mix to pass the same rule.

Returning to the opening question: 34.5 MPa is above the 30 MPa cylinder characteristic and well above $f_{ck} - 4 = 26$ MPa, but it is below a 37 MPa cube characteristic and only just above $37 - 4 = 33$ MPa. Without knowing the specimen type, the number means nothing.

What the test does not tell you

It is not the strength of the structure. The specimen was compacted in a mould on a vibrating table and cured in water at $20 \pm 2^\circ\mathrm{C}$ under EN 12390-2. The concrete in the column was pumped, placed around a congested cage, vibrated by a tired operator and cured by whatever the weather provided. The cube tells you the concrete as delivered was capable of the strength. It does not tell you the concrete as placed achieved it.

It is not a record of the whole pour. A set of cubes is taken from one sample, from one truck, at one moment.

28 days is a convention, not a ceiling. EN 1992-1-1 §3.1.2 gives a maturity relationship for strength with age. With a class N cement, concrete reaches roughly 78% of its 28-day strength at 7 days, and keeps gaining slowly after 28. The 7-day result is a useful early warning, not a verdict.

Temperature history matters more than the date. Cubes cured in a hot site hut and cubes cured properly in a tank will not give the same answer, and the difference is not the concrete's fault.

When a result really is low

If a group genuinely fails, the cubes are the start of the investigation, not the end. The sequence that tends to work:

1. Check the paperwork first: specimen type, age at test, curing record, the machine's calibration date. A surprising share of low results are testing problems. 2. Establish what the low result covers. Which pour, which elements, how much concrete. 3. Assess the structure, not the specimen. Core testing to EN 12504-1 measures the in-situ concrete, and rebound hammer or ultrasonic testing can map variation across an element cheaply once it has been calibrated against cores. 4. Check whether it matters. Many elements are not governed by concrete compressive strength at all. A lightly loaded slab designed for deflection and crack width may be entirely unaffected by a grade shortfall that would be serious in a heavily loaded column.

Step 4 is the one most often skipped, and it is usually the one that decides the cost.

Takeaways

• $f_{ck}$ is a 5% fractile. One result in twenty is expected to sit below it.

• Always check whether a result is a cube or a cylinder value before comparing it to the grade.

• EN 206 judges groups, with the individual criterion at $f_{ck} - 4$ MPa, not at $f_{ck}$.

• A cube measures the concrete as delivered, under ideal curing. It is not a measurement of the structure.

• A failing group calls for cores and an assessment of the actual element, and quite often the element turns out not to be governed by concrete strength.

#concrete #qualitycontrol #eurocode2