Acceptance criteria for concrete cube test results are given in **SANS 10100-2: The structural use of concrete Part 2: Materials and execution of work**.

**Two acceptance criteria are given for general concrete work and hardness testing**, both of which must be complied with:

- The result of any valid test result must not be more than 3 MPa below the specified characteristic strength, e.g. 27 MPa in the case of a grade 30 mix, and
- The average of any three consecutive valid results must exceed the specified characteristic strength by at least 2 MPa, e.g. 32 MPa in the case of a grade 30 mix

In the event of failure, the cause of the failure must be investigated and remedial action taken. Investigation would include checking the sampling and testing procedures, non-destructive testing of the concrete member(s), core testing, and load testing, sometimes all four.

The code also states that if the magnitude of the concrete work, or the sampling frequency, is such that at least 30 valid results for a specific grade of concrete become available in a three month period, the contractor may choose to have results assessed statistically.

In this case the average strength of any 30 consecutive valid results (for each grade) must exceed the specified strength by 1.64 times the standard deviation (SD) of those 30 results. In addition no single valid result may be less than 3 MPa below the specified strength. Again, in the event of failure the cause must be investigated and remedial action taken.

**Concrete cube strength is a random variable** and the test results are influenced by many different factors related to variations in the materials, batching, sampling, concrete testing, equipment and personnel.

Being a random variable, the results of a concrete cube test taken from a particular grade of concrete will, for practical purposes, take the form of a normal distribution when plotted graphically. **This type of distribution is sometimes called a “bell” curve because of its shape**. Like many aspects of concrete technology this is not strictly true – low strength and very high strength concrete test results give skewed distribution curves to the left and right respectively, but for practical purposes a symmetrical distribution is assumed.

The normal distribution curve is characterised by two values: the mean or average value, and the standard deviation which quantifies the spread of the curve either side of the mean value. The requirement that the average strength exceeds the specified (characteristic) strength by 1.64 times the standard deviation implies that slightly fewer than 5% of results will fall below the specified strength.

The only factor under the concrete suppliers control is the standard deviation, the specified strength and the “1.64” being fixed.

The value 1.64 x SD is called the margin. The higher the SD, the higher the margin and the higher the average binder content of the concrete. In other words, **the higher the SD, the more expensive the concrete in terms of binder cost**.

For example, improving control from poor (SD = 7 MPa) to good (SD = 5 MPa) reduces the margin by 1.64 *(7-5) MPa = 3.4 MPa. **A reduction in average strength by 3.4 MPa is equivalent to saving 20 kg of binder** per cubic metre of concrete.

**At current prices this is a saving of roughly R25 / m ^{3}**. Of course one has to balance this against the cost of the additional quality control necessary to reduce the SD and there is an optimum point where it becomes uneconomical to reduce the SD any further. For sophisticated concrete suppliers this point would be an SD somewhere in the region of 2.5 to 3 MPa.

The CUSUM technique is a more sophisticated statistical tool which is used by large concrete suppliers to monitor concrete quality for a range of different mixes. “CUSUM” is an acronym for Cumulative Sum and** the technique involves calculating the cumulative sum of the differences between actual results and targeted or forecast results**. Normally three CUSUM tables and charts are maintained:

- The mean strength CUSUM,
- The correlation CUSUM (correlation between predicted and actual 28 day results)
- The standard deviation CUSUM

The slopes of the CUSUM charts indicate whether the results are over target, on target or under target. Abrupt changes in slope on one or more of the charts indicate that a change has occurred somewhere in the system. The source of the change could be materials related, sampling related or testing related.

Core testing is a form of strength testing that is **commonly resorted to in the case of the failure of cubes to meet their specified strength**. Core drilling, preparation and strength testing is covered in SANS 5865. Acceptance criteria for core test results are given in SABS 0100 Part 2, clause 14.4.3. The average strength result of the cores must exceed 80% of the specified characteristic strength and **no single core result may be below 70% of the specified characteristic strength**.

It is important that the test method is followed exactly. **Experience has shown that test procedures can affect results significantly**. Of particular importance are “squareness” and flatness of the ends of the core. It is interesting that the section of core chosen for testing can significantly affect results. Recent tests carried out on cores from a floor slab consistently gave results 3.5 MPa higher when the test specimen was cut from the bottom of the core rather than the top.

Apart from strength,** core samples can also yield useful information such as degree of compaction, voids content, and distribution of coarse aggregate**. The cores can also be analysed for binder content and slices from the cores can be tested for oxygen permeability, water ‘sorptivity’ and chloride conductivity. They can also be tested for abrasion resistance and depth of carbonation.

The depth of carbonation test must be carried out on a freshly fractured surface, not on the outside surface of the core which is often contaminated with finely ground material from the drilling process.