The specification should define the mixes or grades of concrete required and in what parts of the works each mix or grade is to be used. As mentioned in the preceding section, both BS 5328-2:1997 and BS 8500:2002 describe four classes of concrete mixes designed mixes, prescribed mixes, standard mixes, and designated mixes.

Designed mixes are specified by the purchaser stating the characteristic strength required, maximum size of aggregate and minimum cement content, leaving the supplier to design the mix proportions.

Prescribed mixes are specified by the purchaser stating the proportions of the mix constituents required cement, aggregate, size and type, etc. the purchaser being responsible for the performance of the mix.

Standard mixes are set out in a Section 4 Table 5 of BS 5328-2, and also in BS 8500 where they are called Standardized Prescribed Concretes. They are for concrete of characteristic strengths from 7.5 to 25N/mm2 and BS 5328 gives the mix proportions for size of aggregate, and workability. Table 19.2 reproduces Table 5 of BS 5328-2. The lower grades of characteristic strength 7.5, 10.0 and 15N/mm2 are intended for use in mass concrete filling to strip footings, blinding concrete, and similar.

Designated mixes are for mixes to meet special requirements, such as for sulphate resisting concrete, etc. and also for ready-mix concrete for which the supplier is required to hold a current product conformity certificate, to BS EN 150 9001.

The characteristic strength of a mix is defined as that value of strength below which 5 per cent of¦strength measurements¦are expected to fall. On a statistical basis, cube strength test results on a given mix are found to follow a normal distribution that is 50 per cent of test results are above the mean X, and 50 per cent below. If only 5 per cent of results are to fall below a required value P, then the mean strength X must obviously be higher than P. From the characteristics of a normal distribution curve, the value of X has to be 1.64S higher than P, that is, X (P 1.64S) to achieve not more than 5 per cent results below P, where S is the standard deviation1 of the test results obtained. If only 2.5 per cent of results are to fall below P, then X must be (P 1.96S).

Reliable values of S cannot be obtained unless enough sample results are available. BS 5328 requires a minimum of 30 tests to be made before assessing S; and for n 30 to 100 tests, the value S should be taken where S (0.86 2/n)S, with a minimum value of 6N/mm2. The DoE manual, Design of normal concrete mixes (1988), recommends that for Grades C20 and above, S should be taken as not less than 8N/mm2 until more than 20 test results are available, and a minimum value of 4N/mm2 should apply however many tests are taken. For trial mixes, values of 2S would also normally be taken (rather than 1.64S) to give a further margin of safety. However, a less exacting approach is also allowable as described in the next section.