Statistical analysis - Concrete Mix Design, Quality Control and Specification - page 228

if either the testing process or the regulating analysis system is of less than
the highest standard but does not necessarily have higher variability. Cook's
view is that lower coefficients of variation on high-strength concrete are
obtained simply because the producer is trying harder than with his nor-
mal concrete. This contrasts with the often expressed view that a producer
makes his reputation on his high-strength concrete but his profit on his
low-strength concrete. For this reason, Australian concrete producers are
certainly trying every bit as hard to achieve low variability on their low-
strength concrete. However it may well be the case in the United States,
where specifications often do not allow the producer to derive any financial
benefit (i.e., any cement reduction) from the attainment of lower variability.
The authors' strong advocacy of standard deviation as the measure of
compressive strength variability does not mean that the coefficient of varia-
tion is a useless parameter. Obviously the same standard deviation cannot
apply to such variables as tensile or flexural strength, much less to slump
or density. A 5% to 10% coefficient of variation in anything generally rep-
resents a variable under reasonable control, although, for example, a mod-
ern batch plant can achieve much better than 1% in cement batch weight
(if properly maintained).
9.7 PRACTICAL SIGNIFICANCE OF
STATISTICAL ANALYSIS
The most obvious point emerging from the foregoing is that it is not feasible
to take a quantity of concrete small enough to be regarded as a unit for
purposes of acceptance or rejection, and to represent it by a sufficient number
of test results to assess its quality with reasonable accuracy. It is also eco-
nomically impractical to consider physically rejecting concrete that is only
slightly understrength. Since the future progress of the concrete industry
depends on encouraging reduced variability, it is absolutely essential that
quality be assessed on the basis of a large enough pool of results to enable
not only mean strength but also variability to be accurately assessed. Since
no one should consider rejecting a month's concreting because compliance
testing suggests it is slightly understrength, there is simply no other way to
go than cash penalties or cash incentives. (Although it is feasible for the real
diehards to impose this penalty in the form of increased cement content or
increased testing as noted in Chapter 6.)
The next point is that we do not wish to sit back and watch the contrac-
tor dig his financial grave for a month or so without taking any action.
Even worse not taking appropriate action until the concrete becomes not
just contractually unacceptable but structurally unacceptable. An eventual
cash penalty may bring justice to the situation and may avoid him repeating
his error, but it will not provide the quality of concrete required in the