Quality control - Concrete Mix Design, Quality Control and Specification - page 233

Change point
Figure 10.2 Use of V-mask on cusum chart.
of a downturn (or upturn) can be seen, then it is genuine. For example, if
two or three 7-day results on a strength cusum show a downturn, and that
downturn coincides with a downturn in density and an upturn in either
slump or temperature, then the cause of the strength reduction is fairly
clear, and the downturn is obviously genuine without waiting for a dozen
or more results to provide statistical certainty.
What is required is a multivariable cusum, which is a series of cusum
graphs of all the available information. The ConAd system, now owned
by Command Alkon and retitled “CommandQC”, provides for dozens of
variables including sand grading, cement properties, admixtures, batching
details, delivery truck, site sampler, and tester. Of course it does not help to
have more than eight graphs on the screen at a time but they can be cycled
through (but on the Pareto principle, see later).
Although a full program such as CommandQC or SpectraQEST is very
worthwhile for a substantial producer, Ken Day has produced a small free
program to enable smaller producers to apply the principle and also to
enable specifying engineers to assure themselves, with very little time and
effort, that control is effective.
It should be noted that although independent commercial laboratories
can certainly provide the service (provided they make a point of distributing
all results on the day they are obtained) Day's experience of over 50 years
and many countries has been that the best QC is provided by a laboratory
owned and operated by the concrete producer. Where this is being done and
the honesty or competence of the laboratory is in question, it is generally
too expensive and too difficult to organise for an independent lab to sample
the same truck as the producer's team (and testing different trucks does not
really prove the point). The solution is to require the truck to be sampled
to be selected before its arrival by site staff (not the producer's personnel)
and to require a double set of specimens to be made, with half delivered to
an independent lab. The objective being to check on the producer's testing