Quality control - Concrete Mix Design, Quality Control and Specification - page 238

The cost of a higher margin is easily calculated: Cost of additional mean
strength = Specified strength + 1.65 × SD. With SDs ranging from less than
2 MPa to more than 6 MPa, the producer with poor control has effectively
to supply the next higher grade of concrete at the price of a well-controlled
lower grade.
The cost of better control is not so easily calculated as it is strongly
affected by both the choice of control system and the degree of understand-
ing of several basic concepts. Some of these concepts are
1. Professor J.M. Juran's dictum that QC of any manufactured item
should be aimed at “controlling the mass and not the piece”. Thus
control should be aimed at controlling the whole production of
a plant or even a group of plants rather than individual batches or
isolated deliveries to a particular site of a particular grade or on a
particular day.
2. Pareto's principle of locating the main causes of variability for par-
ticular circumstances and concentrating control on them, rather
than spreading it more thinly over all possible causes. (Pareto was
an Italian economist engaged in trying to discover the sources
of Italy's wealth. He found that in any town, more than half of
the total wealth was under the control of four or five men and he
could get a better answer by first finding these men and then ask-
ing his questions rather than attempting a general survey of the
population.)
3. The purpose of testing anything is not to discover and reject unsat-
isfactory items but to establish the minimum quality level of the
whole. (If one in ten deliveries is tested, then for every individual item
rejected, nine equally defective items will have been accepted.)
4. There is a difference between unsatisfactory and unacceptable con-
crete. Unsatisfactory concrete (that which does not quite meet the
specification but is not dangerously defective) can be detected and
financially penalised on the basis of a statistical analysis of a month's
test results without further investigation. It has to be recognised that,
under perfect control, there is 1 chance in 1,000 of a result 3.09 SD
below the mean. More realistically this can be seen as only 1 chance
in 1000 that such a result is a consequence of normal variability rather
than a particular problem. Again, more realistically, there is 1 chance
in 100 of a result 2.33 SD below the mean and this would be 1 × SD
below the specified strength, so such a result cannot be considered to
be unacceptable.
Any unacceptable concrete has to lead to extensive in situ investiga-
tion of all concrete of that grade during the period and would constitute
a huge problem for both the producer and the site team. (Experience
is that if unsatisfactory concrete is always detected and penalised,