Unchanging concepts - Concrete Mix Design, Quality Control and Specification - page 253

Results
(30 MPa specied strength)

n = 5
( x = 34; σ = 3)

n = 20
( x = 31; σ = 3)

n = 5
( x = 34; σ = 3)

- - -

0
-10
-20
-30
-40
-50

Graph analysis

Figure 11.3 Graphical analysis of run of understrength results, which merits a penalty.

the difference is only $628 and the 30-result basis is reasonably satisfactory
and much simpler to incorporate into a specification.
The assumption is that the concrete supplier would have had to spend
approximately $4.00 m 3 in extra cement on the 400 m 3 (523 yd 3 ) of con-
crete to avoid penalisation (total saving: approximately $2000 in cement
cost), so the net cost to the supplier is approximately $1600. Obviously, the
supplier would prefer to pay this penalty rather than delay the work and
pay the costs of coring and investigating 400 m 3 (523 yd 3 ) of concrete, with
the risk that some or all of it might be rejected.

Importance of quality of testing
It is obvious that the test results forming the basis for a cash penalty should
provide an accurate assessment of the quality of concrete as supplied by the
producer. This is by no means something that is easy or can be taken for
granted. A minimum requirement is that samples should be taken, cured, and
tested by a competent, accredited, and preferably independent organisation.
The best criterion of testing accuracy is the average difference of pairs
of test results from the same sample of concrete. This average difference
should not exceed 1 MPa (145 psi) for normal concrete (specified strength
less than 50 MPa [7246 psi] and possibly excluding very low slump mixes).
It is suggested that the highest of a pair of specimens is likely to be a better
estimate of the true concrete strength than the mean of the pair.
The person responsible for result analysis should be alert for clearly estab-
lished cases of incomplete compaction and improper curing and testing, and
should be prepared to exclude such results from a penalty assessment. The
previously recommended graphical analysis system, including analysis of
related variables such as slump, strength, and testing, has been found valu-
able in distinguishing causes of variability and early detection of problems.