Unchanging concepts - Concrete Mix Design, Quality Control and Specification - page 251

5 MPa (725 psi) (poor control)
• Required mean strength 30 + (1.28 × 5.0) = 36.4 MPa (5275 psi)
• Effective minimum strength 36.4 - (3 × 5.0) 21.4 MPa (3101 psi)
For a k value of 3.0 (preferred), and a specified strength of 25 MPa (3623
psi), the effect would be
2.5 MPa (362 psi)
• Required mean strength 25 + (3 × 2.5) = 32.5 MPa (4710 psi)
• Effective minimum strength 32.5 - (3 × 2.5) = 25 MPa (3623 psi)
5 MPa (725 psi):
• Required mean strength 25 + (3 × 5.0) = 40 MPa (5797 psi)
• Effective minimum strength 40 - (3 x 5.0) = 25 MPa (3623 psi)
The effect of the change would be to worsen the competitive position
of the high-variability supplier and limit the occurrence of occasional low
strengths in the concrete supplied. The low-variability supplier would be vir-
tually unaffected, except for the supplier's improved competitive position.
Figure 11.2 shows the relative situation under exact compliance of a
10% defective criterion for both the high- and low-variability supplier. The
upper graph shows that under the present (U.S.) 10% defective basis, the
low-variability supplier has a reduced incentive and the high-variability
concrete includes some deliveries of very low strength. The lower graph
shows an enhanced competitive position for the low-variability under the
proposed 0.1% defective basis. Both suppliers in this case provide effec-
tively the same minimum strength.
The benefits of low-variability concrete are substantial:
1. Helpful to the concrete placing crew
2. More uniform compaction
3. More uniform appearance
4. More accurately assessed on a given number of test results (possibly
less frequent testing required)
5. Better control of pumping
Influence of change points
The proposed technique assumes that there will be a gradual drift of either
mean strength or variability, and that it will be legitimate to select 30 results
incorporating the worst period. Analysis has shown, however, that changes
are usually step changes rather than gradual drifts. Thus, a specific number
of results constitute the low period and all of them (and no more) should
be analysed to represent the low period rather than taking an arbitrary
30 results. This is too complicated and indefinite for use in a specification