Testing - Concrete Mix Design, Quality Control and Specification - page 131

Te is t i ng
7.1 PHILOSOPHY OF TESTING
It is very important to understand the philosophy of testing. Only
people ignorant of the true situation regard a test result as an accurate
portrayal of the property tested. Unfortunately this tends to include
many people in authority such as specifiers, controllers, and legal people.
First, no test can be perfectly accurate and it is as well to consider
how inaccurate it might be. Second, the sample tested may not be truly
representative of the mass being assessed. For example, standards may
require great care in checking the equipment and following a rigid
procedure to get an accurate sand grading. It may also present clear
rules for obtaining a representative sample. But if you are doing quality
control (QC) on concrete, there is nothing better than doing frequent
rough checks (twice the number of tests in half the time), looking at
the results on a cusum graph of specific surface, and taking a second
sample to confirm if the first one says there has been a change. Another
important point is that it is good to test different parameters whenever
practical so that aberrant results can be weeded out and true change
points confirmed by more than one parameter. On one major project
in the Middle East, the compressive strength reduced at the same time
as the resistivity. Coincidentally the huge building boom at the time
had resulted in a shortage of silica fume that was in the design mix.
The premix company was accustomed to arguing about the vagaries of
compressive strength testing but had no answer to two parameters simul-
taneously changing.
A very important distinction between QC and research is continuity.
A research project, however large and long, must eventually come to
an end, and some very elaborate statistical techniques and great care to
achieve testing accuracy may be of substantial value in reaching an accu-
rate conclusion. QC is a continuing flow of data that may necessitate revised
conclusions from time to time. Many factors may affect the desirable level