Quality control - Concrete Mix Design, Quality Control and Specification - page 242

The other end of the process is the retrieval of data from storage. If
large amounts of data are recorded, then retrieval must be substantially
automated. The largest amount of data is usually batching data. This is
required in full so that cumulative errors and the variability of the process
can be studied. It is not enough to have all the information tabulated so that
the analyst can run their eye down the column to look for exceptions. It is
not even enough to graph the data, revealing exceptions many times more
quickly. It is necessary for the system to be able to retrieve those items, and
only those items, having an error in excess of any nominated amount. It is
also necessary to have cumulative error graphs, showing whether consump-
tion averages that planned.
A small free program called KensQC is provided on Day's website and
described next.
10.6.2 KensQC
KensQC is the program as you see it when you download it from the web-
site http://www.kenday.id.au and install it on your computer.
The program is not really meant to display substantial numbers of results
on graphs but rather to use the cusum graphs to detect change (and its
cause) at the earliest possible stage. However the tabulated data on the
“Multigrade Stats” printout is useful to show change over a period.
The results currently in the program are not an ideal demonstration of
the value of the program, especially since they are only of a single grade,
but some points can be noted:
1. The average pair difference of 1.1 MPa in 28-day test results is a
reasonable but not high standard of testing. However it does show an
improvement after 22/12/2011 from an initially less satisfactory fig-
ure. The earlier period showed several pair differences of 3 to 6 MPa
rather than a figure generally higher than 1.1. As is fairly normal,
such pairs showed one result lower than usual rather than one result
higher than usual, suggesting that the higher result of a pair is more
likely to be the correct value.
2. There is very good agreement between 7-day and 28-day results,
showing that change points can be reliably detected at the earlier age
(see especially 27/12/2011, 28/01/2012, and 17/2/2012).
3. There is some correlation between density and strength, and some
reverse correlation between temperature and density and strength,
showing the effect of temperature on water demand and therefore
strength.
4. It is interesting that temperature shows some correlation with 7-
to 28-day strength gain, suggesting that the 7-day specimens were
slightly less mature in the cooler weather (it is not known whether