Page 49 - 4150
P. 49
methods can be used as a basis to perfect this type of test (comparison
of a variance to a given value).
–Some analysis procedures recommend that the measurements be
repeated twice and to compare the deviation between these values
with a “critical difference” that has been determined by taking the
repeatability of the analysis procedure for its basis. The operator will
have to examine if the difference between the two results is greater
than the critical difference.
Graphic techniques also deserve attention. Studying a chart of
numerical values does not generally make it possible to detect the
abnormal values; graphs, on the other hand, are very rich in
information and something unusual (change of gradient, etc.) in a
curve very often reveals a faulty measurement. These techniques can
be used to verify the calibration of thermocouples by watching out for
the regularity of the calibration curves.
Third method: “monitoring standards” and statistical
supervision of the measurement processes
Statistical control of the measurement processes
The measurement process can be considered as a part of the
production process. For many years, manufacturing companies have
shown interest in the monitoring of the means of production. For the
first time, in 1924, WA Shewhart explained the principle of control
charts. Chiefly adapted to manufacturing processes, they are perfectly
suitable for the monitoring of instruments. The measurement process
is then considered as a production tool that does not make objects, but
which “manufactures” results of measurement.
Measurement process concept
The measurement process is a set comprising of:
– the measuring instrument (means);
– the method of measurement and the measurement procedure
(method);
– the environmental conditions (medium);
– the standards used (means);
– the operators (manpower).
The measurement process provides the results of the
measurements. Just as any manufacturing process, even one that is
47