Instrumentation & Measurement Magazine 23-2 - 83

end customer, one wonders about the currently widespread
strategy of minimizing "customer risk" wherever possible...
This strategy has caused an explosion of "supplier risks" that
are all in fact borne by the end customer. Can we not do better than this? It is our hypothesis that Smart Metrology has a
role to play in much-needed ''risk optimization." In the current era, where we are finally conscious that resources are not
infinite, and where competition is truly international, it is perhaps time to modify certain practices and start to target the
"strict minimum."

The Futility of Measurement
Uncertainties...
In line with the practices of "ISO 9001," the intention of the
NF EN ISO 14253-1 standard [3] was to deal with the problem
of how to take measurement uncertainties into account, a matter that has repercussions for Quality in manufacturing: the
question at issue being how can we be sure that characteristics
or entities comply with production specifications?
The Standard suggests applying the strategy summarized
in Fig. 1. According to this Standard, it is simply a matter of
subtracting measurement uncertainty U from the tolerance
limits in order to define a "Conformity Zone": any entity with
a measured value in that zone can be considered to be "compliant." In addition, the Standard defines two areas of doubt
around the Tinf and Tsup limits, with an interval of width ± U,

Fig. 1. Taking measurement uncertainties into account in accordance with the
ISO 14253-1 Standard.
April 2020

and recommends using a measurement process with a tighter
uncertainty to re-measure any entity falling within that zone
in order to make a decision. Any value measured outside these
zones will obviously be declared "non-compliant."
In practice, conformity decision-making is more likely to
resemble Fig. 2. As measurement uncertainty is generally not
even considered on a daily basis, without, moreover, any apparent disasters resulting, it seems fairly safe to assume that
the tolerances used do correspond to the "conformity zone"
described in NF EN ISO 14253-1. This is not, in fact, surprising.
The tolerances in question have been determined by producing compliant functional entities in an iterative way during
the "perfecting" phase using measured values that are, in fact,
the controlling factor. Bearing this in mind, it seems to us that
to take into account measurement uncertainty would offer a
wonderful opportunity to work towards a more rational expression of functional need. And for anyone able to get beyond
blind belief in accurate measurements, a significant potential for improving industrial productivity opens up. Broader
specifications, in fact adherence to the principle of the "strict
minimum," makes it possible to envision not only lower production costs but also better management of consumption,
as much in terms of energy as raw materials, something very
much needed if we are to have any hope of a sustainable future.
The "Goal of Smart Tolerancing" white paper [4], which
can be downloaded for free, sets out the practices that govern
the allocation of tolerance as well as suggestions for changes.
Currently, the most widespread practice is the so-called
"Worst Case" scenario. This is costly as it takes no account of
the random factors that affect all manufacturing. It assumes,
for example, that the worst of all the pen tops produced will
be paired with the worst of all the pens, whereas it would be
more efficient to ensure an acceptable level of functional "pen/
top" pairs. The "worst case" scenario demands very tight
specifications, and this it is what makes it possible to ignore
measurement uncertainties. The uncertainties are covered to a
great extent by the safety margins inherent in the method, although rarely in a controlled manner.
The potential for improvement is real, therefore, if we can
only rethink our common conceptions. Other seasoned approaches suggest optimization of tolerance allocation as a
means of optimizing production costs. This, however, requires
a dose of mental reprogramming, and this is but one, and not
the least, of the problems encountered. In fact, the white paper [4] concludes that: These methods of "Inertial" or "Process
Tolerancing" have proven their worth. All that is needed is to deploy

Fig. 2. Common practice.

IEEE Instrumentation & Measurement Magazine 83

Instrumentation & Measurement Magazine 23-2

Table of Contents for the Digital Edition of Instrumentation & Measurement Magazine 23-2