Metrology in the real world

To make the right decisions you have to learn to doubt your measurements.

In everyday life we frequently read values from dials and use devices to take measurements. This kind of action has become so commonplace that few of us realise an essential fact: what we are reading has very little chance of matching reality.

For various reasons that will be discussed later, any measure (the act of assigning a value to an observation) is affected by errors. An experimental measurement can therefore only be correctly interpreted if it is associated with an estimate of the probable error named ‘measurement uncertainty’. For example, ‘this tube has a diameter of 10 cm plus or minus 0.5 mm’. In the scope of domestic activities, the impact of this uncertainty can be considered as low. In an industrial environment however, the situation is very different.

Mastering the consequences of this uncertainty is one of the main objectives of metrology - the science of measures that pertains to the theoretical and practical aspects of measurement tasks (the act of measuring). In terms of quality management, metrology involves managing the preparation and execution of measurement tasks in order to guarantee the measurement results, their traceability and their uncertainty. In particular, metrology makes it possible to choose the method and the measuring device appropriate for the required level of accuracy.

Dimitri Vaissière of Endress+Hauser specialises in metrology and measurement uncertainty and is responsible for drafting and editing standards of service for all the Group’s locations. He says one of the reasons metrology is underutilised in industry, is device users don’t doubt their readings.

“Whenever you perform a measurement, you must be aware of what you are observing,” he says. “It is, in fact, ‘reality contaminated by your inability to measure it’. The inability to measure reality is reflected in the measurement uncertainty. This information is crucial for allowing us to make the right decisions. Because what we read is just an indication; it’s not proof.”

Figure 1: The measuring device is not an isolated object; it must be considered in the context of the measurement and, more generally, in the context of the system being measured.

To illustrate this point, let’s take an intentionally simplified example: while setting up a production process, an operator performs an initial test and obtains a reading of 10 from his devices. After making a few changes to the control settings, he repeats the test and this time obtains a reading of 12. Going by his readings, the operator will likely come to the conclusion that the changes he made to the settings are what caused the value to change. But in fact, nothing could be less certain.

If you want to prove the effect of the changes, you first have to know what value the operator would have measured if he hadn’t changed anything - perhaps he would have obtained a 12 even without making changes. In this case, the difference between the two measured values would stem not from the system being measured (the production process), but rather from the system used to perform the measurement (the measurement process).

In order to know whether the two values obtained reveal a change in the behaviour of the production process, the operator must know the range of the variation (uncertainty) around 10 and the extent of the variation around 12. If this variation is equal to ±0.1, one can reasonably suspect that the changes in settings had a significant impact on the production process.

By contrast, if the uncertainty equals ±3, the variation caused by the measurement system is such that we cannot determine whether the changes made had an impact on the process (in one direction or the other).

The same is true for the technician who performs a calibration. Without knowing the calibration uncertainty, it is impossible to interpret the errors attributed to the device undergoing calibration. Without this uncertainty, any corrections or adjustments made will most likely prove inappropriate or even detrimental.

Finally, let’s take the case of a technician who, when performing an annual calibration, regularly finds values that are borderline or even out of tolerance. Quite logically, he will make the decision to calibrate the device every six months instead. While this decision is sure to cost the company more, its efficacy will in no way be certain. This is because the device must always be considered within the context of the measurement or calibration procedure.

Fostering a ‘culture of doubt’ means accepting that our observations are not incontrovertible and thereby adopting a skeptical approach with regard to the measurement procedure and verification methods, with a view to optimising them.

Five key factors of measurement uncertainty

Each factor making up the measurement process has an effect on the measurement results. There are five of these factors: the measurand, the tool, the operator, the method and the environment.

The metrologist’s job is to help users to quantify, and then reduce, the impact of these factors on the uncertainty. And, through an awareness of these five factors, anyone involved in taking measurements (operators or maintenance teams) will be able to interpret the measurements with maximum accuracy.

The measurand: the quantity to be measured

If a diameter (for example) corresponds to an unequivocal mathematical definition, it proves extremely difficult to measure in practice. This is because we are not capable of producing a perfect circle.

But what is the diameter of an object that is not perfectly round? A caliper will encounter a multitude of possible bipoints, each of which will give a different value for the diameter reading, especially since the imperfect form is significant. The measurand is frequently a major factor in the measurement uncertainty. The tool: the measuring device Error in the tool or measuring device is generally the first thing that comes to mind. The device is, in fact, merely one link in the measurement procedure. Sometimes it is the predominant link, but quite often, just the opposite is true. In the diameter measurement example, the caliper generates an uncertainty stemming from the manufacturing defect, not from the tool itself, but in other cases it will be the limits of resolution of the device, which introduces error. The equipment and measurement instruments used are therefore an important element of NATA (National Association of Testing Authorities, Australia) accreditation (see below), and using NATA accredited instruments provides assurance of their accuracy by them meeting NATA accreditation standards.

Figure 2: Deadweight testers are the basic primary standard used for accurate pressure measurement and are widely used for calibration of pressure measurement devices.

The operator: the human factor

In many cases, the operator of the measurement process constitutes an important source of uncertainty: for the diameter measuring example, the way the device is held, the force exerted on the jaws and even the ability to read a graduation are specific to each operator and lead to varying results.

The method: the way in which the measurement is performed

In our diameter-measuring example, the measurement error is impacted by various methodological factors, such as the choice of the number of test points. In order for the results to be usable, they must be obtained under reproducible conditions. Standard procedures for a measurement method are therefore highly important.

The environment: the surroundings in which the measurement occurs

All measurements take place in a specific environment with characteristics that can influence the result: the temperature, pressure, humidity, but also the quality of the lighting, the presence of dust or smoke, etc.

Metrology in the real world

Many manufacturers rely on outside specialists for calibration services. An organisation accredited by NATA is formally recognised as competent to perform specific types of testing, inspection calibration and other related activities (see below). Specialists that have developed standards in compliance with international norms and best practices in statistics and metrology can provide not just calibration, but also all of the materials and methods necessary to ensure the quality of the calibration service - whether it’s carried out on site or at a laboratory.

Figure 3: Endress+Hauser's flow measurement rig at its Cernay calibration plant.

A properly developed operational solution can result in major savings in terms of both the production process and device management - which is how metrology can be extremely useful in the real world.

Usefulness for the process

Vaissière says it’s important for metrologists to engage in dialog with the users of measuring devices.

“The second explanation for the relative underutilisation of metrology in industry is that it is generally approached from the perspective of statistics (perceived as boring) or from the perspective of quality (just as boring), but rarely from the perspective of its usefulness for the process. And yet the real question concerning measurement uncertainty should be: What impact does it have? At what threshold can operators consider that their readings are significant and that intervention in the production process is necessary? And how can the measurement be improved overall? Metrology truly plays a ‘support’ function in that it must reach out to users, listen to them in order to understand their needs and help them make the most of their measurement results.”

The calibration data is a starting point. Besides furnishing proof of the conformity of the devices, it provides a basis for optimising the calibration frequency and, more generally, the measurement itself.

NATA accreditation

NATA is a founding member of, and participates in, the ILAC (International Laboratory Accreditation Cooperation), the APLAC (Asia Pacific Laboratory Accreditation Cooperation) and other international bodies.

From a commercial perspective, accreditation demonstrates technical competence and is required where accuracy is crucial, for example in custody transfer applications or for tax purposes. In addition, many industries specify NATA accreditation for suppliers of testing services and manufacturing organisations also use NATA accreditation to ensure their own in-house laboratories are performing accurate testing.