Words and concepts in Metrology

A decision rule is a predefined rule describing how measurement uncertainty is to be taken into account when making a statement / declaration of conformity. It most commonly concerns whether or not measurement uncertainty is considered in the conformity assessment.

According to ISO/IEC 17025, which applies to accredited laboratories, the decision rule shall be clearly defined, meaning that it shall be selected, communicated, and agreed upon between the laboratory and the customer. Several types of decision rules exist. The most common are:

• Simple acceptance / shared risk: Results may be accepted when the measured value lies within the tolerance limits, even if the measurement uncertainty extends beyond the tolerance limits. The risk is shared between the customer and the laboratory. In practice, this means that measurement uncertainty is not taken into account in the conformity assessment.
• Binary / non-binary decision rules using guard bands: Results may be accepted only if the measurement uncertainty also lies within the tolerance limits, thereby reducing the risk of making incorrect conformity decisions.

A calibration is a comparison between a reference (measurement standard) and the measuring device to be calibrated. During a calibration, we determine how much the device’s indication differs from the reference’s quantity value. This gives us the device's measurement error.

For a calibration to be reliable and to ensure traceability, we must use a measurement standard that has been calibrated by a National Laboratory or by an accredited calibration laboratory.

The calibration protocol tells you how to handle your measuring device and how to correct the device’s indication to compensate for the measurement error.  The protocol also shows the measurement errors and the measurement uncertainties.

Calibration does not mean that the instrument is adjusted, although many instruments can also be adjusted to display the correct value.

A value obtained during the calibration of a measuring instrument that compensates for the measurement error or indication error of the instrument. The correction is equal to the measurement error with the opposite sign. If the measurement error of a scale is −2 kg, the measured value must be corrected by +2 kg in order to obtain the “true” value. Note that, since the measurement error is never known exactly, the correction is associated with uncertainty.

Metrology is the science of measurement and its application. Metrology concerns:

• The definition of measurement units
• Realization of measurement units
• Traceability

There are several subfields within metrology. Scientific/fundamental metrology concerns the definition and the realization of measurement units and metrological traceability at the very highest level. Applied/technical metrology concerns methods and techniques for measuring maintaining traceability in society through chains of calibrations. Legal metrology deals with legal demands and requirements with the aim of protecting health, the environment, and consumers. In Sweden, Swedac is responsible for legal metrology.

Measurement error is the measured value minus the "true" value. If the scale shows 80 kg but the true value is 78 kg, the measurement error is 80-78=2 kg. If the scale shows 78 kg but the true value is 80 kg, the measurement error is 78-80=-2 kg. Since we can never know for sure what the true value is due to measurement uncertainty, we cannot know the exact measurement error.

A device or equipment used to measure one or more physical quantities, such as a tape measure to measure length or a scale to measure mass. All measuring devices are subject to measurement uncertainty. For the measurement device to be reliable, it should be calibrated regularly (considering its area of use).

Measurement uncertainty is the interval around the measured value within which the true value is likely to be found.

Many factors affect measurements: the person measuring, temperature, humidity, what we measure, which measuring device we use etc. Regardless of how good the measuring devices we use are or how ideal the conditions are, it is not possible to know that the measured quantity value is the true value. The measurement uncertainty indicates how much the indication of the measuring device can differ from the true value with a certain probability (usually 95 percent).

A measurement result consists of a measurement value, the measurement uncertainty, and a unit (for example meters or kilograms). The measurement result 5.00 m +- 0.05 m means that the true measurement value, with a certain probability, is found between 4.95 m and 5.05 m.

With very good conditions, methods and measuring devices, we can reach a very low measurement uncertainty. The measurement uncertainty at the National Laboratories is in many cases on the order of parts per million or billion. For most applications, this level of measurement uncertainty is not necessary. The needs of the application should always determine the level of measurement uncertainty required. It is important to plan for the desired level of measurement uncertainty when planning for calibration, as the measurement uncertainty increases with each calibration step.

In quality-assured measurements, it is important not only to look at the indicated value from the measurement device, but also to take the measurement uncertainty into account. This is called a measurement result.

A measurement result consists of a measured value, the measurement uncertainty, and a unit, such as meter or kilogram. The measurement result 5.00 m +- 0.05 m means that the true value, with a certain probability (usually 95 percent), is found between 4.95 m and 5.05 m.

A measurement standard is a reference used in the calibration of measurement devices. Standards are fundamental for maintaining traceability and must always have a lower measurement uncertainty than the measurement device to be calibrated. The standard must in turn be calibrated against another standard with even lower measurement uncertainty. In this way, the traceability chain is maintained all the way up to the realization of the unit. The measurement standard can be physical objects such as a weight or an object with a certain length.

Several different terms are used concerning measurement standards, sometimes with different or overlapping meanings depending on the context:

• Primary standard/national standard. A measurement standard that is calibrated directly against the realization of the unit. In Sweden, RISE as the National Metrology Institute maintains the national standards at its National Laboratories.
• Secondary standard. A measurement standard that is calibrated against a primary standard.
• Working standard/calibration standard. A measuring standard that is used in the daily work of calibration. At the National Laboratories, this means either a secondary standard or a measurement standard that is calibrated against a secondary standard.

In metrology realizing means to bring the measurement unit from its definition into the real world, either by a physical object or based on a physical phenomenon such as a frequency-stabilized laser for the meter. The realization of the measurement unit can then be used as the basis for the traceability chain. The realizations of the units are quality assured through international comparison measurements.

Until 2019, the International Prototype Kilogram, IPK, was both the definition and the realization of the kilogram. The traceability chain for the kilogram went via the Swedish national kilogram, which was flown down to Paris at regular intervals to be calibrated. One of the problems with using a physical object such as the IPK as the definition for the measurement unit is that the object can change due of external effects such as air pollution, but still weigh exactly 1 kg by definition.

Since 2019, the entire international measurement system, SI, is based on defining constants of nature. This means that the definitions of the measurement units are independent of the realizations. For each unit, there are so-called mises en pratique, approved methods, for how the units can be realized. Since the definitions are no longer linked to the realizations, new and better mises en pratique can be developed as technology advances.

By comparing directly with the realization of the unit, we can produce primary standards with the lowest measurement uncertainty. Although it would be possible to do all calibrations directly against the realizations of the units in theory, it would hardly be possible in practice. Most areas of use do not need the level of measurement uncertainty that a direct comparison with the realization can provide. Therefore, secondary standards or working standards that are calibrated against the primary standards are used in daily work at the National Laboratories.

Traceability means that a measurement result can be traced back to the definition of the unit through an unbroken chain of comparisons against suitable standards with specified measurement uncertainties. A measurement without traceability is not reliable.

The traceability chain usually goes via a National Laboratory, either in Sweden or in another country, and is maintained via calibrations: The tape measure at the building site is calibrated against a standard, which in turn is calibrated against a standard at an accredited laboratory, which in turn is calibrated against a working standard at a national measurement site, which in turn is calibrated against a primary standard, which is calibrated against the realization of the unit. The realization is then quality assured via international comparisons. Each calibration step down from the realization of the unit increases the measurement uncertainty.

As a result of the traceability chain, we can expect that a meter is a meter throughout the world, which is a prerequisite for international trade, research, and manufacturing.

As the National Metrology Institute of Sweden RISE is responsible for maintaining and providing traceability in Sweden.

Upon request by the customer, a statement of conformity may be included in a calibration or test certificate. In such cases, an assessment is made as to whether the measurement results of the instrument meet the specified requirements. The requirements are defined by the customer and may, for example, be based on the instrument specifications, standards, regulations, or internal quality requirements.

A statement of conformity usually involves an assessment of whether the measurement result falls within predefined tolerance limits or tolerance requirements. This assessment may result in one of the following outcomes:

• The measurement result (measured value and measurement uncertainty) lies entirely within the tolerance limits.
• The measured value lies within the tolerance limits, but the measurement uncertainty lies partially outside.
• The measured value lies outside the tolerance limits, but the measurement uncertainty lies partially within.
• The measurement result (measured value and measurement uncertainty) lies outside the tolerance limits.

Example

A thermometer is being calibrated. According to the requirements, it shall indicate the correct temperature within ±1.0 °C.

• The reference value from the standard is 100.0 °C
• The thermometer indicates 100.6 °C
• This results in an error of +0.6 °C, which is within the tolerance limit
• The calculated measurement uncertainty is 0.5 °C
• The error of +0.6 °C with a measurement uncertainty of 0.5 °C means that the uncertainty interval extends partially outside the tolerance limits

The applicable limits and requirements for a statement of conformity are determined by the customer and shall be clearly stated in the certificate. The certificate shall also specify which type of decision rule has been applied in the calibration.

Statements / declarations of conformity are governed by ISO/IEC 17025 and ILAC-G8:09/2019, “Guidelines on Decision Rules and Statements of Conformity.”