# Measurement uncertainty estimations: GUM method

Any types of measurement results should be presented with their associated uncertainty. In this post, a method to estimate measurement uncertainties based on Guide to the expression of uncertainty in measurement (GUM) is presented with examples.

Any types of measurement results should be presented with their associated uncertainty. In this post, a method to estimate measurement uncertainties based on Guide to the expression of uncertainty in measurement (GUM) is presented with examples.

GUM is the fundamental reference for uncertainty estimation that is applicable for all types of measurement, for examples, dimensional, geometrical, weight, volume, pressure, viscosity, electric current and temperature measurements.

The principle of GUM is that the uncertainty of a measurement should be calculated by propagating the uncertainty of contributing factors relevant to the measurement. The uncertainty propagation is derived from the model of the measurement.

Contributing factors to measurement uncertainties can be classified into five categories: instrument, operator, work piece, measurement procedure and environment (such as ambient temperature and pressure).

## Definition

Uncertainty is a non-negative parameter that represents the dispersion of a quantity associated or attributed to a measurand (VIM). Measurand is a property to be measured, for example, length, mass, volume and temperature.

An uncertainty value represents the amount of the lack of understanding toward a measurement process. In other words, uncertainty is the quantification of our doubt on a measurement.

The smaller the uncertainty value of a measurement, the larger our understanding or knowledge about the measurement.

It is worth to note that "uncertainty" is only associated to a measurement result and not to an instrument. Meanwhile, "error" is associated to an instrument.

## Eight general steps to estimate measurement uncertainty

There are various methods to estimate measurement uncertainty. In this post, GUM method is explained.

However, there are eight general steps that should be followed to determine measurement uncertainty regardless of what method is used. The general steps are:

1. Think ahead of time about measurement procedures to follow, what measuring instruments to use and safety aspects during measurement processes.
2. Perform the measurement according to a plan
3. Estimate contributing factors that are relevant to the measurement
4. Consider correlation effects among the contributing factors (commonly the correlations are neglected)
5. Calculate measurement results, including systematic error correction
6. Determine the measurement uncertainty (commonly within 95% confidence interval)
7. Express the measurement results with the confidence interval, determined from the uncertainty
8. Record the measurement results and the uncertainty for future improvement

## GUM fundamental formula

GUM method states that the uncertainty of a measurement result must be propagated from the uncertainty of each contributing factor that affects the measurement. To propagate the uncertainty, GUM formula is defined as:

Where $u(Y)$ is the total standard uncertainty value of a measurement $Y$ propagated from its contributing factors. $X_{i}$ is the $i-th$ component of the contributing factors. $u(X_{i},X_{j})$ are the correlation values between “$i$” and “$j$” components.

Commonly, uncertainty is presented within 95% confidence interval. This 95% confidence interval uncertainty is called expanded uncertainty $U$. $U$ is formulated as (assuming Gaussian or normal distribution):

### From where the formula is derived?

To derive the GUM formula, let us assume that a measurement $y$ is formulated as:

The $y$ above can be linearised by using Taylor expansion method as:

Where $C_{i}$ is the coefficients of Taylor expansion. By applying statistical variation property to the above linearised $y$, we get:

Where $var(cy)=C^{2}\mu ^{2}$, so that:

The above equation is the GUM formula as mentioned before. From this equation, GUM method requires a mathematical model that describes a measurement.

## Example 1: A Vernier Calliper measurement

In this example, the simplest measurement process is presented. The example is the length measurement of a gauge block by using a Vernier Calliper.

The measurement process is shown in figure 1. From figure 1, the length measurement is defined as the difference between two measuring scales read from the Vernier Calliper. Hence, the measurement model is formulated as:

Where $X_{1}$ and $X_{2}$ are the two scales read from the calliper. In this case, temperature effect is neglected and not included in the model.

From the measurement (figure 1), we can observe that $X_{1}=0 mm$ and $X_{2}=2.5 mm$. Hence, the gauge block’s length $Y=2.5 mm$.

By applying GUM method to estimate the uncertainty, we can derive:

## The disadvantages of GUM method

Although GUM method is the main reference to follow in determining measurement uncertainty for all types of measurements, this method has several drawbacks that are not always applicable in practice. Some drawbacks of GUM method are:

• Require the analytical or mathematical model of a measurement. In practice, most of real measurements are complex and their analytical measurement models are not available (if not difficult to derive). If the measurement models are not available, we cannot apply partial derivate, required by GUM method, to propagate the uncertainties of contributing factors of the measurement.
• Require function partial derivations of a measurement model. Although the mathematical model of a measurement is available, very often, the model cannot be partially derived.
• Require a complex modelling and calculation, even for simple measurements. The complex modelling and calculation can be a barrier for industrial application or industrial use.

## Conclusion

In this post, GUM method to estimate measurement uncertainties is presented. The GUM is the main reference to calculate measurement uncertainty for all types of measurement results, for example, length, volume, pressure, temperature, weight and humidity measurements.

Three real examples of uncertainty calculation by using GUM method are shown and explained. The examples show in detailed how measurement processes are modelled and how their measurement uncertainties are estimated.

Measurement uncertainty is one of the fundamental concepts in metrology and has important roles in industry.

We sell all the source files, EXE file, include and LIB files as well as documentation of ellipse fitting by using C/C++, Qt framework, Eigen and OpenCV libraries in this link.

We sell tutorials (containing PDF files, MATLAB scripts and CAD files) about 3D tolerance stack-up analysis based on statistical method (Monte-Carlo/MC Simulation).