8. Overview of measurement uncertainty estimation approaches

Brief summary: In this section an overview is given about the main types of approaches that can be used for estimation of measurement uncertainty.

Overview of the approaches for estimating measurement uncertainty
http://www.uttv.ee/naita?id=17704

https://www.youtube.com/watch?v=Ed8S8KN1GlU

The two main approaches that are addressed in this course – the modelling (ISO GUM) and the within-lab validation (Nordtest) approach differ by the extent of detail, by the way the random effects are taken into account and by the status of the obtained uncertainty. Table 8.1 outlines the differences. See also Table 11.1 in section 11.

Table 8.1 Some key differences between the modelling and within-lab validation approach of measurement uncertainty estimation.

Aspect

Modelling (ISO GUM)

Within-lab validation (Nordtest)

Taking random effects into account

Typically random effects are addressed at within-day (short-term) level (see Section 6), using repeatabilities. All effects that within a day are systematic, are taken into account separately. So, usually quite many uncertainty sources are quantified separately.

Random effects are addressed at long-term level, with the aim of incorporating as many effects as possible into the random component of uncertainty, usually quantified as within-lab reproducibility sRW (termed as u(RW) in section 10). This is an important advantage of this approach as explained in section 10.2.

Extent of information needed (or obtained) on different uncertainty sources

As a result of quantifying a number of uncertainty sources separately, the extent of information on different uncertainty sources is usually quite high.

As a result of quantifying a large number of uncertainty sources as within-lab reproducibility, the level of information needed/obtained on different uncertainty sources is low.

Status of the obtained uncertainty

The obtained uncertainty applies to a specific analysis result obtained on a specific day.

The obtained uncertainty is an average uncertainty estimate of the analysis procedure across several a time period of several months.

 

Although ISO GUM (2) is the de facto standard document of measurement uncertainty estimation, especially in physics, the within-lab validation (Nordtest) approach is also gaining popularity. For example, the Nordest approach forms the foundation of the ISO 11352 standard Water quality — Estimation of measurement uncertainty based on validation and quality control data.[1]ISO 11352, Water quality — Estimation of measurement uncertainty based on validation and quality control data, 2012. 

back forward