Model-based measurement uncertainty evaluation,with applications in testing

Basic concepts and implementations of the model-based approach to uncertainty evaluation are discussed. One implementation is that of the law of propagation of uncertainty with the use of the central limit theorem as recommended in the Guide to the expression of uncertainty in measurement (GUM). Another is the propagation of distributions, the subject of a supplemental guide to the GUM, which is at an advanced stage of development. It falls in the category of other analytical and numerical methods indicated in the GUM. Two testing applications are used to illustrate the principles: tennis-ball rebound and tensile strength.

Estimating sampling and analysis uncertainties to assess the fitness for purpose of a water quality monitoring network

In order to verify that the protocols used for water quality monitoring of surface waters within the Long-term Environmental Research Monitoring and Testing System (OPE), located in the north-eastern part of France in relation with a geological disposal for radioactive waste project, are fit for purpose, a validation study was conducted following the methodology described in the Eurachem/citac and Nordtest guidance documents on uncertainty arising from sampling. As one of the objectives of the OPE water monitoring programme was to investigate the spatial and temporal variability of water quality, quality requirements were set to having a measurement variance, including sampling and analytical contributions, less than 20 % of the total variance to minimise the impact of measurement over the observed environmental variability. The replicate method was then selected in order to estimate the measurement uncertainty, including the sampling contribution, as well as the spatial and temporal variability of water quality of surface waters. To minimise costs, a single-split level was selected. Analytical uncertainties were assessed from inter-laboratory data and/or internal quality control data from the last 2–5 years. Finally, ANOVA was applied to the data sets after elimination of outliers. Results showed that for pH, electrical conductivity, turbidity and nitrate, the sampling uncertainty was negligible, whereas for other parameters such as dissolved oxygen, total suspend solids, total organic carbon, nitrite and phosphate, the sampling contribution to the measurement uncertainty was largely significant. For all parameters except calcium, the sampling and analytical protocols were considered fit for purpose.