Evaluation of Uncertainty in Analytical Measurement
The following contribution deals with the contents of two lessons of 45 minutes each. Basic knowledge of the evaluation of measurement uncertainty in analytical chemistry is imparted. After the two lessons the audience should know the new concept accordin
- PDF / 2,753,351 Bytes
- 21 Pages / 439 x 666 pts Page_size
- 21 Downloads / 178 Views
Abstract The following contribution deals with the contents of two lessons of 45 minutes each. Basic knowledge of the evaluation of measurement uncertainty in analytical chemistry is imparted. After the two lessons the audience should know the new concept according to GUM/EURACHEM [1,2] in its fundamentals and be able to calculate the measurement uncertainty of a simple analytical procedure. The content is structured into three parts: The first part starts with a short description of random and systematic influences. Based on this information the old but still applied concept of uncertainty evaluation is presented. That concept distinguishes between the measurement uncertainties "typeA = random errors", and "type B = systematic errors" . Then the teacher leads over to the new concept according to GUM/EURACHEM. It is indicated that the new concept allows the transformation of nonstatistically evaluated uncertainties into standard uncertainties. Thus the calculation of the combined standard uncertainty is possible. In the second paragraph it is shown how to calculate the combined standard uncertainty. In addition the required theoretical knowledge is interposed. The common procedure is demonstrated by means of the corresponding flow chart of the EURACHEM Guide. Each step is illustrated separately. For the identification and the analysis of the uncertainty sources the cause and effect diagram from Ellison and Barwick is initiated. The triangular- and rectangular distribution is presented without deriving the formulas of the variance. The comments about the calculation of the combined standard uncertainty refer only to independent variables. They show the law of propagation of uncertainty and derive two simple mathematical rules for addition/subtraction and multiplication/division. The case of correlated variables is only mentioned without treating. Finally, it is indicated how the results and their measurement uncertainties are put on record correctly. The lectures require only basic mathematical know-how from the students. It is a qualified instrument for the introduction of the measurement uncertainty at universities and technical colleges. However, it has to be taken into consideration that only an introduction can be given in a double lesson and exercises of at least eight hours should follow.
Slide 1 When carrying out a series of measurements, due to randomly fluctuating influence quantities (temperature, humidity, etc.) the individual values are not identical which means that the measured results are scattered around a mean value. This run to run variation of results is very often a normal distribution or an almost normal distribution in analytical chemistry. B. Neidhart et al. (eds.), Quality in Chemical Measurements © Springer-Verlag Berlin Heidelberg 2001
44
M. Rosslein, B. Wampfler
Slide 2 Let us assume that n independent observations qk were carried out under the same experimental conditions but which were subjected to randomly fluctuating influence quantities. How can the values of the measurand q be
Data Loading...
