Evaluation of uncertainties of measurement

Abstract

Each measurement is not perfect and therefore cannot give the "true" value of the quantity to be measured, but only a more or less accurate approximate value, called estimate (of the value). Even if the measurement is repeated under seemingly identical conditions, the measuring instrument with sufficiently high resolution will display values usually differing from one another. The imperfection, or, positively considered, the quality of a measurement is expressed quantitatively by a numerical value, the uncertainty of measurement. The result of a measurement is the more reliable, the smaller the uncertainty. If the specified measurement uncertainty limits are not met during tests and calibrations, the test object will not pass the acceptance test. This chapter describes the definitions and the concept of how the uncertainty of a measurement can be determined. Key words are model function of the measurement, Type A and Type B evaluation methods, standard uncertainties, expanded uncertainty, uncertainty budget and statement of uncertainties. Several examples are given in Appendix B.