Study of approaches to determining the required number of multiple observations

The necessity to determine the minimum number of observations when developing a measurement procedure in accredited test and calibration laboratories is discussed. The methods of evaluating the number of observations when evaluating the expanded measurement uncertainty using the GUM method, the Monte Carlo method, and based on the Law of the expanded uncertainty propagation are considered. In the first case, a nomogram is constructed that allows determining the minimum required number of multiple observations based on the given values of the expanded measurement uncertainty for a probability of 0.9545, the standard deviation of the scattering of the indications of a measuring instrument and the normally propagated standard instrumental uncertainty of type B. In the case of calculating the measurement uncertainty based on the Monte-Carlo method, a normal law and the Student’s law of propagation with given characteristics was modelled, and on its basis, for a probability of 0.95, a diagram to calculate the required number of observations when performing multiple measurements was constructed. The application of the Law of the expanded uncertainty propagation proved to be the most universal for calculating the required number of observations, since it made it possible to obtain approximating expressions for both probabilities and for the normal and uniform laws attributed to the components of type B.