GUMkehyksen

GUMkehyksen, or the GUM framework, refers to the approach described in the Guide to the Expression of Uncertainty in Measurement (GUM). It provides a structured, internationally recognized method for evaluating and expressing the uncertainties attached to measurement results. The aim is to enable meaningful comparison of results and to establish traceability to the SI units.

At the core of the GUM framework is a measurement model y = f(x1, x2, ..., xn), where y is the result, the xi are input quantities that influence the measurement, and c may include fixed constants. Each input quantity has an associated standard uncertainty. Uncertainties are classified as Type A (evaluated from statistical analysis of repeated observations) or Type B (evaluated from other information such as prior data, calibration reports, or expert judgment). If input quantities are correlated, covariances are incorporated.

The standard uncertainty of the result, u(y), is obtained by propagating the input uncertainties through the model. This can be done with the first-order Taylor expansion using sensitivity coefficients (∂f/∂xi) or, for nonlinear or non-Gaussian cases, by a Monte Carlo method often referred to as the GUM Monte Carlo approach. The combined standard uncertainty is then combined with a coverage factor k to yield the expanded uncertainty U = k·u(y), which corresponds to a stated level of confidence (commonly about 95% when k = 2).

Results are reported with the measured value, the standard uncertainty, and the expanded uncertainty, together with information on the uncertainty sources, the model used, and the measurement's traceability. The GUM framework is widely used in laboratories, calibration, and regulatory contexts to ensure transparent, comparable measurement results.