How Accurate are Your Experimental Data? A More Accessible GUM-Based Methodology for Uncertainty Evaluation

Abstract A methodology for evaluating experimental uncertainty is presented. Based on the Guide to the Expression of Uncertainty in Measurement (GUM) in conjunction with a sensitivity analysis, this method readily applies to systems of various degrees of complexity. It consists of three steps: (1) to estimate each uncertainty contribution of the system based on GUM; (2) to determine the sensitivity of the calculated results to variations in each of the input measurands in turn, replacing the partial derivatives of the GUM with a purely numerical approach; and (3) to calculate the overall uncertainty using the error propagation principle. Furthermore, the calculated sensitivity coefficients enable a critical evaluation of the investigated system, allowing the detection of possible targeted improvements. For this reason, the presented method is called “the sensitivity analysis method.” This is applied to three case studies with increasing complexity: a mass calibration procedure, a volume calibration procedure, and a gravimetric densimeter characterized by a multi-parameter nonlinear measuring model. When possible, the results are compared to the GUM uncertainty framework or values available in the literature.