Comparative assessment of ISO/IEC Guide 98-3 and Monte Carlo methods for temperature sensor calibration: A simulation-based study

Abstract

The evaluation of measurement uncertainty is a task in metrology, and remains challenging when measurement models show nonlinearity, or when input quantities follow non-Gaussian probability distributions. Although the ISO/IEC Guide 98-3 (GUM) is widely applied in calibration practice, studies have highlighted situations in which its underlying assumptions may no longer be satisfied. In response to these developments, this paper presents a simulation-based comparative analysis of the GUM framework and the Monte Carlo method recommended in GUM Supplement 1, with a specific focus on temperature sensor calibration. We analysed a set of calibration scenarios, covering linear and nonlinear models, symmetric and asymmetric input distributions, and limited sample sizes. The comparison addresses uncertainty estimates, coverage behaviour, and computational aspects, allowing the practical consequences of each method to be assessed under controlled conditions. The results indicate that while the GUM approach remains suitable for near-linear models with approximately Gaussian inputs, Monte Carlo simulation provides a more faithful representation of the output distribution when these conditions are not met. For readers and practitioners, this study offers methodological insight and practical guidance for selecting an appropriate uncertainty evaluation approach in contemporary temperature sensor calibration, consistent with recent advances reported in the metrological literature.