Coverage interval estimation of the measurement of Gross Heat of Combustion of fuel by bomb calorimetry: Comparison of ISO GUM and adaptive Monte Carlo method

The Guide to Uncertainty of Measurement (GUM) approach and the adaptive Monte Carlo method (MCM) provide two alternative approaches for the propagation stage of the uncertainty estimation. These two approaches are implemented and compared concerning the 95% coverage interval estimation of the measurement of Gross Heat of Combustion (GHC) of an automotive diesel fuel by bomb calorimetry. The GUM approach, which assumes either a Gaussian or a t-distribution for the output quantity (GHC) gives half width intervals of 0.28 MJ kg−1 (Gaussian distribution) or 0.29 MJ kg−1 (t-distribution). On the other hand, MCM, which provides a reliable probability density function of GHC through numerical approximation, gives a half width interval of 0.32 MJ kg−1. Thus, the GUM approach underestimates the calculated uncertainties and coverage intervals by up to 7–12%. The main reasons of these differences are the approximations and the assumptions introduced by GUM approach, i.e. assumption for the GHC probability distribution and overestimation of effective degrees of freedom by the Welch-Satterwaite formula. Moreover, the estimation and the use of sensitivity coefficients and uncertainty budget within GUM and MCM approaches are examined.