Comparison of GUM and Monte Carlo methods for the uncertainty estimation in hardness measurements

Abstract

Monte Carlo Simulation (MCS) and Expression of Uncertainty in Measurement (GUM) are the most common approaches for uncertainty estimation. In this work MCS and GUM were used to estimate the uncertainty of hardness measurements. It was observed that the resultant uncertainties obtained with the GUM and MCS without correlated inputs for Brinell hardness (HB) were ±0.69 HB, ±0.67 HB and for Vickers hardness (HV) were ±6.7 HV, ±6.5 HV, respectively. The estimated uncertainties with correlated inputs by GUM and MCS were ±0.6 HB, ±0.59 HB and ±6 HV, ±5.8 HV, respectively. GUM overestimate a little bit the MCS estimated uncertainty. This difference is due to the approximation used by the GUM in estimating the uncertainty of the calibration curve obtained by least squares regression. Also the correlations between inputs have significant effects on the estimated uncertainties. Thus the correlation between inputs decreases the contribution of these inputs in the budget uncertainty and hence decreases the resultant uncertainty by about 10%. It was observed that MCS has features to avoid the limitations of GUM. The result analysis showed that MCS has advantages over the traditional method (GUM) in the uncertainty estimation, especially that of complex systems of measurement. MCS is relatively simple to be implemented.

Keywords

• uncertainty
• normal distribution
• Monte Carlo
• guideline for uncertainty of measurement
• correlation