Propagation of uncertainties in heat energy calculations in tests for heat- and flame-resistant materials was evaluated. The impact on arc thermal performance value (ATPV) results was also studied and uncertainties calculation methods were compared using the Guide to the Expression of Uncertainty in Measurement (GUM) and the Matrix form covariance propagation. For heat energy calculations, the accuracy of temperature measuring systems and calorimeter copper disk dimensions and mass as required in test method standards for protective clothing against the thermal hazards of an electric arc were considered. For the requested accuracy of temperature meters of 0.75 °C, relative uncertainties in heat energy are in the range of 16% in case of high measured temperature increases, i.e. increases in the range of 500 °C. Use of global mean value for dimensions and mass of the copper disks of the calorimeters can significantly increase errors. Resulting uncertainties have a relevant influence on the determination of the arc thermal performance value via the specified logistic regression algorithm due to the presence of independent variable errors (i.e. errors of incident energy values calculated from high temperature data measured by calorimeters), which are not taken in account in the actual procedures. Also errors on the values of the dependent variable, obtained from the assessment of data (i.e. data calculated from low temperature data measured by calorimeters) by the comparative procedure using the Stoll criterion for skin second degree burn, should be included in uncertainties calculations, as they can also have an important influence on ATPV test results.
Read full abstract