This study aims to assess the uncertainty of the NBR 15575-1 (ABNT, 2013) simulation method regarding the results of the thermal performance rating levels of a dwelling. The assessment was performed by analysing the thermal performance of one dwelling using the EnergyPlus computer programme, for the climate of Florianópolis-SC, Brazil. Essential variables, which are not considered in the standard's simulation method, were analysed for the definition of a typical day, such as the date of the typical summer- and typical winter-day, wind speed and direction, solar irradiation calculation algorithm and sky clearness. These variables were used in a statistical experiment with factorial combination to determine the uncertainty of the results and the sensitivity of the variables for rating thermal performance levels. Simulations were carried out for nine different building envelope models. The results showed that the worst performance levels, both in summer and winter, were obtained for those models without ground contact. The most influential variable for summer and winter analyses, in all cases, was sky clearness. Based on these results, theconclusion was made that the variables ignored by NBR 15575-1 have considerable influence on the thermal performance rating levels of buildings, resulting in the lack of accuracy of the final results.
NBR 15575; Computer simulation; Uncertainty analysis; Sensitivity analysis
