Regression analysis of electric energy consumption of commercial buildings in Brazil

This paper describes a regression analysis performed over parameters related to two commercial building types and three weather files of Brazilian cities. A random sampling technique was applied to reduce the number of simulation runs needed in the parametric analysis. Equations to estimate the electric energy consumption as a function of building parameters were obtained. Despite the high coefficient of determination obtained for each equation, higher than 0.9760, the amount of data estimated with error higher than 20% is significant. The equations are presented as simplified tools to support the designer, as well the simulationist, to get a better understanding of the energy performance of the building. Such an analysis could be carried out before the incursion in a detailed simulation programme. Limitations about the equations are also discussed in the paper.