Abstract
This work proposes a method to manage the complexity of variables involved in building simulation studies and to identify groups of simplified building models suitable to have statistically significant results. The method is described by means of an applicative example, whose aim is the definition of a set of configurations appropriate for the analysis of TRNSYS and EnergyPlus discrepancies in monthly energy needs, hourly peak loads and time of occurrences of hourly peak loads – for both heating and cooling. The proposed procedure for the definition of a reference set of building configurations moves on from the selection of a set of candidate variables describing the building envelope characteristics, paying attention to implications of each choice and to cross-correlations among variables. This is obtained by means of a screening analysis with a simple statistical index (Spearman’s rank correlation coefficient). Two sample sizes are considered in order to evaluate the effects on the selection procedure. For each of the six considered outputs, the most significant group of configuration variables is identified, and the differences among those groups are described.