Abstract
In order to evaluate the energy performance of buildings, both in heating and in cooling periods, the simulation codes can be used. Moreover, in accordance with the technical Standard EN ISO 13790:2008, the simulation codes can be employed for refining the steady-state methods, and particularly the utilization factors estimations, in accordance with the procedure proposed. As the various simulation codes implement different capabilities and refer to different mathematical models and calculation assumptions, the necessary validation steps which are used for diagnostic purposes are not enough to ensure the agreement of the results over a wider range of configurations and conditions. The main dynamic simulation codes have been generally evaluated according to the Standard ANSI/ASHRAE 140:2007 (BESTEST). By this approach the user can choose a software among those successfully tested, giving acceptable deviations between the computed output and the reference values for a selected number of reference buildings defined in the Standard. However the number of those reference building configurations is limited and the considered features are not representative of the common building stock present for instance in Southern Europe. Moreover, as those configurations were selected for diagnostic purposes, they are expected to produce unacceptable biasing when considered with statistical approaches in order to improve the quasi steady state approaches as the one proposed in the technical standard EN ISO 13790:2008. In this work a procedure to identify the main causes of deviation has been developed and has been applied to two well-known dynamic simulation software: TRNSYS (version 16.1) and EnergyPlus (version 7). The approach is based on a factorial plan of comparison aimed to investigate the main variables related to the envelope of the building and its behavior: variations in geometry and boundary conditions (dimensions and orientation of the glazing, amount of dispersing surface) envelope characteristics (walls insulation and heat capacity, insulation and solar transmittance of glazings) internal gains. From the combination of the values of the above variables, more than 1600 different configurations have been obtained for two Italian climatic conditions, each of which providing monthly values for heating and cooling needs and for heating and cooling peak loads. Thanks to the large number of configurations, the monthly heating and cooling energy needs and peak loads have been analysed with inferential statistics, which allowed to evaluate the agreement between the outputs and to characterize the weight of the different variables in causing the deviations found.