Abstract
Building energy balance is strictly connected with incident solar irradiation, whose reliable estimation on envelope surfaces is essential for the evaluation of building solar gains. While diffuse and beam solar components on tilted surfaces are required, meteorological stations usually measure only global radiation on horizontal plane. In the literature, several diffuse irradiance and tilted surface irradiance models are reported, nevertheless, none is suitable for any locality. In this work, we assess the extent to which the choice of solar radiation models affects the predicted energy performance of a set of simplified reference buildings. By means of a full factorial plan, a set of 72 simplified residential buildings is defined by changing the main building features. A full combination of 22 horizontal diffuse irradiance models coupled with 12 irradiance models for tilted surfaces is implemented and used as a preprocessor of solar data used in TRNSYS. The results highlight that the variability of solar radiation in input induces different levels of uncertainty in energy models predictions. Besides, the dispersion of simulation outcomes shows interactions with the building features linked to the transmission of solar radiation inside the building, leading to different uncertainty propagations in BES.