Comparison of Direct Radiation Split Algorithms for Energy Simulation of Buildings

Direct normal radiation (DNI) has great importance for both energy building simulations and solar energy systems. The data is seldom available from measurements but usually is recovered from global radiation data using split algorithms. The present paper analyses the performance of 33 different split radiation models and the error which arises when applied to building energy simulations using generated hourly weather files. The split models have been applied to an observed dataset composed by 525888 points, which comprises global and diffuse radiation on the horizontal plane, related to -year measurements, starting from 2001 with 10-minute time steps. The generated weather files have been employed as input for energy simulations with EnergyPlus on a building generated using DesignBuilder software. We investigated the impact of the weather files in building energy simulation highlighting the performances of four models selected among the 33 models by means of statistical indicators, during different periods of the dataset, since its amplitude allowed us to decompose and analyse 10 different years.