Assessing daylight performance in use: A comparison between long-term daylight measurements and simulations

Climate-Based Daylight Modelling (CBDM) methods have been validated against long-term measurements in laboratory settings and found to exhibit errors small enough to make such assessments useful for daylight performance prediction. However, real occupied spaces are affected by a higher number of uncertainties than laboratory or controlled conditions. This study aims at validating CBDM methods against measurements collected in an occupied classroom space, where a monitoring system based on High Dynamic Range Imaging was installed. Four vertical regions were identified on two of the room's walls, and mean illuminance was calculated for these regions at every time step, both from HDR images and from simulated results. Two simulation methods were evaluated: the 2-phase and the 4-component methods. Sun and sky conditions for the simulations were derived from simultaneous monitored irradiation measurements. Both simulation methods led to moderate over-prediction of HDR-derived results, when considering instantaneous illuminance means and when looking at long-term metrics (cumulative irradiation and Useful Daylight Illuminance). Wall regions exposed to more direct sky- and sunlight were characterised by smaller systematic errors (rMBE = 4%) but similar variance (r² = 0.83) than regions situated at the back of the room (rMBE = 17–34% and rMAE = 27–37%). Further studies are needed to identify and separate the sources of such errors.