Sky model blends for predicting internal illuminance: a comparison founded on the BRE-IDMP dataset

The BRE-IDMP validation dataset contains simultaneous measurements of sky luminance patterns and internal illuminances in two full-size office spaces. This benchmark dataset has been applied previously to test the illuminance predictions from a lighting simulation program under real sky conditions. Sky luminance patterns were mapped into the lighting simulation so that the absolute accuracy of the program could be evaluated without the uncertainties that are introduced when sky models are used. For this follow-on study, the BRE-IDMP dataset is now used to quantify the divergence between the sky model generated luminance patterns and the actually occurring conditions based on the resulting internal daylight illuminances. The internal illuminances were predicted using three 'narrow-range' models (CIE overcast, CIE clear and intermediate) and the Perez All-Weather model. Predictions from the narrow-range models were used to investigate formulations for sky model blends. The illuminance effect of arbitrary sky model blends is reproduced in a post-process of the illuminance predictions from the 'narrow-range' sky model types. The determination of an optimum sky model blend is described. The findings show that relatively simple blends of just two pure sky models (e.g. CIE overcast and intermediate) may be adequate for the prediction of time-varying illuminances founded on climatic test reference year data.