Extract

A significant barrier to routine use in early design of Climate-Based Daylight Modelling (CBDM) is the relatively long time that it takes to run full annual simulations – potentially many hours for more complex models. Quick checks of design ideas are inconvenient, and this discourages examining large numbers of design options or carrying out routine sensitivity analysis. One way of reducing required simulation time is to not simulate all the daytime hours of the year. The downside is that this would seem to undermine the entire point of CBDM. This paper demonstrates that taking smaller samples of random days throughout the year can significantly reduce simulation time with acceptable levels of error in standard daylighting metrics. Simulating only 10 days/month introduces at most ~6% error in performance estimates – a level of error well within accepted tolerances. This can be further improved by using a selection process to filter “representative” samples of hours, enabling simulations using 5 days/month to provide estimates with only ~5% error.

PDF

Download