...designer's reply

In this case the window-to-window wall area ratio required was 35% but the external wall is quite large due to the large floor to ceiling height and a 35% figure will produce different average daylight factors with other arrangements. A 5% average daylight factor figure was selected in order to provide classrooms that could be entirely naturally lit for over 80% of the occupied period. This does require a very large area of glazing but the end result is a considerable reduction in lighting energy. The calculated additional heat loss through this additional glazing was not a major concern as simulations showed that it was significantly offset by the additional useful solar gain benefits.

Glare does not appear to be a problem in the classrooms although direct sunlight must be blocked by the internal blinds. School staff have been asked to keep in mind that blinds block daylight and should be returned to their open position when the glare issue has passed.

  The use of an average daylight factor as a means of daylight design is a rough rule of thumb method as it takes no account of daylight distribution or glare. Specifying a minimum daylight factor takes some account of distribution but still has disadvantages, as does the percentage of window to external wall calculation. All of these methods will produce different real results with different geometries.

  A calculation that combines average daylight factor and uniformity figures would seem to be the most appropriate rule of thumb method of specifying targets although we are not aware of any system that uses this approach.

  I propose that the figure that we, as designer's of energy efficient lighting schemes, are ultimately interested in is kWh/m²/y. It is perhaps time for lighting design software to address this. The software model would include the natural daylight design, the artificial lighting design, switching and blind definitions. The model would then simulate a year's typical lighting data controlling blinds based on incident radiation and switching lighting based on the simulated response of internal photocells. Some of the lighting design packages currently available could be modified with limited effort to offer this feature. Perhaps it is time for daylight design to enter the same level of sophistication as thermal modelling already has?