# Which one to use? Daylight Coefficient or 3-Phase Method?

Hello all,

I am trying to calculate the annual hourly illuminance, radiation and luminance values for a room (6mt x 5mt) with a grid size of 0.5 x 0.5 mt using Radiance 5.3. on Honeybee[+] plugin.

The room is having a single pane clear glass window with an interior roller woven shade (25% open area). I created the BSDF file for this glazing system using LBNL Window7.8.

The window shades are controlled based on different manual blind control algorithms. I am trying to generate a blind control schedule based on different manual blind control algorithms and later calculating the climate-based annual daylighting metrics.

Can anyone suggest, Do I need to use the 3-Phase method to calculate various climate-based daylighting metrics or I can simply use the Daylight Coefficient method?

Any help will be greatly appreciated
Thanks and regards.

Tarun Verma
PhD Scholar
NIT Trichy, India

The daylight coefficient method, with its gross simplification of the overcast sky, is not compatible with most climate-based daylight metrics. Since you wish to model blinds as well, the 3-phase or 5-phase methods will be necessary. Iâ€™m not sure where Honeybee and Ladybug Tools are at currently, with respect to modeling individual window groups though. I know that is planned, but Iâ€™m not sure if itâ€™s implemented just yet (i.e. the ability to control different facades separately in a single model).

1 Like

I think you may be conflating â€śDaylight Coefficientâ€ť with â€śDaylight Factorâ€ť methods, Rob. I used to confuse them all the time, myself. The DF method uses an overcast sky and is, as you say, quite crude. The DC method developed by Tregenza is sometimes called a 2-phase method in relation to the 3-phase method, which adds a stage to the calculation.

In any case, for roller shades, a modified 3-phase method is probably the most efficient annual simulation. Itâ€™s a bit tricky, since controlling roller shades affects the percentage of window coverage, rather than changing out the BTDF as happens in a standard 3-phase calculation. Your 3 phases will look something like this:

1. Compute the interior-to-window-slice view matrices, where you divide the window into horizontal slices at heights corresponding to roller shade control deployments.

2. Compute per-slice daylight (exterior) matrices, or if there are no nearby structures, a single daylight matrix per elevation.

3. Apply dctimestep to compose appropriate combinations of view matrices and BTDF of clear glazing or glazing with roller shades depending on control position to get partial contributions, and finally with sky vectors or matrix from gendaymtx.

The final stage is to add things together using either rmtxop or Excel as is your preference.

The first step is quick because you only need one rfluxmtx run to get all the window slice matrices. The second step is potentially expensive if you need to keep your slices separate due to nearby obstructing exterior geometry, but if you donâ€™t have that, approximating every elevation (compass direction) in a single matrix should be quite quick.

I hope thatâ€™s not too confusing. This assumes a certain familiarity with Sarithâ€™s tutorial, etc.

Cheers,
-Greg

1 Like

OMG Greg is right, of course. I read â€ścoefficientâ€ť, but my brain saw â€śfactorâ€ť. Gregâ€™s response and of course Sarithâ€™s referenced tutorial are the paths forward here.

@Greg_Ward @Rob_Guglielmetti
Thanks for such a nice explanation.