Daylight factor - "too good" results?


I would be really grateful for help in understanding the results I obtained from the Honeybee Legacy Daylight Factor analysis.

I am checking daylight for a building permit in a large housing project. According to a previous Vertical Sky Component analysis, we have a problematic 1st floor of the building (VSC under 15-20%) where the façade is shaded by the proximity of other buildings.

  1. Test
  • Radiance Parameters: -ab 7 -ad 2048 -as 512 -ar 256 -aa 0.1 (default values for the rest).
  • Results: DF values of 0.5, 0.4, and 0.8 in the tested rooms.

It looks very dark but that is pretty much what I expected.
I have then found a study someone else did (in another project) and made a second try with their parameters:

  1. Test
  • Radiance Parameters: -ab 8 -ad 2048 -as 1024 -ar 2048 -aa 0.1 -dt 0.05 -dc 1 -dj 0.65 -ds 0.01 -dr 0 -dp 32 -st 0.01 -lw 0.001 -lr 8 -lw 0.001
  • Results: DF values of 0.9, 0.9, and 1.4 in the chosen rooms.

To my surprise, that looks even “too good” - 0.9, 0.9, and 1.4. The light seemed to go really deep into the room (walls reflections?), which may not look realistic given how deep it is and that there is a balcony over the openings.

Of course, it is great for the project if these results are true, but I would like somebody experienced to confirm if these are results I can trust, given that it will be submitted to the city.
Also, that calculation was extremely time-consuming, so perhaps some of these parameters are unnecessarily inflated?

Thank you!

The radiance parameter in your second test is way more detailed than the default parameter, hence the better result - I wonder how much more time it took to run the second test?

maybe it would be worth conducting a convergence test as suggested below:

1 Like

Thank your for fast reply!
The first test takes 11m (for 3 rooms) while the second 3,3h so there is a huge difference.

The test is a great suggestion, thank you for the link!

I think that is tooooo long, and probably not worth it

Another thing you can do though is to have a smaller _grid_size in LB Generate Point Grid (right now it seems it’s around 0.4-0.5m?) just to have a higher resolution of the analysis grid

Grid size is not affecting the results per room since every test point has lower values in Test1.
But now I have found - I think - the answer to my question so I will post it here in case somebody has a similar issue.

  1. Test
    I was playing with -ar parameter and I went down from to 16. As result I got this “beam” of light that looks similar to the suspicious “reflection” I got in Test1. So now I am pretty sure the high values I got in Test1 were partly caused by artifacts.

  1. Test
    I did the default “high detail level” setting and I got this which look much more natural. Still takes a lot of time though, so the next steps will be to decrease parameters one by one and further testing. The link you gave me before was super helpful so thanks again for it :slight_smile:

In the same time, I have noticed something weird. While changes of -ar affect both time and visual results, in the command shells I always get “-ar 300”. Did anyone have same observation?