BSDF and TensorTree

Hi, thank you for developing this awesome tool. Quick question, does Honeybee Plus have the capacity to simulate light shelf or light louver using BSDF file (like in 3-phase or 5-phase calculation)? If so, do I need to create a geometry for the light shelf or louver or just assign it to a window? I’m not sure if those will be seen as honeybee surface, window surface or window group.

First up, simulating BSDF-based light shelves properly using Radiance, which is what HB[+] uses, is a fairly non-trivial task. So, I hope you have some experience with using Radiance.

You need to assign the BSDF file to a Honeybee surface. If this is just a single simulation, and not a parametric run of multiple lightshelf types, it is better to stick to Daylight Coefficients instead of going the Five Phase route. If you do use the Three/Five Phase method, and the light shelf stretches across the window (eg from exterior to interior), split them into separate internal and external surfaces. Within Radiance parlance, such shading systems can be categorized as out-of-plane shading devices.

Finally, it would be advisable to do a before and after check of illuminance/imaging values, without and with the lightshelf respectively, to ensure that the results make sense. It might be a good idea to place a grid of downward-pointing illuminance sensors on the ceiling to ensure that the light redirection is indeed taking place.

Dear @sarith ,

I am curious to know your thoughts on the following slide from one of Andy McNeels presentation. This makes me wonder that using a Daylight Coefficient based method in combination with a BSDF would not yield correct results?! Nevertheless, in my experience so far, I have got similar results with BSDFs when using DC and 3P methods…

Hi @MohammadHamza,

Andy is right. That’s why for the Daylight Ceofficients Method in HB+ we do (Daylight Coefficients) - (Daylight Coefficients with sun only) + Sunlight Coefficients (this is based on Daysim DDS and my own research where we started using actual sun positions instead of interpolating between spaced out suns)… The part with sunlight coefficients uses “light” instead of “glow” for the solar discs and so the calculations are deterministic. If you are using BSDFs with Daylight Coefficients (in-scene), you should be using Tensor Trees instead of Klems.


Thanks for the explanation Sarith. Very helpful! Could you please maybe elaborate a bit your point about using Tensor Trees? I have 2 questions about it:

  1. So we should use Tensor Trees even with HB+ based DC method ?
  2. What benefit would a Tensor Tree have over Klems, apart from higher resolution / patches? I am assuming a different ray-sampling ? Just a wild guess though…

I think the genBSDF tutorial, and the links contained in it, should give you clearer idea:

thanks sarith. If I assign the BSDF file to a honeybee surface, what should I do with the actual window glass? I’m pretty sure the BSDF file itself does not carry any glazing information. Do I create two honeybee surfaces, one for the glazing one for the BSDF file?

I looked at the output file from OpenStudio radiance measure with lightlouver. Looks like they modeled it as an interior shading device. Does this approach work in HB+? (though I couldn’t find a shading option for windowsurface in HB+). Thanks.

Hi inkmug,

If you are sure that the BSDF is only for the louver and does not include the glazing, then yes you would model them separately as two different honeybee surfaces i.e. one for the BSDF/louvers and one for the glazing.

1 Like