BSDF vs VLT values from Window 7.6 in Honeybee+ Point in Time simulations

Hello everyone,

I am trying to run simulations for LEED v4 Daylight credit option 2, Point in time Illuminance (Since I haven´t been able to run ASE properly for option1 because all results are 0).

The problem is that I exported BSDF files of my window assemblies from LBNL Window 7.6 and when I run the simulation using the BSDF files and compare it with a simulation with a radiance glass material using VLT values from the same Window simulation, results are almost half the percentage of complying area different. Using VLT around 47% vs BSDF 81%. Is this to be expected?, or should I be checking for errors on the BSDF files?.

I´m attaching snapshots of the simulation results and the BSDF files.
Any help would be greatly appreciated.

Thank you
ASFG_Monolitico.xml (138.2 KB)
ASFG_Bib2.xml (138.2 KB)

I am curious, why are you using BSDFs if you can use simple glass material instead? The Klems BSDFs exported through Window 7.x have a known issue related to the handling of peak values in direct sunlight. You can read more about it here.

1 Like

Hello Sarith,

I only used them because I was Trying to simulate ASE and sDA accounting for shades as described in IES LM 83 12, but gave up because of an error in Daylight coefficient and 5 phase simulations where all of my results were 0 for ASE. Then I went on simulating point in time for option 2 of the LEED daylight credit and kept using BSDF´s. I wasn´t aware of such a problem with Window.

Thank you for the clarification. I´ll go through the document you kindly sent.

Do you know if there are any news on the management of full biulding daylight simulations for sDA and ASE. Maybe I should open a new topic for this.

Thank you again

Actually the operation of shades in LM 83 12 doesnt explicitly call for the use of BSDFs. It does, however, ask for assigning of at least one Window group where the shades can be operated based on the amount of direct sun received (this is still work-in-progress in HB[+]). Can you share the project folder that was written out by HB[+] where the values for ASE are zero? The size of the folder prior to the running the simulation should be a few MB or less, so you can probably share it here directly.

I might be able to get back to you within a day.

Sarith (748.4 KB)

I´m attaching the folder before run and a snapshot of the results after.

Thank you very much in advance.


HI Arturo,

I think I might have figured out what is going on. Can you replace all the BSDF materials with standard glass for all the three window groups in your scene, rerun the simulation and check if you still get zero values?
If you are looking for glass-material based equivalents for the BSDFs used in your simulation, the IGDB contained within the LBNL Optics software might have them.
Let’s see if this works. I can explain the rationale in detail later.


Hi Sarith,

Yes, that did the trick!. I´ll look in the IGDB. I just used standard glass as you suggested for now. It´d be fantastic if you can elaborate a bit further on the subject whenever you have time.

Thank you very much. I really appreciate it.

Best Arturo

Hi Arturo,

You were using a Quarter Klems-basis BSDF for the glazing. That could be one of the issues. There is no practical reason to use BSDFs to represent purely specular glazing unless there is some multi-phase calculation invovled. Glass material typically gets stored as a diagonal matrix in the BSDF format (first image below).

The calculation that you are performing is represented by the first row of images, where the rays are directly traced to the sun and sky. BSDFs for purely specular glazing, if used, are required for Three Phase Method and beyond (row 2 and row 3 in the image below).

Detailed info, and links to further reading material, can be found in Chapter 2 of the tutorial from which the above images was taken link.


Hi Sarith,

Thank you very much for the detailed explanation. It makes it a lot clearer. I’ll take the time to dive into the tutorial to get the grip out of it


Hi @sarith,

i experience the same problems: i’d like to simulate sDA+ASE for LEED and have external ventian blinds. therefore i tried to run a 5-phase simulation, that works for sDA but doesn’t have direct illuminance. i replaced the bdsf with the clear.xml from the honeybee samplefiles, then with the #ID03 from the IGDB and an BSDF created with WINDOW7.7. With none of them i get direct illuminance.

Can you please give me a hint on how to fix this (first without the blinds) - or is there a easier way to proof ASE with external blinds? I guess there is no problem when they are closed, but for the times when they are opened, 1000lux can be exceeded with a normal control strategy opening them below 200 w/m² external radiation.

Next step, if direct illuminance works with glass only would be to add the venitan blind to the BSDF as Klems or Tensor-Tree - but i guess that’s another thread :wink:

Thanks in advance!

Hi @HeikoWoehrle, I wish I could suggest a direct solution for this. What follows is mostly an explanation of why things aren’t working :neutral_face: !

With regards to Glass surfaces:
I believe, the absence of direct illuminance when BSDFs are used in-scene, is an “issue” inherent to Radiance. I remember this coming up a while ago, when my advisor at Penn State was trying to implement some workflow with glass-based surfaces in 5-Phase. Additionally, there is also an issue related to Klems-based surfaces being approximations of glass materials. This is something that Greg Ward, the creator of Radiance, had pointed out when I was working on that tutorial for LBNL. In my opinion, if you are just working with glass, it’s better not go the multi-phase route at all.

With regards to Blinds and Complex Geometry:
The Five-Phase Method currently implemented in Honeybee[+] does not support Tensor Tree BSDFs (right @mostapha ?) . So, your best bet might be to just use the standard annual simulation with really high number of ambient bounces and ambient divisions so that you can somewhat brute-force the MonteCarlo algorithm in Radiance to push in adequate amount of rays into your space.

All of the above explanation is specific to working with HB[+]. If you are familiar with command-line Radiance, then of course you can get the simulation to work.