Hello everyone!
I was running a sDA sensitivity analysis when I came across some inconsistencies in the results. I ran the study a few times, changing the variables and simplyfing things that would make the model lighter, but some sort of inconsistency still kept showing up. Let me give an exemple:
In the image above, I tested the model using walls and ceilings with a reflectance of 0,7 and changed the roof transparency. The sDA value for a 0,7 transmitance roof doesn’t make sense. (full results: Design Explorer 2). Most of the inconsistencies tend to show up for higher reflectance values in the interior surfaces, which leads me to think that the problem might be related to the ambiente bounces (wild guess).
We ran the sensitivity analysis using Colibri, and even though I am pretty familiar with these TT toolbox componentes, I thought the problem might be in the simulation automation. But rerunning each individual simulation also brought up strange results (even some diferent ones).
I kind of ran out of ideias of what is going on with this model 
, and since the majority of the results (but not all) make sense, I don’t think there are major modeling errors.
Can anyone shed some light on what might be causing these strange results? I am uploading the grasshopper file (sorry, but everything is in portuguese!).
DaylightModel.gh (1.4 MB)
Thanks!
Vini