I'm doing a daylight analysis for leed 8.1 but using the daylight factor method.  I have come to the point where my radiance parameters have a major impact on the resulting area above 2% df.  For the purpose of df over an analysis grid what radiance parameters should I use to get an accurate result in a reasonable amount of time.  

Thanks in advance for your help.


Views: 3789

Reply to This

Replies to This Discussion

The impact you are talking about, how does it articulate itself?

Trouble has usually got to do with the -ab, -aa and -ar settings combo... the more bounces (ab) the more, well, bounces of light Radiance traces until energy is exhausted. -aa and -ar, combined with scene size, gives you the minimum geometry features that are resolved accurately- see the Daysim manual for a detailed explanation and formula: http://daysim.com/pub/Daysim3.0.Tutorial.pdf

Quoting from page 26:

ambient accuracy (aa=0.1) and ambient resolution (ar=300): The combination of these
two parameters with the maximum scene dimension provides a measure of how fine
the luminance distribution in a scene is calculated. According to page 385 in
Rendering with Radiance, the combination of aa=0.1, ar=300 and a maximum scene
dimension of 100m yields a minimum spatial resolution for cached irradiances of:

(max imum scene dimension x ambient accuracy) / ambient resolution

Hope this helps some,


Excellent, this helps quite a bit,

the only thing I'm still a little fuzzy on is the maximum scene dimension, does this reference to the overall length of my building or does it have to do with the size of render being generated.

Does it hold true that if I have very small facade elements and would like the simulation resolution to be very small that I could increase the ambient resolution so reducing the simulation resolution below 3cm, or equally reduce the ambient accuracy from 0.1 to 0.05 and achieve again a simulation resolution below 3 cm.  In which case is there any benefit in doing one over the other, either in terms of calculation time or accuracy.

And as for ambient divisions, this regulates the smoothness of the transition from light to dark and so need not be very high in a large space with a .4m analysis grid?

Thanks again


scene size means the maximum extent of your scene.. i.e. the diagonal of the ground plane / landform your building sits on.

I don't think that there is much benefit in deciding which parameter to modify, but I generally increase the -ar value instead of further lowering the -aa.. no scientific tests done on this, but it usually felt to do the job faster when doing it that way. I'd leave the -ad value as preset or ramp it up to ~ 2000, but it usually makes no big difference.

Kind regards,


Hi Andrew,

In addition to the settings mentioned above, I would also add that the first couple simulations that I perform on a scene are usually quick with an ab 2 (ambient bounce). If your are getting too much daylight by daylight factor, point in time illuminance (LEED 8.1 simulation), or a climate based method with an ab 2 then their is too much light and I would move on to adjusting the design. But on the other hand, where a borderline amount of light is getting in, increasing the ab will tend to bring more in. Although somewhere after an ab 5 in most architectural models is a point of diminishing returns (where the result doesn't change) with exponentially increasing simulation times. I generally use an minimum ab of 5 for most architectural simulations for the "final" analysis documentation.


Thanks Jeff,

The ambient bounces will effect how deep the light tends to make it into the building as well. When I first started out I was using the default setting and was very confused why light wasn't penetrating very deep into my building and consequently I was no where near the 75% for 8.1.  I think its a very important note for novices on how easily your results can change due to a parameter that is not at first glance easily understood.  

I have since taken Max's advice and the simulation returns exactly what I was hoping it would.

Thanks for your help.


Reply to Discussion


© 2019   Created by jeff niemasz.   Powered by

Badges  |  Report an Issue  |  Terms of Service