[Radiance-general] Exterior light levels and daylight factors

[Rob Guglielmetti]

Hi Caroline,

First off, there is a Desktop-Radiance Yahoo forum, and you should 
definitely ask around on there too, especially for your falsecolor scale 
questions.  I'm not familiar with the Windows version(s) of the various 
Radiance binaries.

Giulio gave you some good leads there regarding sky simulations.  This 
kind of question pops up on here regularly, and indeed I went through 
the exact same experience as you.  I'd read about how accurate Radiance 
could be (read that again, the semantics are important), then did my 
first daylight simulation and was disappointed.  Lots of words have been 
put down on paper about the difference between real, measured skies and 
simulated ones.  Papers with lots of Greek symbols fill the servers at 
universities & research facilities worldwide.  Greg Ward can cut through 
the haze as well as anyone regarding this, but being all smart and 
"mathy", his explanations still confused me for a while.  Eventually I 
got it:

Nature is way too complicated.  To think that one or even a thousand 
formulas could mathematically describe the luminance distribution of the 
nearly infinite number of skies is pure folly.

What has been proven by Dr. Mardaljevic is that Radiance can simulate 
lit environments with a high degree of accuracy.  What he had was an 
incredibly fine dataset of real sky luminance and building interior 
illuminance.  When this data was fed to Radiance, he got very close to 
the same answers.  But most of us do *not* have an incredibly fine 
dataset of real sky luminance and building interior illuminance, we have 
a lat/long, maybe some sky turbidity info, and the woefully inadequate 
suite of sky conditions descriptors "clear, intermediate and overcast". 
  Whoopie. So, it's no wonder that our initial forays into daylight 
simulation with Radiance tend to be confidence shakers.

The truth is Radiance can be as accurate as you want, but most of us are 
starting with input that is inherently inaccurate.  It's a good starting 
point, but you simply can't expect your results to match up to reality 
with any kind of high precision when using these basic CIE skies.  Don't 
get me wrong, the CIE skies can still be used to make valid design 
decisions, and certainly comparative studies of building form and 
material will be quite valid.  But it's when people want to take a light 
meter and have the LCD display show the same number that Radiance spat 
out that the disappointment creeps in.  For pure numeric accuracy, which 
Dr. M proved is possible, you need more measured info about your site. 
My experience with Radiance is that it's very very good, certainly the 
best available tool out there for daylighting analysis.  But I don't 
expect my numbers to be ±5%, or even ±10% sometimes.  But I believe ±20% 
is a fair goal.  Our faith in the numbers is based on experience and 
intuition and we have to work hard to get this point across to our 
clients all the time.  In my view, it's been worth the effort.

Hopefully the Greek symbols types will weigh in on this too; it's an 
interesting topic.

----

      Rob Guglielmetti

e. rpg at rumblestrip.org
w. www.rumblestrip.org


And, while I'm thinking of it, from Ian Ashdown's excellent "Thinking 
Photometrically" Lightfair course notes 
(http://www.helios32.com/Thinking%20Photometrically%20II.pdf):

"One major problem is that the calculations are necessarily based on the 
various CIE or IESNA sky models, which predict illuminances based on 
average sky conditions. It is not uncommon for instantaneous measured 
illuminances to be more than twice or less than half of the mean 
illuminances predicted by clear and overcast sky models; the situation
for partly cloudy skies is even worse.

"As for daylighting calculations, it is likely that Jongewaard (1993) is 
correct – the results are only as accurate as the accuracy of the input 
data. Done with care, it should be possible to obtain ±20 percent 
accuracy in the photometric predictions. However, this requires detailed 
knowledge and accurate modeling of both the indoor and outdoor 
environments. If this cannot be done, it may be advisable to walk softly 
and carry a calibrated photometer.