# visual transmittance

I am doing a series of rtrace daylight calculation in a room with a Southern
facade for several glazings. All glazings are "standard" so that I simply
intend to use the glass modifier to simulate them. My original thinking was
that I carry out the simulation once for a glazing with a transmittance of
80% and then I scale the results for the other glazing types, i.e.

vis transmittance 80% -> transmissivity 87% -> the rtrace calculation yields
at point 1: 100 lx

therefore:

vis transmittance 40% -> transmissivity 44% -> at point 1 illuminance= 100
lx * 44/87 = 51lux

When I tested this assumption, rtrace yielded something like 58lux for the
room with the 40% glazing. I did not expect exact correspondance but an
error of 15% is pretty substantial.

Could anybody explain to me where my error lies?

Christoph

Christoph Tito Reinhart, Dr. Ing., Dipl.-Phys., M.Sc. tel: (613) 993-9703
Research Officer
fax: (613) 954-3733
m 24 Institute for Research in Construction e-mail:
[email protected]
http://www.nrc.ca/irc/ie/light/daysim.html
<http://www.nrc.ca/irc/ie/light/daysim.html>

Hi Christoph !

For some reasons this is an intersting topic for me, too, although I
don't know much about rtrace (never used it so far, I'm working only
with rpict/rview).

Under what circumstances did you perform the calculation ? Direct light
only, or ambient light also? Which were your parameter settings (direct
and ambient ?) Did you perform several consecutive runs with rtrace to
get a mean value ? For this, parameters have to be changed slightly to
"trick" the internal random generators, otherwise you simply would get
the same result again. From my answer so far you can imagine where my
thoughts are heading at: Within the algorithms, a lot of random
fluctuation, ray direction jittering etc. is performed. This leads to
noticeable deviations of results. (In a comparison experiment, I once
observed relative brightness differences between 0 and 8% throughout the
main part of an image, near one shadow boundary up to 18% and higher,
for two runs of the same scene).

Carsten

Hi Karsten,

Under what circumstances did you perform the calculation ?
Direct light

I actually carried out an annual daylight simulation using daylight
coefficients (http://www.nrc.ca/irc/ie/light/daysim.html) so I actually
looked at results for thousands of sky conditions.

"trick" the internal random generators, otherwise you simply would get
the same result again. From my answer so far you can imagine where my
thoughts are heading at: Within the algorithms, a lot of random
fluctuation, ray direction jittering etc. is performed. This leads to
noticeable deviations of results. (In a comparison experiment, I once
observed relative brightness differences between 0 and 8%
throughout the

This is actually an excellent idea Karsten. I will try this and report.

Christoph