[Radiance-general] Simulating light pipe performance

[Lars O. Grobe]

Hi Jia!

I have some unfinished work on this lying around for almost one year 
now, and this topic arises again and again on this mailing list. It is 
true that the standard backwards-raytracing approach of Radiance is not 
suitable for such systems. There are several ways to simulate these if 
you are aware of what is happening, and, more important even, what kind 
of results you need.

In general, if you do not need high angular resolution, the tools 
developed for patch-based sky models are working, and you can even 
calculate the BDSF now using the Radiance-toolchain as long as you are 
more interested in the integrated transmission then in high angular 
resolution. So - if you need to know how many hours a year you can reach 
a certain illuminance on a work plane, that should work. If you want to 
perform glare studies, I would question such results.

For high angular resolution, switching from backwards to forward 
raytracing is more efficient, and here the photon map extension is what 
we have available right now. Unfortunately, the development status is 
not really clear, and at least the latest available versions had some 
bugs that can heavily influence results depending on sky conditions / 
models.

One entirely different approach is to use some optical modeling tool to 
get the transmission distribution or, even easier, to get luminaire 
distributions for certain times of day / dates / weather conditions.

And then, you are free to circumvent the whole ambient calculation and 
try do keep everything in the direct calculation, which works fine with 
such setups. So for calculating horizontal irradiance, you would place a 
fisheye-type camera with a full hemisperical view on you point of 
interest looking up and integrate the radiance - as long as you 
reflector is completely specular w/o roughness.

Cheers, Lars.