I seem to be having a problem calculating daylight factors using Radiance. I ran the daylight analysis for different scenarios varying the number of indirect reflections. I had anticipated the daylight factor should go up with an increased number of indirect reflections, but the results doesn't agree with this (see attachment)
There is a limit for the number of -ab, after exceeding this limit, the change is not great. In some experiments, someone give the conclusion that the number is 5.
Attached image is a experiment about the relationship between DF and the number of -ab.
Anyway, I thought that as the number of indirect reflection/-ab increases, the DF also increases - which is what your image is showing - but the results I gained from Ecotect are suggesting the opposite. On the attached excel spreadsheet, the general pattern is that the average DF decreases as the number of indirect reflection/-ab increases (aside from Scenario 2, where the DF increased for 8 -ab - why is this by the way?).
I hope what I'm asking is clearer now.
I think there is some thing wrong in ecotect`s daylight factor supposition.
To generate an overcast sky with a total horizontal illuminance value of 100 Lux, ECOTECT sets the -b parameter to 0.558659 (100/179, in fact it is also wrong, It should be: 9*100 / 7*179*PI) in gensky command. I think the key lies in this value is too small for actual conditions. If I change the -b value to 22(general conditions:10000lux) in gensky, the DF results will raise following the higher -ab number(that is right).
So I suggest you calculate the illumination, and then divide illumination by 20769.42 (this is illumination corresponding to the defauld -b 47.486034 ECOTECT set in gensky, I don't know why is this value) to get the DF.
Edited by: yunpeng on May 8, 2009 9:28 AM Edited by: yunpeng on May 8, 2009 9:35 AM
If I divide the illumination by 20769.42, the results would be far too small a number (i.e. in the 1x10^-3 region). Should I be dividing by 46.486034 instead?
Attached image is the results I got.
I would like to bring this subject up from it's grave because I'm having the same problem. More bounces gives lower results which is just plain wrong in my world. I can't really trust my results at the moment because they differ a lot depending on how many bounces I choose.
I tested with the solution provided, calculate as illuminance and it seems to work fine, the more bounces the higher the illuminance. But when I divide by 20769 I get quite low DF results, a little bit too low in my opinion. Btw, the formula is in my version of Ecotect is "V * 1,05", not "V * 100" as in yunpengs response, therefore I divided with 207,69 instead.
Does someone have a solution to this? Is the formula correct?