I was checking the SunFile.rad created by using the binned-sun option (+s2) for GenCumulativeSky.exe and found the solid angle specified for each of the sun disk is 0.0533.
void light solar
3 2.822e+008 2.822e+008 2.822e+008
solar source sun
4 -0.034904 -0.991526 0.125133 0.053300
Should the solid angle for the sun disk visible to human eye be around 0.5 degree?
(according to Dr John Mardaljevic's thesis, p.60
Will the solid angle for the sun disk affect rtrace calculation?
Very good observation Ji. I will forward your comment to Darren Robinson who wrote GenCumulativeSky. We have done some validation of the program comparing it monthly measured facade irradiance and to annual Daysim runs and the results were satisfying. I agree that the opening angle of the sun is usually considered to be about half a degree. For example, if you type gensky 6 12 12 on your DOS command line you will get a clear sky in Berkeley with a radiance of 7.18e+006 and a source opening angle of 0.5. The 0.0533 degree used by GenCumulativeSky therefore seems to be wrong. The reason why I suspect is does not matter is that in GenCumulativeSky the direct normal irradiance, taken from the EnergyPlus climate file, is properly divided by the assumed solid angle of the solar disk, i.e. 0.0533. You can see this because the 2.822e+008 is so much larger than the 7.18e+006.
The reason why it does not matter for the calculation which of the two opening angles you are using is that the sun as a light source is so small that Radiance will only find it by using a direct check, i.e. it explicitly verifies whether a point in a scene can "see" a light source by sending a ray directly toward the center of each light source. For larger light sources the geometry of the light source is divided into smaller segments but for a source as small as the sun this is not going to happen.
I agree that this bug in GenCumulativeSky should be remedied. But, I think the program still provides valid results.
Hi, Christoph, Thank you very much for your detailed clarification!
I just found the paper by Diego and you (BuildingSimulation2011) if that's the validation test you mentioned. Will digest the paper and learn!
Ibarra, D., & Reinhart, C. (2011). SOLAR AVAILABILITY: A COMPARISON STUDY OF SIX IRRADIATION DISTRIBUTION METHODS. Paper presented at the Building Simulation 2011.
- Cheers, Ji
As Christoph promised, he passed your query on to me.
When myself and Andy Stone were developing and initially playing around with GenCumulativeSky we checked a number of strategies for modelling the direct solar contribution, including creating a cumulative sky representing all (>4000) suns. Under these conditions, and with the standard solar disk solid angle, there can be considerable overlap and thus prediction errors. Thus the solution to reduce the solar disk solid angle is, perhaps ironically, more accurate. It may however be more appropriate in the case of binned suns (modelled sources representing several suns in their approximate mean position and with their cumulative radiance) to use the standard solar disk solid angle (and thus reduce the source radiance), as in this case the binning can be adjusted to avoid source overlap...but it probably doesn't really matter much.
Hope this helps a little.
Hi, Darren, thank you very much for the clarification!
I understand now that:
1) the size of the sun disk don't affect the sampling for the direct contribution because the sun use the "light" type of material and only the centroid of the sun disk matters
2) sun disk in large solid angle may affect the indirect calculation negatively because a large sun disk may "trick" the ray shooting to the sky (in "glow" type of material) and intercepting with the sun disk to return a zero value, resulting in underprediction of the contribution of the sky (p.61-62 of Dr Mardaljevic's dissertation).
- Cheers, Ji