Specular Reflections with the Daylight Coefficient Method

Hi,
I'm trying to use the daylighting coefficient method (using rtcontrib) to
get annual simulation results for a room equipped with a mirrored
daylighting system (like a light shelf, for example). Is this possible? I
know that by using the -dr option, Radiance will create virtual light
sources to account for specular reflections of direct light from the sun.
However, since the daylight coefficients are calculated using a blank sky
having a uniform luminance distribution, there is no sun for the -dr option
to find. I believe this would make the sky patch coefficients come out lower
than they should be because the specular direct reflection of the sun would
not be included, but my understanding of how the program works may not be
correct.

Can the daylight coefficient method be made to work with specular
reflections in Radiance, or is this not possible given the way the program
is implemented?

Thanks,
Kevin Thuot

Hi Kevin,

The daylight coefficient method usually puts the sun's energy into 1 to 4 sky patches, so the rtcontrib method will work, though it won't give very accurate shadows. The total energy should still be correct if you do it right.

Another approach is to create a set of suns at different positions in the sky corresponding to the expected trajectory throughout the year, and interpolate these results in a separate rtcontrib run. I have done it both ways, and the latter technique tends to produce slightly more accurate shadows. The -dr option will work properly if you do this to calculate reflected beam radiation.

I regret that I don't have a script to provide, but my 2005 presentation covers this in overview:

  http://www.radiance-online.org/radiance-workshop4/cd/website/PDF/Ward_rtcontrib.pdf

Cheers,
-Greg

···

From: Kevin Thuot <[email protected]>
Date: June 1, 2010 6:50:05 AM PDT

Hi,
I'm trying to use the daylighting coefficient method (using rtcontrib) to get annual simulation results for a room equipped with a mirrored daylighting system (like a light shelf, for example). Is this possible? I know that by using the -dr option, Radiance will create virtual light sources to account for specular reflections of direct light from the sun. However, since the daylight coefficients are calculated using a blank sky having a uniform luminance distribution, there is no sun for the -dr option to find. I believe this would make the sky patch coefficients come out lower than they should be because the specular direct reflection of the sun would not be included, but my understanding of how the program works may not be correct.

Can the daylight coefficient method be made to work with specular reflections in Radiance, or is this not possible given the way the program is implemented?

Thanks,
Kevin Thuot

Greg, all:

So what is the protocol for determining the number of patches the sun's energy is distributed across? In Axel Jacob's (excellent) tutorial on rtcontrib three patches are used to represent the sun regardless of whether the Tregenza or the finer Reinhart patch density/distributions are used, but you mention one to four patches being used, Greg. How or when would this number change? Does genskyvec actually make a choice?

- Rob

···

________________________________________
From: [email protected] [[email protected]] On Behalf Of Greg Ward [[email protected]]
Sent: Tuesday, June 01, 2010 10:33 AM
To: Radiance general discussion
Subject: Re: [Radiance-general] Specular Reflections with the Daylight Coefficient Method

Hi Kevin,

The daylight coefficient method usually puts the sun's energy into 1
to 4 sky patches, so the rtcontrib method will work, though it won't
give very accurate shadows. The total energy should still be correct
if you do it right.

Another approach is to create a set of suns at different positions in
the sky corresponding to the expected trajectory throughout the year,
and interpolate these results in a separate rtcontrib run. I have
done it both ways, and the latter technique tends to produce slightly
more accurate shadows. The -dr option will work properly if you do
this to calculate reflected beam radiation.

I regret that I don't have a script to provide, but my 2005
presentation covers this in overview:

        http://www.radiance-online.org/radiance-workshop4/cd/website/PDF/Ward_rtcontrib.pdf

Cheers,
-Greg

From: Kevin Thuot <[email protected]>
Date: June 1, 2010 6:50:05 AM PDT

Hi,
I'm trying to use the daylighting coefficient method (using
rtcontrib) to get annual simulation results for a room equipped with
a mirrored daylighting system (like a light shelf, for example). Is
this possible? I know that by using the -dr option, Radiance will
create virtual light sources to account for specular reflections of
direct light from the sun. However, since the daylight coefficients
are calculated using a blank sky having a uniform luminance
distribution, there is no sun for the -dr option to find. I believe
this would make the sky patch coefficients come out lower than they
should be because the specular direct reflection of the sun would
not be included, but my understanding of how the program works may
not be correct.

Can the daylight coefficient method be made to work with specular
reflections in Radiance, or is this not possible given the way the
program is implemented?

Thanks,
Kevin Thuot

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

Hi Rob,

Genskyvec always picks the three closest patches, regardless of the sky resolution. It would be difficult to use a sky resolution high enough that one patch would be the better choice. I only mentioned the 1-4 range because that's what people have done, not as a recommendation per se.

The important thing is to distribute energy according to which patches are closest, so if the sun really is in the middle of a particular patch, very little energy would be given to neighboring patches. That "very little" should probably be zero in some cases, but I don't think it ever quite is in genskyvec with the weighting function I use.

-Greg

···

From: "Guglielmetti, Robert" <[email protected]>
Date: June 2, 2010 7:37:00 AM PDT

Greg, all:

So what is the protocol for determining the number of patches the sun's energy is distributed across? In Axel Jacob's (excellent) tutorial on rtcontrib three patches are used to represent the sun regardless of whether the Tregenza or the finer Reinhart patch density/distributions are used, but you mention one to four patches being used, Greg. How or when would this number change? Does genskyvec actually make a choice?

- Rob

Thanks dude!

- Rob

···

On Jun 2, 2010, at 9:44 AM, Greg Ward wrote:

Hi Rob,

Genskyvec always picks the three closest patches, regardless of the sky resolution. It would be difficult to use a sky resolution high enough that one patch would be the better choice. I only mentioned the 1-4 range because that's what people have done, not as a recommendation per se.

The important thing is to distribute energy according to which patches are closest, so if the sun really is in the middle of a particular patch, very little energy would be given to neighboring patches. That "very little" should probably be zero in some cases, but I don't think it ever quite is in genskyvec with the weighting function I use.

-Greg

From: "Guglielmetti, Robert" <[email protected]>
Date: June 2, 2010 7:37:00 AM PDT

Greg, all:

So what is the protocol for determining the number of patches the sun's energy is distributed across? In Axel Jacob's (excellent) tutorial on rtcontrib three patches are used to represent the sun regardless of whether the Tregenza or the finer Reinhart patch density/distributions are used, but you mention one to four patches being used, Greg. How or when would this number change? Does genskyvec actually make a choice?

- Rob

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

Genskyvec always picks the three closest patches, regardless of

> the sky resolution. It would be difficult to use a sky
> resolution high enough that one patch would be the better choice.
> I only mentioned the 1-4 range because that's what people have
> done, not as a recommendation per se.

Including yourself, Greg. Your rtcontutor directory that was in Radiance for a while used the four nearest patches. I guess your reasoning to reduce this to just three was something like this:

1 - too crude
2 - still only one-dimensional
3 - just perfect. The real sun will always be in the triangle formed by exactly three patch centres.
4 - Overkill

So in theory, if you we remove the patch intensity of the sky, we can determine the exact position of the sun if the intensity is known, or vice versa by triangulation. This is dictated by Greg's Uncertainty Principle. Think of it as sub-pixel interpolation in image processing.

Cheers

Axel