Annual simulations + BSDF

Hi Everyone,

I¹m trying to figure out how to incorporate clearshade from panelite (
http://www.e-panelite.com/downloads/Panelite_IGU_Oct08.pdf ) into an annual
daylight simulation.

I¹m considering the three-phase method Greg described on slide 11 of in his
workshop #6 talk:
http://www.radiance-online.org/radiance-workshop6/GregWard/WardUsingBTDF_tal
k.pdf

Which in theory seems pretty straight forward:
1 ­ determine daylight coefficients incident on the glazing
2 ­ use a BSDF to define the relationship between incident light directions
and exiting light distribution for the glazing.
3 ­ determine the relationship between exiting light directions and
illuminance in the space.
4 ­ multiply results by many other results to produce an annual illuminance
profile.

However I face a couple of challenges for which I seek advice:

Challenge 1 ­ I don¹t have BSDF data from panelite, they are looking into
having a sample measured but that won¹t help me right now. I was thinking
that I could build a radiance model of panelite and use rtconrib to generate
a BSDF of sorts. Is this a terrible idea?

Challenge 2 ­ Writing a rtcontrib expression to separate contributions
according to exiting direction from the portal. Can I use directions from
treganza.cal? Is there some reason that this isn¹t a good idea? Klems
directions has been mentioned. I have no knowledge of klems directions. Is
that a better alternative?

And I guess the overall questions: Are there more challenges I have yet to
discover? Is anyone who has done this willing to share experiences? Would
I be better off creating a model with thousands of cylinders between glass
panes and using rtcontrib + brute force? Is there an even better method
that I haven¹t discovered trolling through archives?

Thanks,
Andy

···

____________________________________________________________
Electronic mail messages entering and leaving Arup business
systems are scanned for acceptability of content and viruses

Hi Andrew,

With funding from Southern California Edison through the Heschong-Mahone Group and separate funding from LBNL, I've been working on the 3-phase method described. We have a few tools in the HEAD version of Radiance now, which I'll be talking about next week at the Radiance workshop. Unfortunately, I haven't prepared any presentation materials yet, but if you make it to Boston, I will have something by then and we can have a nice conversation.

The lack of BTDF data is a serious issue, and could be addressed with appropriate modeling. I don't know how well rtcontrib would work for this, as I've never tried it, but it's an interesting suggestion. The folks in the Windows group at LBNL have been using TracePro to generate their BTDFs when they aren't taking measurements.

It's an ambitious project you have, and not something you're going to sort out in an afternoon. What's your deadline?

Cheers,
-Greg

···

From: Andrew McNeil <[email protected]>
Date: October 15, 2009 12:17:02 PM PDT

Hi Everyone,

I’m trying to figure out how to incorporate clearshade from panelite ( http://www.e-panelite.com/downloads/Panelite_IGU_Oct08.pdf ) into an annual daylight simulation.

I’m considering the three-phase method Greg described on slide 11 of in his workshop #6 talk:
http://www.radiance-online.org/radiance-workshop6/GregWard/WardUsingBTDF_talk.pdf

Which in theory seems pretty straight forward:
1 – determine daylight coefficients incident on the glazing
2 – use a BSDF to define the relationship between incident light directions and exiting light distribution for the glazing.
3 – determine the relationship between exiting light directions and illuminance in the space.
4 – multiply results by many other results to produce an annual illuminance profile.

However I face a couple of challenges for which I seek advice:

Challenge 1 – I don’t have BSDF data from panelite, they are looking into having a sample measured but that won’t help me right now. I was thinking that I could build a radiance model of panelite and use rtconrib to generate a BSDF of sorts. Is this a terrible idea?

Challenge 2 – Writing a rtcontrib expression to separate contributions according to exiting direction from the portal. Can I use directions from treganza.cal? Is there some reason that this isn’t a good idea? Klems directions has been mentioned. I have no knowledge of klems directions. Is that a better alternative?

And I guess the overall questions: Are there more challenges I have yet to discover? Is anyone who has done this willing to share experiences? Would I be better off creating a model with thousands of cylinders between glass panes and using rtcontrib + brute force? Is there an even better method that I haven’t discovered trolling through archives?

Thanks,
Andy

Hi Greg,

Thanks for your reply. Unfortunately I¹m not going to Boston. Now I have
yet another reason to wish I was going.

I¹m glad you¹ve been able to further develop the 3-phase method. I see
increasing demand for this type of analysis for complex façade shading
systems. This is the fourth time I¹ve started down this route ­ the other
three times stopped abruptly for various reasons. Twice before with
panelite products and once with an engineered scattering window film. The
email I sent yesterday was a partially complete left in my drafts folder
from two months ago - the last time this I was looking into this approach.

I don¹t have a hard deadline yet. The goal for now is to develop a working
model and then focus on accuracy later. Ideally I¹d have something that
works in the next 2 weeks.

Hopefully Panelite will have measured BTDF data in a few months that we slot
into the analysis. If measured BTDF data doesn't become available then
we¹ll focus on improving the accuracy of the simulated BTDF.

So I guess my first step is approximating a BTDF with Radiance. I was
thinking I could a uniform sky above the panelite model as input and a
hemisphere of points below as sensors. With rtcontrib I could determine the
light output in a direction for each input direction. The input directions
could be tregenza sky divisions (?). The output could be phi/lambda
directions (?). I imagine the key to a decent BTDF is lots of samples for
each output direction. Maybe I create a grid of 1000(00?) points in a plane
just below the panelite and use a script to create a hemisphere of sensor
points centered at each grid point.

I know very little about BTDF¹s and even less about creating them so I could
use any tips, tricks, guidance, watch its, how are you going to account
for?s and examples that anyone is willing to offer.

Thanks,
Andy

···

On 10/15/09 1:37 PM, "Greg Ward" <[email protected]> wrote:

Hi Andrew,

With funding from Southern California Edison through the Heschong-Mahone Group
and separate funding from LBNL, I've been working on the 3-phase method
described. We have a few tools in the HEAD version of Radiance now, which
I'll be talking about next week at the Radiance workshop. Unfortunately, I
haven't prepared any presentation materials yet, but if you make it to Boston,
I will have something by then and we can have a nice conversation.

The lack of BTDF data is a serious issue, and could be addressed with
appropriate modeling. I don't know how well rtcontrib would work for this, as
I've never tried it, but it's an interesting suggestion. The folks in the
Windows group at LBNL have been using TracePro to generate their BTDFs when
they aren't taking measurements.

It's an ambitious project you have, and not something you're going to sort out
in an afternoon. What's your deadline?

Cheers,
-Greg

From: Andrew McNeil <[email protected]>

Date: October 15, 2009 12:17:02 PM PDT

Hi Everyone,

I¹m trying to figure out how to incorporate clearshade from panelite (
http://www.e-panelite.com/downloads/Panelite_IGU_Oct08.pdf ) into an annual
daylight simulation.

I¹m considering the three-phase method Greg described on slide 11 of in his
workshop #6 talk:

http://www.radiance-online.org/radiance-workshop6/GregWard/WardUsingBTDF_tal
k.pdf

Which in theory seems pretty straight forward:
1 ­ determine daylight coefficients incident on the glazing
2 ­ use a BSDF to define the relationship between incident light directions
and exiting light distribution for the glazing.
3 ­ determine the relationship between exiting light directions and
illuminance in the space.
4 ­ multiply results by many other results to produce an annual illuminance
profile.

However I face a couple of challenges for which I seek advice:

Challenge 1 ­ I don¹t have BSDF data from panelite, they are looking into
having a sample measured but that won¹t help me right now. I was thinking
that I could build a radiance model of panelite and use rtconrib to generate
a BSDF of sorts. Is this a terrible idea?

Challenge 2 ­ Writing a rtcontrib expression to separate contributions
according to exiting direction from the portal. Can I use directions from
treganza.cal? Is there some reason that this isn¹t a good idea? Klems
directions has been mentioned. I have no knowledge of klems directions. Is
that a better alternative?

And I guess the overall questions: Are there more challenges I have yet to
discover? Is anyone who has done this willing to share experiences? Would
I be better off creating a model with thousands of cylinders between glass
panes and using rtcontrib + brute force? Is there an even better method
that I haven¹t discovered trolling through archives?

Thanks,
Andy

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

____________________________________________________________
Electronic mail messages entering and leaving Arup business
systems are scanned for acceptability of content and viruses

Hi Andrew,

I guess I should write a program or script to create a BTDF file for an arbitrary shading system. It's not particularly difficult to do this using rtcontrib, but I need to find time to do it. The next couple of weeks are crazy between the workshop and the HDR Imaging book deadline, so I don't think I'll get to it before then. I won't be much help with your deadline, unfortunately.

-Greg

···

From: Andrew McNeil <[email protected]>
Date: October 16, 2009 9:34:13 AM PDT

Hi Greg,

Thanks for your reply. Unfortunately I’m not going to Boston. Now I have yet another reason to wish I was going.

I’m glad you’ve been able to further develop the 3-phase method. I see increasing demand for this type of analysis for complex façade shading systems. This is the fourth time I’ve started down this route – the other three times stopped abruptly for various reasons. Twice before with panelite products and once with an engineered scattering window film. The email I sent yesterday was a partially complete left in my drafts folder from two months ago - the last time this I was looking into this approach.

I don’t have a hard deadline yet. The goal for now is to develop a working model and then focus on accuracy later. Ideally I’d have something that works in the next 2 weeks.

Hopefully Panelite will have measured BTDF data in a few months that we slot into the analysis. If measured BTDF data doesn't become available then we’ll focus on improving the accuracy of the simulated BTDF.

So I guess my first step is approximating a BTDF with Radiance. I was thinking I could a uniform sky above the panelite model as input and a hemisphere of points below as sensors. With rtcontrib I could determine the light output in a direction for each input direction. The input directions could be tregenza sky divisions (?). The output could be phi/lambda directions (?). I imagine the key to a decent BTDF is lots of samples for each output direction. Maybe I create a grid of 1000(00?) points in a plane just below the panelite and use a script to create a hemisphere of sensor points centered at each grid point.

I know very little about BTDF’s and even less about creating them so I could use any tips, tricks, guidance, watch its, how are you going to account for?s and examples that anyone is willing to offer.

Thanks,
Andy

Hi Greg,

I¹ve written a script that does this, it¹s been running over the weekend.
I¹ll share the script if the results are consistent with expectations ( I
give it a 40% chance ).

I decided to use both phi/theta input and output directions, because that¹s
consistent with the window 6 BSDF specification that mkillum uses.

I look forward to seeing your presentation slides for the conference next
week.

Andy

···

On 10/18/09 11:26 AM, "Greg Ward" <[email protected]> wrote:

Hi Andrew,

I guess I should write a program or script to create a BTDF file for an
arbitrary shading system. It's not particularly difficult to do this using
rtcontrib, but I need to find time to do it. The next couple of weeks are
crazy between the workshop and the HDR Imaging book deadline, so I don't think
I'll get to it before then. I won't be much help with your deadline,
unfortunately.

-Greg

From: Andrew McNeil <[email protected]>

Date: October 16, 2009 9:34:13 AM PDT

Hi Greg,

Thanks for your reply. Unfortunately I¹m not going to Boston. Now I have
yet another reason to wish I was going.

I¹m glad you¹ve been able to further develop the 3-phase method. I see
increasing demand for this type of analysis for complex façade shading
systems. This is the fourth time I¹ve started down this route ­ the other
three times stopped abruptly for various reasons. Twice before with
panelite products and once with an engineered scattering window film. The
email I sent yesterday was a partially complete left in my drafts folder
from two months ago - the last time this I was looking into this approach.

I don¹t have a hard deadline yet. The goal for now is to develop a working
model and then focus on accuracy later. Ideally I¹d have something that
works in the next 2 weeks.

Hopefully Panelite will have measured BTDF data in a few months that we
slot into the analysis. If measured BTDF data doesn't become available then
we¹ll focus on improving the accuracy of the simulated BTDF.

So I guess my first step is approximating a BTDF with Radiance. I was
thinking I could a uniform sky above the panelite model as input and a
hemisphere of points below as sensors. With rtcontrib I could determine the
light output in a direction for each input direction. The input directions
could be tregenza sky divisions (?). The output could be phi/lambda
directions (?). I imagine the key to a decent BTDF is lots of samples for
each output direction. Maybe I create a grid of 1000(00?) points in a plane
just below the panelite and use a script to create a hemisphere of sensor
points centered at each grid point.

I know very little about BTDF¹s and even less about creating them so I
could use any tips, tricks, guidance, watch its, how are you going to
account for?s and examples that anyone is willing to offer.

Thanks,
Andy

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

____________________________________________________________
Electronic mail messages entering and leaving Arup business
systems are scanned for acceptability of content and viruses