Finding contribution of different glass roof panels using r(t)contrib

Dear all,

I've been following some of the discussions on the mailing list with keen interest and gratitude. Now, I'd like to hear your thoughts on something I'm working on. (Hopefully, I'll be able to start answering other people's questions soon too!)

I'm working on a building with a large glass roof, with plants and people inside. The goal is to find optimal light transmittances for different parts of the roof so it is relatively uniformly lit throughout the year. The main worry is getting enough light for the plants, yet limiting irradiance to not require excessive cooling.

Here's what I am thinking at the moment:

* Break up dome into N separate panels using Rhino/Grasshopper, assign each panel to a layer, and then export layers as materials using DIVA

* Generate a materials file with N identical materials with material_i as name

* Run r(t)contrib for the different modifiers (using -M and a modifier file) and points, using a cumulative sky generated by DIVA, to find to each point by each panel

* Visualise output data using Rhino/Grasshopper

* Find (near) optimum light transmittances based on 2-3 criteria (still to be defined, but likely UDI and uniformity). With the available data is should a purely numerical exercise, no simulations involved.

I've tested the separate steps but not yet the whole workflow. I may need to extend the simulation by doing an hourly simulation to obtain useful daylight illuminances, probably by creating genskyvec+rtcontrib-loop, like in Axel Jacobs' rtcontrib tutorial. That may get computationally intensive, in which case I may resort to using EC2.

A few questions:

* Any experiences with something similar? Ways to simplify /improve my workflow? Which errors will I likely run into?

* obj2rad generates an interpolated .rad file using texfunc.cal and something called 'M-nor'. This means I can't use my material as a modifier for rcontrib. For now, I have exported my model as simple planes so that the materials can be used as modifiers, but is there a better way?

* Should I use a CPU-optimised instance on EC2? What are some ways to properly do parallel computing on EC2? Any examples?

* An alternative is using one material for the dome and creating a custom binning for rtcontrib. Does this have any computational advantages?

Any help greatly appreciated!

Cheers,
Reinier Zeldenrust
Environmental Designer
Atelier Ten
Building Services Engineers + Environmental Design Consultants
CIBSE Building Services Consultancy of the Year 2011
[image001][image003]<http://www.amazon.co.uk/Green-House-Sustainable-Gardens-Singapore/dp/1935935569/ref=sr_1_2?ie=UTF8&qid=1351789637&sr=8-2>
19 Perseverance Works
38 Kingsland Road
London E2 8DD
T +44 (0)20 7749 5950
F +44 (0)20 7729 5388
atelierten.com<http://www.atelierten.com/>
Atelier Ten is an accredited ISO 9001 company and is pleased to be a founding member of the UKGBC. This e-mail (and any attachments) may contain privileged and/or confidential information. If you are not the intended recipient of this e-mail you must not disclose, copy, distribute, disseminate or take any action in reliance on it. If you have received this message in error please notify the sender and then delete it from your system. For the 12-months to the end of October 2012, with Clear we have offset our expected CO2 emissions in respect of our business energy usage within our UK offices; commuting; work-related travel; office vehicles; couriers including supplies and deliveries; waste from our UK office and the production of our project reports. Click here<http://www.atelierten.com/about-us/clear.asp> for more information

Company Registration Number: 255 2224. Company Registered Address: 19 Perseverance Works, 38 Kingsland Road, London, E2 8DD

···

________________________________________________________________________
This e-mail has been scanned for all viruses by Star. The
service is powered by MessageLabs. For more information on a proactive
anti-virus service working around the clock, around the globe, visit:


________________________________________________________________________

Hi Reinier,

You should probably take a look at the "three-phase method" which breaks flux transfer down into three discrete steps:

1. sky to glazing exterior
2. transmission through glazing
3. glazing interior to built environment

Glazing transmission (2) is accounted for via a BSDF representation. The nice thing about this is that if you want to change glass type/performance you do not end up having to re-run a complex simulation, it is a matter of recomputing a matrix calculation. In other words, the material behavior of the glazing is a separate step and does not actually take that long. Andy McNeil has a great tutorial on using this method:

I recently worked on a project using this method. The main goal was to evaluate daylight based on hourly weather data for the locale for plant growth within a selected exhibition environment. Using this approach, was a big help when we needed to look at some other glazing options.

This does requires some setup work particularly if you are looking at rooftop glazing. You will also want to think carefully about how far you want to break down the glazing panels, whether individual panels or grouping of some sort.

If you are calculating sensor points you may find that you can do it on your workstation, however AWS/EC2 is definitely worth considering as an option. I did use AWS/EC2 for final runs as I did not want to have to wait, however I was generating images...

Hope this helps.

-Jack

···

--
Jack de Valpine
President

Visarc Incorporated
www.visarc.com

channeling technology for superior design and construction

On 7/15/2013 10:53 AM, Reinier Zeldenrust wrote:

Dear all,

I've been following some of the discussions on the mailing list with keen interest and gratitude. Now, I'd like to hear your thoughts on something I'm working on. (Hopefully, I'll be able to start answering other people's questions soon too!)

I'm working on a building with a large glass roof, with plants and people inside. The goal is to find optimal light transmittances for different parts of the roof so it is relatively uniformly lit throughout the year. The main worry is getting enough light for the plants, yet limiting irradiance to not require excessive cooling.

Here's what I am thinking at the moment:

·Break up dome into N separate panels using Rhino/Grasshopper, assign each panel to a layer, and then export layers as materials using DIVA

·Generate a materials file with N identical materials with material_i as name

·Run r(t)contrib for the different modifiers (using --M and a modifier file) and points, using a cumulative sky generated by DIVA, to find to each point by each panel

·Visualise output data using Rhino/Grasshopper

·Find (near) optimum light transmittances based on 2-3 criteria (still to be defined, but likely UDI and uniformity). With the available data is should a purely numerical exercise, no simulations involved.

I've tested the separate steps but not yet the whole workflow. I may need to extend the simulation by doing an hourly simulation to obtain useful daylight illuminances, probably by creating genskyvec+rtcontrib-loop, like in Axel Jacobs' rtcontrib tutorial. That may get computationally intensive, in which case I may resort to using EC2.

A few questions:

·Any experiences with something similar? Ways to simplify /improve my workflow? Which errors will I likely run into?

·obj2rad generates an interpolated .rad file using texfunc.cal and something called 'M-nor'. This means I can't use my material as a modifier for rcontrib. For now, I have exported my model as simple planes so that the materials can be used as modifiers, but is there a better way?

·Should I use a CPU-optimised instance on EC2? What are some ways to properly do parallel computing on EC2? Any examples?

·An alternative is using one material for the dome and creating a custom binning for rtcontrib. Does this have any computational advantages?

Any help greatly appreciated!

Cheers,

*Reinier Zeldenrust
*Environmental Designer

Atelier Ten
Building Services Engineers + Environmental Design Consultants
CIBSE Building Services Consultancy of the Year 2011

image001image003 <http://www.amazon.co.uk/Green-House-Sustainable-Gardens-Singapore/dp/1935935569/ref=sr_1_2?ie=UTF8&qid=1351789637&sr=8-2>

19 Perseverance Works
38 Kingsland Road
London E2 8DD
T +44 (0)20 7749 5950
F +44 (0)20 7729 5388

atelierten.com <http://www.atelierten.com/>

Atelier Ten is an accredited ISO 9001 company and is pleased to be a founding member of the UKGBC. This e-mail (and any attachments) may contain privileged and/or confidential information. If you are not the intended recipient of this e-mail you must not disclose, copy, distribute, disseminate or take any action in reliance on it. If you have received this message in error please notify the sender and then delete it from your system. For the 12-months to the end of October 2012, with Clear we have offset our expected CO_2 emissions in respect of our business energy usage within our UK offices; commuting; work-related travel; office vehicles; couriers including supplies and deliveries; waste from our UK office and the production of our project reports. Click here <http://www.atelierten.com/about-us/clear.asp>for more information

Company Registration Number: 255 2224. Company Registered Address: 19 Perseverance Works, 38 Kingsland Road, London, E2 8DD

________________________________________________________________________
This e-mail has been scanned for all viruses by Star. The
service is powered by MessageLabs. For more information on a proactive
anti-virus service working around the clock, around the globe, visit:
http://www.star.net.uk
________________________________________________________________________

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

I am not sure if the 3phase method is the best in this case, since the
accepted basis (145 Klems' division) is not very good for specular
fenestration systems (such as glazings).

Maybe a possible solution would be to make separate DC analysis for the
different fenestration groups (i.e. make all glazings black, except the one
to be analyzed) and somehow scale them by transmissivity...?? I mean:

For each glazing group:
     - Make every other glazing black, and this glazing with 100%
transmittance
     - Calculate DC.
end

then, the lighting levels would be

Lux= a1*DC1+a2*DC2+a3*DC3...

Does this make any sense?

···

2013/7/15 Jack de Valpine <[email protected]>

Hi Reinier,

You should probably take a look at the "three-phase method" which breaks
flux transfer down into three discrete steps:

   1. sky to glazing exterior
   2. transmission through glazing
   3. glazing interior to built environment

Glazing transmission (2) is accounted for via a BSDF representation. The
nice thing about this is that if you want to change glass type/performance
you do not end up having to re-run a complex simulation, it is a matter of
recomputing a matrix calculation. In other words, the material behavior of
the glazing is a separate step and does not actually take that long. Andy
McNeil has a great tutorial on using this method:

http://www.radiance-online.org/learning/tutorials/Tutorial-ThreePhaseMethod.pdf

I recently worked on a project using this method. The main goal was to
evaluate daylight based on hourly weather data for the locale for plant
growth within a selected exhibition environment. Using this approach, was a
big help when we needed to look at some other glazing options.

This does requires some setup work particularly if you are looking at
rooftop glazing. You will also want to think carefully about how far you
want to break down the glazing panels, whether individual panels or
grouping of some sort.

If you are calculating sensor points you may find that you can do it on
your workstation, however AWS/EC2 is definitely worth considering as an
option. I did use AWS/EC2 for final runs as I did not want to have to wait,
however I was generating images...

Hope this helps.

-Jack

--
Jack de Valpine
President

Visarc Incorporatedwww.visarc.com

channeling technology for superior design and construction

On 7/15/2013 10:53 AM, Reinier Zeldenrust wrote:

Dear all,****

** **

I’ve been following some of the discussions on the mailing list with keen
interest and gratitude. Now, I’d like to hear your thoughts on something
I’m working on. (Hopefully, I’ll be able to start answering other people’s
questions soon too!)****

** **

I’m working on a building with a large glass roof, with plants and people
inside. The goal is to find optimal light transmittances for different
parts of the roof so it is relatively uniformly lit throughout the year.
The main worry is getting enough light for the plants, yet limiting
irradiance to not require excessive cooling.****

** **

Here’s what I am thinking at the moment:****

· Break up dome into N separate panels using Rhino/Grasshopper,
assign each panel to a layer, and then export layers as materials using DIVA
****

· Generate a materials file with N identical materials with
material_i as name****

· Run r(t)contrib for the different modifiers (using –M and a
modifier file) and points, using a cumulative sky generated by DIVA, to
find to each point by each panel****

· Visualise output data using Rhino/Grasshopper****

· Find (near) optimum light transmittances based on 2-3 criteria
(still to be defined, but likely UDI and uniformity). With the available
data is should a purely numerical exercise, no simulations involved.****

** **

I’ve tested the separate steps but not yet the whole workflow. I may need
to extend the simulation by doing an hourly simulation to obtain useful
daylight illuminances, probably by creating genskyvec+rtcontrib-loop, like
in Axel Jacobs’ rtcontrib tutorial. That may get computationally intensive,
in which case I may resort to using EC2.****

** **

A few questions:****

· Any experiences with something similar? Ways to simplify
/improve my workflow? Which errors will I likely run into?****

· obj2rad generates an interpolated .rad file using texfunc.cal
and something called 'M-nor'. This means I can't use my material as a
modifier for rcontrib. For now, I have exported my model as simple planes
so that the materials can be used as modifiers, but is there a better way?
****

· Should I use a CPU-optimised instance on EC2? What are some
ways to properly do parallel computing on EC2? Any examples?****

· An alternative is using one material for the dome and creating
a custom binning for rtcontrib. Does this have any computational advantages?
****

** **

Any help greatly appreciated!****

** **

Cheers,****

*Reinier Zeldenrust
*Environmental Designer****

Atelier Ten
Building Services Engineers + Environmental Design Consultants
CIBSE Building Services Consultancy of the Year 2011****

[image: image001][image: image003]<http://www.amazon.co.uk/Green-House-Sustainable-Gardens-Singapore/dp/1935935569/ref=sr_1_2?ie=UTF8&qid=1351789637&sr=8-2>
****

19 Perseverance Works
38 Kingsland Road
London E2 8DD
T +44 (0)20 7749 5950
F +44 (0)20 7729 5388****

atelierten.com <http://www.atelierten.com/>****

Atelier Ten is an accredited ISO 9001 company and is pleased to be a
founding member of the UKGBC. This e-mail (and any attachments) may contain
privileged and/or confidential information. If you are not the intended
recipient of this e-mail you must not disclose, copy, distribute,
disseminate or take any action in reliance on it. If you have received this
message in error please notify the sender and then delete it from your
system. For the 12-months to the end of October 2012, with Clear we have
offset our expected CO2 emissions in respect of our business energy usage
within our UK offices; commuting; work-related travel; office vehicles;
couriers including supplies and deliveries; waste from our UK office and
the production of our project reports. Click here<http://www.atelierten.com/about-us/clear.asp>for more information

Company Registration Number: 255 2224. Company Registered Address: 19
Perseverance Works, 38 Kingsland Road, London, E2 8DD****

** **

________________________________________________________________________
This e-mail has been scanned for all viruses by Star. The
service is powered by MessageLabs. For more information on a proactive
anti-virus service working around the clock, around the globe, visit:
http://www.star.net.uk
________________________________________________________________________

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

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

Sorry for sending a second e-mail, I sent it by mistake.

a1, a2 and etc... are the transmissivities in my example (1 equals air, 0
equals black). But I am not sure if that linear scaling and adding is
valid...?

I am interested to see the solution you use!

THANKS

German

···

2013/7/17 Germán Molina Larrain <[email protected]>

I am not sure if the 3phase method is the best in this case, since the
accepted basis (145 Klems' division) is not very good for specular
fenestration systems (such as glazings).

Maybe a possible solution would be to make separate DC analysis for the
different fenestration groups (i.e. make all glazings black, except the one
to be analyzed) and somehow scale them by transmissivity...?? I mean:

For each glazing group:
     - Make every other glazing black, and this glazing with 100%
transmittance
     - Calculate DC.
end

then, the lighting levels would be

Lux= a1*DC1+a2*DC2+a3*DC3...

Does this make any sense?

2013/7/15 Jack de Valpine <[email protected]>

Hi Reinier,

You should probably take a look at the "three-phase method" which breaks
flux transfer down into three discrete steps:

   1. sky to glazing exterior
   2. transmission through glazing
   3. glazing interior to built environment

Glazing transmission (2) is accounted for via a BSDF representation. The
nice thing about this is that if you want to change glass type/performance
you do not end up having to re-run a complex simulation, it is a matter of
recomputing a matrix calculation. In other words, the material behavior of
the glazing is a separate step and does not actually take that long. Andy
McNeil has a great tutorial on using this method:

http://www.radiance-online.org/learning/tutorials/Tutorial-ThreePhaseMethod.pdf

I recently worked on a project using this method. The main goal was to
evaluate daylight based on hourly weather data for the locale for plant
growth within a selected exhibition environment. Using this approach, was a
big help when we needed to look at some other glazing options.

This does requires some setup work particularly if you are looking at
rooftop glazing. You will also want to think carefully about how far you
want to break down the glazing panels, whether individual panels or
grouping of some sort.

If you are calculating sensor points you may find that you can do it on
your workstation, however AWS/EC2 is definitely worth considering as an
option. I did use AWS/EC2 for final runs as I did not want to have to wait,
however I was generating images...

Hope this helps.

-Jack

--
Jack de Valpine
President

Visarc Incorporatedwww.visarc.com

channeling technology for superior design and construction

On 7/15/2013 10:53 AM, Reinier Zeldenrust wrote:

Dear all,****

** **

I’ve been following some of the discussions on the mailing list with keen
interest and gratitude. Now, I’d like to hear your thoughts on something
I’m working on. (Hopefully, I’ll be able to start answering other people’s
questions soon too!)****

** **

I’m working on a building with a large glass roof, with plants and people
inside. The goal is to find optimal light transmittances for different
parts of the roof so it is relatively uniformly lit throughout the year.
The main worry is getting enough light for the plants, yet limiting
irradiance to not require excessive cooling.****

** **

Here’s what I am thinking at the moment:****

· Break up dome into N separate panels using Rhino/Grasshopper,
assign each panel to a layer, and then export layers as materials using DIVA
****

· Generate a materials file with N identical materials with
material_i as name****

· Run r(t)contrib for the different modifiers (using –M and a
modifier file) and points, using a cumulative sky generated by DIVA, to
find to each point by each panel****

· Visualise output data using Rhino/Grasshopper****

· Find (near) optimum light transmittances based on 2-3 criteria
(still to be defined, but likely UDI and uniformity). With the available
data is should a purely numerical exercise, no simulations involved.****

** **

I’ve tested the separate steps but not yet the whole workflow. I may need
to extend the simulation by doing an hourly simulation to obtain useful
daylight illuminances, probably by creating genskyvec+rtcontrib-loop, like
in Axel Jacobs’ rtcontrib tutorial. That may get computationally intensive,
in which case I may resort to using EC2.****

** **

A few questions:****

· Any experiences with something similar? Ways to simplify
/improve my workflow? Which errors will I likely run into?****

· obj2rad generates an interpolated .rad file using texfunc.cal
and something called 'M-nor'. This means I can't use my material as a
modifier for rcontrib. For now, I have exported my model as simple planes
so that the materials can be used as modifiers, but is there a better way?
****

· Should I use a CPU-optimised instance on EC2? What are some
ways to properly do parallel computing on EC2? Any examples?****

· An alternative is using one material for the dome and creating
a custom binning for rtcontrib. Does this have any computational advantages?
****

** **

Any help greatly appreciated!****

** **

Cheers,****

*Reinier Zeldenrust
*Environmental Designer****

Atelier Ten
Building Services Engineers + Environmental Design Consultants
CIBSE Building Services Consultancy of the Year 2011****

[image: image001][image: image003]<http://www.amazon.co.uk/Green-House-Sustainable-Gardens-Singapore/dp/1935935569/ref=sr_1_2?ie=UTF8&qid=1351789637&sr=8-2>
****

19 Perseverance Works
38 Kingsland Road
London E2 8DD
T +44 (0)20 7749 5950
F +44 (0)20 7729 5388****

atelierten.com <http://www.atelierten.com/>****

Atelier Ten is an accredited ISO 9001 company and is pleased to be a
founding member of the UKGBC. This e-mail (and any attachments) may contain
privileged and/or confidential information. If you are not the intended
recipient of this e-mail you must not disclose, copy, distribute,
disseminate or take any action in reliance on it. If you have received this
message in error please notify the sender and then delete it from your
system. For the 12-months to the end of October 2012, with Clear we have
offset our expected CO2 emissions in respect of our business energy
usage within our UK offices; commuting; work-related travel; office
vehicles; couriers including supplies and deliveries; waste from our UK
office and the production of our project reports. Click here<http://www.atelierten.com/about-us/clear.asp>for more information

Company Registration Number: 255 2224. Company Registered Address: 19
Perseverance Works, 38 Kingsland Road, London, E2 8DD****

** **

________________________________________________________________________
This e-mail has been scanned for all viruses by Star. The
service is powered by MessageLabs. For more information on a proactive
anti-virus service working around the clock, around the globe, visit:
http://www.star.net.uk
________________________________________________________________________

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

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