query about accurately modelling glazing

Hi Zack, Daniel,

The BRTDfunc definitions delivered by desktop radiance are a bit difficult. they do allow for different reflectance from front and back faces, but there are some major problems with it.

firstly, to get it to work, you need two glass polygons for each window, one for the front and one for the back.

secondly, what does this do to the transmittance? physically, the transmittance going one way should be the same as the transmittance going the other way. should the (rtrns, gtrns, btrns) variables then be the same for both front and back panes? or does it not matter, as long as the product of their transmittances is equal to that of the combination.

finally, there are no functions built in. that means that there is no dependence on angle of incidence for either transmittance or reflectance. this is unrealistic, as transmittance generally reduces and reflectance generally increases with increasing angle of incidence.

so, i never use the BRTDfunc materials delivered by desktop radiance. if i understand correctly, these definitions are provided by Optics 5 (or 4, or something like that). is this right? if it is, then i wouldn't use the Optics output either.

i like the look of Zack's definition using glazing.cal. might have to give that a burl. what i have been doing is creating 'glass' primitives and setting the r,g,b components and refractive index to give me the desired normal reflectance and transmittance, and hoping that the resulting angular dependence is about right. have created a spreadsheet with Greg's equations to determine the (r,g,b,n) parameters.

Phil.

[email protected] 15/01/2004 12:11:56 pm >>>

Hi Daniel,

My question is this. To what extent will my simplified way of
describing the transmittance affect the accuracy of my results?

Defining glass using the BRTDFunc method just lets you more accurately
define the front and back reflectance of the glass. Defining glass with
just the glass primitive uses a default front and back reflectance.

I
don't completely understand the workings of the BRTDfunc (even after reading the manual page), and
think I
would have trouble measuring any variable of glazing other than
transmittance of the specimens I am sampling in the real world (Rendering with Radiance
talks briefly of measuring glazing transmittance,
but not things like reflectance.

I recommend looking at Optics 5 from LBNL.

http://windows.lbl.gov/materials/optics5/

It has a very extensive database of just about every type of glass and
glass compositions available and allows you to build up various double
pane combinations and save it all as a Radiance input file. Rather than
measuring a sample in the real world and defining it in Radiance, you
could probably find the glazing in this database and use Optics 5 to
define it.

I presume also that transmittance + reflectance (+ absorptance) of
glazing can never exceed 1. In Desktop Radiance, the description of
Generic clear glass lbnl (clear3.rad) states a transmittance of
89.90% and a reflection of 82.0%. Is my understanding fundementally
wrong, or is this a typo?

I've always been a little confused how desktop radiance defines glass.
The front RGB reflectance is actually defined with this line under
"clear3_front"

0.07428 0.08322 0.08556

and the back RGB reflectance is defined with this line under "clear3_back"

0.07567 0.08418 0.08538

So the reflectances are roughly 8% which does obey the rule.

I am not completely clear what the other RGB values are but my guess is
that they modify the transmittivity defined under clear3_glass somehow.
Whenever I've manually defined glass I've used this form (from
"glazing.cal") instead, its just a little clearer to me.

mod BRTDfunc my_glazing
    10 rrho grho brho
        rtau gtau btau
        0 0 0
        glazing.cal
    0
    18 0 0 0
        0 0 0
        0 0 0
        FRRHO FGRHO FBRHO
        BRRHO BGRHO BBRHO
        RTAU GTAU BTAU

    where:
        FRRHO FGRHO FBRHO is front normal spectral reflectance
        BRRHO BGRHO BBRHO is back normal spectral reflectance
        RTAU GTAU BTAU is normal spectral transmittance

Hope this helps!

Zack

···

--
Zack Rogers
Staff Engineer
Architectural Energy Corporation
2540 Frontier Avenue, Suite 201
Boulder, CO 80301 USA

tel (303)444-4149 ext.235
fax (303)444-4304

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

Hi Zack and Phil.

Thank you very much for your most helpful suggestions. My experiments are in two parts. Firstly,
I am attempting to simulate as accurately as possible the lighting conditions under a real atrium
roof canopy, and comparing with illuminance measurements measured directly under the roof. In
terms of modelling it accurately, I think Optic5 (and associated database) will be a useful way to
derive transmittance information. I have still not quite decided whether to take the information
(calculated transmittance of multi-layer configurations) from the export and create a simple
'glass' primitive, or to use the BRTDFunc generated. Perhaps I will experiment and see whether it
makes much of a difference.

The second part involves a systematic varying of the transmittance of the glazing in the
simulation. Here I believe it would be appropriate to use the simple glass primitive, and vary the
r,g,b,n parameters (though I think n should stay constant at 1.52 (or whatever it is...), and I
think it said in the manual page that having only the r,g,b arguments has Radiance use the default
n, fine for my needs). I thought that the BRTDFunc may have included some function depending on
angle of incidence that would make it more accurate than the glass primitive, but as it doesn't, I
will stick with the glass (as I am dealing with overcast skies, the 'grazing angle' changes in
transmittance may not be as critical (though still a factor) as if I were looking at direct light).

Once again many thanks

Dan Lash
PhD researcher
Sheffield Hallam University, UK

[email protected] writes:

···

Hi Zack, Daniel,

The BRTDfunc definitions delivered by desktop radiance are a bit difficult. they do allow for
different reflectance from front and back faces, but there are some major problems with it.

firstly, to get it to work, you need two glass polygons for each window, one for the front and one
for the back.

secondly, what does this do to the transmittance? physically, the transmittance going one way
should be the same as the transmittance going the other way. should the (rtrns, gtrns, btrns)
variables then be the same for both front and back panes? or does it not matter, as long as the
product of their transmittances is equal to that of the combination.

finally, there are no functions built in. that means that there is no dependence on angle of
incidence for either transmittance or reflectance. this is unrealistic, as transmittance
generally reduces and reflectance generally increases with increasing angle of incidence.

so, i never use the BRTDfunc materials delivered by desktop radiance. if i understand correctly,
these definitions are provided by Optics 5 (or 4, or something like that). is this right? if it
is, then i wouldn't use the Optics output either.

i like the look of Zack's definition using glazing.cal. might have to give that a burl. what i
have been doing is creating 'glass' primitives and setting the r,g,b components and refractive
index to give me the desired normal reflectance and transmittance, and hoping that the resulting
angular dependence is about right. have created a spreadsheet with Greg's equations to determine
the (r,g,b,n) parameters.

Phil.

[email protected] 15/01/2004 12:11:56 pm >>>

Hi Daniel,

My question is this. To what extent will my simplified way of
describing the transmittance affect the accuracy of my results?

Defining glass using the BRTDFunc method just lets you more accurately
define the front and back reflectance of the glass. Defining glass with
just the glass primitive uses a default front and back reflectance.

I
don't completely understand the workings of the BRTDfunc (even after reading the manual page), and
think I
would have trouble measuring any variable of glazing other than
transmittance of the specimens I am sampling in the real world (Rendering with Radiance
talks briefly of measuring glazing transmittance,
but not things like reflectance.

I recommend looking at Optics 5 from LBNL.

http://windows.lbl.gov/materials/optics5/

It has a very extensive database of just about every type of glass and
glass compositions available and allows you to build up various double
pane combinations and save it all as a Radiance input file. Rather than
measuring a sample in the real world and defining it in Radiance, you
could probably find the glazing in this database and use Optics 5 to
define it.

I presume also that transmittance + reflectance (+ absorptance) of
glazing can never exceed 1. In Desktop Radiance, the description of
Generic clear glass lbnl (clear3.rad) states a transmittance of
89.90% and a reflection of 82.0%. Is my understanding fundementally
wrong, or is this a typo?

I've always been a little confused how desktop radiance defines glass.
The front RGB reflectance is actually defined with this line under
"clear3_front"

0.07428 0.08322 0.08556

and the back RGB reflectance is defined with this line under "clear3_back"

0.07567 0.08418 0.08538

So the reflectances are roughly 8% which does obey the rule.

I am not completely clear what the other RGB values are but my guess is
that they modify the transmittivity defined under clear3_glass somehow.
Whenever I've manually defined glass I've used this form (from
"glazing.cal") instead, its just a little clearer to me.

mod BRTDfunc my_glazing
   10 rrho grho brho
       rtau gtau btau
       0 0 0
       glazing.cal
   0
   18 0 0 0
       0 0 0
       0 0 0
       FRRHO FGRHO FBRHO
       BRRHO BGRHO BBRHO
       RTAU GTAU BTAU

   where:
       FRRHO FGRHO FBRHO is front normal spectral reflectance
       BRRHO BGRHO BBRHO is back normal spectral reflectance
       RTAU GTAU BTAU is normal spectral transmittance

Hope this helps!

Zack

--
Zack Rogers
Staff Engineer
Architectural Energy Corporation
2540 Frontier Avenue, Suite 201
Boulder, CO 80301 USA

tel (303)444-4149 ext.235
fax (303)444-4304
www.archenergy.com

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

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

Hi again,

I'm afraid i don't have on hand a good reference for transmittance being the same either way through a glazing system. i like the explanation given on that web site. such a glazing system would probably break the third law of thermodynamics.

about your comparison below... the desktop radiance glass material file contains definitions of three different materials (clear3_glass, clear3_front, clear3_back). but, when DR creates the associated geometry files, they refer only to one of these materials (clear3_glass in this example). so, the created BRTDfunc materials are not used. that means that you will not get different reflectances from different sides of the materials, and that you will get a normal reflectance that doesn't necessarily match what is measured. the only way to get both the transmittance and reflectance you are after with the glass primitive is to alter the refractive index. you still don't get different reflectances for the different sides. but, normally, you are only concerned with one side of the glass (eg. creating visualisation or predicting illums inside and not outside), so just use the reflectance of that side (eg inside face).

the BRTDfunc you created with the inbuilt functions will likely work better than the glass primitive with an altered refractive index. the new forms (glass1.cal, glass2.cal and glaze.csh) will probably be better yet.

Phil.

[email protected] 16/01/2004 3:35:40 am >>>

Hello,

secondly, what does this do to the transmittance? physically, the transmittance going one way should be the same as the transmittance going the other way. should the (rtrns, gtrns, btrns) variables then be the same for both front and back panes? or does it not matter, as long as the product of their transmittances is equal to that of the combination.

This is true for a symetric glass composition (ie. single pane no low-e,
double pane, clear, no low-e) but not true for non-symetric glass which
is often the case for glass with low-e and glass with one of the panes
tinted. I believe the front and back transmittance is always really
close, (ie. it can't be 10% in one direction and 90% in the other) but
they can vary slightly. This is what desktop radiance and optics 5
definition method allows you to do.

finally, there are no functions built in. that means that there is no dependence on angle of incidence for either transmittance or reflectance. this is unrealistic, as transmittance generally reduces and reflectance generally increases with increasing angle of incidence.

I do think Desktop Radiance and Optics 5 both make use of the angular
transmittance function. This is part of the glass primitive. And in my
example, the BRTDFunc calls glazing.cal which provides the angular
transmittance function. Also, from my comparisions a while ago now, the
two methods gave me identical results. That is:

void glass clear3_glass
0
0
3 0.92189 0.98612 0.972
                       
void BRTDfunc clear3_front
10
    0.84636 0.90553 0.89251
    0.07428 0.08322 0.08556
    0 0 0
    .
0
9 0 0 0 0 0 0 0 0 0
   
void BRTDfunc clear3_back
10
    0.84636 0.90553 0.89251
    0.07567 0.08418 0.08538
    0 0 0
    .
0
9 0 0 0 0 0 0 0 0 0

is equivalent to:

void BRTDfunc clear3_glass
    10 rrho grho brho
          rtau gtau btau
          0 0 0
          glazing.cal
    0
    18 0 0 0
          0 0 0
          0 0 0
          0.07428 0.08322 0.08556
          0.07567 0.08418 0.08538
          0.84636 0.90553 0.89251

Also, you might want to check out glass1.cal and glass2.cal which is
part of the newest HEAD radiance release. It just distinguishes between
single pane and double pane which have different angular dependance
functions. Also, check out glaze.csh in the latest release, I was just
informed of this and have not checked it out personally, but I
understand it helps create these definitions.

So, I think the BRTDFunc methods can provide greater accuracy than the
glass primitive.

I hope I have not mistated any of this, anyone please correct me if so.

Regards,
Zack

···

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