Hello,
I've forwarded some discussion I was part of regarding Radiance and
Optics5. I apologize its a little messy. Basically, I concluded that
Optics 5 does produce fairly accurate glazing definitions for most
applications. Anymore accuracy, particularly accuracy at higher
incidence angles, would require the use of a modified glazing.cal
function.
Hope this helps!
Zack
>Mike Rubin wrote:
> > > O5 Team:
> > > I just want to make sure this doesn't get lost
> > > and appears in the bug list at the appropriate
priority level.
> > >
> > > I talked to Zack Rogers about these two problems:
> > >
> > > 1) The "RGB transmissivities" should be easy
> > > to check. As I understand they are simply the
transmittance
> values at 3
> > > particular wavelengths in the visible or perhaps
averaged over 3 bands
> > > in the visible. If this is correct, then the numbers
must be wrong as
> > > Zack says. I cannot check it however because at the
moment
> Optics5 gives
> > > an error when attempting to export a Radiance file.
Saill
> could this be
> > > broken because of the change to the spectral
averaging dll or is it a
> > > completely self contained method specified by Chas?
> > > Note: I am not saying to jump right in and fix it,
> > > just give some idea of what the problem might be and
how much
> effort to
> > > fix for now.
> > >
> > > 2) The angle dependence problem I cannot figure out
without more
> > > information. Chas specifed the "glass material"
output type
> whose angle
> > > dependece is interpreted by the "glazing.cal"
function. Presumably he
> > > had good reason for doing so, although Zack says
Chas is aware of the
> > > deficiencies of this method with regard to angle
dependence. I get the
> > > impression that a new .cal function would be needed
in
> Radiance to do it
> > > better. I don't really know what all this means or
how it is done now.
> > > If Chas or John could explain, maybe we could
suggest a solution. Mike
> > >
> Greg Ward wrote:
>
> > I have little to add except that Zack seems to have
made a good analysis
> > of the situation. The three values given to the
"glass" material type
> > are the transmission (internal only, ignoring
reflections) of the
> > glazing at normal incidence. The values may be
averaged over wavelength
> > bands corresponding roughly to red, green, and blue
phosphors. (It is
> > really impossible to get the spectral part of things
exactly right in
> > Radiance, so this part is approximate.) The final
values should be
> > between zero and one for each primary. A value of 1.0
corresponds to a
> > lossless glazing, which does not mean the
transmittance is 100% -- it
> > will be closer to 90% (depending on the index of
refraction) due to
> > reflections at the first and second interfaces.
> >
> > As for #2, Zack is correct that a .cal file is
preferable to relying on
> > the "glass" primitive for coated and multiple
glazings. The original
> > "glazing.cal" was based on an older version of the
Window program, and
> > is long overdue for a rewrite. If a new .cal file is
written, I would
> > like to check it over, and it would be a good idea to
use the "BRDTfunc"
> > primitive type rather than "glass".
> >
> > -Greg
Rebecca Powles wrote:
I have been in communication with Chas and Judy Lai on
this - Chas sent me
some very helpful information which helped me to decide
that there was an
error in the way we were calculating these numbers, and
the code has since
been changed by myself and Richard Versluis based on Chas'
comments. The
updated code will be used in the next version of Optics5
which we expect to
release next week.
I am not sure how to test the results except by forwarding
some examples to
you for checking and comment in this email.
Please review the attached files and send any comments to
me.
The files represent:
Single Glazings
6mm clear glass : CLEAR_6.rad
applied film on 6mm bronze glass :
SLCLBZ06.rad
Multiple Glazings
6mm clear glass / 6mm clear glass :
Clear6Double.rad
6mm bronze glass / 6mm low-e glass :
BronzeLowe.rad
Just for the record, here is how we are doing the
calculations now for
single glazing layers and glazing systems (multiple
glazing):
First we determine the color co-ordinates of the T/Rf/Rb
spectra using an
equal-energy ('pure white') source and the CIE 1964 10
degree observers from
ASTM E308
These color co-ordinates are transformed into a RGB gamut
with:
Red x,y = 0.640, 0.330
Green x,y = 0.290, 0.600
Blue x,y = 0.150, 0.060
White x,y = 0.333, 0.333
Finally, two types of Radiance primitive are output - a
glass primitive and
a BRTDfunc primitive.
The glass primitive has 'adjusted' transmittance RGB
values since this
primitive uses as input the transmittance without
interreflections taken
into account. The effect of interreflections is
subtracted from the total
transmittance assuming a constant index of refraction of
according to:
tn =
(sqrt(.8402528435+.0072522239*Tn*Tn)-.9166530661)/.0036261119/Tn
Both primitives also adjust the RGB values so that they
scale to the total
photopic transmittance or reflectance by using the
following scaling factor:
1/(0.265074126*R+0.670114631*G+0.064811243*B) applied to
each RGB value.
The .rad file output by Optics5 contains both primitives
regardless of
whether a single glazing layer or a glazing system has
been selected. The
glass primitive is probably not appropriate for glazing
systems (I guess).
Some reference information is output as a header (comment
lines) in the
Radiance file. For single layers, this includes the
filename, product name,
NFRC ID, Glazing Type, Coated Side and Thickness of the
layer as well as the
Photopic (visible) averaged T/Rf/Rb values. For glazing
systems, some of
these fields do not apply and they are left blank. For
glazing systems, the
thickness is set to -999.
To use (in the next version of Optics5):
Load one or more layers into the glazing system.
Either select a layer by pressing one of the buttons #1,
#2, etc.
or
Select the glazing system by pressing the button 'Glazing
System'
Choose the menu option File|Create Radiance File, select a
filename and
press OK.
Rebecca Powles
Visiting Postdoctoral Fellow Physicist
Lawrence Berkeley National Laboratory
1 Cyclotron Rd, MS 2-300
Berkeley CA 94720
RCPowles@lbl.gov
ph: 1 (510) 486 7802
fax: 1 (510) 486 6099
Zack Rogers wrote:
Rebecca,
I took a second to look over these rad files and for the
most part they seem accurate. A few questions and comments (anyone,
please correct
me if I am wrong):
1) The incidence angle data in Optics 5....does this data
come from actual measurements performed by the manufacturer or does
Optics 5 come
up with these values using an equation? I am assuming this
data is true, measured by the manufacturer, and am using it to compare
the results of
these rad files. What are the options of
assymetric...symmetric etc. for? It seems like this setting doesn't
change any of the characteristics.
A side note: it would be a nice feature to have an option
that allows the user to view the data for all incident angles in a
single graph. Like a
toggle that would switch from the view of all wavelengths at
a single incident angle to all incident angles at either a given
wavelenth, or showing
three typical RGB wavelengths or just an overall average of
all wavelengths.
2) As I'm sure you know, the normal transmittance seems to
be right on with Optics5 normal transmittance. One question....how come
we are
making the radiance definition in 3 parts like so:
void glass CLEAR_6_glass
0
0
3 0.934 0.973 0.957
void BRTDfunc CLEAR_6_front
10
0.077 0.081 0.082
0.858 0.893 0.879
0 0 0
.
0
9 0 0 0 0 0 0 0 0 0
void BRTDfunc CLEAR_6_back
10
0.077 0.081 0.082
0.858 0.893 0.879
0 0 0
.
0
9 0 0 0 0 0 0 0 0 0
instead of simply one definition of:
void BRTDfunc CLEAR_6_2_glass
10 rrho grho brho
rtau gtau btau
0 0 0
glazing.cal
0
18 0 0 0
0 0 0
0 0 0
0.077 0.081 0.082
0.077 0.081 0.082
0.858 0.893 0.879
As far as I can tell these two radiance definitions give
very similar results. The second definition, besides being a little
more compact, then gives
the option to use a different .cal function if necessary.
Below I tried to look in more detail at whether or not a different .cal
function is
necessary.
3) As far as using the same glazing.cal function for all
glazings, it seems like for the most part this is okay since most
glazings have very similar
trans vs. incident angle characteristics. See the attached
graph.
The glazing .cal functions, both the clear and the bronze,
are supposed to be in bold. (I apologize if this is illegible. Let me
know and I can just
send you the excel spreadsheet I created this in) I tried
to get both extremes, from the quickest cut-off (Double bronze), to the
slowest
(Starphire) although I didn't spend too much time looking
for these extremes. From the graph, the two glazing.cal functions do
not grossly
mispredict ( I know thats not a word but you get the point)
the transmittance. The worse case possibly occuring at 60deg incidence
where it
could mean a difference of at most +/- 250 fc. It seems
like a specialized .cal function for each glazing might be overkill but
it seems like
modifying the glazing.cal file to include another one or two
functions might do the trick. Right now, the break between the clear
function and the
bronze function is a normal trans=0.645. It does seem that
the relationship between normal transmittance and incident angle
drop-off is pretty
linear, that is the lower the normal trans the quicker the
drop-off. And so, for example, another break could be added with trans
= 0.5 and then
trans = 0.35 (or whatever break will give the best fits).
Then the same glazing.cal function could be used for all glazings and
the error would be
reduced significantly. And I guess that if there are
coatings or films with very different trans vs. angle characteristics, a
special .cal function
may still be in order. I am personally hoping to find a
glazing system that doesn't significantly drop in trans until 80deg or
so.
I hope this all makes sense. Please let me know if you have
any questions/comments/clarifications etc. Overall, I am really excited
to have this
new functionality in Optics 5. For any Radiance modeler, it
gives the ability to compare numerous glazing systems with the assurance
that the
results are relatively accurate. Any update on when the new
version will be released?
Regards,
Zack
Charles Ehrlich wrote:
Hi all,I think that there is one other person who needs to be
involved in this discussion. It was Eleanor Lee who originally brought
to my attention the need for
better accuracy glazings in Radiance. While most uncoated
glazings have similar angular transmittance properties, I think it was
coated glazings that introduce
angle-dependence issues. I think Eleanor can contribute some
greater understanding of the issue to this discussion.The reason I
created two BRTDfuncs is
because the reflectance is different for front and back. While
it is reasonable to define a different front/back reflectance within a
single BRTDfunc, the other
limitation we are working with is Desktop Radiance and AutoCAD.
The orientation of the glazing determines the direction in which the
light flows into a building,
i.e., from outside to inside. This is impossible to reconcile
with DesktopRadiance's need to use glazing orientation to define the
direction of the flow of
illumination. Currently Desktop Radiance only uses the first
entity type--glass. The other two are for future use. But regardless,
two will be necessary to work
within the AutoCAD framework.I am not sure I follow Zack's
suggestion of creating a different Radiance file for different types of
glazings depending upon normal
transmittance. Is it true that the fall-off is a function of the
transmittance? It also does not really make sense on a conceptual level
to transfer the burden of
accurately rendering these glazings (Radiance) upon the modeling
tool (Optics). A single glazing.cal file makes the Optics-->Radiance
interface much cleaner
and straightforward. So, I advocate an improved glazing.cal
file.There has already been work done on what the new glazing.calf
function should look like based
upon total solar transmittance. Perhaps Eleanor remember's the
citation?Thank you Zack for your excelent work! I would like to get a
copy of your Excel
spreadsheet.-Chas
Zack Rogers wrote:
Chas, thanks for your comments! I've attached the excel file I
used for my comparison. It is crudely notated so please ask me if there
are any questions.
I'm still not totally clear why two BRTDfuncs are needed since you
can still specify different front and back reflectances with one but I
am not at all familiar
with the Desktop Radiance to ACAD relation.
As far as the relationship between transmittance and fall-off; I
just noticed a trend in the glazings I analyzed that showed this
somewhat linear relationship. The
fact that the glazing.cal function uses the clear equation for
0.645 and the Bronze for Trans<0.645 also suggested to me that
atleast someone else thought
that there was this relationship. My suggestion was to still
maintain a single glazing.cal function but with the addition of another
one or two divisions, ie Trans
<0.5 and Trans <0.35, if this relationship proves to be true
enough. But I agree, it seems like a coating or film could have a major
impact on the angle-dependant
transmittance and in those cases a special *.cal file would be
necessary to accurately model in Radiance. By the way, if anyone knows
of such coatings, in
particular coatings that increase or maintain transmittance at high
incident angles, I would be very interested. This could be very useful
in daylighting
applications.
Regards,
Zack
···
--
Celebrating 20 Years of Improving Building Energy Performance
Zack Rogers
Daylighting Designer/Engineer
Architectural Energy Corporation
2540 Frontier Avenue, Suite 201
Boulder, CO 80301 USA
tel (303)444-4149 ext.235
fax (303)444-4304