Radiance and Ecotect

radiance-general
1

Can anyone help here?

I'm trying to model daylight factors on the vertical surface of a 10m high
warehouse rack. The reflectances are fairly low, 0.3 for the rack and 0.6
for the other surfaces. I've compared daylight factor results from Ecotect
and Radiance and found big differences in the magnitude and range of values
across the surface. In one case, Ecotect is giving values varying between
about 7.5% and 1%, the corresponding values from Radiance are between about
2.5% and 0.25%.

Ecotect's export to Radiance facility appears to incorrectly set the
transmissivity of the glazing material in the .rad output. Setting the
glazing transparency (I guess this is the same as tranmsmittance) in Ecotect
to 0.88 and using the 'clean' glass preset I would expect to see
transmissivity values of 0.96 in the .rad file output. I'm actually getting
values of 0.66. Even after correcting the .rad file I still find the output
values to be considerably smaller than those from Ecotect.

I'm using the high quality preset for Ecotect's calculation (further
increasing the quality appears to give slightly larger daylight values), and
an 'indirect reflections' value of 5 for Radiance, all other parameters are
the defaults for Ecotect's Radiance interface.

I suppose my real question here is which values should I trust? Perhaps
neither.

Nick

2

Nick,
Ecotect uses the split-flux method for daylight factor calculation. Radiance
calculates the daylight factor using backward raytracing. Thus, if the
surface reflectance values is accurate, the Radiance method should be more
accurate. There is the argument that if the surface reflectance value is
slightly off, then the raytracing method magnifies the error with each
bounce of light, where with the split-flux method, the error is not
magnified.

With regard to glass definition, Ecotect has been known not to transfer
glazing properties correctly into Radiance definition. If the discussion
forum was still online, then you would be able to follow a forum discussion
regarding the difficulties in translating Ecotect glass to Radiance
definition. The way I get around this problem is by using Optics (not
Window, as I had previously mentioned) from LBNL to create a Radiance
definition of a commercially available glass, and use the Optics definition
to replace the Ecotect glass. If you want a step-by-step method, email me
off the forum.

So, to answer your ultimate question, I generally trust Radiance results
over Ecotect results. But then again, this is just my opinion. The
difference between Ecotect and Radiance results are simply because of the
algorithm used. This does not necessarily mean that one is wrong and the
other is right.

Hope this helps

John

···

On 5/9/06 6:00 AM, "[email protected]" <[email protected]> wrote:
Can anyone help here?

I'm trying to model daylight factors on the vertical surface of a 10m high
warehouse rack. The reflectances are fairly low, 0.3 for the rack and 0.6
for the other surfaces. I've compared daylight factor results from Ecotect
and Radiance and found big differences in the magnitude and range of values
across the surface. In one case, Ecotect is giving values varying between
about 7.5% and 1%, the corresponding values from Radiance are between about
2.5% and 0.25%.

Ecotect's export to Radiance facility appears to incorrectly set the
transmissivity of the glazing material in the .rad output. Setting the
glazing transparency (I guess this is the same as tranmsmittance) in Ecotect
to 0.88 and using the 'clean' glass preset I would expect to see
transmissivity values of 0.96 in the .rad file output. I'm actually getting
values of 0.66. Even after correcting the .rad file I still find the output
values to be considerably smaller than those from Ecotect.

I'm using the high quality preset for Ecotect's calculation (further
increasing the quality appears to give slightly larger daylight values), and
an 'indirect reflections' value of 5 for Radiance, all other parameters are
the defaults for Ecotect's Radiance interface.

I suppose my real question here is which values should I trust? Perhaps
neither.

Nick

3

Nick,

John's recommendations are on the nose. It took me a while to come to the
same conclusions, but using Optic to create the .rad description for the
glazing is the simplest way to go. You just need to be certain that Optic is
providing the description for your complete window build-up. I vaguely
remember recent posts regarding Optic and this topic.

Given the calculation speed, and greater transparency of the settings using
radiance, I gave up on using Ecotects internal calcs a long time ago. With a
minimum of tweaks, you can also get Radiance to do VSC or Sky component
calcs too.

Hope that is useful

Nickd

Message: 2

···

Date: Tue, 09 May 2006 23:51:53 -0400
From: John An <[email protected]>
Subject: [Radiance-general] Re: Radiance and Ecotect
To: <[email protected]>
Message-ID: <C086DB99.186D%[email protected]>
Content-Type: text/plain; charset="US-ASCII"

Nick,
Ecotect uses the split-flux method for daylight factor calculation. Radiance
calculates the daylight factor using backward raytracing. Thus, if the
surface reflectance values is accurate, the Radiance method should be more
accurate. There is the argument that if the surface reflectance value is
slightly off, then the raytracing method magnifies the error with each
bounce of light, where with the split-flux method, the error is not
magnified.

With regard to glass definition, Ecotect has been known not to transfer
glazing properties correctly into Radiance definition. If the discussion
forum was still online, then you would be able to follow a forum discussion
regarding the difficulties in translating Ecotect glass to Radiance
definition. The way I get around this problem is by using Optics (not
Window, as I had previously mentioned) from LBNL to create a Radiance
definition of a commercially available glass, and use the Optics definition
to replace the Ecotect glass. If you want a step-by-step method, email me
off the forum.

So, to answer your ultimate question, I generally trust Radiance results
over Ecotect results. But then again, this is just my opinion. The
difference between Ecotect and Radiance results are simply because of the
algorithm used. This does not necessarily mean that one is wrong and the
other is right.

Hope this helps

John

On 5/9/06 6:00 AM, "[email protected]" <[email protected]> wrote:
Can anyone help here?

I'm trying to model daylight factors on the vertical surface of a 10m high
warehouse rack. The reflectances are fairly low, 0.3 for the rack and 0.6
for the other surfaces. I've compared daylight factor results from Ecotect
and Radiance and found big differences in the magnitude and range of values
across the surface. In one case, Ecotect is giving values varying between
about 7.5% and 1%, the corresponding values from Radiance are between about
2.5% and 0.25%.

Ecotect's export to Radiance facility appears to incorrectly set the
transmissivity of the glazing material in the .rad output. Setting the
glazing transparency (I guess this is the same as tranmsmittance) in Ecotect
to 0.88 and using the 'clean' glass preset I would expect to see
transmissivity values of 0.96 in the .rad file output. I'm actually getting
values of 0.66. Even after correcting the .rad file I still find the output
values to be considerably smaller than those from Ecotect.

I'm using the high quality preset for Ecotect's calculation (further
increasing the quality appears to give slightly larger daylight values), and
an 'indirect reflections' value of 5 for Radiance, all other parameters are
the defaults for Ecotect's Radiance interface.

I suppose my real question here is which values should I trust? Perhaps
neither.

Nick

------------------------------

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

End of Radiance-general Digest, Vol 27, Issue 7
***********************************************

4

John wrote : So, to answer your ultimate question, I generally trust

Radiance results over Ecotect results. But then again, this is just my
opinion. The difference between Ecotect and Radiance results are simply
because of the algorithm used. This does not necessarily mean that one
is wrong and the other is right.

I largely agree with what John wrote except that I would add that once
the material properties have been properly assigned in your Ecotect and
Radiance models, Radiance should be your benchmark. In case you are
close to a window, the simulation results of both programs should (and
do) largely coincide, but if you use an L-shaped room with a single
window in one of the 'end' walls and run an Ecotect DF simulation,
you'll see that as soon as a sensor does not 'see' the window any more,
the DF become zero which is of course physically incorrect. On the other
hand, Ecotect is somewhat faster than Radiance so for the 'easy' cases
it might be your preferred choice.

In the end you should think twice before using the daylight factor as
its capability to improve your design is somewhat limited. You might
want to look at dynamic metrics instead such as daylight autonomy,
useful daylight index, or continuous daylight autonomy using either SPOT
or DAYSIM. The new Daysim calculates all three metrics for you and you
can load the results back into Ectoect.

One useful feature that might be of interest is that the export function
from Ecotect to Radiance can be set so that Ecotect checks whether a
file <layer name>.rad exists in the "Material Export Directory" (Menu
option: FILE >>> USER PREFERENCES). If so, it uses the content of the
*.rad file as the material description for the layer in the Radiance
model.

Christoph

5

Thanks very much for your suggestions.

I looked briefly at outputting glazing data from Optics. It's probably the
way to go if I want to model a 'real' glazing type. I don't really
understand how the BRTDfunc is used; it sets the surface reflectance as well
as transmittance doesn't it? At the moment I'm only considering a
'notional' glazing so I'm just calculating the transmissivity and editing
the Radiance file output from Ecotect by hand. For typical glass types, how
much difference would using the BRTDfunc definitions make to my internal
daylight calculations?

As far as reliability of results is concerned, It seems Radiance (in the
hands of an expert user) is more capable of realistic results than Ecotect.
I've noticed that Ecotect's speed of calculation is influenced quite
significantly by the number of windows. It doesn't appear to make much
difference to Radiance's calculation time. I guess it's down to the
differences in calculation method but it does make me wonder how well
Ecotect handles models with many windows.

On the other hand, I find Radiance doesn't seem to behave quite as expected.
For example, for an equal glazed area I would expect fewer, larger
rooflights to give the same average %DF across the whole floor as more,
smaller rooflights. In this case, Ecotect seems to behave as I expect, but
Radiance suggests fewer, larger rooflights give the better average %DF. I
wonder why this should be. Maybe my assumption is incorrect.

Nick