[Radiance-general] R: Radiance-general Digest, Vol 84, Issue 6

[Thomas Bleicher]

Hello again.

On Mon, Feb 7, 2011 at 7:50 AM, caria.roberto at tiscali.it
<caria.roberto at tiscali.it> wrote:
> R: I know I can get the external illuminace from epw file but no all
> locations have these data values (if you open ITA_Cagliari Elmas
> you can see it is that)

Well, you mentioned London in your first email. The EPW
file for London contains illuminance data.

> I think if I consider the Illuminance values taken from epw file
> I can use gensky and insert them as -B and -R parameters
> (dividing them to 179, before) and add +s when I have both
> values and -c when I have only the diffuse Illuminance.
> (Can you confirm it?)

For Sicily +s might actually be appropriate :). If you have some
cloudiness you can also use +i and -i. Sometimes you also get
a "sky cover" index in the EPW file which you could use to pick
the right sky model.

>> Now the interesting bit:
>>
>> If you look at your results or the global horizontal values
>> you will see that the Daysim values are roughly 25%-30%
>> higher than those calculated in Radiance. However, they
>> are consistently higher (which is good).
>
> Why is good to get higher values?

Higher (or lower) values as such make no difference. But if
the difference is the same for various sky situations you can
assume that it does not depend on your input and is due to
the different algorithms the two applications use. Is says
nothing about the correctness of each program, though. You
have to work that out by comparing the calculated results
with real measurements.


> I did it and there is a small difference between epw and
> daysim data illuminance (about 2-3%), so I suppose I can
> use illuminance values provided by Daysim.
>
> So, using the same input (Irradiance values comes from
> epw file) the sky created by Daysim works better than a
> sky generated by gendaylit?

To my knowledge Daysim also uses gendaylit (Perez model).
They may adjust their values to match the recorded input in a
second step which you would have to do manually if you were
to do the same calculations in Radiance.

> Iin which conditions can I use gendaylit?

When you use Radiance and gendaylit you will find that in
early morning and evening hours the results from gendaylit
are way off. This is either a shortcoming in the algorithm or
just a problem of the implementation. If that happens you
can use gensky for this particular situation; it's better than
dropping the record completely from the evaluation.

Side note: I remember reading something about this being
a bug in gendaylit that was fixed with a patch. But that was
in the 90s and I don't know if this patch is included in the
commonly available gendaylit source code.

> So, if i want to show particular illuminance condition in
> radiance (i.e. 12:00 on 21 March) by mean a falsecolor
> image with illuminace values, is better to use gensky
> with +s or -c depending on if there is direct and diffuse
> illuminance or only diffuse one, using the illuminance data
> values from daysim or epw (if I have it)?

There are two topics here:

1) How do you get illuminance values if the EPW file
   doesn't have them.

2) Which sky model (and options) do you use to match
   a given set of irradiance/illuminance values?

First point:

The only reliable way to get illuminance is to measure it
on site under the real world conditions. Point. If you don't
have that you need the sky efficacy to convert irradiance
to illuminance but that changes for any given location and
sky condition!

Therefore if you don't have measured illuminance values
you have to assume a reasonable sky efficacy. The built-in
value in Radiance is 179 lm/W. This value is only used at
the end to convert calculated irradiance to illuminance (the
magic "*179" in the command line).

If you are more interested in a relative comparison over the
year and not so much in absolute illuminance values at a
particular date you don't have to worry much. You can still
compare the illuminance images even if they are say 15%
too bright.

However, if you want to calculate the energy savings for a
daylight liked artificial lighting system that switches off at
500 lux you'd better get a set of real world measurement to
have some data you can use as calibration points.

Now the second point:

No. If you use Radiance you have to check that the irradiance
from your sky model matches that of the EPW for every given
sky. It does not matter if you use gendaylit or gensky. Both
work with irradiance input ("-P" for gendaylit and "-r" and "-B"
for gensky). Gendaylit is commonly used because it creates
a more sophisticated sky distribution than gensky. You also
have to chose to add a sun or not (and which model you want
to use in the case of gensky.

You now calculate the irradiance of a single point on a plane
and compare that with the "global horizontal irradiance" from
the EPW file. If the two values differ this is just a sign that
the sky model does not fit perfectly the actual atmospheric
conditions. Modify your sky description to compensate for
the difference and check that the global irradiance matches
your calculations.

If you don't have illuminance values you are set. Repeat the
above process for all dates of the year to find matching sky
descriptions which you can use in a Radiance simulation.

If you do have illuminance values you calculate the sky efficacy
from the ratio of the irradiance and illuminance values and use
this factor to convert the rendered irradiance picture to a falsecolor
illuminance picture (or anywhere else where you want to calculate
illuminance values).

If you have a good estimate for the sky efficacy you should
also use that instead of the built in value.

> If my weather file has not illuminance data values, I must
> use Daysim to generate the illuminace external values
> and use them in Radiance to make analysis (by means gensky);
> I can`t use gendaylit (starting from Irradiance value provided
> by the weather file) because it provides illuminance values
> are not correct compared to measured ones.
> Can you confirm it?

No. If you think the results of Daysim are accurate enough
then you should use Daysim to calculate your images if it
is not too limited for your purpose. It will save you all the
steps mentioned above and for a simulation covering the
whole year it might also be faster.

There is no need to use Daysim to calculate illuminance
values. You (and I) have no idea which conversion factors
are used to get to these values.

If you use Radiance you can (and should) modify the sky
descriptions generated by gensky or gendaylit until they
match best the available data. Then you can use the sky
in a simulation and be confident that the results will match
the reality within the limits of the simulation.

The replication of real world atmospheric conditions is quite
tricky. Don't expect your results to be too accurate. For the
type of evaluation that's usually done on the basis of climate
data files this inaccuracy does not matter much (cumulative
or relative evaluations). But don't depend too much on the
results of a single calculation out of a whole year's worth of
data.

Regards,
Thomas