unit of irradiance using epw as input

radiance-general
1

Hi, may I ask if the unit of irradiance is *Wh/m2* if I use the "epw
weather file + gendaylit/gencumulativesky + rtrace" method to get the
irradiance of a given point for a particular hour in a year or annual total
irradiance?

The three radiation variables in epw file are using the unit of Wh/m2, so I
assume the output of rtrace simulation using epw file as input should have
the same unit. Is it correct?

Thanks!

2

Hi David,

The EPW weather file should contain total solar irradiance values in watts/sq.meter. Radiance also reports values in watts/sq.meter, but only over the visible spectrum. Therefore, we normally include an "efficacy" calculation that converts the EPW's radiometric values into these semi-photometric values. (It's a bit confusing, but suffice it to say that Radiance's units have a relation to photometric units that employs a predefined efficacy/conversion value of 179 lumens/watt.)

If you want Radiance to actually give you total solar irradiance, then you should employ the -O1 option of gendaylit and gendaymtx. This should forgo the usual correction factors.

Cheers,
-Greg

···

From: David Davidson <[email protected]>
Subject: [Radiance-general] unit of irradiance using epw as input
Date: April 26, 2015 12:22:55 AM PDT

Hi, may I ask if the unit of irradiance is Wh/m2 if I use the "epw weather file + gendaylit/gencumulativesky + rtrace" method to get the irradiance of a given point for a particular hour in a year or annual total irradiance?

The three radiation variables in epw file are using the unit of Wh/m2, so I assume the output of rtrace simulation using epw file as input should have the same unit. Is it correct?

Thanks!

3

Dear Greg,

Thank you very much!

Hope you and the list can shed light on the following question about *Spectral
Fraction* value in Fluent solar load model based on your explanation:

In * Fluent *(CFD simulation software) menu for its *solar load
model*, the *spectral
fraction *is defined as the* fraction of incident solar radiation in the
visible part of the solar radiation spectrum*, which is calculated as *V
(the visible incident solar radiation)* divided by *V+IR (the total
incident solar radiation , visible plus infrared).*

I'd like to ask if the following calculation using Radiance is correct
(using Singapore's epw weather file, Mar-20, 3pm as an example):

*Step 1:* get the global horizontal irradiance (full spectrum, in wh/m2)

*gendaylit 3 20 15 -a 1.37 -o -103.98 -m -120 -W 447.0 378.0 -O 1 >
./gendaylit.skyoconv ./gendaylit.sky ./gendaylit_sky.rad >
./gendaylit_sky.oct*

*echo 0 0 0 0 0 1 | rtrace -w -h -I+ -ab 1 ./gendaylit_sky.oct | rcalc -e
"$1=$1*0.265+$2*0.670+$3*0.065" >
./gendaylit_global_horizontal_irradiance_in_Wh-m^2.txt*

The content of the gendaylit_sky.rad file, which describes the sky
geometry, is:

skyfunc glow sky_glow

0

0

4 1 1 1 0

sky_glow source sky

0

0
4 0 0 1 180

The output is: *771.593 (wh/m2)*,

* which is close to 799 wh/m2 (global_horizontal_radiation) reported in the
epw file.*
*Step 2: *
*get the global horizontal irradiance (integrated over the visible
spectrum, in wh/m2) *
The three commands are basically the same except that the [-O] option of
gendaylit is set as [-O 0]*:*
*gendaylit 3 20 15 -a 1.37 -o -103.98 -m -120 -W 447.0 378.0 -O 0 >
./gendaylit_visible.sky*

*oconv ./gendaylit_visible.sky ./gendaylit_sky.rad >
./gendaylit_sky_visible.oct*

*echo 0 0 0 0 0 1 | rtrace -w -h -I+ -ab 1 ./gendaylit_sky_visible.oct |
rcalc -e "$1=$1*0.265+$2*0.670+$3*0.065" >
./gendaylit_global_horizontal_irradiance_visible_in_Wh-m^2.txt*

The output is: 465.338 (wh/m2), which corresponds to global horizontal
illuminance of 465.388*179(lm/w) = 83304 (lux), which is close to 89300lux
(global_horizontal_illuminance) as reported in epw file.

So, the Spectral Fraction for Mar-20, 3pm is *465.338 / *

*771.593 = 0.603*
Is it correct?

···

On Mon, Apr 27, 2015 at 1:56 AM, Greg Ward <[email protected]> wrote:

Hi David,

The EPW weather file should contain total solar irradiance values in
watts/sq.meter. Radiance also reports values in watts/sq.meter, but only
over the visible spectrum. Therefore, we normally include an "efficacy"
calculation that converts the EPW's radiometric values into these
semi-photometric values. (It's a bit confusing, but suffice it to say that
Radiance's units have a relation to photometric units that employs a
predefined efficacy/conversion value of 179 lumens/watt.)

If you want Radiance to actually give you total solar irradiance, then you
should employ the -O1 option of gendaylit and gendaymtx. This should forgo
the usual correction factors.

Cheers,
-Greg

*From: *David Davidson <[email protected]>

*Subject: *[Radiance-general] unit of irradiance using epw as input

*Date: *April 26, 2015 12:22:55 AM PDT

Hi, may I ask if the unit of irradiance is *Wh/m2* if I use the "epw
weather file + gendaylit/gencumulativesky + rtrace" method to get the
irradiance of a given point for a particular hour in a year or annual total
irradiance?

The three radiation variables in epw file are using the unit of Wh/m2, so
I assume the output of rtrace simulation using epw file as input should
have the same unit. Is it correct?

Thanks!

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

4

Hi David,

This looks correct to me. There is probably an easier way to get the result based directly on the output of gendaylit, but I'm not familiar enough with this program to say which parameters in the file correspond to global irradiance.

Cheers!
-Greg

···

From: David Davidson <[email protected]>
Subject: Re: [Radiance-general] unit of irradiance using epw as input
Date: May 4, 2015 7:43:07 AM PDT

Dear Greg,

Thank you very much!

Hope you and the list can shed light on the following question about Spectral Fraction value in Fluent solar load model based on your explanation:

In Fluent (CFD simulation software) menu for its solar load model, the spectral fraction is defined as the fraction of incident solar radiation in the visible part of the solar radiation spectrum, which is calculated as V (the visible incident solar radiation) divided by V+IR (the total incident solar radiation , visible plus infrared).

I'd like to ask if the following calculation using Radiance is correct (using Singapore's epw weather file, Mar-20, 3pm as an example):

Step 1: get the global horizontal irradiance (full spectrum, in wh/m2)
gendaylit 3 20 15 -a 1.37 -o -103.98 -m -120 -W 447.0 378.0 -O 1 > ./gendaylit.sky

oconv ./gendaylit.sky ./gendaylit_sky.rad > ./gendaylit_sky.oct

echo 0 0 0 0 0 1 | rtrace -w -h -I+ -ab 1 ./gendaylit_sky.oct | rcalc -e "$1=$1*0.265+$2*0.670+$3*0.065" > ./gendaylit_global_horizontal_irradiance_in_Wh-m^2.txt

The content of the gendaylit_sky.rad file, which describes the sky geometry, is:
skyfunc glow sky_glow
0
0
4 1 1 1 0

sky_glow source sky
0
0
4 0 0 1 180

The output is: 771.593 (wh/m2), which is close to 799 wh/m2 (global_horizontal_radiation) reported in the epw file.

Step 2: get the global horizontal irradiance (integrated over the visible spectrum, in wh/m2)

The three commands are basically the same except that the [-O] option of gendaylit is set as [-O 0]:
gendaylit 3 20 15 -a 1.37 -o -103.98 -m -120 -W 447.0 378.0 -O 0 > ./gendaylit_visible.sky

oconv ./gendaylit_visible.sky ./gendaylit_sky.rad > ./gendaylit_sky_visible.oct

echo 0 0 0 0 0 1 | rtrace -w -h -I+ -ab 1 ./gendaylit_sky_visible.oct | rcalc -e "$1=$1*0.265+$2*0.670+$3*0.065" > ./gendaylit_global_horizontal_irradiance_visible_in_Wh-m^2.txt

The output is: 465.338 (wh/m2), which corresponds to global horizontal illuminance of 465.388*179(lm/w) = 83304 (lux), which is close to 89300lux (global_horizontal_illuminance) as reported in epw file.

So, the Spectral Fraction for Mar-20, 3pm is 465.338 / 771.593 = 0.603

Is it correct?

On Mon, Apr 27, 2015 at 1:56 AM, Greg Ward <[email protected]> wrote:
Hi David,

The EPW weather file should contain total solar irradiance values in watts/sq.meter. Radiance also reports values in watts/sq.meter, but only over the visible spectrum. Therefore, we normally include an "efficacy" calculation that converts the EPW's radiometric values into these semi-photometric values. (It's a bit confusing, but suffice it to say that Radiance's units have a relation to photometric units that employs a predefined efficacy/conversion value of 179 lumens/watt.)

If you want Radiance to actually give you total solar irradiance, then you should employ the -O1 option of gendaylit and gendaymtx. This should forgo the usual correction factors.

Cheers,
-Greg

From: David Davidson <[email protected]>
Subject: [Radiance-general] unit of irradiance using epw as input
Date: April 26, 2015 12:22:55 AM PDT

Hi, may I ask if the unit of irradiance is Wh/m2 if I use the "epw weather file + gendaylit/gencumulativesky + rtrace" method to get the irradiance of a given point for a particular hour in a year or annual total irradiance?

The three radiation variables in epw file are using the unit of Wh/m2, so I assume the output of rtrace simulation using epw file as input should have the same unit. Is it correct?

Thanks!

5

Hello Greg and David,

I am currently running simulations for the full solar spectrum. I understand, that (as written above from David), the following calculation is applied when considering the visible spectrum: “$1=$1*0.265+$2*0.670+$3*0.065” (in this case when using the -O0 option, only considering the visible spectrum).
Though, I’m trying to understand, how the same spectral weighting could be applied to full spectrum (as shown by David in the above post with using the -O1 option). I thought, that these coeffiecients (0.265, 0.670, 0.065) were calibrated Radianc-internally only for the visible spectrum, and the -O1 option would already return the correctly weighted results. Do you think in the wrong direction here?

Best,
Rafael

6

Hi Rafael,

The coefficients David gave are to derive photopic response to RGB colors. To get something closer to the radiometric average, just average the 3 channels together. There is also the 179 lumens/watt factor, which is needed for converting to luminance or illuminance but not radiometric units.

I hope this helps.
-Greg

7

Hi Greg,

Thank you very much for your explanation. Yes, I understand that the 179 conversion factor is for lumens/watt. As I’m interested in the extra-portion of information, the spectral discretization brings with it (for spectrally differentiated material-reflections and illuminants), I would assume that for simulations for the visible spectrum of radiation, using the above-mentioned coefficients, even for usage of radiometric metrics would be the most accurate.

In the current setup I have the SPD data of the LED light sources. I used MGF2RAD to derive the correct r-g-b weights of that light source. Then I was wondering, if I should scale the three values (which by default together give 1, for material reflectances), so that the sum of those r-g-b weights MGF2RAD would be 3? I would then insert these scaled r-g-b values into the r-g-b fields of the .rad file of the light source.
I thought that by doing so, I would get the extra benefit of having the correct weight applied to each channel, combining it with the mentioned coefficients of David’s post (for converting values back from the RADIANCE color space to photometric/radiometric values).

And that’s where I concluded that those r-g-b coefficients would only work for irradiance conversion based on simulations within the visible spectrum and that simulation results for the entire solar spectrum (like -O1 gendaxmtx information) would be slightly biased when using the same coefficients.

Best regards,
Rafael

8

Hi Rafael,

If you have the spectral power distribution for the LED source, you should be able to enter that directly to Radiance using the new “spectrum” type or similar. You need to be careful about normalization, but the current (6.x) version of Radiance fully supports spectral rendering and averaging to compute irradiance over whatever spectrum you specify.

The use of the photopic coefficients is not recommended if you want to match radiometric data. Again, averaging the component coefficients (RGB or spectral values) is how Radiance expects you to integrate its output. For RGB output, you would use (R+G+B)/3 to compute irradiance out of rtrace -I.

Cheers,
-Greg

9

Hi Greg,

Thank you, I will do, as you suggested.
The idea was to check how much better the hyperspectral approach gets compared to the 3-channel one. That’s why I was trying to push the standard 3-channel simulations to its limits. Does that normalization procedure as described above make sense (r+g+b = 3, after receiving r-g-b weights through -cspec)?

Best, Rafael

10

Hi Rafael,

Yes, a uniform irradiance spectrum would yield RGB=(1,1,1) using MGF “cspec” with a 1.0 multiplier. You may try it to check.

For uniform (grey) surfaces, it doesn’t matter what color space you use or how many spectral samples you simulate over the visible range. It only matters for non-grey color interactions, whether that is a bluish sky through greenish glass onto a reddish carpet or different spectral paints on various surfaces.

Best,
-Greg