DF calculated in Radiance and Relux

Per_Haugaard · March 29, 2008, 7:21am

Dear Radiance Users,

In order to double check the DF which I calculated in Radiance I also calculated these for the exact same room with all the same settings (reflectance, transmittance, dimensions, areas of the pane, reference plance, reductions factor = 1 in Relux and city (Oslo)) in Relux .

Surprisingly since Relux is based on Radiance the DF which I calculated in Radiance are about 1.7 higher than the ones calculated in Relux (DF calculated in Radiance = 17 and the ones calculated in Relux = 15,3 0.6m from the facade).

I have used the folowing gensky and rtrace command in Radiance:

gensky
6 21 12 -c -a 59.9 -o -10.8 -m -15.00 -B 55.866

rtrace -w -h -I+ -ab 20 -aa 0.1 -ad 512 -as 64 -ar 128 DF_OfficeLL.oct
< DFcoordinates.inp | rcalc -e
'$1=($1*0.265+$2*0.670+$3*0.065)*179/10000*100' > OfficeLL.log

The same issue arrise when I include external shading (trans) to my model?

Does anybody has any insight of the calculation settings in Relux or have I made a beginners mistake of my input in Radiance?

I appreciate any help and comments on this

Best regards

Per Haugaard

···

______________________

Per Haugaard,
Griffenfeldsgade 33, 2.tv
2200 København N
Denmark

mobil: +45 26 39 06 40
E-mail: perhaugaard@yahoo.dk

Find din nye laptop på kelkoo.dk. Se de gode tilbud her - http://dk.yahoo.com/r/pat/mm

Thomas_Bleicher1 · March 29, 2008, 9:42am

Dear Radiance Users,

In order to double check the DF which I calculated in Radiance I also calculated these for the exact same room with all the same settings (reflectance, transmittance, dimensions, areas of the pane, reference plance, reductions factor = 1 in Relux and city (Oslo)) in Relux .

Surprisingly since Relux is based on Radiance the DF which I calculated in Radiance are about 1.7 higher than the ones calculated in Relux (DF calculated in Radiance = 17 and the ones calculated in Relux = 15,3 0.6m from the facade).

Are you sure Relux is based on Radiance? According to my knowledge only the visualisation
add-on 'Relux Vision' is Radiance based. The calculation part uses another approach (radiosity
I think).

Anyway a difference of 70 % is something to worry about. I have compared Dialux with
Radiance in the past and the results differed by 10-15%. Have you double checked your
geometry?

I have used the folowing gensky and rtrace command in Radiance:

gensky 6 21 12 -c -a 59.9 -o -10.8 -m -15.00 -B 55.866

rtrace -w -h -I+ -ab 20 -aa 0.1 -ad 512 -as 64 -ar 128 DF_OfficeLL.oct < DFcoordinates.inp | rcalc -e '$1=($1*0.265+$2*0.670+$3*0.065)*179/10000*100' > OfficeLL.log

'-ab 20' - impressive! I usually don't go higher than 6. If your afraid of missing some
tiny value in the third decimal you should use '-av' to set a background value. One
way to find this is described in RwR using rvu, another is to check the output of the
gensky program for the 'Ground ambient level' and adjust that for your indoor situation
(1-2% of the outdoor level).

If you have a detailed geometry (small windows and openings) you should also
increase '-ad' (4096), '-ar' (512) and '-as' (256). The speed penalty here is less than
that of an increase in '-ab' so I tend to be pretty bold about these.

I have noticed that Radiance seems to add to your rtrace results the higher you
set your '-ab'. I'd have expected the values to approach an upper limit where a further
increase makes no noticeable difference. In reality I could add a significant percentage
to the DF simply by increasing the '-ab' value. So I assume that your extreme
difference might be a result of the very high '-ab' setting.

Regards,
Thomas

···

On 29 Mar 2008, at 07:21, Per Haugaard wrote:

Per_Haugaard · March 31, 2008, 9:46am

Dear Thomas and other Radiance users,
Thank you very much for your reply. I used the following rtrace command instead:

rtrace -w -h -I+ -ab 6 -aa 0.1 -ad 1026 -as 256 -ar 512 DF_OfficeLL.oct < DFcoordinates.inp | rcalc -e '$1=($1*0.265+$2*0.670+$3*0.065)*179/10000*100' > OfficeLL.log
which reduced the calculated DF in radiance to pretty much match the ones calculated in Relux.
Then I changed the transmittance for the glazing from 0,72 to 0,58 and made a new octree and re-calculated the DFs. I here after used the above rtrace command.

However, this had no influence on the calculated DF at all. I have also tried to alter the '-ad', '-as' and the '-ar' with limit results.

How can that be?
Below I have listed my window.rad command:

···

#
# An emissive window
#

# synlig rudetype til illum
void glass window_glass_58
0
03 .63 .63 .63
# hullet i vaeggen
window_glass_58 polygon window5
0
0
12
               8.116 16.8 0.9
               11.1 16.8 0.9
               11.1 16.8 2.5 8.116 16.8 2.5
window_glass_58 polygon window6
0
0
12
               5.133 16.8 0.9
               8.066 16.8 0.9
               8.066 16.8 2.5
               5.133 16.8 2.5

window_glass_58 polygon window7
0
0
12
               2.1 16.8 0.9
               5.081 16.8 0.9
               5.081 16.8 2.5
               2.1 16.8 2.5

The data used for the pane with a transmittance of 0,72 is similar exept from void glass window_glass wich is set to 3 0.78 0.78 0.78.

Since there should be nowhere else where I should change the settings for the zone when I alter the transmittance, how can the daylight
factors be similar for to dirrerent types of panes???
I apreciate any input on this.

Best regards
Per Haugaard
______________________
On 29 Mar 2008, at 07:21, Per Haugaard wrote:

Dear Radiance Users,

In order to double check the DF which I calculated in Radiance I
also calculated these for the exact same room with all the same
settings (reflectance, transmittance, dimensions, areas of the
pane, reference plance, reductions factor = 1 in Relux and city
(Oslo)) in Relux .

Surprisingly since Relux is based on Radiance the DF which I
calculated in Radiance are about 1.7 higher than the ones
calculated in Relux (DF calculated in Radiance = 17 and the ones
calculated in Relux = 15,3 0.6m from the facade).

Are you sure Relux is based on Radiance? According to my knowledge
only the visualisation
add-on 'Relux Vision' is Radiance based. The calculation part uses
another approach (radiosity
I think).

Anyway a difference of 70 % is something to worry about. I have
compared Dialux with
Radiance in the past and the results differed by 10-15%. Have you
double checked your
geometry?

I have used the folowing gensky and rtrace command in Radiance:

gensky 6 21 12 -c -a 59.9 -o -10.8 -m -15.00 -B 55.866

rtrace -w -h -I+ -ab 20 -aa 0.1 -ad 512 -as 64 -ar 128
DF_OfficeLL.oct < DFcoordinates.inp | rcalc -e '$1=($1*0.265+
$2*0.670+$3*0.065)*179/10000*100' > OfficeLL.log

'-ab 20' - impressive! I usually don't go higher than 6. If your
afraid of missing some
tiny value in the third decimal you should use '-av' to set a
background value. One
way to find this is described in RwR using rvu, another is to check
the output of the
gensky program for the 'Ground ambient level' and adjust that for
your indoor situation
(1-2% of the outdoor level).

If you have a detailed geometry (small windows and openings) you
should also
increase '-ad' (4096), '-ar' (512) and '-as' (256). The speed penalty
here is less than
that of an increase in '-ab' so I tend to be pretty bold about these.

I have noticed that Radiance seems to add to your rtrace results the
higher you
set your '-ab'. I'd have expected the values to approach an upper
limit where a further
increase makes no noticeable difference. In reality I could add a
significant percentage
to the DF simply by increasing the '-ab' value. So I assume that your
extreme
difference might be a result of the very high '-ab' setting.

Regards,
Thomas

______________________

Per Haugaard,
Griffenfeldsgade 33, 2.tv
2200 København N
Denmark

mobil: +45 26 39 06 40
E-mail: perhaugaard@yahoo.dk

Trænger du til at se det store billede? Kelkoo giver dig gode tilbud på LCD TV! Se her http://dk.yahoo.com/r/pat/lcd

Thomas_Bleicher1 · March 31, 2008, 7:34pm

Dear Thomas and other Radiance users,

Thank you very much for your reply. I used the following rtrace command instead:
rtrace -w -h -I+ -ab 6 -aa 0.1 -ad 1026 -as 256 -ar 512 DF_OfficeLL.oct < DFcoordinates.inp | rcalc -e '$1=($1*0.265+$2*0.670+$3*0.065)*179/10000*100' > OfficeLL.log

which reduced the calculated DF in radiance to pretty much match the ones calculated in Relux.

Still doesn't explain why a very high '-ab' value results in a DF that doesn't look right.

I always thought that Radiance would produce more accurate results the longer it runs.

Then I changed the transmittance for the glazing from 0,72 to 0,58 and made a new octree
and re-calculated the DFs. I here after used the above rtrace command.
However, this had no influence on the calculated DF at all.

Without knowing the geometry it's hard to say. If your windows are the only openings
you should see a reduction in the DF. It's also possible that your window polygons are
not where you expect them to be. If you're on a Unix system you can use rvu to get
a visual impression of the room and with the 't' keystroke you can trace a ray to it's
intersection with the geometry. It tells you which polygon and material you've hit.
You can use it to verify your material changes are now part of the new octree.

Under Windows you can render a picture to control that the windows are in place.
To get an equivalent of the 't' you could use rtrace with a carefully selected origin and
direction and a modified output option.

I have also tried to alter the '-ad', '-as' and the '-ar' with limit results.

These options will only have an effect if your geometry is very detailed and you
might miss a window with the initial '-ad' setting. Given the window polygons
below I'd say you have a few very large windows, so you don't have to change
these values.

Hth,
Thomas

···

On 31 Mar 2008, at 10:46, Per Haugaard wrote: