Measuring materials

Recently, Visarc Jack said:

I think that Rob G. could give everyone a really superb summary on how to setup a measuring environment for glazing materials.

Then he went on to say:

Depending on the application is might be worth checking out glaze.csh in the latest release. This calculates models for both single and double surface glazing materials. I am not sure how it compares to in terms of the accuracy of the underlying model. Note however that glaze.csh is built around a handful (5 or 6) sets of reflectance and transmittance values as it "database". Some of these were actually measure by me and looking back on it, I would prefer to measure the "pvb" materials under a more precise protocol than I did originally (again Rob G. would be the real resource here).

Ha. Well, Greg's the real resource here. All I did was come up with a cockamamie scheme to use a luminance meter and an illuminance meter to try and figure out a diffuse glass sample we had; Greg's the one who provided the brains to both verify that my scheme would actually work, and to implement the procedure. The materials we measured were also mundane enough to be modeled with just a trans; we didn't need to use BRTDfunc. But here's what we did:

I had a couple of glass samples, an ellipsoidal reflector spotlight ("leko"), a Minolta illuminance meter and a Minolta luminance meter. What I did was aim the spotlight at the glass, take an illuminance reading incident on the front of the glass, then a luminance reading on the back of the glass. Dividing the luminance by Pi, then by the illuminance, gave me a reasonable estimate of transmittance. We took luminance readings at 15 degree intervals from nadir to +- 60 degrees, for incident angles of 0-45 (also in 15 degree increments). There's a picture of the measurement rig here:

The transmissions were then hastily emailed to Greg Ward, who did some trial & error with trans parameters, using bgraph to visualize the transmission models until the trans definitions matched the measured dataset. Greg sent me a valid trans description, but I of course wanted to know how he accomplished this. Greg's email that explained what he did included this shell script:

#!/bin/csh -f


# Plot one or more files and display them
set dest=/tmp/plot$$
bgraph $* | meta2tga -x 700 -y 700 -o $dest
open -a Preview ${dest}*.tga

... but I could never get this to work on my end. If you or anyone else on the list wants to have a look at this and help me out, I'd appreciate it. When Greg first sent me this stuff, I really didn't get it at all. I just used his trans material definition. But in reviewing it all for this email, I understand it now. I just lack the understanding of bgraph to fully troubleshoot this script on my computer. But it's a very useful technique! WHat he did was take each dataset I sent him and used that for the "Adata", then used a mathematical model based on the trans parameters for the second dataset, the "B". He played with the Ts Td and roughness (alpha) parameters and used bgraph to see how they matched up. Here's a sample bgraph input file; this is for the 15 degree incident angle (reflected in the theta variable) dataset:

-60 0.086210394
-45 0.103198913
-30 0.174652116
-15 0.244228975
0 0.363959998
15 0.424915818
30 0.304880523
45 0.201275914
60 0.146151974

Unfortunately, I don't have any copies of the graphs here on my laptop, but basically this input file and the script above creates a polar transmission plot showing two curves, one for the measured data and one for the mathematical plot of the Radiance trans performance, based on the input parameters in the input file. Greg played with the T parameters until the curves looked similar. It's one of those applications of the Radiance suite that shows the beauty and power of Radiance. I just wish I could get the script to run on my own machines, so I could experiment on my own.

I hope this all makes sense, and sheds some light on what we did.

    Rob Guglielmetti