CIE Unified Glare Rating calculation - error in findglare

radiance-dev
1

The documentation for RADIANCE Visual Comfort Calculation
(http://radsite.lbl.gov/radiance/refer/Notes/glare.html) states:

The Guth position index is used by this formula as well, and its calculation
is described in [Levin75].

This is true for IESNA Visual Comfort Probability (VCP) calculations, but
it is *not* true for CIE UGR calculations. Quoting from CIE 117-1995,
"Discomfort Glare in Interior Lighting," Section 4.5: "The position index
is found by interpolating the data of Table 4.1 [Table of position
indices]."

The equation of Levin, R. E. 1975, "Position Index in VCP Calculations,"
Journal of the IES, pp. 99-105, is based on:

Committee on Recommendations of Quality and Quantity of Illumination of the
IES, "Outline of a Standard Procedure for Computing Visual Comfort Ratings
for Interior Lighting, Report No. 2 (1966), Illuminating Engineering, Vol.
61, No. 10, October 1966, p. 643.

which shows a plot of the Guth position index.

As noted by Levin, "The maximum difference between the basic Guth plot and
Equation (6) is about 12 percent." An error plot shows that errors in excess
of 5 percent occur over most of the range of the Guth position plot.

This issue aside, the RADIANCE documentation states:

If a single view direction is selected, findglare samples uniformly on the
projected hemisphere.

Unfortunately, CIE 117-1995, Section 4.5, states: "It is recommended that
luminaires with T/R values outside the range of the table (0 to 3) be
ignored. It is further noted that some positions in the table at large H/R
values are void. These correspond to positions hidden from observers by
eyebrows and foreheads, and accordingly to luminaires which do not add to
the UGR."

The entries of Table 4.1 define an elliptical field of view measuring
approximately 62 degrees vertical by 73 degrees horizontal.

The problem is that by ignoring the shielding effect of eyebrows and
forehead in the Guth position index (which regrettably assumes a Caucasian
physiognomy), the UGR values for narrow-beam luminaires with small luminous
areas seen at high viewing angles can be grossly overestimated. In one
study, the difference was 23 versus 14.

Such errors typically occur within a range of less than 30 cm [12 inches]
for the observer position, but it remains that findglare does not appear to
be in full compliance with the requirements and recommendations of CIE
117-1995. To do so would require replacing the Levin analytic equation with
an interpolation of the Guth position indices table published therein.

(This analysis is based on the above-referenced Visual Comfort Calculation
document, dated 1992. I have not examined the current source code for
findglare to see whether it still applies.)

Ian Ashdown, P. Eng., FIES

President

byHeart Consultants Limited

http://www.helios32.com <http://www.helios32.com/>

CONFIDENTIALITY NOTICE: This entire communication, including without
limitation any attachments, is intended for the use of the recipient to
which or whom it is addressed, and may contain confidential, personal,
and/or privileged information. Please contact us immediately if you are not
the intended recipient of this communication, and do not copy, distribute,
or take action relying on it. Any communication received in error, or
subsequent reply, should be deleted or destroyed.

2

Hi Ian,

Thanks very much for your detailed analysis of the algorithm used in findglare and glareval. You are of course correct that the Guth position index is applied broadly even where a particular glare metric calls for a different treatment. This was done as an effort-saving measure in what turned out to be a rarely visited cul-de-sac in the Radiance roadmap. I am not sure at this point if anyone relies on these calculations in their regular practice.

Personally, I never held much faith in the various glare metrics. As important as this topic is, there seems to be a lot of disagreement as to what constitutes a good glare metric or what it is even supposed to measure. The notion of human comfort is very ill-defined and has a wide range of variance from one individual to the next and even one moment to the next for the same individual.

This is an aside from the central topic you raise, which is whether or not findglare and glareval adhere to the specified calculations they claim to support. I must concur with your assessment that they do not, at least in the case of UGR. I remember when I wrote these tools about 23 years ago that I was more focused on VCP than any of the others, which were added mostly for comparison purposes and not as definitive implementations. In making a general interface for scene analysis (findglare), I was forced to compromise some of the more peculiar specifications in UGR and the other glare metrics. At the time, this was still an active area of research, and our particular interest at the EPFL was in the application of glare analysis to daylighting situations, which were problematically different from typical electric lighting scenarios the index creators had in mind.

If there is strong interest in developing an accurate UGR calculation in Radiance, I am happy to help whoever wants to work on it.

Cheers,
-Greg

···

From: "Ian Ashdown" <[email protected]>
Date: February 18, 2014 11:54:35 PM GMT+01:00

The documentation for RADIANCE Visual Comfort Calculation (http://radsite.lbl.gov/radiance/refer/Notes/glare.html) states:

The Guth position index is used by this formula as well, and its calculation is described in [Levin75].

This is true for IESNA Visual Comfort Probability (VCP) calculations, but it is *not* true for CIE UGR calculations. Quoting from CIE 117-1995, “Discomfort Glare in Interior Lighting,” Section 4.5: “The position index is found by interpolating the data of Table 4.1 [Table of position indices].”

The equation of Levin, R. E. 1975, “Position Index in VCP Calculations,” Journal of the IES, pp. 99-105, is based on:

Committee on Recommendations of Quality and Quantity of Illumination of the IES, “Outline of a Standard Procedure for Computing Visual Comfort Ratings for Interior Lighting, Report No. 2 (1966), Illuminating Engineering, Vol. 61, No. 10, October 1966, p. 643.

which shows a plot of the Guth position index.

As noted by Levin, “The maximum difference between the basic Guth plot and Equation (6) is about 12 percent.” An error plot shows that errors in excess of 5 percent occur over most of the range of the Guth position plot.

This issue aside, the RADIANCE documentation states:

If a single view direction is selected, findglare samples uniformly on the projected hemisphere.

Unfortunately, CIE 117-1995, Section 4.5, states: "It is recommended that luminaires with T/R values outside the range of the table (0 to 3) be ignored. It is further noted that some positions in the table at large H/R values are void. These correspond to positions hidden from observers by eyebrows and foreheads, and accordingly to luminaires which do not add to the UGR."

The entries of Table 4.1 define an elliptical field of view measuring approximately 62 degrees vertical by 73 degrees horizontal.

The problem is that by ignoring the shielding effect of eyebrows and forehead in the Guth position index (which regrettably assumes a Caucasian physiognomy), the UGR values for narrow-beam luminaires with small luminous areas seen at high viewing angles can be grossly overestimated. In one study, the difference was 23 versus 14.

Such errors typically occur within a range of less than 30 cm [12 inches] for the observer position, but it remains that findglare does not appear to be in full compliance with the requirements and recommendations of CIE 117-1995. To do so would require replacing the Levin analytic equation with an interpolation of the Guth position indices table published therein.

(This analysis is based on the above-referenced Visual Comfort Calculation document, dated 1992. I have not examined the current source code for findglare to see whether it still applies.)

Ian Ashdown, P. Eng., FIES
President
byHeart Consultants Limited
http://www.helios32.com

3

(I'm reposting this as it seems to have gotten stuck in the mail queue since yesterday).

Hi Ian,

Thanks very much for your detailed analysis of the algorithm used in findglare and glareval. You are of course correct that the Guth position index is applied broadly even where a particular glare metric calls for a different treatment. This was done as an effort-saving measure in what turned out to be a rarely visited cul-de-sac in the Radiance roadmap. I am not sure at this point if anyone relies on these calculations in their regular practice.

Personally, I never held much faith in the various glare metrics. As important as this topic is, there seems to be a lot of disagreement as to what constitutes a good glare metric or what it is even supposed to measure. The notion of human comfort is very ill-defined and has a wide range of variance from one individual to the next and even one moment to the next for the same individual.

This is an aside from the central topic you raise, which is whether or not findglare and glareval adhere to the specified calculations they claim to support. I must concur with your assessment that they do not, at least in the case of UGR. I remember when I wrote these tools about 23 years ago that I was more focused on VCP than any of the others, which were added mostly for comparison purposes and not as definitive implementations. In making a general interface for scene analysis (findglare), I was forced to compromise some of the more peculiar specifications in UGR and the other glare metrics. At the time, this was still an active area of research, and our particular interest at the EPFL was in the application of glare analysis to daylighting situations, which were problematically different from typical electric lighting scenarios the index creators had in mind.

If there is strong interest in developing an accurate UGR calculation in Radiance, I am happy to help whoever wants to work on it.

Cheers,
-Greg

···

From: "Ian Ashdown" <[email protected]>
Date: February 18, 2014 11:54:35 PM GMT+01:00

The documentation for RADIANCE Visual Comfort Calculation (http://radsite.lbl.gov/radiance/refer/Notes/glare.html) states:

The Guth position index is used by this formula as well, and its calculation is described in [Levin75].

This is true for IESNA Visual Comfort Probability (VCP) calculations, but it is *not* true for CIE UGR calculations. Quoting from CIE 117-1995, “Discomfort Glare in Interior Lighting,” Section 4.5: “The position index is found by interpolating the data of Table 4.1 [Table of position indices].”

The equation of Levin, R. E. 1975, “Position Index in VCP Calculations,” Journal of the IES, pp. 99-105, is based on:

Committee on Recommendations of Quality and Quantity of Illumination of the IES, “Outline of a Standard Procedure for Computing Visual Comfort Ratings for Interior Lighting, Report No. 2 (1966), Illuminating Engineering, Vol. 61, No. 10, October 1966, p. 643.

which shows a plot of the Guth position index.

As noted by Levin, “The maximum difference between the basic Guth plot and Equation (6) is about 12 percent.” An error plot shows that errors in excess of 5 percent occur over most of the range of the Guth position plot.

This issue aside, the RADIANCE documentation states:

If a single view direction is selected, findglare samples uniformly on the projected hemisphere.

Unfortunately, CIE 117-1995, Section 4.5, states: "It is recommended that luminaires with T/R values outside the range of the table (0 to 3) be ignored. It is further noted that some positions in the table at large H/R values are void. These correspond to positions hidden from observers by eyebrows and foreheads, and accordingly to luminaires which do not add to the UGR."

The entries of Table 4.1 define an elliptical field of view measuring approximately 62 degrees vertical by 73 degrees horizontal.

The problem is that by ignoring the shielding effect of eyebrows and forehead in the Guth position index (which regrettably assumes a Caucasian physiognomy), the UGR values for narrow-beam luminaires with small luminous areas seen at high viewing angles can be grossly overestimated. In one study, the difference was 23 versus 14.

Such errors typically occur within a range of less than 30 cm [12 inches] for the observer position, but it remains that findglare does not appear to be in full compliance with the requirements and recommendations of CIE 117-1995. To do so would require replacing the Levin analytic equation with an interpolation of the Guth position indices table published therein.

(This analysis is based on the above-referenced Visual Comfort Calculation document, dated 1992. I have not examined the current source code for findglare to see whether it still applies.)

Ian Ashdown, P. Eng., FIES
President
byHeart Consultants Limited
http://www.helios32.com