Glare is the most widely known and researched cause of visual discomfort in daylit environments. It is also a very difficult phenomenon to measure. Some people are more tolerant of glare conditions than others and the parameters affecting glare will always be different in each office.

Assessment Methods

Since the late 1950s researchers have been trying to make the difficult link between the physical characteristics of the glare in the space (the luminance of the window and its surroundings, the solid angle of the glare source, the illuminance etc) with the subjective response of the occupants (how much glare is perceived).

Several glare calculations / formulas have been developed – many of which are tailored towards artificial lighting.

1.1 Glare Ratings for Artificial Lighting

Several discomfort glare indices have been developed to assess glare from artificial light sources. These include:

• The British Glare Index (BGI) developed by Hopkinson
• The CIE Glare Index (CGI) proposed by Einhorn
• The Unified Glare Rating (UGR) where Sorenson combined the “best” aspects of the British Glare Index and the CIE Glare Index
• The J-Index developed by Meyer
• Stationary Virtual Reality (SVR) Method created by Sick and Wienold

However, because these glare indices were developed for artificial lighting they are less reliable for daylit environments. Osterhaus conducted experiments to compare the results of these glare indices with the subjective glare perception of test subjects. , It also investigated the precision of the Daylight Glare Index and simply measuring vertical illuminance at the eye of the subject. These will be discussed later. The tests were conducted in a laboratory environment, where 32 participants were placed in front of an artificially-lit window-like glare source and asked to rate the glare level of 39 different lighting conditions. Regression analysis (both linear and logarithmic) revealed that the three indices above had poorer correlation with the subjective rating than both the Daylight Glare Index and the illuminance measurement at the eye.

The J-Index and the Stationary Virtual Reality method were reviewed by Velds in her PhD thesis. The J-Index is designed for measuring comfort levels at the VDT and was still in development at the publication date. The index appears to be not solely confined to glare. The Stationary Virtual Reality Method uses computer generated stereographic images viewed in 3D which allows exact replication of the environment between subjects, who assess preferred luminance values. However, the validation of this technique had not been completed at the time of the PhD publication either, and appears to have since been abandoned.

None of the glare indices not specific to daylit environments have provided reliable data under daylight conditions, which is why specific daylight glare indices have been developed.

1.2 Glare Ratings for Daylighting

The glare indices primarily designed for artificial lighting are, not surprisingly, unsuited to daylit environments. They do not allow for the large area of the glare source that is a window. Several glare ratings have been developed to consider the glare caused by daylight. These include:

• The Daylight Glare Index (DGI) developed by Hopkinson and modified slightly by Chauvel
• The New Daylight Glare Index (DGIN), a reworking of the above by Nazzal and Chutarat
• The Visual Comfort Evaluation Method (VCE) and User Acceptance Studies designed by Velds
• The Predicted Glare Sensation Vote (PGSV) proposed by Iwata and
• Vertical illuminance at the eye suggested by Osterhaus

1.2a Daylight Glare Index

The Daylight Glare Index (Hopkinson) was the first formula used to calculate the perceived glare from daylight specifically. It was then slightly modified by Chauvel by replacing the luminance of the source with the luminance of the window in the bottom line of the equation. Chauvel’s DGI is accepted as being more reliable in daylit spaces than the indices used for artificial lighting. The DGI is given by the formula:

DGI = 10 log10 0.478

Where
Ls = the average luminance of each glare source in the field of view (cd/m2)
Lb = the average luminance of the background excluding the glare source (cd/m2)
Lw = the average luminance of the window (cd/m2)
s = the solid angle of the source seen from the point of observation (sr)
Ω = the solid angle subtended by the source, modified for the position of the light source with respect to the field of view and Guth’s position index P (sr)
n = the number of glare sources

The index produces results between just imperceptible and just intolerable:

Mean subjective assessment of glare DGI
Just imperceptible 16
18
Just acceptable 20
22
Just uncomfortable 24
26
Just intolerable 28

Several studies have been conducted using the Daylight Glare Index, which give a good indication of the accuracy of its prediction.

Boubekri and Boyer tested different window sizes with 40 office workers and asked them to assess the level of glare on a semantic scale. When the perceived glare was compared to the calculated Daylight Glare Index, it revealed that the DGI consistently over-estimated the level of glare present. However, it was also noted that there was a very good view and the workers were used to windowless environments, which may have distorted the results.

In the study by Osterhaus , discussed above in section 3.1.1, the correlation between the subjective glare rating of the occupants and DGI was analysed. In contrast to the results of Boubekri and Boyer, the predictions of the DGI in this experiment under-estimated the perceived level of glare. Furthermore, the regression analysis showed a weak correlation between the subjective glare rating and the DGI for both linear and logarithmic regression. It should be noted, however, that this experiment was performed in a laboratory environment in front of an artificial window, without view, which may have affected the perceived glare levels.

As part of the International Energy Agency (IEA) Task 21 “Daylight in Buildings” project, Aizlewood assessed a variety of blind systems using the DGI. He did not attempt to validate the DGI – he simply used it to assess the reduction of glare using the blinds. Of more interest is the method used to calculate the DGI. Rather than measure the required luminances (which could be difficult) Ls, Lb and Lw were derived so that they could be measured using an illuminance meter instead:

Ls = where: E(Shielded) = the vertical illuminance at the sensor with shielding cone
Ø = the configuration factor of the glare source with respect to the measurement point

Lb =

NB: If the shielding cone is shaped such that the shielded sensor sees the whole window but nothing else, then Lw is the same as Ls. This results in a “cone” with an irregular pyramidal shape.

This methodology allows the DGI to be calculated needing only an illuminance meter, which may make it more feasible for building inspectors and / or building owners to perform the calculation as they do not have to purchase a luminance meter.

Wilson suggests that the Daylight Glare Index / ‘Cornell formula’ remains “the only substantive equation to use for estimating discomfort glare for daylight”. Whilst Nazzal and Chutarat, Iwata and Tokura, Velds and Osterhaus have all suggested methods designed to supersede the DGI, which shall be discussed.

1.2b New Daylight Glare Index

Nazzal and Chutarat have proposed a new formula and protocol for measuring discomfort glare from daylight. They cite the lack application for direct sunlight in existing glare indices, the unpredictable variability of subjective responses due to personal situations, the difficulties of using CCD cameras to map luminances and the misrepresentation of artificial sky experiments as reasons why a new model was needed. These are all very valid points. The new daylight glare index (DGIN) uses the same basic parameters as Chauvel’s method – size of light source, luminance and position of the light source in field of view. However, the calculation of the solid angles and luminances differ in the DGIN. The solid angles are modifies to include the effect of observation position and the calculation has greater emphasis on the immediate surround luminance.

The basis for creating the new model and the changes made are sound; however there is a lack of any reliable testing regime to check that the DGIN does correlate with the glare perceived by occupants. The new method was trialled in a test room in Helsinki, Finland, where the physical measurements were taken. This then provided the basis for Radiance (a lighting simulation software package) simulations to be performed of the case study in Helsinki, as well as a control simulation run in Fort Worth, Texas. The comparison between the perceived glare from the original DGI and the DGIN revealed that the new method consistently suggested higher levels of glare.

Whilst the DGIN appeared to have great potential as an improvement on the method of glare assessment, it failed to provide any reasonable validation. It simply compared it to the original DGI and proved that the new method calculated higher levels of glare and suggested that “the new DGIN procedure appears to yield sensible and consistent glare values”. There is no justification as to why the new procedure is an improvement – simply testing scenarios on a computer and making a comparison to one other method does not provide enough evidence. It has not been compared to the subjective response of occupants in a real-life situation and thus appears to lack credibility.

1.2c Visual Comfort Evaluation (VCE) Method and User Acceptance Studies

As part of her PhD work, Velds developed two methods of assessing visual comfort in daylit environments, with a particular focus on glare. The Visual Comfort Evaluation (VCE) Method was developed for use by architects and lighting designers to test the implications of various design strategies. , "…the Visual Comfort Evaluation Method is a method to appoint the degree of discomfort glare in daylit situations under reproducible conditions, through assessment of subjects in a large scale model placed in front of an artificial sky."

The VCE method does not allow assessment of existing offices, which this report is primarily concerned with. It is basically a modeling technique to allow experimentation with daylighting systems and assess occupant reactions at a 1:5 scale, whilst also gathering objective data.

The User Acceptance Studies, however, are more suited to the purpose of this report. Two identical rooms were set up in Berlin to conduct the studies. One room housed five illuminance meters, which were logged continuously in combination with the sky luminance logging. The other room was where the 23 test subjects worked for approximately half a day on their own work (some VDT- based, some paper-based), whilst occasionally completing questionnaires regarding their visual comfort which ‘popped up’ on the computer. Having completed each questionnaire they then took a picture of the room using a CCD camera. This set-up allowed a direct link between the subjective glare rating and the luminous conditions of the room, which yielded some very interesting results:

• “A high correlation with subjective glare ratings was found for the direct vertical illuminance at the eye, for glare assessments behind the VDU [facing wall perpendicular to window] as well as towards the window without daylighting systems, but it is not the sole parameter in the glare assessment.”

• “A differentiation should be made between visual comfort criteria when working behind a VDU or working on a horizontal task. The glare assessment behind the VDU is only slightly more critical than those for the horizontal tasks. More significant is the acceptance of this degree of discomfort glare by subjects, which is lower when working behind the VDU. This means that, even though the orientation of the VDU is towards the East wall [with the window in the South wall] and the visible light source is smaller, these assessments are more critical in the evaluation of discomfort glare in a room.”

The illuminance measurements were used to create maximum and minimum external illuminances (based on the daylight factors of the specific case study) within which would be a visually comfortable environment (for that building), free of glare (without any shading device ie: blinds):

VDU – Emin, hor = 3 300 lux
VDU – E max, hor = 25 500 lux
Horizontal task – E min, hor = 3 400 lux
Horizontal task – E max, hor = 32 500 lux

She also notes that visible sky luminance is critical to the glare experienced at the front of the room near the window, but less so at the back of the room where there is far less visible sky. The thesis suggests that the illuminance level at the eye could be used as an indicator of glare when the occupant is facing the window; however, it appears not to be a valid method when they look away from the window.

Whilst the User Acceptance Studies do not provide a specific glare assessment method, they do contribute some very relevant information which could be used in the creation of a visual comfort assessment protocol.

1.2d Vertical Illuminance at Eye

During the investigation conducted by Osterhaus (as described earlier) the relationship between vertical illuminance and subjective glare rating was analysed. Of all the glare indices it was found to have the strongest correlation. This is an encouraging relationship as measuring the illuminance at the eye is a relatively quite and simple task to complete when assessing a building. However, it must be remembered that the results were from a laboratory experiment with an artificial glare source. Nevertheless, the hypothesis that vertical illuminance is a good indicator of discomfort glare is supported to some extent by Velds and Nazzal above. At this stage there is not sufficient research to support this theory although future work may provide stronger evidence for it.

Recommendations for Glare Assessment

It is clear that the glare indices designed for artificial lighting conditions are not valid for use in daylit offices. Other glare rating methods have been developed to specifically address daylighting, which are more reliable. The Daylight Glare Index still remains the most common and trusted method of determining the level of glare. The derivation process used by Aizlewood allows the entire index to be calculated using only an illuminance meter, which makes it far more practical for actually implementing.

However the DGI is not the perfect system. The problems with the DGI raised by Nazzal and others are severe limitations for its effective use in practice. It does not cope well with direct sunlight or variable personal responses. Furthermore, it is only a spot measurement and thus several calculations would be needed throughout the room and during different sky conditions. Secondly, it is an assessment of daylight glare alone – how the calculation is affected by the use of both artificial light and daylight together is unknown. It is very rare for an office building to use only daylight, so the glare assessment method has to be able to cope with both light sources. Unfortunately, the glare indices developed later do not provide a definite answer. The new daylight glare index is unproven. The Visual Comfort Evaluation Method is not suited to use in existing buildings. The User Acceptance Studies was very well conducted and contributed some very relevant information; however, it only provided suggestions as to the types of things to measure for assessing visual comfort. It did not clearly outline a protocol. Lastly, the vertical illuminance at the eye could potentially be incorporated into an assessment method but would need to be partnered with other assessment measures in order account for all variables.

2.1 Possible Measurements Needed in Assessment Procedure

At the current point in time the original daylight glare index is the most useful calculation of glare available. This should be used as the basis for the glare assessments. As the DGI can only be used for spot calculations, the process should be repeated at only a few of the worst-affected workstations in the office (as indicated by the occupants), preferably at a time and on a day when the sky conditions generate the most glare.

Secondly, a grid of horizontal daylight factors throughout the whole space at the working plane should be measured. These can then be used to provide an indication as to whether or not glare reduction measures for the front of the space (such as blinds) will have too detrimental an effect on the illuminance at the rear of the space away from the window.

Lastly, a simple observation of whether or not there are facilities available to the occupants that might reduce the glare should be made. These facilities may be shading devices or even rearrangement of computer screens or office furniture. It must also be noted if the occupants have control over such facilities. It is not uncommon for automated control systems to be sabotaged (an obvious hint of dissatisfaction) if there is no manual control/override.

3.2.2 Future Research Needed

The suggested procedure above is not perfect. Further research is needed to develop a method which considers the entire room at once, not just a single location with a particular field of view. Furthermore, a glare index which considers both artificial and daylight needs to be developed. However, until that stage, the assessment measures that are available have to be utilised.