Glare and visual comfort

On a clear October morning Holm takes the students to the reading room of the city's central library, a building with a wall of south-facing glass. Half the readers have retreated to tables away from the windows; a student near the glass has propped a notebook against his laptop screen; the blinds on the east side are down although the sun left them an hour ago. 'This room has excellent daylight by every annual metric,' Holm says. 'Look at what people are doing. Daylight that people cannot use because it hurts their eyes is not daylight; it is a problem they solve with blinds and electric light.' The students spend twenty minutes mapping where people sit and where the blinds are, and return to the studio with a new respect for glare.

What glare is

Glare is visual discomfort or loss of visibility caused by excessive brightness or excessive contrast in the field of view. Lighting researchers distinguish two kinds. Disability glare reduces the ability to see, as when the low sun in a windshield hides the road; it is caused by light scattered in the eye, which veils the image. Discomfort glare causes annoyance without necessarily reducing visibility, as when a bright window behind a speaker makes the audience squint. Discomfort glare is the usual concern in buildings. It depends on the brightness of the glare source, its size and position in the field of view, and the brightness of the surroundings to which the eye is adapted.

The traditional glare indices, such as the Unified Glare Rating used for electric lighting, were developed for small, bright sources such as lamps against a darker ceiling. Windows are large sources whose brightness varies across their area and over time, and those indices predict daylight glare poorly. In 2006 Jan Wienold and Jens Christoffersen published the daylight glare probability (DGP), derived from experiments in which people sat in offices with real windows and rated their discomfort. DGP estimates the probability that a person in a given position and view direction is disturbed by glare. It is now the most widely used daylight glare metric and is referenced in EN 17037.

Daylight glare probability

DGP combines two terms: the vertical illuminance at the eye, which captures how bright the whole view is, and a sum over glare sources of their luminance, size and position, which captures contrast. It is calculated from a high dynamic range image of the view, a fisheye picture with the luminance of every pixel, either rendered by Radiance or photographed with a calibrated camera.

Here E_v is the vertical illuminance at the eye in lux, L_s the luminance of each glare source in candelas per square meter, ω_s its solid angle and P its position index, which reduces the weight of sources far from the line of sight. When no direct sun or very bright source is in view, the contrast term is small, and Wienold proposed a simplified form, DGPs, that uses the vertical illuminance alone. It is much faster to calculate for every hour of a year, which makes annual glare studies practical.

Comfort classes and annual glare

Wienold's classes, used in EN 17037 and in most glare studies, divide DGP into four bands: below 0.35 glare is imperceptible, 0.35 to 0.40 perceptible, 0.40 to 0.45 disturbing, and above 0.45 intolerable. A single DGP value applies to one position, one view direction and one moment. To judge a design, the analyst runs many hours and positions, and summarizes them. A common annual summary is glare autonomy: the share of occupied hours in which DGP stays below a threshold, often 0.40, at a given seat and view. EN 17037 sets its own requirement: DGP must exceed the chosen comfort class in no more than 5 in 100 of the occupied hours.

The students calculate DGPs for a reading desk in the hub, first for a few eye illuminances and then for every occupied hour of a year, with a crude model of how bright the window view is. They compare a desk facing the window with one turned 90 degrees, and with blinds that people lower when glare becomes disturbing.

# Glare at a reading-room desk: the simplified daylight glare probability from vertical eye illuminance
import math, random
def dgps(ev):
    """Simplified DGP (Wienold 2009), valid when no direct sun reaches the eye."""
    return 6.22e-5 * ev + 0.184
def band(d):
    return "imperceptible" if d < 0.35 else "perceptible" if d < 0.40 else "disturbing" if d < 0.45 else "intolerable"
for ev in (1500, 3000, 4500, 6000):
    print(f"Ev {ev:5d} lux -> DGPs {dgps(ev):.2f} ({band(dgps(ev))})")
def year(seat_ev_scale, blinds):
    random.seed(7)                     # the same sky for every option
    bad = 0; hours = 0; drawn = 0
    for day in range(365):
        for hour in range(8, 18):
            hours += 1
            sky = max(0.0, random.gauss(1.0, 0.45)) * (0.6 + 0.4 * math.cos(2 * math.pi * (day - 172) / 365))
            ev = 7000 * sky * seat_ev_scale * math.sin(math.pi * (hour - 7) / 11)
            if blinds and dgps(ev) >= 0.40:
                ev *= 0.3; drawn += 1
            if dgps(ev) >= 0.40:
                bad += 1
    return 1 - bad / hours, drawn / hours
for name, scale, blinds in (("desk facing the window", 1.0, False),
                            ("desk turned 90 degrees", 0.55, False),
                            ("facing window, with blinds", 1.0, True)):
    ga, drawn = year(scale, blinds)
    print(f"{name:28s} glare autonomy {ga:.2f} of occupied hours; blinds down {drawn:.2f}")

Ev  1500 lux -> DGPs 0.28 (imperceptible)
Ev  3000 lux -> DGPs 0.37 (perceptible)
Ev  4500 lux -> DGPs 0.46 (intolerable)
Ev  6000 lux -> DGPs 0.56 (intolerable)
desk facing the window       glare autonomy 0.69 of occupied hours; blinds down 0.00
desk turned 90 degrees       glare autonomy 0.90 of occupied hours; blinds down 0.00
facing window, with blinds   glare autonomy 1.00 of occupied hours; blinds down 0.31

An eye illuminance of 1,500 lux is comfortable, 3,000 lux is perceptible, and 4,500 lux is already intolerable. The desk facing the window keeps DGPs below 0.40 for only 0.69 of occupied hours, which means a reader there would be disturbed by glare for almost a third of the year. Turning the desk 90 degrees, so that the window is to the reader's side rather than in front, raises glare autonomy to 0.90 with no change to the building at all. Blinds remove glare entirely in this model, but they are down for 0.31 of occupied hours, during which daylight and the view are lost and the lights probably go on.

Callum draws the lesson Holm hoped for: the first tool against glare is the layout of the room, not the shading. Desks with windows to the side, screens facing away from bright glass, and reading positions in the shadow of a pier or an alcove all cut glare cheaply. Shading comes second, and it is most effective when it blocks direct sun and the brightest part of the sky while keeping the view.

Luminance ratios and contrast

Glare is relative. A window of a given brightness is less glaring in a room with bright walls and ceiling, because the eye adapts to a higher level. Lighting designers use luminance ratios as rules of thumb: about 1 to 3 between a task and its immediate surroundings, and about 1 to 10 between the task and the far field, with brighter walls around windows to soften their contrast. Splayed, pale window reveals, deep window jambs that catch light, and pale ceilings all reduce contrast. The reading room's deep timber window surrounds, designed in second year for their warmth, turn out to raise contrast around the glass; the students propose a paler finish on the reveals.

Shading and glare control

Shading devices differ greatly in how they handle glare. Exterior overhangs and fins block direct sun but leave the bright sky in view. Interior roller shades of a given openness factor reduce the brightness of the whole window; darker fabrics preserve the view better than light ones, but transmit less daylight. Venetian blinds can be tilted to block sun and redirect light to the ceiling. Electrochromic and other dynamic glazing tints the glass in response to sun and glare. Each needs to be modeled properly: for daylight and glare, as a complex fenestration system with measured optical data; for energy, as a change in solar gain and visible transmittance that depends on how it is controlled.

Holm stresses control. A shade that is never lowered does nothing; one that is lowered and left down for weeks removes daylight. Automated shading driven by sensors, with occupant override, usually performs better than manual shading, but only if it is commissioned and maintained. Simulation should represent the realistic control, not the ideal one.

Glare from electric light and other sources

Glare is not only a daylight problem. Poorly shielded luminaires, bright reflections in screens and glossy surfaces, and the contrast between a bright screen and a dark room all cause discomfort. Electric lighting glare is assessed with the Unified Glare Rating (UGR); limits such as UGR 19 for offices and classrooms are common in lighting standards. Reflections on screens depend on the position of luminaires and windows relative to the screen and the viewer, and a quick check in a rendered view of each workstation can avoid them.

Glare in other building types

Different uses need different glare criteria. In a gymnasium, glare from windows can be dangerous for players tracking a ball; in a museum, contrast and reflections matter for viewing art; in a hospital room, patients lie looking at the ceiling and the window from angles a desk-based metric ignores; in a daycare, children nap on the floor. Wren notes that the hub's playrooms need a glare study at floor level for the nap area and at the staff's supervision point, not only at a desk. Holm agrees, and adds that view direction matters as much as position: glare studies should test the directions people actually look.

Exercise 7: glare at the critical seats

Each student carries out a glare study for one room in their studio project.

• Identify the critical positions and view directions.
• Calculate DGP at representative hours with images that include direct sun.
• Calculate annual glare autonomy with DGPs where no sun is in view.
• Compare at least one layout change with one shading change.

Self-critique checklist

• Have I tested the views people actually take, at the right eye height?
• Have I used full DGP where direct sun or bright sources are in view?
• Have I modeled realistic shading control?
• Did I consider layout before shading?