Glare & Visual Comfort

Glare is not one phenomenon. Disability glare is the objective kind: a bright source scatters light within the eye and lays a veiling luminance over the retinal image, physically reducing your ability to see the task - think of a low sun washing out a screen so you genuinely cannot read it. It is directly measurable and, at the extreme, a safety issue (driving into the sun).

Discomfort glare is subtler and, in interiors, usually the bigger design problem. Here you can still see - but the visual scene is unpleasant, distracting, tiring. It is the nagging brightness of a window in the corner of your eye that makes you squint, reposition, or reach for the blind. It is subjective and driven by contrast: a bright patch against a relatively dark surround is far more uncomfortable than the same brightness in an already-bright scene. This is why a sunlit window is fine in a bright, light-walled room but punishing in a dim one with a screen - and why simply making a space darker overall can make its glare worse, by deepening the contrast a bright source stands against. Because discomfort glare is what drives occupants to defeat a daylighting design - blinds down, lights on, the daylight you carefully modelled thrown away - it is the kind our metric (DGP) targets. Both matter, but managing discomfort glare is where daylighting design is usually won or lost.

Everything in Lessons 5.1-5.3 measured illuminance: lux arriving at a surface, like a desk. Glare cannot be judged that way, because glare is about what the eye sees - and that is luminance, measured in candela per square metre (cd/m2): the brightness of a surface or source in a given direction. A desk can sit at a perfectly comfortable 400 lux while a slice of bright sky behind the monitor blazes at 8000-12000 cd/m2 and ruins the view. Illuminance says the task is well lit; luminance reveals the glare source.

So glare analysis works from a luminance image - the point-in-time render Radiance produces (Lesson 5.2) - not an illuminance grid. The tool captures the scene from the occupant's eye position and view direction (often a fisheye, to include the whole field of view), computes the luminance of every pixel, and then looks for bright sources against their surround. Two quantities matter: the absolute luminance of the source (a raw sun disc reflection is brutal) and its contrast with the background. This is why glare is inherently view-dependent: move the desk, turn the chair, and the same room can go from comfortable to intolerable. It is also why you must tell the simulation where the person sits and which way they face - a step the illuminance metrics never needed.

The metric that consolidated glare research is Daylight Glare Probability (DGP), developed by Wienold and Christoffersen and computed by Radiance-based tools (evalglare in Honeybee, and inside ClimateStudio). DGP predicts the fraction of occupants likely to be disturbed by glare in a given view, on a 0-1 scale, and cleverly combines two ingredients: the overall vertical illuminance at the eye (how bright the whole scene is) and the luminance and size of individual glare sources relative to the background. That pairing is what makes it robust - it captures both 'the whole window is dazzling' and 'one hot spot is stabbing'.

The interpretation bands are worth memorising: DGP < 0.35 is imperceptible glare, 0.35-0.40 perceptible, 0.40-0.45 disturbing, and > 0.45 intolerable. A common design target for workspaces is keeping DGP below about 0.40 for most occupied hours in the main view. Because DGP depends on the sky and sun, it is evaluated for specific moments (a clear winter afternoon is often worst-case) or, increasingly, annually as enhanced simplified DGP (DGPs) across the year - the same climate-based spirit as sDA. Older indices exist - the Daylight Glare Index (DGI) and the unified UGR for electric lighting - but for daylight, DGP is the current standard, and the one you will quote.

Glare control is not about less daylight; it is about managing peak luminance and contrast while keeping the useful light - and, conveniently, the same devices usually cut solar heat too, tying this lesson to Module 6. The main levers: Orientation and aperture - a north light (south, in the southern hemisphere) gives even, largely glare-free daylight; low east and west sun is the hardest to shade and the worst for glare and heat. External shading - overhangs, fins and a brise-soleil stop sun before it enters, cutting glare and cooling load at once (an overhead is highly effective for high midday sun; vertical fins for low lateral sun). Operable internal shading - venetian blinds, roller shades and, increasingly, automated or switchable ('electrochromic') glazing - handle the residual and the changing sun, which is exactly why the three-phase method (Lesson 5.3) exists: to model an operable blind across the year. Interior strategy - light finishes and a lightshelf raise the background luminance, which paradoxically reduces discomfort glare by lowering the contrast, and push daylight deeper.

The honest tension is that these often trade against sDA: a deep overhang that kills ASE and glare also trims daylight, and the right depth is exactly the kind of number simulation resolves - test overhang depths against sDA, ASE and DGP together, not one at a time. Balancing daylight, glare and solar heat simultaneously is the essence of good passive design, and simulating all three is how you find the sweet spot rather than guessing it.

Because glare depends on the sun's exact position, a single DGP number is only a snapshot, and the honest question is how often a view is uncomfortable. Early practice picked a plausible worst case - typically a clear-sky low-sun afternoon with the occupant facing toward the window - and reported DGP there; if that survived, most milder moments would too. But the same climate-based spirit that gave us sDA now gives us annual glare: tools compute an enhanced simplified DGP (DGPs) or run the fast daylight-coefficient machinery to evaluate DGP across many hours, then report the fraction of occupied time a view exceeds a threshold (say, DGP > 0.40). That turns 'is it glary right now?' into 'how many working hours is this desk uncomfortable?' - a far more useful design number, and one that finally lets glare be balanced against sDA on the same annual footing.

Keep the limits in view, though. DGP was derived largely for office-type views and tasks, so applying it to a gallery, an atrium or a devotional space needs judgement, not blind trust in the 0.40 line. It predicts the share of people disturbed, so a low DGP means most - not all - occupants are comfortable. It depends on getting the eye position, view direction and field of view right, and on honest source luminances, so a careless camera set-up can flatter or condemn a design unfairly. And like every result in this course it is decision-support: superb for comparing a shaded option against an unshaded one, but not a guarantee that no one will ever squint. Used comparatively, with sensible views and an annual read, DGP is the best glare predictor available - and the tool that stops a beautiful, high-sDA room from being ruined the day it is occupied.