Skylight Design: Photometric Characteristics

This study analyzes the photometric performance of a sample of commercially-available 4’ by 4’ skylights, of the type commonly used to light warehouses, grocery stores and big box retail stores. Though skylight photometry has been measured earlier on scale model and residential sized skylights, this study is the first time that the luminous distribution of commercial sized skylights has been measured and recorded in the IESNA LM-63-95 photometric file format. These files are freely available at www.h-m-g.com. An innovative goniophotometer was used to take measurements under real skies in both clear and overcast conditions. This paper analyses the measured photometric data to investigate the way in which skylight efficiency and light distribution are influenced by common design variables such as skylight shape, glazing material properties, light well depth and light well surface reflectance. These variables determine not only the luminous performance of the skylight (as described by its luminous efficiency, the degree of asymmetry of the light, and the uniformity of illuminance in the building beneath) but also its uniformity over time, since the light distribution of some skylights changes significantly during the course of the day while others have a fairly unchanging distribution. The idea of an “ideal” skylight that would produce a reasonably constant level of illumination under a wide variety of sky conditions is discussed. The tested skylights do not approach this ideal performance; this paper describes a framework within which skylight manufacturers could improve and test their existing products. Types of Skylight Tested Three shapes of skylight were tested: dome, pyramid and compound arch (see Figure 1). The dome and compound arch shapes are formed by blow forming or stretching over a frame a heat-softened flat plastic sheet, while the pyramid shape is formed by joining four or more flat sheets. The plastics we tested included: acrylic, structured polycarbonate, PET (polyethylene tetrachloride) and fiberglass insulating panels. Some of the glazings diffused light via the use of pigments, refraction due to surface and shape of the glazing cross section and others used fibers. Transparent plastic and glass skylights were also tested. Figure 1 The three shapes of skylight tested: dome, pyramid and compound arch Each skylight was tested on top of a generic non-splayed square light well 1.2 m (4’) in width and length, and either 0.3m (1’), 0.9m (3’) or 1.8m (6’) in depth. The walls of the light well were lined with either white-painted sheetrock or highly specular aluminum. Photometric Test Method To measure the photometric distribution of a skylight, a large photometer with unobstructed access to daylight was required. Since no existing device of this type could be found, one was specially constructed IESNA Annual Conference 2004, Tampa Paper #16 – Skylight design: Photometric characteristics for this project (see Figure 2), and has been described in detail in a previous paper that also discusses sources of error. The photometer resembles a small room with a 2’ by 2’ (0.6m by 0.6m) square aperture on top and a series of fixed mirrors and photocells inside. Skylights that were 2’ by 4’ or 4’ by 4’ could be tested one section at a time. The skylight photometer is subject to all the sources of error of a regular photometer, and to several further sources of error incurred by the inherent variability of the light source. Sources of Experimental Error Measurements were taken under real skies that are non-uniform and that change constantly; this induces two types of error. Firstly, the condition of the sky was slightly different from one set of measurements to another. We intended to conduct the tests under perfectly overcast or perfectly clear skies, since these are extreme conditions of the sky, and many other sky conditions can be approximated as a combination of overcast and clear. Also, overcast and clear are fairly common sky conditions in many climates. Unfortunately, real skies are almost never either perfectly overcast or perfectly clear, so to ensure that the skies under which measurements were taken were as close as possible to ideal conditions, one of two criteria for sky ratio had to be met before measurements were taken. Sky ratio is the ratio of diffuse horizontal illuminance to global (diffuse plus direct solar) horizontal illuminance. Sky ratio had to be less than 0.25 for clear skies and greater than 0.85 for overcast skies. However, cloud locations may have varied during each test cycle such that the sky ratio criterion may not have been met for the entire duration of each cycle. Furthermore, 0.25 does allow some degree of cloudiness in a mainly clear sky (see Figure 3). Although sky ratio was much less than 0.25 during most of the measurements, the measurements taken at low sun angles had sky ratios that approached 0.25 because the horizontal illuminance from the sun is proportional to the sine of solar elevation. For future measurements, a more robust criterion for sky clearness might be the ratio of normal beam illuminance to diffuse illuminance. A threshold value for this ratio would have to be chosen to ensure both that there is an unobstructed view of the sun, and that cloud cover is not more than a few percent. Secondly, sky conditions change constantly; during the course of one set of measurements the sun moves a short distance across the sky, and the pattern of luminance of an overcast sky may change. To minimize the resulting error, the time taken for the measurements had to be as short as possible, so the number of measurements taken was reduced to the bare minimum: 22.5 degree increments of azimuth and 10 degree increments of elevation were used. These increments are larger than the 15 degree increments of azimuth and 5 degree (or 2.5 degree) increments of elevation typically used in luminaire photometry. Other sources of error are more fully described in Domigan et al. 85