Canadian Architect


Light from Above

Roof windows are shown to give at least twice as much light as other configurations studied.

October 1, 2003
by Marie-Claude Dubois

Imagine three identical rooms located on the top floor of a building. These three rooms have the same dimensions (3.25 m by 3.85 m), same floor-to-ceiling height (2.5 m), identical interior finishes (see Table I below), and possess only one south-oriented window of identical size (0.8 m by 1.3 m) located exactly 1 metre above the floor. The only difference between these three rooms is the window-roof configuration (see above):

the first room has an ordinary vertical window (flat roof, a)

the second has a dormer window (sloped roof, b)

the third has a roof window, c)

What is the impact of these various window-roof configurations on the quality of natural light inside the rooms? This question was raised at the Danish Building and Urban Research (DBUR) Institute. The Daylight in Buildings researchers’ group at DBUR initiated a small study with a methodology based on computer simulation. The program used was the Radiance Lighting Simulation System, which has many advantages over other programs:

it has been validated by a number of studies

it allows the production of high quality renderings with no limitation on the point of view and type of projection generated

it makes possible the analysis of renderings in detail since the original luminance or illuminance values “behind” each pixel can be retrieved from any image

it allows the study of many types of skies (overcast, uniform, intermediate, sunny) at any time and latitude

According to the time and budget allowed for the study, the following limitations were imposed:

south orientation (window)

location: Copenhagen (latitude 55.4N; longitude 12.35E)

study of one overcast sky (CIE) and one perfectly sunny sky (for the equinox: 21 September, at 12.00 and 15.00 hours (true solar time)

Some of the results obtained under overcast sky conditions are summarized below (see top, facing page), extracted from a full report (see note [6] below for full report).

The roof window gives at least twice as much light as the other configurations studied.

the roof window produced a significantly higher daylight factor (DF)–approximately twice as high–on a horizontal plane (0.7m above the floor) than the vertical and dormer windows. Morevover, the DF was unacceptably low in the case of the dormer window (<2% for the whole room area), nearly acceptable with the vertical window (2% for 50%; >5% for 15% of the room area)

the vertical and dormer windows yielded similar DF distributions

the roof window yielded a better DF distribution compared to the two other windows (see distribution of DF % on a horizontal plane, chart below). This variation might be preferable since previous research indicated that people prefer an interior to have a measure of visual lightness combined with a degree of visual interest (visual interest applied to the non-uniformity of the light pattern).

The roof and dormer windows have linings with an intermediate luminance allowing a softer transition between the high sky luminance and the luminance of inner surfaces. The dormer and roof windows have linings that have a luminance between the high sky luminance and the luminance of the other surfaces in the room. These linings therefore create a form of luminance transition between the surfaces in the field of view. The luminance ratios “sky linings:south wall-slopes” were the most favourable with the roof window (100:10.4 1.2-1.8. In comparison, it was 100:3.6:0.4 with the dormer window). However, in this case, the linings may not be sufficiently large to allow for a good luminance transition, which is obvious in renderings showing an interior view (see below). The luminance of the walls, ceiling and floor was the highest with the roof window and the lowest with the dormer window. (see Figure 3, below)

In summary, the study generally indicates that a roof window gives at least twice as much light, under overcast sky conditions, than the vertical and dormer windows. This is important considering the fact that overcast sky conditions prevail approximately 75% of the time in Denmark (and 60% in Quebec). Moreover, the linings of the roof window allow a luminance transition from the high sky luminance to the luminance of the main interior surfaces (walls, floor, and ceiling). The roof window therefore seems preferable from the point of view of the interior lighting designer. However, other equally important aspects such as e.g. thermal comfort, overall energy consumption, view to the outside, appearance of the interior space, exterior architectural expression, construction costs, water tightness, and others have not been considered in this study.


[1] Ward Larson, G. & Shakespeare, R. (1998). Rendering with Radiance: The Art and Science of Lighting Visualization. Morgan Kaufmann Publishers. San Francisco (CA). 664 pages.

[2] Aizlewood M., Laforgue P, Mittanchey R., Carroll W., Hitchcock R. (1998). Data Sets for the Validation of Daylighting Computer Programs: Proceedings of the Daylighting ’98 Conference, International Conference on Daylighting Technologies and Energy Efficiency in Buildings, May 11-13. Ottawa (Canada). pp. 157-164.

[3] Jarvis D., Donn M. (1997). Comparison of Computer and Model Simulations of a Daylit Interior with Reality: Proceedings of Fifth International IBPSA Conference / Building Simulation 97, Sept. 8-10. Prague (Czech Republic). Vol III (9).

[4] Mardaljevic J. (1999). Daylight Simulations: Validation, Sky Models and Daylight Co-efficients. PhD thesis. De Montfort, Leicester (UK): De Montfort University, Inst. of Energy and Sustainable Development.

[5] Ubbelohde S., Humann C. (1998). A Comparative Evaluation of Daylighting Software: Superlite, Lumen Micro, Lightscape and Radiance: Proceedings of the Daylighting ’98 Conference, International Conference on Daylighting Technologies and Energy Efficiency in Buildings, May 11-13. Ottawa (Canada). pp. 157-164.

[6] Dubois M.C., Grau K., Traberg-Borup S. & Johnsen K. (2003). Impact of three window configurations on daylight conditions: Simulations with Radiance. Internal report. April 2003. Danish Building and Urban Research (By og Byg), Hrsholm, Denmark. 35 pages.


[8] CIBSE (1999). Daylighting and Window Design. Lighting Guide LG10:1999. The Chartered Institution of Building Services Engineers. London, UK.

[9] Loe, D.L. (1997). Task and Building Lighting: The Link between Lighting Quality and Energy Efficiency. Procedures of the Right Light 4 Conference. Nov. 19-21, Copenhagen (Denmark). Vol. 1. pp. 11-15.

[10] Loe, D.L., Mansfield, K.P. & Rowlands, E. (1994). Appearance of lit environment and its relevance in lighting design: experimental study. Lighting Res. Technol. 26(3). 119-133.

Marie-Claude Dubois worked as a Senior Researcher in the energy and indoor climate division of the Danish Building and Urban Research Institute in Hoersholm, Denmark from May 2001 to April 2003. VELUX A/S was a collaborator in the 2003 study. She is now Adjunct Professor at the Laval University School of Architecture and part of GRAP (groupe de recherche en ambiances physiques) a research unit that studies issues in thermal comfort and lighting.

Table 1. Reflectance (Rvis) and transmittance (Tvis) of inner surfaces and glazing properties.
Surfaces/ Material/color Rvis Tvis
elements sample (%) (%)
Walls, slopes Painted light grey 57.0 0.0
Floor Chestnut wood 37.9 0.0
Ceiling Painted pure white 83.5 0.0
Door Painted grey 38.8 0.0
Glazing Double 14.0 78.0

Print this page

Related Posts

Have your say:

Your email address will not be published. Required fields are marked *