Energy and Buildings 138 (2017) 347–354

Contents lists available at ScienceDirect

Energy and Buildings

journal homepage: www.elsevier.com/locate/enbuild

Analysis of daylight metrics of side-lit room in Canton, south China: A

comparison between daylight autonomy and daylight factor
Yu Bian a , Yuan Ma b,∗
a
School of Architecture, State Key Laboratory of Subtropical Building Science, South China University of Technology, Guangzhou 501640 Guangdong, China
b
Low-Carbon Ecological Urban & Rural Research Center, College of Architecture & Urban Planning, Guang Dong University of Technology, Guangzhou
510090, Guangdong, China

a r t i c l e i n f o a b s t r a c t

Article history: To ensure sufficient daylight in rooms, daylight performance metrics are the basic references to guide
Received 28 January 2016 building design or to benchmark a building against another in terms of daylighting in a room. Daylight
Received in revised form factor (DF) is the most commonly accepted daylight performance metric, but it has limitations in evalu-
15 December 2016
ating the daylighting of a room space in a real daylight climate, as defined under CIE standard overcast
Accepted 18 December 2016
Available online 24 December 2016
sky, while daylight autonomy (DA) is a climate-based performance metric which takes into consideration
the regional daylight climate. Based on long-term continuous measurements of daylight illuminance in a
test room under real climate, combined with scale model tests under an artificial sky and computational
Keywords:
Daylight performance metrics simulations, the quantitative relationship between monthly average daylight illuminance, DF and DA are
Daylight factor holistically analyzed in this paper. The result shows that a monthly average daylight illuminance above
Daylight autonomy 300 lx in a room located in Canton requires a DF of no less than 1.8% for north-facing space. Finally, the
Side-lit window depth of DA300lx [50%] daylit area for four cardinal directions was proposed in comparison with DF.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction criteria of performance metric, especially under regional daylight

climate from which building daylighting design will benefit.
Building performance metrics are supposed to be “quality mea- This article relates a commonly used daylight metric (daylight
sures” of buildings with respect to their energy efficiency, safety, factor, DF) to a newer, more complex daylight metric (daylight
quality of design, etc. [1]. Criteria of performance metric are vital for autonomy, DA) with the objective of providing simple DF-based
judging whether or not a building design meets a certain functional design guidance in Canton, China so as to achieve adequate day-
requirement. Moreover, daylighting environments are notoriously light levels within interior space, particularly, for those who do not
difficult to evaluate, for which a series of methods have been pro- have the ability or time to conduct more DAYSIM simulations.
posed including static metrics as well as dynamic metrics, parts of Daylight factor (DF) is the most common metric used in actual
which are not easily accessible for non-professional daylight simu- practice and/or guidelines, which is the ratio of internal illuminance
lation designers. Side-lit windows are the most widely used glazing to external horizontal illuminance under an overcast sky defined
openings in various buildings which ensure direct transmission of by the CIE luminance distribution. Although this is a fair model
daylight into buildings, and well-designed side-lit windows could of sky brightness under certain types of dull weather conditions,
reduce electrical lighting energy consumption and enhance indoor it is not particularly common [3]. Significant amongst the various
environmental quality. It is estimated that huge amount of energy reasons for this may be the lack of realism of the standard predictive
has been wasted associated with electric lighting during typical method: the DF approach [4].
daytime working hours (8:00–18:00) partly because the imperfect
design of side-lit windows [2], so it is meaningful to analyze the
daylight performance metrics of side-lit windows and discuss the
1. The CIE standard overcast sky is merely an idealist sky model:
Overcast sky type is not unique. The CIE overcast sky is applicable
when the complete sky canopy is covered with uniform dark
∗ Correspondence to: Guang Dong University of Technology, Yuexiu District, clouds representing heavy overcast sky only [5].
Guangzhou, China. 2. DF is assessed under overcast conditions, no account is made of
E-mail address: aryubian@163.com (Y. Ma). illuminance from sun and non-overcast skies, and so the daylight

http://dx.doi.org/10.1016/j.enbuild.2016.12.059
0378-7788/© 2016 Elsevier B.V. All rights reserved.
348 Y. Bian, Y. Ma / Energy and Buildings 138 (2017) 347–354

factor is invariant to building orientation and the location of the

room [6].

Daylight autonomy (DA), on the other hand, is a climate-based

daylight performance metric which factors in the daylight climate
of the building site and façade orientation. DA is represented as a
percentage of annual daytime hours that a given point in space is
above a specified illumination level. It is a major innovation since it
encompasses specific weather conditions of the geographic loca-
tion on an annual basis [7]. DA uses work plane illuminance as
an indicator of whether sufficient daylight is rendered in a space
so that an occupant can work by daylight alone. Required mini-
mum illuminance levels for different space types can be directly
taken from reference documents such as “China National Standard
of Building Daylighting Design”. DA300lx [50%] represents the per-
centage of the work plane area that exceeds 300 lx for at least 50%
of the annual time.
In this paper, under the real daylight climate condition in Can-
ton, the relationship between DF and monthly average daylight
illuminance on work task plane were studied and as a conclusion Fig. 1. Inside view of test room.
the criteria of DF was proposed. By comparing the measured data
in the test room and the software simulation, the results showed
Table 1
that the deviation between simulation and measurement is less Photometric description in test room.
than 20%, which is an acceptable result, indicating the validity of
Item Photometric description
the simulation in terms of accuracy. Based on the simulated results,
the depth of DA300lx [50%] daylit area in four cardinal orientations GLAZING Double glazing with Low-E
coating: transmittance Tvis = 0.45
were studied, and the comparison between DA300lx [50%], DF, aver-
INTERIOR WALL Lambertian diffuser with a 0.85
age daylight illuminance in a specific month and annual daylight reflectance, White
illuminance were proposed in various orientations. These results CEILING Lambertian diffuser with a 0.85
could be adopted as design references for the building daylighting reflectance, White
design, especially for those non-profession daylighting simulation CURTAIN Lambertian diffuser with a 0.66
reflectance, Medium gray
designers who are familiar with DF as the performance metric.
FLOOR 0.15 specular component, 0.42
reflectance, Dark gray ceramic tile
PROJECTOR SCREEN Lambertian diffuser with a 0.78
2. Methodology reflectance, White

2.1. Test room measurement

To study the daylight performance metrics in Canton, a qualified

side-lit room space was selected as the test room where the mea-
surements towards daylight illuminance distribution under real
daylight climate were taken place. According to Prof. C. F Reinhart:
for cities in low latitude in the north hemisphere (e.g. Phoenix,
United State), the north facing facade receives little to no direct
sunlight [8,9]. A north facing room was selected in this research as
a baseline to study the DF criteria and used as the test room to test
and verify the simulation program.
The test room adopted in this study is a due north facing side-lit
room located in Canton, with open surroundings. The dimensions
are 7.6 m*3.0 m*2.7 m, the window head height (WHH) is 2.2 m. The
room depth is 7.6m, more than 3 times the WHH, which related to
the rules of thumb. The window-to-wall ratio (WWR) of the test Fig. 2. Plan of test room & photometers arrangement.
room is 45%. The outer frame and mullion width are 0.05 m result-
ing in a frame factor of 18% of the rough opening area of the window
(Fig. 1). There are 3 sets of fluorescent grille lamp installed on photometers are wire linked and a string of 24 photometers collect
the ceiling, but turned off during the measurement period. Optical illuminance data every 10 min simultaneously and then recorded
properties of all room components are listed in Table 1. in computer. The measurement is processed daily from 8:00AM to
The arrangement of photometers (Sensor Maker/Model: Konica 6:00PM, to be consistent with the occupied hours of a typical office
Minolta T-10, Range: 0.01–299,900 lx along with automatic range room. This occupancy schedule is in agreement with the IESNA’s
switching, Accuracy ±5%) is represented in Fig. 2. A total of 24 pho- new Lighting Measurement IES LM-83-12 which promotes climate
tometers are settled on the axis of symmetry of the test room and based daylighting metrics [11].
on two equidistant axes at 1 m. Therefore, the photometers are The study included the daylight illuminance data collected from
located with a spacing of 1/2 WHH and at a height of a typical work April 2013 to April 2015, except a few hours or days that the pho-
task plane (above the interior floor of 0.75m). The arrangement tometers were halt or broken down, a total of 39,000 sets of data
of photometers had referenced Ho and Chiang’s research work of based on intervals of 10 min measurement were analyzed in this
daylight level distribution in a classroom in Taiwan [10]. All the study.
 Y. Bian, Y. Ma / Energy and Buildings 138 (2017) 347–354 349

Fig. 4. inside view of scale model for DF measurement.

ers of 1 mm acrylic glass (maker: Mitsubishi, T = 0.70) and 1 layer

of 1 mm acrylic glass (maker: Mitsubishi, T = 0.90) was used in
the model as window glass. multi-way photometers with 8 micro
lux sensors (Sensor Maker/Model: Konica Minolta T–10 M, Range:
0.01–299,900 lx along with automatic range switching, Accuracy
±5%) arranged in one row were used to measure the illuminance
in the scale model (Fig. 4). All the 24 study points were measured
for 3 rounds, and the outcome of DF obtained are shown in Fig. 5.
All the DF analyzed in this study come from the measurements in
Fig. 3. DF measurement within scale model in wall mirrored artificial sky. the scale model.

2.2. Accessing daylight factors in test room 2.3. Daysim simulation

Following the study by Navarro, Sendra and Barros, one fre- 2.3.1. The description of computer model
quently used approach to assessing DF within buildings is by means The computer model is created similar to the test room, with tri-
of the scale model, in which a wall-mirrored artificial sky simulator als in this stage aiming to verify the daylighting simulation program
is applied to reproduce the CIE overcast sky [12,13]. According to “Daysim”, and the model was also adopted to study the daylight
Jiangtao Du’s research work on accessing the DF in an atrium build- autonomy distribution with various orientations.
ing with the measurement in the wall mirrored artificial sky, the Depending on the shape, size and surface reflectance of the test
accuracy of the artificial sky has been previously contrasted [14]. room, the computer model for the analysis of DF is defined as a room
As can be observed in Fig. 3, a wall mirrored artificial sky was of 3.0 m wide by 7.6 m deep by 2.7 m high. The side-lit window is in
applied, which represents an ideal overcast sky. Its design corre- rectangular shape, with 0.05 m thick mullion and a glazing trans-
sponds with that of a parallelepiped model of the artificial sky, mittance of 0.45. All the optical properties of indoor surfaces are
where a constant luminance reflector emits light inside a cube. similar with those in the test room, as listed in Fig. 6. The positions of
Upon the reflector, there are 32 fluorescent tubes providing a study points are arranged in accordance with those of photometers
homogeneous emission of light flux. The walls are mirrors simulat- in the test room. The study points are represented in Fig. 6.
ing the location of the horizon in the infinite. The scale model in this
study is placed on a platform whose height is the bottom of the cube. 2.3.2. Verification by measurement
The actual width of the interior of artificial sky is 2400 mm; the Radiance has been verified in a number of lighting environments
width based on a 1:15 scale model in the artificial sky is 500 mm; [15,16]. Those investigations showed that Radiance simulations
The distance between the window aperture and the closet wall of could achieve a high accuracy in typical daylit spaces through com-
the artificial sky is 950 mm. parison between measurement and theoretical analysis. A recent
The scale model was made similar to the test room in order study [17] using Radiance as a benchmark to verify a general
to quantify the DF on the study points, the reflectance of each approach to computing daylight coefficient sets for rooms employ-
indoor surface in accordance with the actual test room, 2 lay- ing multiple dissimilar components. These investigations further

Fig. 5. Measured DF on study points.

350 Y. Bian, Y. Ma / Energy and Buildings 138 (2017) 347–354

Fig. 6. Computer model.

Fig. 7. Comparison between test room measured (m) and Daysim simulated (s) daylight illuminance.

indicated the status of Radiance in the field of daylighting simu- Fig. 7 compared the measured (m) and simulated (s) result of
lation. Daysim, a computational tool for dynamically calculating average daylight illuminance distribution in December and annual
annual illuminance profiles, has adopted the Radiance algorithms average daylight illuminance distribution; the values are listed in
and added two other functions (daylight coefficient and Perez sky Table 2. comparison between measured and simulated daylight
model) [18]. In this paper, for side-lit room daylighting, Daysim was illuminance in the test room shows, the deviation between sim-
regarded as an efficient approach. ulation and measurement is obvious (−37%–22%) in the area close
Daysim/Radiance simulation parameters have a direct influ- to window, but agreement is particularly critical for values beyond
ence on the simulation result, therefore, these values need to be WHH of 1.0 around the 300 lx and this is where the disagree-
calibrated so that the effect of complex fenestration devices will ment between simulation and measured values are acceptable
be accurately represented throughout the depth of the reference (−12.9%–22.4%).
office [11]. This study compared Daysim simulated values with Based on the two years measurements of daylight illuminance
on-site measured illuminance distributions. However, as the mea- distribution in test room, the measured DA300lx was shown in Fig. 8.
surements occurred in a real room under a real weather, but the As can be seen from the comparison between the measured and
simulations were processed under a typical weather, with weather Daysim simulated DA, the mean deviation ≤20%, and at 0.84 times
file drawn from China Standard Weather Data (CSWD), it is obvi- the WHH from the façade, the simulated DA300lx [50%] demon-
ously that deviation existed in this type of comparison.
 Y. Bian, Y. Ma / Energy and Buildings 138 (2017) 347–354 351

Table 2
Daysim simulated and test room measured illuminance (lx).

Distance to façade Simulated annual Measured annual Difference between Simulated average Measured average Difference between
in multiples of average daylight average daylight simulation and daylight daylight simulation and
WHH illuminance illuminance measurement illuminance in Dec illuminance in Dec measurement

0 1258.9 1084.3 16.1% 768.8 625.6 22.9%

0.5 611.2 594.7 2.8% 467.2 396.3 17.9%
1 330.5 379.3 −12.9% 265.3 216.8 22.4%
1.5 232.1 271.1 −14.4% 159.4 161 −1.0%
2 152.1 196.4 −22.6% 96.3 120.4 −20.0%
2.5 100.5 152.8 −34.2% 61.8 96.3 −35.8%
3 96.7 120.8 −20.0% 49.2 78.1 −37.0%

Fig. 8. Comparison between test room measured (m) and Daysim simulated (s) Daylight autonomy.

strates to be, 15.1% more than what the measured demonstrates illuminance in the test room were shown in Fig. 9. As can be con-
at 0.73 times the WHH from facade. cluded from the result, the monthly daylight illuminance in Canton
Based on comparisons with the test room measurement, the varies significantly, with the average daylight illuminance in April,
simulations prove to be a satisfying variation of DA. Daysim can June, July, August, September and October surpassing the annual
achieve a −37%–22% agreement result in the process of predicting average level and falling under the latter in other months. The day-
daylight levels in rooms located in Canton. The divergence could be light performs obviously higher in June, July, August and September
mainly explained by the weather data divergence between the real than in other months, among which the daylight illuminance in
weather and typical weather data (source: CSWD), geometric and July is the highest in a typical year, which is 43% more than the
photo metric divergence between the test room and scale model, annual average level, and the daylight performs weakly in January
and some errors that may have occurred in the measuring processes and December, with the weakest in December–the average daylight
[19,20]. illuminance in December is less than half of that in July, about 63%
To conclude the Daysim simulations have been verified at cer- of annual level, meaning that the daylight condition in December
tain level (−37%–22%) by the measurements for a north façade is the worst in a typical year in Canton, so the average daylight
exposed largely to diffuse sky condition, and can therefore be used illuminance in December was adopted to define the criteria for the
for further analysis of buildings under construction and those yet daylight factor metric.
to be constructed in Canton. To define the criteria of daylight factor metric in Canton, the
average daylight illuminance in December was selected to compare
with the DF, because the daylight illuminance in December is the
3. Analysis of criteria of performance metrics
lowest in a typical year in Canton, south China. Fig. 10 and Table 3
shows the relationship between average daylight illuminance in
3.1. Criteria of daylight factor in Canton, south China

At present, the DF represent the most widely used metric in Table 3

the evaluation of daylighting. An accurate criteria of daylight factor Comparing DF with monthly or annually average daylight illuminance.
based on rigorous experiment under the regional daylight climate is Average daylight illuminance in Daylight Annual average daylight
the baseline to guide the building daylighting design. In this study, December(lx) factor(%) illuminance(lx)
that criteria is developed from more than 38,000 sets of daylight 75 0.5 130
illuminance data, which were collected from 2013 to 2015. 150 0.9 297
According to the statistics of daylight illuminance on 24 study 300 1.8 533
points in the test room for 2 years, the relative values(monthly 500 3.0 826
750 4.5 1273
average value/annually average value) of monthly average daylight
352 Y. Bian, Y. Ma / Energy and Buildings 138 (2017) 347–354

Fig. 9. Relative value of monthly average daylight illuminance measured in test room.

Fig. 10. Relationship between average daylight illuminance (test room measured) and DF (scale model tested).

Table 4 illuminance of no less tan 300 lx, the required daylight factor in a
Comparing daylight factor and daylight autonomy in four cardinal directions.
north facing room space should be no less than 1.8%. Correspond-
Orientation ingly, for a monthly daylight illuminance of no less than 500 lx, the
threshold of daylight factor in a north facing room space should be
Metric North East South West
up to 3.0%.
DA300lx 50% 50% 50% 50%
DF (%) 1.52 1.35 1.20 1.35
Distance to façade in 0.84 0.90 1.05 0.90 3.2. Comparison between DA and DF in side-lit room in Canton
multiple of WHH
Based on observances of the sky luminance distribution in
Canton, south China, DF could not well reflect the real daylight con-
December/annual and DF, as we can see from Table 4, in Canton, dition in south China as defined under diffused skylight, therefore
the average daylight factor in the test rooms have a proportional the climate-based metric DA is more applicable in South China,
relationship with the daylight level on the work task plane, average andthe DA values studied in this article are that in unoccupied
daylight illuminance Eavg = 166.67*DF. To have a monthly daylight space.
 Y. Bian, Y. Ma / Energy and Buildings 138 (2017) 347–354 353

Fig. 11. Daysim simulated DA 300 lx for varying façade orientation.

The Daysim simulations had been verified by the measurements

in chapter 2.3. The computer model adopted to study the daylight
autonomy for four cardinal orientations is identical with the one in
the verification study but with varying orientations. Similarly, the
study points arranged in the computer model are also identical with
the arrangement in the test room. With this model, the daylight
illuminance were generated via Daysim simulation and then the
illuminance data above 300 lx was processed to ratio of occurrence.
Fig. 11 shows the daylight autonomy distributions along the
center lines of the test room facing four cardinal directions in Can-
ton. Up to a distance of 1/2 window head height, the daylight
autonomies are nearly identical, showing that right near a window
there is plenty of daylight for any façade orientation. Differences
appear deeper in space, with the south-facing space receiving most
daylight, the DA300lx [50%] appeared at 1.05 times the window head
height from façade. The west-facing space receives almost iden-
tical daylight as its east-facing counter part due to the assumed
occupancy pattern from 8:00 to 18:00, and the DA300lx [50%] in
west-facing space and east-facing space appeared at 0.90 times the Fig. 12. Scale model tested DF in artifical sky.
window head height from façade; The north facing space receives
the minimum daylight in four cardinal directions, the DA300lx [50%]
appeared at%0.84 times the window head height from façade. in Canton, and also could be used as a reference in other projects
Fig12 shows DF distribution in the test room, and the distance to based in south China area.
façade where the DA300lx [50%] occurs for the four orientations are
indicated. DF for an overcast sky is not dependent on orientation, 4. Conclusion
so the relationship between distance to façade and DF for various
orientations are identical. Some conclusions that can be drawn from this study include:
To compare the DF and DA in Canton, the DF which measured
with a scale model in the artificial sky and the Daysim simulated (1) It has been found that there is an agreement (deviation <20%)
daylight autonomy (DA300lx [50%])were adopted in this section of between the test room measured DA and the simulated value
the study. The results shown in Table 4 indicates that, for a north from the Daysim for a north facing side-lit room under the day-
facing façade in Canton, DA300lx [50%] pronounced at the 0.84times light climate in Canton, South China, particularly within the
of window head height where the daylight factor is 1.36%, for the distance from façade of window head height.
east facing façade the DA300lx [50%] pronounced at the 0.90 times (2) The test room measured daylight levels in December is the low-
of window head height where the daylight factor is 1.35%, for the est in an actual year, which is 63% of the annual average daylight
south facing façade the DA300lx [50%] pronounced at the 1.20 times level in Canton, China.
of window head height where the daylight factor is 1.20%, for the (3) The scale model tested DF levels have a proportional relation-
west facing façade the DA300lx [50%] pronounced at the 0.90 times ship with the test room measured daylight level on the work
of window head height where the DF is 1.35%. For most of architect task plane. The former is derived from a parallelepiped arti-
and parts of building engineering consultant, DA is not a familiar ficial sky, and a monthly average daylight level of no less than
daylight performance metric to access, so the corresponding DF 300 lx requires a daylight factor of 1.8%, correspondingly, 500 lx
was listed in this study to help them analyze the daylight cases requires a daylight factor of 3.0% for a north facing façade.
354 Y. Bian, Y. Ma / Energy and Buildings 138 (2017) 347–354

(4) DA is a more applicable daylight performance metric than DF, [3] P.R. Tregenza, The daylight factor and actual illuminance ratios, Lighting Res.
as the former factors in the regional daylight climate and façade Technol. 12 (2) (1980) 64–68.
[4] J. Mardaljevic, A. Nabil, The useful daylight illuminance paradigm: a
orientation. DA distribution in the test room varies significantly replacement for daylight factors, Energy Build. 38 (2006) (905-903).
in four cardinal directions in the distance from façade sur- [5] T. Muneer, Solar irradiance and illuminance models for Japan I: sloped
passing 1/2 WHH in Canton. In this case study, the test room surfaces, Light. Res. Technol. 27 (1995) 209–222.
[6] A. Nabil, J. Mardaljevic, Useful daylight illuminance: a new paradigm for
with a double glazing low-E window, the glazing transmittance assessing daylight in buildings, Lighting Res. Technol. 37 (1) (2005) 41–59.
of which is 0.45. The simulation results indicate that south- [7] C.F. Reinhart, O. Walkenhorst, Dynamic RADIANCE-based daylight
facing space receives most daylight, with the DA300lx [50%] simulations for a full-scale test office with outer venetian blinds, Energy
Build. 33 (7) (2001) 683–697.
pronounced at 1.05 times the window head height from façade.
[8] Christoph F. Reinhart, Daylighting Handbook, 2013, pp. 91–95.
The west-facing space receives almost identical daylight with [9] C.F. Reinhart, A simulation-based review of the ubiquitous
its east-facing, with both pronounced at 0.90 times the window window-head-height to daylit zone depth rule of thumb, in: Proceedings of
Buildings Simulation 2005, Montreal, Canada, 2005, pp. 15–18.
head height from façade; The north facing space receives the
[10] Ming-Chin Ho, Che-Ming Chiang, Optimal sun-shading design for enhanced
minimum daylight in four cardinal directions, the DA300lx [50%] daylight illumination of subtropical classrooms, Energy Build. 40 (10) (2008)
pronounced at 0.84 times the window head height from façade. 1844–1855.
(5) Comparing the scale model tested DF under artificial sky and [11] Christoph F. Reinhart, J. Alstan Jakubiec, Diego Ibarra, Definition of a reference
for standardized evaluations of dynamic façade and lighting technologies,
daysim simulated DA in Canton, for a north facing façade in Proceedings of BS2013 (2013) 3645–3652.
Canton, DA300lx [50%] pronounced at where the daylight factor [12] J. Navarro, J.J. Sendra, C. Barros, Design and calculation method for natural
is 1.52%, for the east facing façade the DA300lx [50%] pronounced lighting in architecture, in: Proceedings of the 3rd European Conference on
Architecture: Solar Energy in Architecture and Urban Planning, Florence, Italy,
at where the daylight factor is 1.35%, for the south facing façade May, 1993, pp. 371–373.
the DA300lx [50%] pronounced at where the daylight factor is [13] I. Acosta, J. Navarro, J.J. Sendra, Towards an analysis of daylighting simulation
1.20%, for the west facing façade the DA300lx [50%] pronounced software, Energies 4 (7) (2011) 1010–1024.
[14] Jiangtao Du, Steve Sharples, The variation of daylight levels across atrium
at where the daylight factor is 1.35%; walls: reflectance distribution and well geometry effects under overcast sky
conditions, Sol. Energy 85 (2011) 2085–2100.
These conclusions are obviously limited to measurement in the [15] J. Mardaljevic, Validation of a lighting simulation program under real sky
conditions, Lighting Res. Technol. 27 (4) (1995) 181–188.
test room with an unobstructed window and free from shading
[16] M. Fontoynont, P. Laforgue, R. Mitanchey, M. Aizlewood, J. Butt, W. Carroll,
facilities under the real weather in actual years (2014–2015). The et al. IEA SHC Task 21: validation of daylighting computer programs -ECBCS
test room with various glazing, façade types and blind systems Annex 29; Nov (1999).
[17] A. Laouadi, C.F. Reinhart, D. Bourgeois, Efficient calculation of daylight
should also be investigated to find the average daylight levels in
coefficients for rooms with dissimilar complex fenestration systems, J. Build.
daylit rooms. These issues will be studied in future work. Perform. Simul. 1 (2008) 3–15.
[18] C.F. Reinhart, J. Wienold, The daylighting dash board: a simulation-based
design analysis for daylit spaces, Build. Environ. 46 (2) (2011) 386–396.
Acknowledgments
[19] J. Du, S. Sharples, Assessing and predicting average daylight factors of
adjoining spaces in atriumbuildings under overcast sky, Build. Environ. 46
The work was supported by China National Key Research and (2011) 2142–2152.
Development Program (Project No. 2016YFC0700205), National [20] M. Aizlewood, J. Butt, K. Isaac, P. Littlefair, Daylight in atria: a comparison of
measurement, theory and simulation, in: Proceedings Lux Europa,
Natural Science Foundation of China (Project No. 51208205), Netherlands: Amsterdam, 1997.
autonomous research project of State Key Laboratory of Subtropical
Building Science (2015ZC15). We are thankful to Dr. Alstan Jakubiec
for his valuable advises and improving the quality of paper.

References

[1] C.F. Reinhart, C.J. Mardaljevic, Z. Rogers, Dynamic daylight performance

metrics for sustainable building design, Leukos 3 (1) (2006) 1–20.
[2] A. Shehabi, N. DeForest, A. McNeil, et al., The light harvesting potential of
dynamic daylighting windows, Energy Build. 66 (2013) 415–423.