You are on page 1of 15

Energy & Buildings 287 (2023) 113001

Contents lists available at ScienceDirect

Energy & Buildings


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

A study on daylighting metrics related to the subjective evaluation of


daylight and visual comfort of students in China
Binyan Liu a,b, Yubo Liu a,b,⇑, Qiaoming Deng a,b, Kai Hu a,b
a
School of Architecture, South China University of Technology, Guangzhou, China
b
State Key Laboratory of Subtropical Building Science, China

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

Article history: Uneven daylighting distribution in classrooms is a common issue in many countries, despite meeting the
Received 13 December 2022 standard daylighting factor (DF). The issue has been particularly acute in China. With the development of
Revised 13 March 2023 dynamic daylighting metrics, they have been incorporated into daylighting standards in various coun-
Accepted 16 March 2023
tries. The introduction of dynamic daylighting metrics in China is an opportunity to improve the class-
Available online 21 March 2023
room lighting environment, considering that the Chinese regulations have not been updated for a long
time. However, there is a dearth of research investigating the validity of dynamic daylighting metrics
Keywords:
in Chinese school buildings. Therefore, a correlation analysis of dynamic daylighting metrics and stu-
Daylighting metrics
Subjective evaluation
dents’ subjective evaluation has been carried out. This one-year study included field surveys and illumi-
Daylight simulation nance measurements collected in 18 classrooms in a Chinese school; the dynamic and static daylighting
Correlation metrics of these classrooms were simulated. The spatial daylight autonomy (sDA) at 450 lx for 50% of
annual hours (sDA450/50%; a dynamic metric) and the proportion of area with an illuminance
(Ep) > 300 lx (a static metric) were highly correlated with student evaluations. Additionally, sDA450/50%
50% was preferred by student evaluations. Annual sunlight exposure (ASE) was not correlated with stu-
dent evaluations in north-facing classrooms in the Guangzhou area. However, a correlation was found
between perceived blackboard glare and the percentage of space where disability glare probability
(DGP) exceeded a threshold of 0.35 for>5% of the usage time (sDGPexceed). Consequently,
sDA450/50%50% is a recommended metric for China’s school-building daylight standards, while
sDA250/50% 75% can be used simultaneously to improve the effective daylighting area of classrooms.
Ó 2023 Elsevier B.V. All rights reserved.

1. Introduction DF of 5% in classrooms. Later, to simplify calculations, the average


daylight factor (DFavg) was developed by Lynes and subsequently
Students spend one-third of their time at school. Good daylight improved by Crisp and Paul Littlefair. In 1992, the BSI issued
in a classroom not only improves the performance of students [1] ‘‘Lighting for Buildings–Part 2: Code of Practice for Daylighting”.
but may also prevent the onset of myopia in children [2,3]. Addi- This code replaced the 1945 version and changed the daylight stan-
tionally, using daylight illumination in a classroom may increase dard from the DF to the DFavg [6]. Subsequently, Climate-based
student satisfaction and save energy [4]. Therefore, the design of daylight modelling (CBDM) was first proposed by Mardaljevic at
good lighting in schools is an important and complex challenge the 2006 CIBSE National Conference [7]. The main principle of
for designers. Daylighting metrics play an important role in the CBDM is to design buildings for specific regional climate character-
classroom lighting environment. Designers refer to these metrics istics based on regional weather data. An increasing number of
to determine the proper size, orientation, and position of classroom researchers have also noted the shortcomings of traditional static
windows. daylighting metrics [8–11] and proposed a series of dynamic day-
In recent years, advances in daylighting metrics have led to the lighting metrics. Commonly used and representative dynamic day-
development of static and dynamic metrics. The British Standards lighting metrics include daylight autonomy (DA) [12] and useful
Institute (BSI) issued the first basic building codes in 1945 [5], stip- daylight illuminance (UDI) [13]. Spatial daylight autonomy (sDA)
ulating a minimum daylight factor (DF) of 2% and recommending a and annual sunlight exposure (ASE) [14] are the most recently
developed indicators. In 2012, the IES published the natural light-
⇑ Corresponding author. ing standard IES-LM-83-12, in which the recommended daylight-
E-mail address: liuyubo@scut.edu.cn (Y. Liu). ing indicators were sDA of 300 lx for 50% of annual hours

https://doi.org/10.1016/j.enbuild.2023.113001
0378-7788/Ó 2023 Elsevier B.V. All rights reserved.
B. Liu, Y. Liu, Q. Deng et al. Energy & Buildings 287 (2023) 113001

(sDA300/50%), corresponding to daylighting quality and ASE of in eight classrooms in five Italian universities and calculated a
1,000 lx for 250 h (ASE1,000/250h), corresponding to visual comfort set of objective metrics (daylight, circadian, and energy-related).
[14]. This standard was the first human-factor, evidence-based Significant correlations were observed for all daylighting metrics:
daylight standard approved by the IES [15]. In 2014, sDA and among them, the DFavg and ASE had higher correlations than the
ASE1,000/250h were codified by the latest version of Leadership in DA, sDA, and UDI [32]. In studies of Asian schools, Kong Z et al.
Energy and Environmental Design (LEED) (v4), which stipulated explored instantaneous and long-term lighting metrics based on
that within the total daily use area, 2 points are allotted for subjective evaluations within higher-education buildings in tropi-
sDA300/55% and 3 points are allotted for sDA300/75%; areas with cal weather and recommended 150–250 lx horizontal or vertical
ASE1,000/250h should not exceed 10% of the total area [16]. In illuminance as the lighting thresholds for classroom access to day-
2018, the European Union published EN17037:2018, the latest light [33,34]. Literature reviews have revealed a lack of consensus
international standard for natural lighting. The standard included on what lighting metrics to adopt for specific geographical loca-
the following four aspects: daylight provision, sunlight, view, and tions, climates, building types, and occupants.
glare. Daylight provision is measured with sDA and further divided In the study of daylighting in Chinese primary and secondary
into 3 levels based on the illumination threshold, which differs school classrooms, Bian Yu et al. studied the main variables affect-
from the recommendations of the IES. Glare is assessed with the ing classroom text legibility through a 10-month test. They noted
daylight glare probability (DGP). It is recommended that critical that luminance contrast is the dominant factor [35]. In another
glare situations exceeding 5% of the reference time (DGPt) should study, he used the contrast ratio over 1.3 on the blackboard, which
be limited [17]. is expressed by the ‘area ratio with (Probability of C  1.3)  75%’
In China, students’ seats in the classroom are usually fixed and as the indicator that has applicability in classrooms [21]. They are
do not change for at least several months. Additionally, there is a daylighting metrics studies aimed explicitly at predicting class-
high number of students per classroom, usually approximately room blackboard glare. Anxiao Zhang et al. adopted UDIavg for day-
45 students, and a high density of students. According to our pre- light evaluation purposes to optimize the window design
vious research, when students are in class, artificial lighting is parameters of school buildings in the cold climate of China [36].
always used during working hours due to the closure of curtains, Jin Ma and Qingxin Yang used the sDA and ASE as indicators to
which is used to control glare issues near windows. The same prob- optimize annual daylighting performance for atrium-based class-
lem has arisen in classrooms in Greece [18] and Malaysia [19]. rooms of primary and secondary schools in Nanjing, China [37].
Although the DF in some classrooms is up to standard, the lighting Yizhe Xu et al. took UDI as one of the multiple goals to optimize
distribution of classrooms is uneven, and glare [20]. The current the performance of primary and secondary school classrooms in
lighting standards in China still rely on static daylighting metrics southern China [38]. These studies aimed at exploring the applica-
(the DF and DGI). The emergence of dynamic daylighting metrics tion of dynamic daylighting metrics in design. However, there is a
provides a good opportunity for classrooms to improve daylighting lack of validation on the effectiveness of these dynamic daylighting
design and to identify appropriate daylighting indicators for use in metrics in terms of user evaluation about adequacy, glare, and sat-
China [21]. Some researchers have attempted to assess the applica- isfaction in primary and secondary schools in China. Consequently,
bility of the newly proposed daylighting standards. For example, it is necessary to revise daylight standards based on users’ expec-
Sepúlveda A et al. evaluated the validity of daylight standards tations in each geographic region. Therefore, given the reasons pre-
EN17037:2018 in specific local contexts, namely, cold climates viously mentioned, the main objectives of the current study are as
[22], and the building regulation concerning daylight provision in follows:
Sweden is considered to be updated. Five daylight performance cri-
teria from EN17037:2018 and Building Research Establishment 1. To provide technical details and methods that can be replicated
Environmental Assessment Methodology (BREEAM) as well as the in subsequent studies to widen the scope of research;
UDI were tested [23]. In addition, many researchers have explored 2. To provide suggestions on daylighting metrics that predict the
the correlation between users’ visual comfort and daylighting met- subjective evaluations of primary and secondary school stu-
rics [24–26]. Some studies have also discussed threshold values for dents in Guangzhou, China.
daylighting metrics based on subjective evaluation. Amir Nezam-
doost and Kevin Van Den Wymelenberg performed an eight-year 2. Methodology
human-factor daylighting field study and reported that the ASE
threshold should be improved [27]. Nastaran Seyed Shafavi et al. The research methodology included five steps. First, the reliabil-
proposed that UDI300–3,000/50% and ASE better predicted daylight ity of the Rhino model was determined by illuminance measure-
availability and visual discomfort than previous daylighting met- ment and simulation validation; second, a subject survey of the
rics [28]. students of the classrooms of interest was conducted; third, the
Several articles have investigated daylighting metrics and visual classrooms in the previous step were simulated to calculate the
comfort in school buildings. Aimilios Michael et al. assessed the daylighting metrics; and fourth, statistical analysis was used to
natural lighting performance and visual comfort of school build- compare the subjective survey results to objective daylighting
ings in Cyprus through a questionnaire-based survey and static metrics obtained via simulation. Finally, the correlations between
and dynamic simulations. The questionnaire results were consis- the daylighting metrics and student responses were assessed. A
tent with the simulation results [29]. Lourenço et al. discussed flowchart of the methods is presented in Fig. 1.
the light use patterns in Portuguese school buildings and suggested
that the current standards are not in line with users’ actual behav- 2.1. School description
iors and preferences, which are more diverse [30]. Zahra S.
Zomorodian et al., in a one-year subjective survey of four class- The school, which was designed and completed in 2017, was
rooms in two LEED buildings, reported high correlations of located in Guangzhou, China (Fig. 2). Guangzhou has a humid sub-
dynamic metrics (sDA and ASE) with occupant assessments and tropical climate influenced by the East Asian monsoon; thus, build-
showed that a vertical illuminance-based metric (sDGPexceed) was ing designs must provide shading and ventilation. Elementary
more compatible than horizontal illuminance-based metrics school buildings have a maximum of four floors according to Chi-
(ASE) in fitting the scope of perceived discomfort glare [31]. Verso nese building codes. Most classrooms in Guangzhou are designed
V R M L et al. investigated students’ subjective daylight judgment with a north–south orientation and equipped with windows on
2
B. Liu, Y. Liu, Q. Deng et al. Energy & Buildings 287 (2023) 113001

Fig. 1. Flow chart of the research methodology.

Fig. 2. Photograph of LuoGang Middle School.

both sides (i.e., windows on the north and south sides). LuoGang Data were recorded every five minutes for a total of 12  85 =
Middle School also uses windows on each side of the classroom 1,020 data points. In order to obtain the horizontal illumination
for lighting except on the fourth floor (top floor), which has extra value of the desk due to daylight, the measuring procedures
roof monitors, as presented in Fig. 3. involved the opening of the curtain on the side window and the
deactivation of the classroom lighting system. To include the light
2.2. Illuminance measurement and simulation validation reflectance of all surfaces and the light transmittance of windows
in the simulations, we approximated the reflectance of surfaces
The accurate prediction of illumination requires accurate simu- with two LX metres (JTG01) with an accuracy of ± 4% (0.1–
lation; thus, the Rhino model must be calibrated using measured 100,000 lx). One metre was directed towards the surface, and the
illuminance values. The illuminance data were collected on 25 other was oriented away from it. As a result, the reflectance of
September 2021 at 09:00–16:00 local time. The test classroom the surface was approximated according to the light absorbed by
was on the 4th floor. It had two side windows on the south wall the surface and the reflected light [22,34]. The reflectance and
and one on the north wall as well as a roof monitor facing north. transmittance values are presented in Table 1.
Twelve measurement points were arranged on the reference plane Rhino 7.0 and ver. 0.0.68 with Ladybug and Honeybee tools
at desk level (elevation of 0.75 m, following the regulations of the were used for point-in-time and annual daylight simulations. To
Chinese national standard). The distribution of measurement facilitate comparison with the measured data, the simulation data-
points in the plane is shown in Fig. 4. set had the same date and periods as the measured data. Using the

3
B. Liu, Y. Liu, Q. Deng et al. Energy & Buildings 287 (2023) 113001

Fig. 3. Representative classroom on the fourth floor (1). Representative classroom for the first to third floors (2).

Fig. 4. Locations of sensors (placed on desks) (1). Interior view of the classroom (2).

Table 1
Values of light reflectance of surfaces and light transmittance of windows.

Wall Floor Ceiling Window frame Grey wall (outside) White wall (outside) Window Ground Blackboard
0.5 0.4 0.5 0.5 0.3 0.4 0.6 0.2 0.2

tools mentioned above, a grid of sensors (12  12) was arrayed on dow frame, furniture, and so on. Therefore, we needed to identify
the reference plane, which was then divided into 12 sections cor- important variables for the simulation results and estimate
responding to the 12 measurement points. Each section contained whether including these model details would affect the simulation
12 sensors. The average simulation results of one set of 12 sensors results. The normalized mean bias error (NMBE) and coefficient of
represents the value of points 0–11 (Fig. 5). variation of the root mean square error (CVRMSE) of the simulated
Generally, the more detailed a model is, the more accurate the value and the measured value were compared to determine
simulation. In contrast, simplified models improve the speed of whether a model detail should be included. The following equa-
calculating simulation results. Some model details may greatly tions were used to calculate these values:
influence the simulation results, such as the reference plane, win- PN
i¼1 ðmi  siÞ
NMBEð%Þ ¼  ð1Þ
Nm
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1ffi
u0P
u N ðmisiÞ2
uB C
u@ i¼1 A
t N

CVRMSEð%Þ ¼  ð2Þ
m
where si and mi are the illuminance values obtained from simula-

tions and measurements, respectively; m is the average value of
the measured illuminance; and N is the number of values
compared.
Fig. 5. Grid of sensors (12  12) in Grasshopper (1). The reference plane was According to Chinese regulations, the distance between the
divided into 12 equal parts corresponding to 12 measured points (2). edge of desks in the front row and the blackboard should be at least
4
B. Liu, Y. Liu, Q. Deng et al. Energy & Buildings 287 (2023) 113001

2.2 m; additionally, the distance between the edge of desks in the on desk arrangement and inclusion of window frames) in later
last row and the opposite wall should be>1.1 m. Therefore, the simulations.
actual use area was smaller than the floor area within a classroom.
The scope of the reference plane needs to be adjusted to match the 2.3. Subjective survey
actual use area (i.e., adjusted considering the desk arrangement
regulations of China). Usually, the reference plane is assumed to We referred to [39,40] to devise an appropriate questionnaire.
be offset from the wall by 1200 to 2400 (according to IES LM-83-12 The present study sought to identify appropriate daylighting met-
[14]). Therefore, both sizes of the reference plane were constructed rics according to point-in-time simulations, annual illuminance
and used in the simulation and the calculation of NMBE and simulations, and questionnaire responses.
CVRMSE values; both sets of values were used to compare the Because of the COVID-19 pandemic, there was limited opportu-
results (Table 2). The results showed that the error of the reference nity to conduct field investigations at schools. Therefore, the ques-
plane offset based on the desk arrangement was smaller than the tionnaire was distributed online. Since some students were too
error based on the floor plan. young to understand the questions, the questionnaire was admin-
The window frame is an important part of the window details. istered to students in grades 3–5 (i.e., students 8–11 years of age)
According to IES LM-83-12, a window detail>200 in any dimension from 18 classes. The red numbers in Fig. 6 indicate the classes
should be calculated and summed into an ‘‘occlusion area”. The included in the present study and their locations in the building.
occlusion area is then translated into a reduction factor applied These students were asked to answer the online questionnaire
to the visible light transmission (VLT) of the specified glass. This (consisting of items rated on a 7-point scale) during one week in
method may be appropriate for building simulations when win- December 2021.
dow frame details are not known. However, for buildings previ- The questionnaires collected two types of information as fol-
ously constructed, the dimensions of the existing window frames lows: items 1–9 collected general information (age, sex, eyesight,
should be used. Then, the NMBE and CVRMSE values were used seat number, etc.), and items 10–23 collected information on light-
for error comparison. As shown in Table 3, the window frames ing perceptions (daylight sufficiency, glare, and daylight satisfac-
greatly influenced the results. tion). Items 10–23 were answered on a 7-point Likert scale and
The furniture can be modeled according to IES LM-83-12; how- focused on the present conditions during the assessments (point-
ever, it should be noted that this may increase the complexity and, in-time) or over an entire year (annual) in terms of three major
consequently, reduce the simulation speed. With the test room, we aspects: daylight sufficiency, glare, and daylight satisfaction. The
used NMBE and CVRMSE values to compare the simulations of fur- question topics are shown in Table 5. A total of 499 valid question-
nished and unfurnished classrooms. naires were received.
Overall, comparison of the simulation and measurement results
indicated reduced error in models constrained in size within the
2.4. Evaluation criteria & simulation
reference plane that included window frames and furniture
(Table 4). However, as furniture exerted only a small influence
Using the latest guidelines regarding annual climate-based day-
on the results and was substantially more labour-intensive, we
lighting metrics and performance criteria, studies have been car-
decided to include only the first two alterations (i.e., offset based
ried out in different countries [14,17]. The results indicate that

Table 2
Comparison of NMBE and CVRMSE values for different reference plane offset distances.

0 1 2 3 4 5 6 7 8 9 10 11
Offset based on the floorplan NMBE 0.59 0.04 0.14 0.44 0.43 0.31 0.10 0.00 0.39 0.07 0.24 0.23
CVRMSE 0.63 0.20 0.23 0.48 0.49 0.35 0.21 0.25 0.45 0.17 0.29 0.29
Offset based on desk arrangement NMBE 0.40 0.02 0.14 0.06 0.01 0.13 0.27 0.32 0.50 0.20 0.08 0.67
CVRMSE 0.45 0.20 0.23 0.19 0.19 0.22 0.32 0.36 0.56 0.26 0.18 0.69

Table 3
Comparison of NMBE and CVRMSE values with and without window frames.

0 1 2 3 4 5 6 7 8 9 10 11
With window frames NMBE 0.18 0.17 0.05 0.18 0.18 0.02 0.06 0.01 0.25 0.02 0.10 0.29
CVRMSE 0.27 0.26 0.19 0.26 0.27 0.17 0.18 0.18 0.35 0.16 0.19 0.32
Without window frames NMBE 0.4 0.02 0.14 0.06 0.01 0.13 0.27 0.32 0.50 0.20 0.08 0.67
CVRMSE 0.45 0.20 0.23 0.19 0.19 0.22 0.32 0.36 0.56 0.26 0.18 0.69

Table 4
Comparison of NMBE and CVRMSE values with and without furniture.

0 1 2 3 4 5 6 7 8 9 10 11
Unfurnished NMBE 0.18 0.17 0.05 0.18 0.18 0.02 0.06 0.01 0.25 0.02 0.1 0.29
CVRMSE 0.27 0.26 0.19 0.25 0.27 0.18 0.18 0.18 0.35 0.16 0.19 0.32
Furnished NMBE 0.16 0.19 0.06 0.19 0.19 0.03 0.05 0.00 0.23 0.00 0.11 0.27
CVRMSE 0.25 0.27 0.19 0.26 0.27 0.18 0.18 0.18 0.33 0.16 0.20 0.31

5
B. Liu, Y. Liu, Q. Deng et al. Energy & Buildings 287 (2023) 113001

Fig. 6. Locations of classrooms included in this study.

Table 5
Data collected and question topics included in the questionnaire.

General information Age Class Sex Eyesight Study time Floor Seat
Daylight sufficiency 1. Sufficiency of daylight in the classroom (when the curtains are fully open and the lights are off).
2. The ability to use daylight as the source of illumination.
3. Sufficiency of daylight on the desk (when the curtains are fully open and the lights are off).
Glare 4. Does the daylight ever cause glare strong enough to annoy you?
5. How disturbing is the brightness of your desk when you write?
6. How disturbing is the brightness of the blackboard?
7. Can you read the words on the blackboard?
Daylight 8. How satisfied are you with the daylight in the classroom?
satisfaction 9. How satisfied are you with the daylight on your desk?
10. How satisfied are you with the daylight on the blackboard?

the performance of the same metrics in different studies is dissim- tglare


fDGP; exceed ¼ ð4Þ
ilar or even contradictory [28]. This discrepancy may be due to the tref
influence of regional culture on individual judgement of daylight
provision and glare. To research daylighting metrics in China, we
summarize the dynamic and static metrics of daylight provision where tglare represents the time of year when vertical illuminance
and the dynamic metrics of glare. These metrics were drawn from (EV) is greater than a certain value (e.g., 2,668 lx), and tref is the
the standards of China, the US, the UK, etc., and from the recom- hours of daylight between 08:00 and 18:00 throughout the year.
mendations of experts and researchers. The maximum acceptable exceeding time in the whole year was
The whole building model was constructed in Rhino according 5% (fDGPexceed  5%).
to the detailed model verified in Section 2.2. Rhino and Grasshop- Because the classroom we studied had a wide corridor to the
per were used for static and dynamic daylight simulations, and the south, which provided shade, there was almost no direct light in
metrics in Table 6 were used as outputs. The results of all class- the classroom. Therefore, in this simulation, we chose a lower
rooms were automatically saved via Python. Then, the results of threshold for DGP partitioning in the specification. We set DGP val-
the classrooms corresponding to the survey were selected and ues to 0.35 or EV over 2,668 lx, which is the high recommendation
saved in Excel. for glare protection. In this study, the calculation was based on
EN17037:2018 introduced a new metric for glare, i.e., the per- hourly DGP values for 72 different perspectives at each position
centage of occupied hours with a DGP above a threshold value in the building. The default view height was 1.2 m above the fin-
(fDGPt) should not exceed 5%. This value was based on the concept ished floor (eye height for a seated observer). The frequency of dis-
proposed by [41]. The DGP was calculated based on a simplified turbing glare was visualised in Rhino using 72 directional pie
equation (Eq. (3). slices, with colour indicating frequency from 0 to 5% (Fig. 7). In this
jDGPsjwienold ¼ 6:22  105  Ev þ 0:184 ð3Þ study, we evaluate the results using sDGPexceed, which is the per-
centage of the space where the vertical illuminance is>2,668 lx
The value of fDGPexceed was calculated based on the equation in>5% of the occupation time. A room’s sDGPexceed is the average
below (Eq. (4): sDGPexceed calculated from the sDGPexceed results of all orientations
on the reference plane.
6
B. Liu, Y. Liu, Q. Deng et al. Energy & Buildings 287 (2023) 113001

Table 6
Summary of static and dynamic metrics of daylight provision and the dynamic metrics of glare.

Daylighting metric Metric source Thresholds in standards


Daylight Static Daylight factor Overcast conditions LEED v2 75% of the area DF  2%
provision metrics Average daylight factor Outdoor illumination China Architecture & DFavg  3%
of 15,000 lx Building Standard for
daylighting design of
buildings
Illuminance 9 am to 3 pm LEED v3 75% of the area E(250–5,000 lx)
Average Outdoor illumination China Architecture & Eavg  450 lx
illuminance of 15,000 lx Building Standard for
Uniformity Minimum/average daylighting design of When using a skylight for lighting,
buildings uniformity  0.7
Dynamic sDA300/50% LEED v4 sDA300/50%>55%, 2 points;
metrics sDA300/50%>75%, 3 points
IES LM 83–12 sDA300/50%>55% (nominally accepted)
sDA300/50%>75% (preferable)
EN17037:2018 sDA300/50%>50% & sDA100/50% >95% (minimum)
sDA500/50% >50% & sDA300/50% >95% (medium)
sDA750/50% >50% & sDA500/50% >95% (high)
UDI300–3,000 [13] UDI300–3,000/50%
Glare Dynamic ASE1,000/250h LEED v4 ASE1,000/250h  10%
metrics
IES LM 83–12 ASE1,000/250h  10% (minimum)
ASE1,000/250h  7% (medium)
ASE1,000/250h  3% (high)
DGP values Wienold and DGP values < 0.45
Christoffersen, 2006
sDGPexceed EN17037:2018 DGPe< 5%

on the desk) as well as the average values on these three questions


from an individual [(10 + 11 + 19)avg]. Satisfaction was represented
by the values of questions 12 (overall daylighting satisfaction), 18
(blackboard daylighting satisfaction), and 21 (desk daylighting sat-
isfaction). Glare was represented by the values of questions 13
(glare), 16 (blackboard glare), and 17 (legibility of words on the
blackboard).
Second, the classroom average of all valid responses to the
questions mentioned in the first step was calculated. Some ques-
tions had two values, the annual value and the point-in-time value.
Third, the daylighting metrics shown in Table 6 were simulated.
Among them, daylight provision correlated with 29 annual and 16
point-in-time metrics. Glare correlated with two annual metrics,
while satisfaction correlated with all of the above metrics.
Finally, the correlation between the simulation daylighting
metrics of 18 classes and the subjective assessments was calcu-
lated, as shown in Fig. 8. The correlation analysis was completed
in SPSS software (version 26.0, SPSS Inc).

3. Results and discussion

3.1. Survey results

3.1.1. Analysis of daylighting adequacy, glare and daylighting


satisfaction survey results
Fig. 7. A sample of the sDGPexceed simulation results for a classroom. To explore the survey results regarding the annual overall day-
lighting adequacy in further detail, the number of students who
responded with each rating (i.e., 1–7 points) was determined,
2.5. Correlation analysis and a box plot was created to display the proportion of participants
with each rating score (Fig. 9). Comparison of the means of each
The Spearman correlation coefficient (r) was calculated to group indicated that most students responded with a rating of 5.
assess whether there was a correlation between daylighting met- Comparison of the medians of each group indicated that most stu-
rics and subjective assessments. dents responded with scores of 4 and 7. Overall, the proportion of
The following steps were included in the analysis. students with ratings of 4–7 was greater than that of students with
First, the questionnaire data were divided into three categories: ratings of 1–3. Thus, most students considered the daylighting of
daylight provision, satisfaction, and glare. Daylight provision was classrooms acceptable or more than acceptable. The results of the
represented by questions 10 (daylighting adequacy), 11 (ability one-week assessment showed a similar trend to those of the
to use daylight as illumination), and 19 (sufficiency of daylighting annual assessment.
7
B. Liu, Y. Liu, Q. Deng et al. Energy & Buildings 287 (2023) 113001

Fig. 8. Correlation analysis steps.

Fig. 9. Overall daylighting adequacy (annual).

Glare was assessed with question 13 (‘‘Does the daylight ever Regarding satisfaction with daylighting, the results of question
cause glare strong enough to annoy you?”) and question 20 18 (‘‘How satisfied are you with the daylighting on the black-
(‘‘How disturbing is the brightness of your desk when you write?”); board?”) and question 21 (‘‘How satisfied are you with the day-
students provided high scores on these questions, with most lighting on your desk?”) were relatively consistent with the glare
responding with ratings of 6 and 7 points, respectively. assessment results on the blackboard and desk.
For question 16 (‘‘How disturbing is the brightness of the black- The box plots show the satisfaction of students in terms of
board?”) and question 17 (‘‘Can you read the words on the black- classroom daylight (question 12: ‘‘How satisfied are you with the
board clearly?”), the results were similar; the proportion of daylighting in the classroom?”) over one week and over one year
students providing ratings of 4–5 was slightly lower than that of (Fig. 11). According to the mean and median, most students pro-
students providing ratings of 6–7 (Fig. 10). On the whole, students vided ratings of 4 and 7. The proportion of students providing
did not appear to suffer from classroom glare, as there were only a scores of 4–5 was almost equal to that of students providing scores
few complaints.

8
B. Liu, Y. Liu, Q. Deng et al. Energy & Buildings 287 (2023) 113001

Fig. 10. Percentage of students reporting glare on the blackboard on a weekly (left) or annual (right) basis.

Fig. 11. Satisfaction with daylight in the classroom on a weekly (left) and annual (right) basis.

Fig. 12. Daylighting adequacy according to classroom.

of 6–7. Thus, student satisfaction with classroom daylight was tion of scores in these classrooms, we concluded that daylighting
between medium and excellent. adequacy varied according to the floor and presence of surround-
ing shade.
3.1.2. Comparative analysis of daylighting adequacy, glare, and A box plot was generated to compare blackboard glare evalua-
daylighting satisfaction between classrooms tion across classrooms (Fig. 13). The results showed that being
The box plot of the overall daylighting adequacy evaluation of on a higher floor and having no southern shade caused blackboard
18 classrooms is shown in Fig. 12. Comparison of the means indi- glare, as has been previously established. Notably, the mean and
cated that classrooms 401–403 and 505 had better daylighting median values for classrooms 501 and 505 were low, meaning that
adequacy. Comparison of the medians suggested that classrooms classrooms with no northern shade also had some blackboard
402, 403, 505, and 504 had better daylighting adequacy. However, glare.
the interquartile range of classroom 504 was the largest, which Fig. 14 shows the daylighting satisfaction of different classes.
indicates that its evaluation was highly variable. From the distribu- The daylighting satisfaction of students in classrooms 501 and
9
B. Liu, Y. Liu, Q. Deng et al. Energy & Buildings 287 (2023) 113001

Fig. 13. Glare according to classroom.

Fig. 14. Daylighting satisfaction according to classroom.

301 was relatively high when their glare assessment was low and quacy or satisfaction. UDI300–3000/50% was moderately correlated
their daylighting adequacy was high. Thus, we hypothesized that with daylighting adequacy (r = 0.482) and overall daylighting ade-
adequacy had a stronger influence on daylighting satisfaction than quacy (r = 0.519), as previously reported [28].
glare. More accurate predictions can be obtained when using different
thresholds for DA and sDA (e.g., 100, 200, 300. . ., 900, and 1000 lx).
3.2. Correlation results The correlation of sDA with daylighting adequacy, the ability to use
daylight as illumination, and the sufficiency of daylighting on the
To determine which metrics accurately predicted student desk were increased by increasing the thresholds above 300 lx.
assessments, we performed correlation analyses between student The strongest correlations were obtained when the sDA threshold
responses to questions 10–23 and daylighting metrics. was between 400 and 500 lx, for maximum values of sDA450/50%. In
particular, the overall sufficiency (daylighting adequacy, ability to
use daylight as illumination and sufficiency of daylighting on the
3.2.1. Correlations of students’ assessment of annual daylight provision
desk) were strongly correlated (0.6 < r < 0.8) with sDA450/50%
with satisfaction and dynamic metrics
(r = 0.631). In other words, sDA450/50% precisely predicted user eval-
The overall daylighting adequacy of classrooms was assessed
uation of daylighting provision. This result is similar to that of an
with questions 10 (daylighting adequacy), 11 (the ability to use
IES study, which reported a correlation between classroom day-
daylight as illumination), and 19 (sufficiency of daylighting on
lighting metrics and subjective surveys [15]. We also found a mod-
the desk). Daylighting satisfaction was assessed with question 21
erate correlation between sDA700/50% and desk daylighting
(desk daylighting satisfaction) and question 12 (overall daylighting
satisfaction as well as a moderate correlation (r = 0.599) between
satisfaction).
sDA450/50% and overall daylighting satisfaction.
Most of the variables were moderately correlated (0.4 < r < 0.6)
In general, sDA450/50% was the most relevant metric for predict-
(Table 7), however, there were no significant correlations of the
ing subjective evaluations.
spatial average UDI100–2,000 or UDI300–3,000 with daylighting ade-
10
B. Liu, Y. Liu, Q. Deng et al. Energy & Buildings 287 (2023) 113001

Table 7
Correlation coefficient of students’ assessment of annual daylight provision, satisfaction, and dynamic metrics. The largest values among the significant values in each column are
shown in bold.

Dynamic metric Daylighting Ability to use daylight Sufficiency of Adequacy + illumination + desk Desk daylighting Overall daylighting
Daylighting adequacy as illumination daylighting on the sufficiency (annual) satisfaction satisfaction
evaluation (annual) (annual) desk (annual) (annual) (annual)
UDI300-3000 0.429 0.413 0.436 0.439 0.403 0.328
UDI300-3000/50% 0.482* 0.452 0.440 0.519* 0.441 0.444
UDI100-2000 0.288 0.293 0.188 0.287 0.325 0.378
sDA100/50% 0.167 0.126 0.062 0.133 0.259 0.373
DA100 0.314 0.319 0.209 0.314 0.349 0.398
sDA200/50% 0.276 0.317 0.178 0.287 0.327 0.378
DA200 0.375 0.39 0.287 0.394 0.398 0.434
sDA250/50% 0.492* 0.469* 0.430 0.523* 0.501* 0.448
DA250 0.443 0.423 0.334 0.450 0.438 0.461
sDA300/50% 0.499* 0.484* 0.464 0.544* 0.463 0.444
DA300 0.434 0.446 0.367 0.467 0.439 0.468*
sDA400/50% 0.536* 0.542* 0.527* 0.603** 0.429 0.545*
DA400 0.491* 0.477* 0.443 0.530* 0.465 0.501*
sDA450/50% 0.561* 0.573* 0.545* 0.631** 0.513* 0.599**
DA450 0.509* 0.487* 0.473* 0.553* 0.467 0.521*
sDA500/50% 0.543* 0.552* 0.492* 0.596** 0.477* 0.579*
DA500 0.517* 0.491* 0.488* 0.563* 0.465 0.532*
sDA600/50% 0.471* 0.498* 0.406 0.515* 0.468 0.475*
DA600 0.518* 0.500* 0.488* 0.566* 0.46 0.538*
sDA700/50% 0.530* 0.518* 0.423 0.552* 0.538* 0.496*
DA700 0.521* 0.512* 0.472* 0.565* 0.468* 0.534*
sDA750/50% 0.448 0.465 0.289 0.449 0.389 0.419
DA750 0.516* 0.512* 0.458 0.558* 0.472* 0.527*
sDA800/50% 0.526* 0.511* 0.351 0.519* 0.511* 0.437
DA800 0.514* 0.515* 0.444 0.553* 0.475* 0.519*
sDA900/50% 0.439 0.462 0.292 0.446 0.447 0.418
DA900 0.508* 0.516* 0.423 0.542* 0.481* 0.507*
sDA1000/50% 0.403 0.394 0.211 0.376 0.376 0.343
DA1000 0.502* 0.515* 0.404 0.532* 0.486* 0.497*

*, p < 0.05; **, p < 0.01.

Table 8
Correlation coefficient of student point-in-time assessments of daylight provision, satisfaction, and static metrics. The largest values among the significant values in each column
are shown in bold.

Static metrics Daylighting Daylighting Ability to use Sufficiency of Adequacy + illumination Desk daylighting Overall daylighting
evaluation adequacy daylight to as daylighting on + desk sufficiency satisfaction satisfaction
(weekly) illumination the desk (weekly) (weekly) (weekly)
(weekly) (weekly)
% of area with Ep > n lux 50 0.298 0.168 0.170 0.246 0.269 0.332
100 0.533* 0.466 0.330 0.504* 0.451 0.510*
150 0.656** 0.578* 0.446 0.639** 0.499* 0.690**
200 0.606** 0.557* 0.473* 0.622** 0.575* 0.623**
250 0.588* 0.507* 0.431 0.582* 0.538* 0.589*
300 0.593** 0.578* 0.485* 0.629** 0.637** 0.579*
350 0.504* 0.39 0.356 0.479* 0.483* 0.463
400 0.535* 0.547* 0.456 0.583* 0.606** 0.570*
450 0.544* 0.459 0.421 0.544* 0.549* 0.529*
500 0.615** 0.537* 0.474* 0.620** 0.579* 0.560*
550 0.578 0.560* 0.484* 0.616* 0.612* 0.622*
600 0.535* 0.447 0.478* 0.560* 0.539* 0.567*
DFavg 0.577* 0.502* 0.408 0.566* 0.518* 0.575*
% of area with DF > 2% 0.654** 0.588* 0.455 0.645** 0.487* 0.694**
% of area with EP-LEED > 250 (at 0.406 0.399 0.36 0.443 0.519* 0.468
09:00)
Uniformity 0.464 0.363 0.392 0.468* 0.414 0.236

*, p < 0.05; **, p < 0.01.

3.2.2. Correlations of student point-in-time assessment of daylight nese metric, was not correlated with most of the assessments and
provision with satisfaction and static metrics was only moderately correlated with daylighting adequacy. There-
The DFavg is one of the main metrics of daylight standards in fore, these two metrics (the DFavg and uniformity) were inadequate
China and was moderately correlated with subjective reports of day- for predicting student subjective responses in China.
lighting adequacy and satisfaction (Table 8). However, none of these The correlation between student evaluations and Ep was stron-
variables were strongly correlated with user responses; in other gest for Ep values in the range of 150–300 lx. The area with DF > 2%
words, the DFavg was unable to predict student assessments of day- was most strongly correlated with student-reported daylighting
lighting adequacy and satisfaction. Uniformity, the other major Chi- sufficiency and satisfaction.

11
B. Liu, Y. Liu, Q. Deng et al. Energy & Buildings 287 (2023) 113001

Table 9
Correlation coefficient of students’ assessment of annual glare, satisfaction and dynamic metrics. The largest values among the significant values in each column are shown in
bold.

Dynamic metrics Daylighting Glare Legibility of words on the Blackboard glare Blackboard satisfaction Overall daylighting
evaluation (annual) blackboard (annual) (annual) (annual) satisfaction (annual)
ASE 0 0 0 0 0
Average sDGPexceed 0.128 0.141 0.491* 0.154 0.330

*, p < 0.05; **, p < 0.01.

3.2.3. Pearson correlations between students’ assessment of annual acceptable (3–4), and unacceptable (1–2). The x-axis in Fig. 15 rep-
glare and dynamic metrics resents the illumination threshold, which ranges from 100 to
The results are shown in Table 9. Because the ASE simulation 1,000 lx. The y-axis represents the percentage of space with this
results of all classrooms were zero, the correlation coefficient illumination level during 50% of the year.
between the analyzed ASE and student evaluations was also zero. The logarithmic dotted lines represent the relationship between
These classrooms had no direct sunlight because the window on the percentage of the area exceeding each illuminance threshold
the south side had a 3.5-metre-wide corridor that provided shade. and the corresponding subjective evaluation scores, similar to the
However, this finding also reflects the inadequacy of ASE as an method used in [27,28]. The orange trend line shows the transition
indicator of glare: when there is no direct light in classrooms, between preferable and acceptable; the logarithmic equation is
ASE does not reflect student evaluations of glare. In addition, the shown in Fig. 15 above the line (R2 = 0.89). The blue trend line
average sDGPexceed was moderately correlated with blackboard shows the transition between acceptable and unacceptable; the
glare, which was not as well predicted by other variables. logarithmic equation is shown below the line (R2 = 0.88). In other
words, given a set threshold of illumination, the percentage of the
area (i.e., classroom) with values above the orange trend line
4. Revision of acceptable metric thresholds reflects preferred values (i.e., higher subjective evaluations), while
the percentage of the area with values between the orange trend
4.1. Annual daylighting metrics line and the blue trend line reflects acceptable to preferable results.
Conversely, the percentage of the area with values below the blue
To determine the most acceptable illumination threshold of trend line reflects unacceptable results.
sDAn lux/50% based on user evaluations, the overall classroom day- According to the LEED v4 and IES LM83-12 standards, sDA300/50%
lighting adequacy results were divided into three categories should be at least 55% (although above 75% is preferable). As
according to ratings on a 7-point Likert scale: preferable (5–7), shown in Fig. 15, an illuminance of at least 53.9% is acceptable,

Fig. 15. Percentage of the area exceeding certain illuminance thresholds, plotted by student-reported daylighting adequacy.

12
B. Liu, Y. Liu, Q. Deng et al. Energy & Buildings 287 (2023) 113001

similar to the metrics in the two standards. However, the preferred metrics is suggested to ensure the uniformity of lighting conditions
illuminance in this study was at least 68.9%, which is slightly lower throughout the classroom.
than the previous two criteria.
The EN17037:2018 standard states that the medium level of
4.2. Point-in-time daylighting metrics
daylight provision is sDA500/50%>50% and sDA300/50%>95%, and the
minimum level is sDA300/50%>50% and sDA100/50%>95%. Using the
Similarly, to determine the acceptable minimum illuminance
logarithmic formula displayed in Fig. 15, we calculated that the
threshold for Ep > n lux according to user evaluations, we simu-
acceptable level is sDA500/50%>32.3%, sDA300/50%>53.9%, and
lated the point-in-time results as depicted in Fig. 16. The x-axis
sDA100/50%>100%, and the preferred level is sDA500/50%>45.34%,
represents the illuminance threshold (50–600 lx), and the y-axis
sDA300/50%>68.9%, and sDA100/50%>100%. The results fall between
represents the percentage of space with a certain illuminance
the medium and minimum levels according to the EN17037:2018
value.
standard.
The orange trend line shows the transition between preferable
In the present study, the metric that most strongly correlated
and acceptable; the logarithmic equation is shown in Fig. 16 above
with student evaluations was sDA450/50%. Based on the logarithmic
the line (R2 = 0.93). The grey trend line shows the transition
lines in Fig. 15, for an illumination threshold of 450 lx, the accept-
between acceptable and unacceptable; the logarithmic equation
able area percentage is at least 37.6%, and the preferred area per-
is below the line (R2 = 0.80).
centage is at least 50.2%. Considering the importance of natural
According to the orange trend line shown in Fig. 16, the pre-
light to children, we suggest that substantial consideration regard-
ferred percentage of space is at least 26.3% when the illuminance
ing methods to increase daylighting metrics to promote better
is 300 lx, much lower than the 75% in the LEED v3 guidelines. There
school-building design is merited. Therefore, we choose
are several reasons for this phenomenon. First, the high density of
sDA450/50% 50% as the annual daylighting metric. In addition,
school buildings in China may lead to a relatively uneven daylight
50% of the area encompassed more effective space than 37.6%.
environment. Second, the sample size in this study was limited,
When a larger area threshold, e.g. 75%, is needed to make the room
and many of the values (percentage of space of Ep > n lux) were
look more efficient or to distribute daylighting more evenly, the
small. Third, Chinese children are generally well-behaved; thus,
illuminance threshold is approximately 250 lx, calculated accord-
they may hesitate to provide a negative evaluation in surveys.
ing to the equation of the orange trend line in Fig. 15. However,
as shown in Table 7, sDA250/50% appears only moderately correlated
with subjective preference because the data were clustered. That 4.3. Comparisons with previous research
is, the median was close to the maximum value, indicating that
the sample was not normally distributed [27]. Therefore, it con- Two previous articles have used the same research methodol-
cluded that sDA450/50% 50% can be used as the metric that best ogy for threshold analysis. One study was a survey of 22 spaces
predicts subjective evaluation, while sDA250/50% 75% can be used (12 offices, 5 classrooms, and 5 other spaces) by Amir Nezamdoost
to control 3/4 of the classroom area. The combined use of these two et al. [27]. The other study was an examination of 20 architectural
office spaces by Nastaran Seyed Shafavi et al. [28]. Results are

Fig. 16. Percentage of area exceeding certain illuminance thresholds, plotted by student-reported daylighting adequacy.

13
B. Liu, Y. Liu, Q. Deng et al. Energy & Buildings 287 (2023) 113001

Fig. 17. Comparison of acceptable and preferred daylight levels in current study and the previous studies, annual (1), point-in-time (2).

shown in Fig. 17. In terms of the results of the annual simulations,  In terms of daylight provision metrics, sDA450/50% was strongly
the trend lines for the preferred space in this study and in Nastaran correlated with daylighting adequacy; according to the illumi-
Seyed Shafavi et al. are relatively similar, with most of the differ- nance threshold chart, individuals preferred sDA450/50% of at
ences falling within 5 percent. The difference between our results least 50%. The illuminance threshold was in accordance with
and those of Amir Nezamdoost et al. is approximately 10 percent. the current national standard in China. Thus, sDA450/50%50%
However, comparison of the point-in-time simulation results is a recommended metric for China’s school-building daylight
revealed the lower limits of the acceptable and preferred ranges standards, while sDA250/50% 75% can be used to improve the
differ significantly among the three studies. Specifically, the effective daylighting area of classrooms. The two metrics can
threshold results of this paper are much lower than those of the be used in combination to maintain the uniformity of the
other two studies. In this study, the students were satisfied with classroom.
low illumination levels, possibly driven by student reluctance to  Regarding glare, there was a moderate relationship between the
say that they were dissatisfied. In addition, these three papers var- average value of sDGPexceed and blackboard glare. No evidence
ied in terms of research objects and geographical locations; thus, showed that ASE had a relationship with students’ glare assess-
user evaluations of daylight may have varied. It is necessary to ment. Further investigation of different types of classrooms,
revise daylight standards based on users’ expectations in each geo- such as those with windows on opposing walls, is needed. As
graphic region. for glare metrics, more detailed studies are required.

4.4. Metric application prospect The limitations of this study include the number and age of stu-
dents involved, the classroom design (i.e., window placement), and
We further compared annual (Fig. 15) and point-in-time met- the geographic location (research took place in only the Guangz-
rics (Fig. 16). The point-in-time findings are quite different from hou area). Therefore, similar studies with a larger sample size
previous studies, possibly due to the uneven daylight distribution and age range and with more classroom types may obtain more
and the use of the DFavg [20]. On the other hand, the DFavg and accurate results.
the percentage of area with Ep > n lux are both static metrics that Future studies should expand the sample size, age range, and
are unable to represent the annual daylight environment. In addi- classroom types included by replicating the methods of this study
tion, given that school buildings have long-term use, dynamic day- and developing truly human-centric daylight standards for China.
lighting metrics are more appropriate. With the rapid development of ML algorithms, building lighting
The use of dynamic daylighting metrics has also increased in simulation results can be applied during the early stage of design.
other countries. Researchers have conducted similar studies and Thus, designers can fully understand the natural lighting in a class-
even attempted to apply daylighting metrics in actual building room when making decisions regarding window design (e.g., the
design. Moreover, with the development of machine-learning inclusion of roof monitors, clerestories, or skylights).
(ML) algorithms, dynamic daylighting metrics are relatively easy
to obtain in the early stage of design and also reduce simulation
time. The designers do not need to know how to use daylighting
simulation software. Thus, the annual daylight performance can CRediT authorship contribution statement
be evaluated at the conceptual stage of design to support decisions
regarding the fenestration design of buildings [42]. Binyan Liu: Conceptualization, Methodology, Validation, Inves-
tigation, Writing – original draft, Formal analysis, Data curation.
5. Conclusion and future work Yubo Liu: Writing – review & editing, Project administration.
Qiaoming Deng: Resources, Project administration. Kai Hu: Soft-
In the present study, we simulated daylighting conditions in 18 ware, Data curation.
classrooms in a Guangzhou middle school and administered ques-
tionnaires to students to determine the correlation between day-
lighting metrics (daylighting adequacy, satisfaction, and glare)
and student subjective evaluations. Subsequently, according to Data availability
the questionnaire results, metrics were ranked and appropriate
thresholds were determined. The conclusions of this study are: Data will be made available on request.
14
B. Liu, Y. Liu, Q. Deng et al. Energy & Buildings 287 (2023) 113001

Declaration of Competing Interest [22] A. Sepúlveda, F. De Luca, T. Varjas, J. Kurnitski, Assessing the applicability of
the European standard EN 17037: 2018 for office spaces in a cold climate,
Build. Environ. 225 (2022).
The authors declare that they have no known competing finan- [23] I. Bournas, Daylight compliance of residential spaces: Comparison of different
cial interests or personal relationships that could have appeared performance criteria and association with room geometry and urban density,
Build. Environ. 185 (2020).
to influence the work reported in this paper.
[24] N.S. Shafavi, Z.S. Zomorodian, M. Tahsildoost, M. Javadi, Occupants visual
comfort assessments: A review of field studies and lab experiments, Sol.
Acknowledgments Energy 208 (2020) 249–274.
[25] M. Fakhari, V. Vahabi, R. Fayaz, A study on the factors simultaneously affecting
visual comfort in classrooms: A structural equation modeling approach, Energ.
The work was supported by the National Natural Science Foun- Build. 249 (2021).
dation of China under Grant No. 51978269 & No. 51978268. [26] R. Viula, T. Hordijk, in: Testing the predictive power of visual discomfort from
glare metrics in the near-window and near-wall zones of the daylit classroom
environment, International Commission on Illumination, CIE, 2019, pp. 282–
References 294.
[27] A. Nezamdoost, K. Van Den Wymelenberg, A daylighting field study using
[1] L. Heschong, Daylighting and human performance, ASHRAE J. 44 (6) (2002) 65– human feedback and simulations to test and improve recently adopted annual
67. daylight performance metrics, J. Build. Perform. Simul. 10 (5–6) (2017) 471–
[2] M. He, F. Xiang, Y. Zeng, J. Mai, Q. Chen, J. Zhang, W. Smith, K. Rose, I.G. Morgan, 483.
Effect of time spent outdoors at school on the development of myopia among [28] N.S. Shafavi, M. Tahsildoost, Z.S. Zomorodian, Investigation of illuminance-
children in China: a randomized clinical trial, JAMA 314 (11) (2015) 1142– based metrics in predicting occupants’ visual comfort (case study:
1148. Architecture design studios), Sol. Energy 197 (2020) 111–125.
[3] Z. Zhou, T. Chen, M. Wang, L. Jin, Y. Zhao, S. Chen, C. Wang, G. Zhang, Q. Wang, [29] A. Michael, C. Heracleous, Assessment of natural lighting performance and
Q. Deng, Y. Liu, I.G. Morgan, M. He, Y. Liu, N. Congdon, C.-W. Pan, Pilot study of visual comfort of educational architecture in Southern Europe: The case of
a novel classroom designed to prevent myopia by increasing children’s typical educational school premises in Cyprus, Energ. Build. 140 (2017) 443–
exposure to outdoor light, PLoS One 12 (7) (2017) e0181772. 457.
[4] D. Guyon, Daylight Dividends Case Study: Smith Middle School, Chapel Hill, [30] P. Lourenço, M.D. Pinheiro, T. Heitor, Light use patterns in Portuguese school
NC, J. Green Build., 1 (1) (2006) 33-38. buildings: User comfort perception, behaviour and impacts on energy
[5] W. Wu, E. Ng, A review of the development of daylighting in schools, Light. Res. consumption, J. Clean. Prod. 228 (2019) 990–1010.
Technol. 35 (2) (2003) 111–124. [31] Z.S. Zomorodian, M. Tahsildoost, Assessing the effectiveness of dynamic
[6] British Standards Institution, BS-8206-2: 1992 – Lighting for Buildings – Part metrics in predicting daylight availability and visual comfort in classrooms,
2: Code of practice for daylighting, BSI, London, 1992. Renew. Energy 134 (2019) 669–680.
[7] J. Mardaljevic, Examples of Climate-Based Daylight Modelling, CIBSE National [32] V.L. Verso, F. Giuliani, F. Caffaro, F. Basile, F. Peron, T. Dalla Mora, L. Bellia, F.
Conference: Engineering the Future, 2006. Fragliasso, M. Beccali, M. Bonomolo, Questionnaires and simulations to assess
[8] D. Bourgeois, C.F. Reinhart, G. Ward, Standard daylight coefficient model for daylighting in Italian university classrooms for IEQ and energy issues, Energ.
dynamic daylighting simulations, Build. Res. Inf. 36 (1) (2008) 68–82. Build. 252 (2021).
[9] M. Jamnický, Building information modeling with static and dynamic daylight [33] Z. Kong, J.A. Jakubiec, Instantaneous and long-term lighting design metrics for
analysis, in: Advanced Materials Research, Trans Tech Publ, 2014, pp. 255-258. higher education buildings in a tropical climate, East & West 48 (58.6) (2019)
[10] J. Mardaljevic, L. Heschong, E. Lee, Daylight metrics and energy savings, Light. 83.87.
Res. Technol. 41 (3) (2009) 261–283. [34] Z. Kong, J.A. Jakubiec, Instantaneous lighting quality within higher educational
[11] S. Kleindienst, M. Bodart, M. Andersen, Graphical representation of climate- classrooms in Singapore, Front. Arch. Res. 10 (4) (2021) 787–802.
based daylight performance to support architectural design, Leukos 5 (1) [35] Y. Bian, J. Luo, T. Luo, T. Leng, Contrast demand on the blackboard in typical
(2008) 39–61. secondary school classrooms of China: Effects of daylight reflections on text
[12] C.F. Reinhart, J. Mardaljevic, Z. Rogers, Dynamic daylight performance metrics legibility, Energ. Build. 261 (2022).
for sustainable building design, Leukos 3 (1) (2006) 7–31. [36] A. Zhang, Q. Huang, Y. Sun, Q. Zhang, Multi-objective window optimization of
[13] A. Nabil, J. Mardaljevic, Useful daylight illuminances: A replacement for school buildings for thermal and daylight performance in the cold climate of
daylight factors, Energ. Build. 38 (7) (2006) 905–913. China, in, in: Proceedings of 33rd PLEA International Conference: Design to
[14] IESNA, IES LM- 83-12, Approved Method: IES Spatial Daylight Autonomy (sDA) Thrive, 2017, pp. 4461–4468.
and Annual Sunlight Exposure (ASE). New York, NY, USA, (2012). [37] J. Ma, Q. Yang, Optimizing Annual Daylighting Performance for Atrium-Based
[15] L. Heschong, Heschong Mahone Group, Daylight Metrics PIER Daylighting Plus Classrooms of Primary and Secondary Schools in Nanjing, China, Buildings 13
Research Program, Technical Report, California Energy Commission, 2011. (1) (2022) 11.
[16] US Green Building Council (USGBC), LEED Reference Guide for Building Design [38] Y. Xu, C. Yan, H. Qian, L. Sun, G. Wang, Y. Jiang, A novel optimization method
and Construction, LEED V4, Technical Report, USGBC, Washington DC, USA, for conventional primary and secondary school classrooms in southern China
2013. considering energy Demand, Thermal Comfort and Daylighting, Sustainability
[17] European Comission, BS EN 17037:2018: Daylight in Buildings, 2018. https:// 13 (23) (2021) 13119.
www.en-standard.eu/bs-en-17037-2018-daylight-in-buildings/. [39] S.S. Korsavi, Z.S. Zomorodian, M. Tahsildoost, Visual comfort assessment of
[18] T. Theodosiou, K. Ordoumpozanis, Energy, comfort and indoor air quality in daylit and sunlit areas: A longitudinal field survey in classrooms in Kashan,
nursery and elementary school buildings in the cold climatic zone of Greece, Iran, Energ. Build. 128 (2016) 305–318.
Energ. Build. 40 (12) (2008) 2207–2214. [40] W. Wei, An investigation into the relationship between daylighting quality and
[19] N.L.N. Ibrahim, A.D. Mathalamuthu, V. Ponniah, M.W.M. Shafiei, R. Ismail, quantity for school buildings in Hong Kong, The Chinese University of Hong
Illuminance uniformity using Public Works Department (PWD) standard Kong (Hong Kong), 2003.
design for public schools classroom design in Malaysia, J. Adv. Res. Fluid Mech. [41] J. Wienold, Dynamic daylight glare evaluation, in: Proceedings of Building
Thermal Sci. 52 (2) (2018) 205–214. Simulation, Citeseer, 2009, pp. 944-951.
[20] P. Tsikra, E. Andreou, Investigation of the energy saving potential in existing [42] Y. Han, L. Shen, C. Sun, Developing a parametric morphable annual daylight
school buildings in Greece. The role of shading and daylight strategies in visual prediction model with improved generalization capability for the early stages
comfort and energy saving, Procedia Environ. Sci. 38 (2017) 204–211. of office building design, Build. Environ. 200 (2021).
[21] Y. Bian, Y. Chen, J. Luo, Y. Sun, Contrast-based evaluation of the reflected
daylight glare on the blackboard in a classroom with complex fenestration
systems, J. Build. Eng. 64 (2023).

15

You might also like