Journal of Building Engineering 33 (2021) 101631

Contents lists available at ScienceDirect

Journal of Building Engineering

journal homepage: http://www.elsevier.com/locate/jobe

Comprehensive evaluation of thermal and energy performance of radiative

roof cooling in buildings
Jianheng Chen , Lin Lu *
Department of Building Services Engineering, The Hong Kong Polytechnic University, Hong Kong

A R T I C L E I N F O A B S T R A C T

Keywords: This paper newly established a radiative roof cooling performance model to conduct thorough parametric studies
Radiative roof cooling on pivotal factors governing achievable equilibrium temperature and net cooling power in building roofs. A
Solar reflectance detailed building model integrated with radiative roof cooling considering typical climate regions in China was
Thermal emittance
correspondingly developed to evaluate building thermal response and energy efficiency. The results indicate that
Thermal response
Building energy saving
compared with broadband and transmittance based adaptive emitters as roof surfaces, atmospheric window
based selective emitter can achieve the minimum sub-ambient equilibrium temperature. Net cooling powers are
reduced by 13–33.6 W/m2 and sub-ambient equilibrium temperatures are weakened by 7.1–13.5 ◦ C with at­
mospheric precipitable water vapor increasing from 2.3 to 10 mm. The cooler degree of roof top temperature can
reach over 30 ◦ C under radiative roof surface which is conducive to avoiding thermal shock. The roof bottom
temperature reduction noticeably reaches up to 13.8 ◦ C for non-air-conditioned buildings, guaranteeing a more
thermally comfortable built environment. The annual energy savings turn out to be positive for five typical cities,
where cooling load dominant regions exhibit the most saving potentials with the saving rates of 7.1% for Hong
Kong and 9.3% for Kunming respectively. With the application of a radiative roof cooling surface, the heat
resistance layer can be removed from the thermal perspective for cost savings. The study sheds light on the
promising benefits of radiative roof cooling for achieving energy efficient buildings in China.

1. Introduction potentials will be enhanced accordingly. Basically, the traditional way

to improve roof thermal insulation includes the addition of thick thermal
Due to the skyrocketing growth in social economy and urbanization, resistance layers with low conductivities to reduce the heat transmission
building industry accounts for near 1/3 of total energy consumption and through roof layers, which, however, is not cost-effective and sometimes
contributes to 40% of carbon emissions [1]. Moreover, the growing unsafe in regard to the flammability of numerous thermal insulation
desire for a better indoor built environment has increased the energy materials. Technically, the exterior roof surface constantly interacts
consumption in heating, ventilation and air-conditioning systems to take with external environment in radiative forms, which include, for
up around 65% of total energy usage in building industry which will be instance, the reflection of incident solar power, the emission of infrared
the main energy users in the years to come [2]. In particular, for growing thermal radiation, the absorption of incoming atmospheric radiation,
commercial buildings, where energy consumption was predicted to be and so forth. The process significantly influences the roof heat gain, thus
10–20 times larger than that in residential buildings [3], the energy affecting building energy consumption. If the roof radiative heat transfer
saving appears more significant for stakeholders [4]. Therefore, seeking can be controlled to optimize the achievable heat gain by minimizing
an effective and cost-efficient way to address the building energy issues heat absorption and maximizing heat dissipation for cooling purpose,
turns out to be urgent facing the building industry. the building energy efficiency will be greatly improved, which is bene­
Actually, roof surfaces are the main receivers of solar heat gain ficial to the society both economically and environmentally.
regarding buildings with high roof-to-wall ratios, which may account for Radiative cooling has been broadly investigated for several decades
5%–10% of total building energy consumption and more than 40% of in regard to its applications in building cooling as well as freezing water
energy usage for top-floor buildings [5]. Thereby, if the roof thermal desalination, and its feasibility has been already successfully verified
insulation performance is signieficantly improved, the energy saving [6–8]. The nocturnal radiative cooling usually entails the near

* Corresponding author.
E-mail address: [email protected] (L. Lu).

https://doi.org/10.1016/j.jobe.2020.101631
Received 18 February 2020; Received in revised form 20 June 2020; Accepted 27 June 2020
Available online 3 July 2020
2352-7102/© 2020 Elsevier Ltd. All rights reserved.
J. Chen and L. Lu Journal of Building Engineering 33 (2021) 101631

blackbody emitter which emits radiation strongly across the entire [39–41], the surface heat flux which may exert great impacts on the
thermal spectrum. Through the highly emissive property, the heat can achievable radiative cooling potential should be analyzed in detail by
be released continuously into the atmosphere, thus achieving passive considering corresponding components interacting with the roof
cooling. While daytime radiative cooling is much more challenging due surface.
to the absorption of solar radiation, which could totally undermine or Considering a building with a flat roof, the energy flows and ex­
even eliminate the cooling capacity brought by the emitter-based heat changes acting on the exterior roof surface are shown in Fig. 1, where
dissipation. Promisingly, radiative cooling during daytime period has four components conducting heat exchanges include the incident solar
seen rapid development due to the advent of photonic fabrication radiation Psun, the downward atmospheric radiation Patm absorbed by
methods and metamaterials [9–15], the products of which are capable of the surface, the outgoing infrared thermal radiation Prad for heat dissi­
highly reflecting incident solar radiation and meantime selectively pation, and the non-radiative heat transfer with the external environ­
emitting the thermal radiation. By using a thermal photonic approach ment Pnonrad.
[10,16], the spectrally selective profile of a radiator consisting of seven Assuming a roof surface at temperature T whose spectral and angular
layers of HfO2 and SiO2 was fabricated, which enables to reflect 97% of emissivity is εr(λ, θ), when the surface is exposed to a daylight sky, it is
solar energy whilst emit selectively in the atmospheric transparency subject to both solar irradiance and atmospheric thermal radiation
window. By applying the photonic radiator, 4.9 ◦ C of sub-ambient under ambient air temperature Tamb. Taking into account all the heat
temperature and 40.1 W/m2 of cooing power have been reached. To exchange processes, the net cooling power Pnet of a radiative roof surface
improve the scalable-manufactured ability, Zhai et al. [14,17] proposed which is equivalent to the surface heat flux can be defined as:
a glass-polymer based metamaterial with randomized dielectric micro­
Pnet (T) = Prad (T) − Patm (Tamb ) − PSun − Pnonrad (1)
spheres to fulfill the radiative cooling potential. The radiator is equipped
with the infrared emissivity greater than 0.93 and solar reflectance of where the power radiated out from the surface per unit area is:
96% with the cooling power of 93 W/m2 being demonstrated.
∫ π2 ∫∞
With the spiral development of new radiative cooling materials, their
/

Prad (T) = 2π sin θ cos θdθ IBB (T, λ)εr (λ, θ)dλ (2)
integrations with building’s roof systems are capable of contributing to 0 0
considerable benefits regarding the possible cooling load savings and
concomitant capital and operational cost reductions in air-conditioning Here:
systems. Before the emergence of the spectrally selective radiators, the IBB (T, λ) is the spectral radiance of a blackbody at temperature T,
benefits of radiative cooling combined with roof systems have been W⋅sr− 1 m− 3, which can be expressed as:
realized and corresponding cool roof schemes have been carried out in
2hc2 1
America [18,19] and then got widespread expansion in some countries IBB (T, λ) = (3)
λ5 ehc/λkB T − 1
including Britain [20], Greece [21], Singapore [22–24], France [25],
Italy [26], Japan [27], Malaysia [28], and so forth. Cool roof is a good
Here:
choice to make buildings stay cooler in a hot summer [29] and thereby
provide more thermally comfortable built environment [30–32]. Some
h is Planck’s constant, 6.62607004 × 10− 34 J∙s;
novel cool roof systems were proposed for building cooling. For
kB is the Boltzmann constant, 1.38064852 × 10− 23 J/K;
instance, Yew et al. [28,33] proposed a new cool roof system integrated
c is the speed of light in vacuum, 3.0 × 108 m/s;
with a moving air cavity to strengthen the cooling ability of attic air
λ is the wavelength, m.
temperature. The benefits from cool roof include the reduction in
cooling energy consumption, carbon emission, air pollutants, and urban
heat island problems owing to the passive thermal management way
[21,22,32,34–36]. However, investigations on radiative roof cooling in
China are still insufficient [37], which requires further research
considering its varying climatic characteristics [38]. Therefore, the
paper, which aims to thoroughly investigate the radiative roof cooling
potentials and its impacts on the thermal and energy performance in
buildings under varying climates, first conducted extensive parametric
studies based on an established cooling power performance model to
epitomize the key factors exerting crucial impacts on the achievable
cooling performance. According to the results of parametric analysis
which elucidate the importance of spectral properties of radiative roof
surfaces and atmospheric conditions on the achievable cooling perfor­
mance, a building model was accordingly developed with a special focus
on China considering its varying climate zones to investigate and eval­
uate in detail the impacts of radiative roof cooling on building thermal
and energy performance. The study provides insights into the applica­
tion of radiative roof cooling in buildings in China.

2. Theoretical study of radiative roof cooling

2.1. Model establishement

The heat transfer between the exterior roof surface and the outside
atmosphere is regarded as a boundary condition for the overall heat
transfer model of roof structures. Due to the complex and time-varying Fig. 1. Energy flows through a radiative roof surface, including four compo­
external environment directly facing the roof surface, including, for nents: incoming solar radiation Psun, downward atmospheric radiation Patm,
instance, the stochastic convective heat exchange, the incident solar outgoing infrared thermal radiation Prad and non-radiative heat ex­
radiation and incoming atmospheric radiation absorbed by the roof change Pnonrad.

2
J. Chen and L. Lu Journal of Building Engineering 33 (2021) 101631

Based on the Kirchhoff’s radiation law, the surface’s absorptivity

equals to its emissivity. Therefore, the amount of incident atmospheric
radiation that is absorbed by the roof surface per unit area can be
expressed as:
∫ π2 / ∫∞
Patm (Tamb ) = 2π sin θ cos θdθ IBB (Tamb , λ)εr (λ, θ)εa (θ, λ)dλ (4)
0 0

where, εa (θ, λ) is the angle and spectrum dependent emissivity of the

atmosphere, which can be obtained by Ref. [7]:

εa (θ, λ) = 1 − [1 − εa (0, λ)]1/cosθ = 1 − [t(λ)]1/cosθ (5)

where, εa (0, λ) is the spectrum dependent emissivity in the normal di­

rection of the surface, and t(λ) is the atmospheric transmittance in the
zenith direction [42], and θ is the zenith angle. The atmospheric
transmittance is highly dependent on the local weather conditions, like
precipitable water vapor. Thereby, regarding the calculation of atmo­
spheric radiation absorbed by a radiative surface, the indigenous cli­ Fig. 3. Relations between atmospheric transmittance and precipitable
matic conditions should be taken into account accordingly. The water vapor.
distributive characteristics of atmospheric transmittance are presented
in Fig. 2 which takes the location of Cerro Pachon in Chile under the
Pnonrad (T, Tamb ) = hc (Tamb − T) (7)
condition of Air Mass 1.5 and precipitable water vapor column of 10 mm
as a reference [43]. The wavelength range from 8 to 13 μm is regarded as
hc is the combined non-radiative heat transfer coefficient which com­
the atmospheric transparency window, which exhibits the largest
bines the effects of heat conduction and convection interacting with the
transmittance for thermal radiation. Coincidently, the atmospheric
radiative surface, like the surrounding wind flow based heat exchange.
transparency window falls within the peak thermal radiation of the
The coefficient can reflect the collective effects of conductive and
blackbody spectral irradiance defined by Planck’s law under Tamb = 30
convective heat exchange between the radiative surface and ambient
◦
C. The characteristic elucidates the potential that terrestrial bodies at
air.
the ambient temperature can dissipate heat to outer space through the
The established heat transfer model with a focus on the exterior roof
atmospheric window, the existence of which provides the possibility for
surface, as illustrated in Equation (1), basically correlates the net cooling
terrestrial objects to radiate more heat than it absorbs from downward
power Pnet (T) with its surface temperature T. If the surface under a sub-
atmospheric radiation for passive cooling [39].
ambient temperature radiates more heat onto outer space than it gains
Additionally, the increase of atmospheric precipitable water vapor
from the absorption of incident solar irradiance, downward atmospheric
will definitely reduce the sky transmissivity, and thus increase the
thermal radiation and non-radiative heat exchanges, the net cooling
downward atmospheric radiation. The impacts of precipitable water
power shall turn out to be positive and the radiative surface can be
vapor on the atmospheric transmittance taken from Cerro Pachon [43]
regarded as the daytime cooling radiator. The achievable net cooling
are presented in Fig. 3 with a close-up on the atmospheric transparency
power Pnet (T) is defined when the surface temperature reaches the
window.
ambient air temperature Tamb. If the net power reduces to zero, the
The incident solar power that is absorbed by the roof surface per unit
radiative surface would reach a minimum steady-state equilibrium
area can be defined as:
temperature Teq. Therefore, two pivotal parameters are basically adop­
∫∞
ted to evaluate the cooling performance of a radiative surface, namely
PSun = IAM1.5 (λ)εr (λ, θSun )dλ (6)
0
the cooling power Pnet (T = Tamb) when the surface temperature main­
tains at the ambient air temperature, and the equilibrium temperature
where, IAM1.5 (λ) is the radiation of AM 1.5 spectrum [44] to represent Teq (Pnet = 0) when cooling power reaches zero. The following calcu­
solar illumination as shown in Fig. 4. lation will base on the two parameters to compare the cooling capacity
The power lost due to the convection and conduction between the of the roof surface with different spectral properties.
roof surface per unit area and external environment is expressed as:

Fig. 2. Atmospheric spectral transmittance and blackbody spectral irradiance. Fig. 4. Spectral emissivity of three ideal radiative cooling surfaces.

3
J. Chen and L. Lu Journal of Building Engineering 33 (2021) 101631

2.2. Calculation assumption making a dent in the achievable passive cooling potential [17,39,
45–48]. In this paper, the weather data with varying precipitable
In order to theoretically analyze and evaluate the achievable net water vapors are taken from Cerro Pachon [43] to calculate the
cooling power and equilibrium sub-ambient temperature of a roof sur­ corresponding atmospheric radiations. Due to the fact that radiative
face, the following assumptions are considered before performing cooling is particularly meaningful in tropical zones where ambient
parametric study on the cooling power performance. air temperature remains slightly changed [49], the ambient air
temperature is considered as 30 ◦ C for simplification.
1) Three kinds of ideal emitters as roof surfaces are chosen, namely 3) Although ideal emitters are assumed to totally reflect all the incident
broadband emitter, atmospheric window based selective emitter, solar radiation, the incoming sunlight will more or less be absorbed
and transmittance based adaptive emitter respectively. The broad­ in actual application scenarios. Thereby the paper also considers the
band emitter is equipped with an emissivity like a blackbody within solar absorption rate in some cases for comparative analysis.
the entire emission band over the atmosphere except the main solar
spectral band from 0.3 to 3 μm. While the atmospheric window based 2.3. Parametric study
selective emitter comprises unity emissions only within the wave­
lengths between 8 and 13 μm, selectively spanning over the atmo­ 2.3.1. Analysis of cooling power components
spheric transparency window [39]. Furthermore, the transmittance As indicated in Equation (1), the components with significant im­
based adaptive emitter is also equipped with selective properties for pacts on cooling power performance include the radiative power
spectrum absorption while its emissivity spectrum adapts to the emitted out from the surface, the absorbed atmospheric radiation, the
specific atmospheric transmittance. Thereby, for ideal roof surfaces, intensity of solar absorption rates and the non-radiative heat transfer. To
the solar absorptance remains 0 and the main difference among three obtain the net cooling power, the subcomponents are calculated
ideal emitters is the middle and far-infrared spectral emissivity respectively based on the proposed assumptions with results illustrated
profiles as shown in Fig. 4. in Fig. 5.
2) The parameters of non-radiative heat coefficient and atmospheric As illustrated in Fig. 5(a), the power radiated out by three emitters
transmittance are analyzed regarding their effects on the radiative for heat dissipation varies depending on the temperature difference
cooling performance. The atmospheric transmittance is mainly between the ambient air and the radiative surfaces. When the emitters’
affected by the precipitable water content, the increase of which temperatures are larger than the ambient air inducing the negative
could basically enlarge the absorption of atmospheric radiation, thus values of Tamb-T, the larger temperature difference of |Tamb-T|

Fig. 5. Subcomponents of net radiative cooling power. (a) Radiative power Prad being radiated out from the emitters versus temperature difference Tamb-T. (b)
Incoming atmospheric radiation absorbed by the emitters Patm. (c) Incident solar power absorbed by emitters Psun versus sunlight absorption rate εr(λ, θsun). (d) Non-
radiative heat exchanges Pnonrad between the emitters and external environment versus temperature difference Tamb-T and heat coefficient hc.

4
J. Chen and L. Lu Journal of Building Engineering 33 (2021) 101631

corresponds to a higher radiative power from the emitters. While if the 2.3.2. Spectral selective property
emitters’ temperatures are reduced below the ambient air incurring the The comparison results among three emitters with different spec­
positive Tamb-T, the radiative power turns out to be smaller under a trum selective properties are illustrated in Fig. 6, which is calculated by
bigger temperature difference. Among three emitters, the broadband only considering the components of radiative power and absorbed at­
emitter with a wider range of infrared absorptive wavelength (3 μm to mospheric radiation. Additionally, sunlight absorption of 5% for each
∞) exhibits a much higher radiative power, while the selective emitter emitter is also presented for comparison. The results distinctly illustrate
presents the least radiative power owing to its narrowest spectrally se­ the characteristics of each emitter in terms of their achievable net
lective wavelength (8–13 μm). Therefore, compared with selective and cooling power changing with the surface temperature. The selective
adaptive emitters, the broadband emitter presents a stronger potential to emitter exhibits the lowest equilibrium temperature but also the mini­
dissipate more heat onto exterior atmosphere. mum net cooling power when its surface temperature equals to that of
However, the atmospheric radiation absorbed by the emitters pre­ ambient air Tamb. However, the broadband emitter presents a totally
sents a totally different characteristic from the radiative power, as opposite result with the highest net cooling power at the ambient surface
shown in Fig. 5(b). Broadband emitter absorbs the maximum incoming temperature while the least capability to achieve a low sub-ambient
atmospheric radiation from the overhead sky due to its widest wave­ equilibrium temperature. Due to the spectrum selectivity falling be­
length to maximize the absorptive capacity, which conversely weakens tween the broadband and selective emitters, the adaptive emitter shows
its radiative cooling potential to some extent. The selective emitter with the cooling capacity in between. The absorption of sunlight, which in­
the narrowest wavelength range for infrared absorption obtains a min­ curs more additional negative heat into emitters, reduces both the net
imum atmospheric radiation among three emitters, the characteristic of cooling power and sub-ambient equilibrium temperature.
which indicates its strength in achieving the maximum net cooling Due to the fact that more interests are focused on the sub-ambient
power in sub-ambient temperature ranges where the radiative power temperature range, the relevant results are given at length in Fig. 7,
emitted from the three emitters maintains minor differences while the where different components in Equation (1) are respectively considered.
disparities of atmospheric radiation hold dominance in the cooling The absorbed solar power Psun herein is calculated by adopting the
performance. The adaptive emitter maintains in-between powers of both surface absorption rate of 5% with considering the currently reported
the radiative and atmospheric radiation, and thereby it is preferable in daytime radiative cooling materials with solar reflections ranging from
middle temperature ranges. 90% to 97% [10,14,51,52], and the non-radiative heat transfer Pnonrad is
The AM 1.5 solar spectrum with the total solar radiation power of obtained based on the heat coefficient hc of 3 W/(m2⋅K) so as to differ­
1000.4 W/m2 is utilized in the calculation of absorbed solar power for entiate the features of three emitters. The results exhibit the features that
emitters under various solar absorption rates and the results are illus­ the broadband emitter has the largest net cooling power while the se­
trated in Fig. 5(c). The reflectivity of incoming solar radiation is crucial lective emitter, though with the minimum cooling power at T = Tamb,
for surfaces to reduce the absorbed heat gain, which is beneficial to shows the advantage of the lowest equilibrium temperature. The adap­
achieve positive net cooling power. Current studies on radiative cooling tive emitter remains in between for both achievable net cooling power
have experimentally achieved the net cooling power below 100 W/m2 and equilibrium temperature. Basically, when more components are
by using different radiators with high solar reflectance, for instance, involved in the interaction with the emitters, both the net cooling power
40.1 W/m2 obtained by the photonic cooling material with 97% of Pnet (T = Tamb) and the equilibrium temperature Tamb-T (Pnet = 0) turn
incident sunlight being reflected [10], 93 W/m2 for a glass-polymer out to decrease. The net cooling power with only considering the
hybrid metamaterial by reflecting 96% of solar irradiance [14], and negative incoming atmospheric radiation turns out to be 128.4, 120.3,
96 W/m2 for a hierarchically porous polymer coating with the solar 109.5 W/m2 for broadband, adaptive and selective emitters respectively
reflectivity of 96% [15]. The weak cooling power is easily counteracted (Fig. 7(a)), whilst if the absorbed solar energy of 50 W/m2 is added up,
by the incoming absorption of solar energy. Therefore, ensuring a highly the results reduce to 78.4, 70.3, 59.5 W/m2 accordingly (Fig. 7(b)).
solar reflective profile for radiators is crucial for achieving passive Among three emitters, according to the required range of net cooling
cooling under sub-ambient temperature scenarios. power and sub-ambient temperature, the appropriate emitter type can
The non-radiative power lost owing to the convective and conductive be chosen as illustrated in the shaded area in Fig. 7. Based on particular
heat transfer between emitters and external environment plays a vital application scenarios and requirements, the three emitters can satisfy
role in the cooling power performance. When the exterior surface is the need for different cooling capacity requirements. With more com­
maintained under sub-ambient temperature, the heat is transferred from ponents comprising the cooling power, the shaded areas where the
surroundings into emitters inducing a positive non-radiative power as broadband and adaptive emitters dominate become increasingly larger
presented in Fig. 5(d), thereby entailing more radiative power to offset
the extra unwanted heat flow, thus weakening the achievable net
cooling power. However, if the surface temperature maintains above
ambient air, the consequent negative non-radiative power released from
the emitters to outside will otherwise contribute to the positive net
cooling power. Since the passive radiative cooling is more desired in the
scenario of sub-ambient temperature ranges, and the maintaining of sub-
ambient temperature difference is also preferred, the eradication of non-
radiative heat coefficient is of great significance to minimize the non-
radiative heat transfer and thus enlarge the systematic cooling capac­
ity. Currently, the material of low density polyethylene film, which is
transparent to almost all radiative wavelengths, has been widely applied
atop radiators as a wind shield in experimental systems to suppress the
direct contact of outside air with radiators, thereby minimizing the
corresponding heat transfer coefficient [10,45,49,50]. Regarding the
potential large-scale applications of radiative cooling materials on
building roof surfaces, the practical scenarios may restrict the integra­
tion of wind shields with roof surfaces. Therefore, impacts of
non-radiative heat transfer on the passive cooling potential shall be
investigated so as to provide the related theoretical guidance. Fig. 6. Net cooling power versus surface temperature.

5
J. Chen and L. Lu Journal of Building Engineering 33 (2021) 101631

Fig. 7. Comparison results of the three emitters regarding the net cooling power Pnet obtained by considering different components, namely the power radiated out
from the surface Prad, absorbed atmospheric radiation Patm, solar power Psun, and non-radiative heat transfer Pnonrad, respectively. (a) Pnet = Prad- Patm. (b) Pnet = Prad-
Patm- Psun. (c) Pnet = Prad- Patm- Pnonrad. (d) Pnet = Prad- Patm- Pnonrad- Psun.

whilst the dominant area of the selective emitter gradually dwindles and
even vanishes, the results of which can be observed by the variations
from Fig. 7(a)–7(d) plotted under a gradually smaller scale due to the
reduced cooling power and equilibrium temperature range. The results
indicate that the broadband emitter is equipped with superior net
cooling power at small ranges of sub-ambient temperature while the
selective emitter is more desirable under high sub-ambient temperature
conditions whereas its contribution to the net cooling power is accord­
ingly limited, and the adaptive emitter could provide available options
in between. The advantages of the selective emitter gradually disappear
with more powers involved as illustrated in Fig. 8 regarding the equi­
librium temperature differences among three emitters. Under the
circumstance where Pnet = Prad- Patm, the selective emitter could achieve
the equilibrium sub-ambient temperature over 63.2 ◦ C, which is 40.5 ◦ C
and 22.1 ◦ C higher than the broadband and adaptive emitters, respec­
tively. However, if the emitters interact with more powers, for instance,
Pnet = Prad-Patm-Pnonrad-Psun, the selective emitter’s achievable equilib­
rium temperature drastically decreases to 11.2 ◦ C and the differences
among three emitters remain negligible. Therefore, passive radiative
Fig. 8. Equilibrium temperature difference Tamb-Teq against involved
cooling potential could be more pronounced by adopting systems which
power components.
are highly insulated to the incident solar irradiance and surrounding
non-radiative heat transfer.
under the atmospheric condition with precipitable water vapor of 10
mm. As demonstrated, the negative cooling power represents heat
2.3.3. Non-radiative heat transfer coefficient
flowing from the external environment into the roof surface, the process
The influence of non-radiative heat exchange on net cooling power
of which constrains the fulfilment of heat removal from the building
Pnet and equilibrium temperature Tamb-Teq is illustrated in Fig. 9 and
envelope to external atmosphere. For the scenario under sub-ambient
Fig. 10 respectively. The results in Fig. 9 were obtained based on the
temperature (Tamb-T>0), the cooling capacity is naturally restrained
selective emitter without considering the effects of incoming sunlight
due to the higher external temperature exerting additional convective

6
J. Chen and L. Lu Journal of Building Engineering 33 (2021) 101631

highest sub-ambient equilibrium temperature followed in turn by

adaptive and broadband emitters. However, the differences become
unnoticeable if hc is larger than 10 W/(m2 ⋅K), which indicate the
importance of eliminating non-radiative heat transfer to differentiate
characteristics of different emitters.

2.3.4. Atmospheric transmittance

Atmospheric radiation is highly dependent on local sky trans­
mittance, which is basically related to the amount of precipitable water
vapor existing in the atmosphere. Essentially, the increase in the at­
mospheric precipitable water vapor will aggrandize the atmospheric
infrared absorptive ability, thus by the same token inducing a larger
emittance of downward radiation. Combined with different spectrally
selective properties of three emitters, the net cooling powers influenced
by the varying atmospheric radiations exhibit different characteristics as
illustrated in Fig. 11. The increase in water vapor column enlarges the
absorbed atmospheric radiation which exerts the maximum impact on
the broadband emitter and the least influence on the selective emitter.
Fig. 9. Influence of non-radiative heat transfer coefficient hc on net cooling
The corresponding net cooling powers of three emitters decrease in
power Pnet for selective emitter. response to the increase in water vapor column. Specifically, the net
cooling power reduces by 33.6, 28.8 and 13.0 W/m2 and the equilibrium
temperature decreases by 7.1, 13.5 and 13.3 ◦ C for broadband, adaptive
heat transfer on emitters. Thereby the non-radiative heat coefficient hc
and selective emitters respectively when water vapor column increases
exhibits great impacts on the achievable net cooling power. The increase
from 2.3 to 10 mm as presented in Table 1. Therefore, the cooling ca­
of hc can significantly reverse the cooling power from positive to nega­
pacity which depends on atmospheric conditions may be greatly
tive values, dramatically undermining the system’s cooling potential.
inhibited in real application scenarios where the absorption of sunlight
Regarding the desired application scenario under sub-ambient temper­
and non-radiative heat transfer on the cooling surface are sometimes
ature, the obtained positive net cooling powers (0–108 W/m2) are
inevitable to deteriorate the cooling potential.
stringently limited to a certain range of sub-ambient temperature and
The above parametric analysis illustrates the significant impacts of
non-radiative heat transfer coefficient, which indicates the necessary of
emitters’ spectral properties, non-radiative heat transfer and local at­
adopting appropriate shielding methods to reduce non-radiative effects
mospheric transmittance on the achievable cooling potential of radiative
so as to guarantee a positive cooling power under the sub-ambient
roof surface. The external roof surface cooling will exert noticeable
operating temperature.
impacts on the overall building energy savings by heat transmission
As illustrated in Fig. 10, the sub-ambient equilibrium temperature
through roof structures. Hence, in terms of utilizing radiative roof
exhibits a rapid decreasing trend at the initial stage with decreasing
cooling to achieve energy savings in buildings, the above influencing
rates slowing down as the heat transfer coefficient increases. The in­
factors including the atmospheric transmittance which varies from re­
crease in the sunlight absorption reduces the achievable sub-ambient
gion to region, and the different roof surface spectral properties should
equilibrium temperature due to more extra solar powers being
be particularly taken into account for building energy consumption
balanced by outgoing radiative power to maintain the equilibrium state.
analysis. Specifically, in consideration of China which has diverse
When sunlight is totally blocked, the selective emitter could achieve the
climate patterns due to its vast territory spanning over several climatic
sub-ambient temperature over 63 ◦ C, which sharply decreases below 10
regions resulting in varying location-based atmospheric radiations and
◦
C after a small increase in hc. The temperature differences caused by
solar irradiances into building roof surfaces, special investigations are
solar absorption are more prominent under smaller hc while become
entailed to evaluate the building energy savings in different climatic
gradually indistinctive as hc increases. Based on the comparisons of
zones in China by integrating radiative roof cooling characteristics,
three emitters in Fig. 10(b), the selective emitter shows the feature of the
which can be fulfilled by defining proper building model combined with

Fig. 10. Influence of non-radiative heat transfer coefficient hc on achievable equilibrium temperature Tamb- Teq. (a) Selective emitter under different sunlight ab­
sorption rates. (b) Comparisons of three emitters with 3% sunlight absorption.

7
J. Chen and L. Lu Journal of Building Engineering 33 (2021) 101631

Fig. 11. Influence of atmospheric conditions on absorbed atmospheric radiation and net cooling power by considering two components of radiative power and
atmospheric radiation. (a) Impacts on atmospheric radiation. (b) Impacts on net cooling power.

Table 1
Net cooling power and sub-ambient temperature under different water vapor columns.
Pnet (10 mm)/W/m2 Pnet (2.3 mm)/W/m2 Tamb-Teq (10 mm)/◦ C Tamb-Teq (2.3 mm)/◦ C ΔPnet (2.3–10 mm)/W/m2 Δ(Tamb-Teq) (2.3–10 mm)/
◦
C

Broadband emitter 128.4 162 22.7 29.8 33.6 7.1

Adaptive emitter 120.3 149.1 41.2 54.7 28.8 13.5
Selective emitter 109.5 122.5 63.4 76.7 13.0 13.3

specific weather data for building energy simulation as presented in ventilation, and other energy flows in buildings [19]. The established
Section 3. model is illustrated in Fig. 12.
The established three-dimensional building model has the dimension
3. Building model development of 8 × 6 × 3.5 m (length × width × height). Due to the fact that building
thermal response and energy performance may rely on many factors, for
3.1. Building model and weather data instance, local specific climatic conditions, building envelopes, COP of
HVAC system and its control strategies, the building model is required to
According to the aforementioned theoretical analysis of radiative be representative of particular architecture typology in different cli­
cooling on the achievement of sub-ambient temperature and net cooling matic zones. Nevertheless, for a better focus on the climatic effects and
power for exterior roof surfaces, the parameters of which constitute an detailed comparisons, the same building prototype assumed to be
indispensable boundary condition for the overall roof heat transfer located across several regions in China was adopted interacting with
model, it is insightful to investigate more in-depth potential benefits varying climatic conditions. In the model definition, the ideal load air
based on a building prototype regarding its thermal response and energy conditioning system which is capable of removing heat with 100% ef­
savings. Conceptually, benefits of radiative roof cooling are highly ficiency to meet the building thermal requirements was defined. The
dependent on specific atmospheric conditions. Considering multi-region heating and cooling control points were maintained at 20 ◦ C and 24 ◦ C,
varying climate patterns in China, in this paper the dynamic interactions respectively. The model also considered space infiltration and the in­
between different outdoor climates and building performance are ternal heat gain from people, lights, equipment, and so forth, with de­
evaluated by a numerical method for a typical one-story medium-sized tails given in Tables 2 and 3. The radiative properties of roof surface
commercial building based on EnergyPlus, which is a whole building were defined by setting the appropriate thermal and solar absorptance
energy simulation program that models heating, cooling, lighting, for exterior roof layers with the wavelength ranging from 2.5 μm to ∞

Fig. 12. Schematic of the modeling building prototype. (a) Building dimensions. (b) Building profile.

8
J. Chen and L. Lu Journal of Building Engineering 33 (2021) 101631

Table 2
Material property of the modeling building.
Building Layer Materials (from exterior to Thickness Density (kg/ Thermal conductivity (W/ Specific heat (J/ Dimension: Length × width
structure interior) (m) m3) (m⋅K)) (kg⋅K)) (m)

Wall Wood 0.009 530 0.14 900 8 × 3.5

Fiberglass 0.066 12 0.04 840
Plasterboard 0.012 950 0.16 840
Roof Roof deck 0.019 530 0.14 900 8×6
Fiberglass 0.118 12 0.04 840
Plasterboard 0.01 950 0.16 840
Floor HF–C5 0.1015 2243 1.73 837 8×6
Window Clear glass 0.006 / 0.9 / 3×2
Air 0.0032 1.23 0.03 /
Clear glass 0.006 / 0.9 /

maximum monthly average temperature may bring more challenges for

Table 3 thermally insulated building design. The southwest city of Kunming
Input parameters for thermal control of the building model.
under the temperate climate zone exhibits the smallest yearly temper­
Items Parameters ature fluctuations. The completely different climate characteristics
Air conditioning system Ideal load air system across the five regions will undoubtedly impose different solar heat
Thermostat control Heating temperature: 20 ◦ C gains, atmospheric radiations and convective heat exchanges on
Cooling temperature: 24 ◦ C external building envelopes including roof surfaces, thereby exerting
Internal load Number of people: 6
Lights: 1000 W
significant impacts on the thermal response and energy consumptions of
Electric equipment: 456 W local buildings.
Infiltration rate: 0.9 ACH
3.2. Model validation
for incoming thermal radiation, 0.3–2.537 μm for incident solar power
In order to elucidate the effectiveness and reasonableness of the
and 0.37–0.78 μm for visible spectrum.
obtained results, the established building model was validated by the
Regarding the atmospheric conditions, basically according to yearly
reported experimental data in Ref. [55], where a cool coating was
statistical mean temperatures taken from the coldest and hottest months
applied on a concrete roof surface for a building under tropical climate
as the main index, the design criteria to satisfy the building thermal
in Singapore and extensive temperature measurements were conducted
requirements in China are based on five typical climatic zones, namely
regarding the building’s roof top and ceiling surfaces. For a fair com­
severe cold, cold, hot summer and cold winter (HSCW), hot summer and
parison, the building specification of the model was adjusted accord­
warm winter (HSWW) and temperate zones with Harbin, Beijing,
ingly to make the roof structure similar to the experimental building in
Shanghai, Hong Kong and Kunming selected as the representative cities
Ref. [55]. The reported building roof was composed of three layers,
respectively [53]. The location map is illustrated in Fig. 13 [37].
namely cool coating (0.50 mm thickness), concrete structure (100 mm
The typical weather data for the above five representative cities are
thickness) and plasterboard (10 mm thickness). Hence, the roof mate­
presented in Fig. 14, which illustrates the monthly global horizontal
rials and layers in the building model were revised with the same
radiations, dry bulb temperatures, relative humidities, and wind speeds
physical properties to the given experimental parameters. The reported
and directions [54]. In terms of the main climatic features of five cities,
on-site measured daily weather data including the outdoor air temper­
Harbin, located in the severe cold climate zone, exhibits the largest
ature, incident solar radiation, as well as the local wind speed, and so
temperature variations with the minimum monthly average tempera­
forth, were integrated into the model so as to conduct daily building
tures. By comparison, the HSWW-based city of Hong Kong with the
thermal simulation with a focus on the temperature variations of roof
top and ceiling surfaces. Since the experimental building was not
air-conditioned, the corresponding building model removed the existing
air-conditioning system to reach utmost approximation to the reported
building prototype.
By considering the spectral properties of the reported cool coating,
the daily simulation results from the above developed model integrated
with the reported experimental building and weather information are
presented in Fig. 15. Compared with the experimental data, the simu­
lation results agree quite well with the variations of both roof top and
ceiling temperatures (Fig. 15(a)). The error analysis (Fig. 15(b)) shows
the differences between the reported experimental data and simulated
values are within ±10% indicating the effectiveness of using the
developed model to conduct the building simulation for radiative roof
cooling investigation.
Additionally, three indicative parameters are introduced to give
more quantitative comparisons between the numerical simulation re­
sults and the experimental data, namely mean bias error (MBE), root
mean square error (RMSE) and index of agreement (d) [56], which are
defined by the following equations:

1 ∑ 24
( )
MBE = Tmodel,i − Texp,i (8)
24 i=1
Fig. 13. Five typical climate zones and representative cities in China.

9
J. Chen and L. Lu Journal of Building Engineering 33 (2021) 101631

Fig. 14. Weather data in five typical cities of China. (a) Monthly global horizontal radiation. (b) Monthly dry-bulb temperature. (c) Monthly averaged relative
humidity. (d) Wind speed and direction in twelve months, where the wind speed values are presented in a cumulative mode.

Fig. 15. Model validation in comparison to the cool concrete rooftop and ceiling temperatures from the reported experimental data in Refs. [55]. (a) Rooftop and
ceiling temperature variations from the model and reported experimental data. (b) Error analysis of the model.

√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
∑24 ( )2 Based on the Equations from (8) to (10), the MBEs for the surface
i=1 Texp.i − Tmodel,i temperature of rooftop (Troof) and ceiling (Tceiling) are 0.74 and − 0.18 ◦ C
RMSE = (9)
24 respectively, and the RMSEs are 1.25 and 1.81 ◦ C respectively. Either
MBEs or RMSEs turn out to be small enough relative to the daily roof
∑24 ( )2
i=1 Texp,i − Tmodel,i temperature variations as presented in Fig. 15 (a). The MBEs and RMSEs
d = 1 − ∑24 ( ⃒ ⃒ ⃒ ⃒)2 (10)
⃒ ⃒ ⃒ ⃒ either for rooftop or ceiling temperatures only account for less than 3.2%
i=1 Tmodel,i − T exp + Texp,i − T exp
and 7.3% of the daily average experimental temperatures, respectively.
where, Tmodel and Texp are the rooftop or ceiling temperatures from the The accuracy at this level is sufficient to show the effectiveness of the
model and reported experiment, respectively, ◦ C.Texp are the daily numerical simulation. Moreover, the indices of agreement d for Troof and
average rooftop or ceiling temperatures from the experiment, ◦ C. Tceiling are over 0.98 and 0.87 respectively, which indicate the good

10
J. Chen and L. Lu Journal of Building Engineering 33 (2021) 101631

match of the results between the model and the reported experiment. 4.1. Thermal response of roof surface

4. Results and discussion The roof top daily temperature variations across several typical cities
are illustrated in Fig. 16, where the solar reflectivity ranges from 0.2 to
Based on the officially available categories of the certified roof 0.95 while the thermal emittance remains at 0.9. The rooftop temper­
products in Cool Roof Rating Council (CRRC) [57] and ENERGY STAR atures during daytime periods are generally higher than that of outdoor
[58], the roof radiative parameters are composed of solar reflectance air due to the absorption of incident sunlight whist below ambient air
and thermal emittance which cover numerous combinations from small during the nighttime owing to the high thermal emittance for heat
to large values to satisfy different application demands. Therefore, in dissipation. Exceptionally, with the increase of solar reflectivity to 0.95,
order to obtain comparable results, the paper opted for a broad range of the highly reflective roof surface basically decreases to the sub-ambient
solar reflectivities and thermal emissivities so as to adequately reflect temperature throughout the whole day, realizing the passive daytime
spectrally radiative characteristics of the officially certified roof coating radiative cooling. Regarding the daily peak temperature variations as
products. The results herein could practically illustrate benefits of in­ presented in Fig. 16(f), in comparison to the roof surface with the solar
tegrated radiative roof surface on buildings’ thermal performance and reflectivity of 0.2, the parameter of which corresponds to the cool roof
energy savings under various climate regions. coating products with grey or brown colors as listed in the CRRC rated
products directory [57], the increase in solar reflective ability obviously

Fig. 16. Daily rooftop surface temperature on 15 July. (a) Beijing. (b) Harbin. (c) Hong Kong. (d) Kunming. (e) Shanghai. (f) Peak temperature reduction in
comparison to the rooftop with the solar reflectivity of 0.2.

11
J. Chen and L. Lu Journal of Building Engineering 33 (2021) 101631

steps up the peak temperature reduction. For instance, the building reflectivity due to the decreased requirement of cooling loads from the
prototype located in Hong Kong with a peak temperature of 30.6 ◦ C air-conditioning system. Among five cities, HSWW-based Hong Kong
under the solar reflectivity of 0.95 can achieve cooler degree of 30.9 ◦ C and temperate Kunming with cooling loads dominant throughout the
than that with the solar reflectivity of 0.2. The maximum rooftop cooler year exhibit more prominent reduction in energy consumptions. By
degrees are 33.8, 33.4, 30.9, 29.4, and 32.8 ◦ C for Beijing, Harbin, Hong contrast, for severe cold (Harbin), and cold (Beijing) climatic zones, over
Kong, Kunming, and Shanghai respectively. The significant temperature which heating loads prevail, the energy savings which are significantly
reduction in exterior roof surface beneficially reduces thermal shock to suppressed still turn out to be positive, which indicates the potential
roofing materials so as to guarantee a longer lifespan of roofing systems. benefits from radiative roof cooling applications. By comparison, heat­
Additionally, a cooler peak-time surface could possibly decrease the ing loads predominantly prevail over Harbin followed by Beijing, whilst
peak cooling load demands and thus reduce the burden of urban power cooling loads dominate in Hong Kong and then Kunming. Since different
supply. climatic patterns demand corresponding building loads to maintain
The monthly temperature differences of the rooftop surface between buildings’ thermal comfort, the benefits from radiative roof cooling vary
the solar reflectivity of 0.2 and 0.95 are presented in Fig. 17. Due to the across several regions.
increase in the roof solar reflective ability, more heats are blocked out of As presented in Fig. 19(f), the obtained cooling load saving is
the roof surface. Thereby, all the temperature differences turn out to be generally higher than the heating load penalty for each typical city,
positive. While the maximum temperature differences for the five inducing annually positive energy savings. The cities of Kunming and
typical cities are mainly concentrated in the hotter months, for instance, Hong Kong top the two largest energy saving rates (9.3% and 7.1%
April for Kunming, May for Shanghai, June for Beijing and Harbin, and respectively) followed in turn by Shanghai (4.1%), Beijing (3.3%) and
July for Hong Kong, which indicate more prominent benefits could be Harbin (0.6%). Even though heating load penalties are unavoidable in
achieved under hotter weather conditions regarding the roof thermal winter period due to the reduced heat gain into buildings, the cooling
performance improvement. Unlike the daily temperature difference load savings turn out to overwhelmingly surpass the extra heating
which could reach up to over 30 ◦ C cooler for peak-time temperature, penalties. Meanwhile, considering the basically higher utility costs of
the monthly averaged temperature differences between the solar electricity for cooling power than that of coals or natural gases for
reflectivity of ρg = 0.2 and ρg = 0.95 are generally less than 12 ◦ C and heating supply, the radiative roof cooling technique presents a prom­
vary differently from month to month. ising prospect to be integrated with building systems, especially for
The thermal responses in terms of roof bottom temperature of the buildings in HSWW-based and temperate climatic regions where much
building without air conditioning systems settled in Hong Kong are greater energy saving potentials can be anticipated.
exhibited in Fig. 18, the prototype of which removes the fiberglass Due to the fact that Hong Kong, Kunming and Shanghai located in
insulation beneath the rooftop layer so as to achieve more obviously the southern part of China are not provided with central public heating
comparable results. As illustrated in Fig. 18(b), the daily maximum services in winter, if not considering the individual heating loads from
temperature difference could reach up to 13.8 ◦ C between roof surfaces air-conditioning systems, the overall heating load penalties will be
with the solar reflectivity of 0.95 and 0.2, which indicates the significant eliminated from the total energy consumptions. Thus, the energy saving
impacts of surface radiative properties on the building thermal perfor­ potentials will be improved as illustrated in Table 4 for the three cities
mance contributing to varying temperature variations in the roof bottom with or without heating load penalties. The enhancement is more
surface. The temperature reductions in non-air-conditioned buildings’ noticeable for Kunming and Shanghai with energy saving rate im­
ceiling surfaces can reduce both the convective and radiant heat trans­ provements of 6.6% and 5.8% respectively. The comparison results
fer, leading to a more thermally comfortable built environment for oc­ indicate that a larger portion of energy savings can be possibly achieved
cupants during hot days. in practical scenarios. By and large, for typical climatic regions in China,
the radiative roof cooling achieved by spectrally reflective surfaces is
conducive to achieving energy-efficient buildings and meantime may
4.2. Energy saving analysis further contribute to a more sustainable society.

The annual energy consumptions and savings benefited from higher 4.3. Discussions on influencing factors
solar reflective roof surfaces for five cities are exhibited in Fig. 19. The
total energy consumptions which are composed of heating and cooling (1) Roof thermal resistance
loads in five typical cities decrease gradually under a higher solar
The existence of the fiberglass with a low thermal conductivity of
0.04 W/(m⋅K) as the roof heat insulation layer could exert great impacts
on the heat transmission, the achievable roof bottom temperature and
building energy consumption. The roof bottom temperature variations
with or without the fiberglass layer are illustrated in Fig. 20, which
shows the daily temperature variations under several solar reflectivities.
Due to the low conductivity of fiberglass insulation layer to suppress
the heat transfer through roof structure, the roof bottom temperature is
obviously reduced for a small solar reflective surface, whilst as the
surface solar reflectivity increases, the temperature differences of two
roof structures gradually remain unnoticeable, especially for ρg = 0.95.
The results indicate the effectiveness of applying radiative roof cooling
to achieve the equal thermal insulation performance. By applying roof
surfaces with high solar reflectance, the roof bottom temperature can be
potentially reduced to the same degree as that with a thick insulation
layer which can be possibly removed from the thermal perspective for
cost savings in buildings. It is worth mentioning that the unnoticeable
temperature differences during the nighttime between the roof struc­
Fig. 17. Monthly rooftop surface temperature differences between solar tures with or without the fiberglass are mainly caused by the hot sum­
reflectivity of ρg = 0.2 and ρg = 0.95. mer weather condition with high outdoor air temperatures, which

12
J. Chen and L. Lu Journal of Building Engineering 33 (2021) 101631

Fig. 18. Daily roof bottom surface temperature without fiberglass insulation layer on 15 July in Hong Kong. (a) Roof bottom surface temperature. (b) Temperature
differences under various reflective roof surfaces in comparison to that with solar reflectivity of 0.2.

Fig. 19. Annual building energy consumptions and savings in five typical cities. (a) Beijing. (b) Harbin. (c) Hong Kong. (d) Kunming. (e) Shanghai. (f) Annual
building loads between solar reflectivity of ρg = 0.2 and ρg = 0.95.

13
J. Chen and L. Lu Journal of Building Engineering 33 (2021) 101631

Table 4
Comparisons of annual energy savings with and without considering heating load penalty.
City With heating load penalty Without heating load penalty
2
Energy saving/kWh/m Energy saving rate/% Energy saving/kWh/m2 Energy saving rate/% Saving rate improvement/%

Hong Kong 10.2 7.1 10.5 8.4 1.3

Kunming 10.3 9.3 10.6 15.9 6.6
Shanghai 7.1 4.1 8.2 9.9 5.8

increases to over 35% in Hong Kong and over 32% in Kunming. The
differences of energy saving rates for roof structures with or without
fiberglass are prominent except for the severe cold climatic city of
Harbin where the heating loads dominate over a long duration. The
comparative results indicate the radiative roof cooling system is effec­
tive in most climatic regions in China, and especially efficient for those
with dominant cooling loads. In particular, the building prototype with a
simple roof structure of low thermal resistance is more preferable to be
integrated with the radiative roof cooling application. For instance, the
buildings with highly conductive metal roof and on-site temporary
simplified buildings with light steel framework as envelope materials
will be good candidates for the decision-making.

(2) Surface spectral property

Fig. 20. Comparisons of daily roof bottom temperature with or without the Theoretically, solar reflectance and thermal emittance are two key
fiberglass insulation layer on 15 July in Hong Kong. parameters influencing the solar absorption and heat dissipation of roof
surfaces, which thereby can impose noticeable impacts on buildings’
induce a higher exterior roof surface temperature than that of the cooling and heating loads. As illustrated in Fig. 22 where the fiberglass
interior roof surface due to the radiative heat transfer process, thus insulation layer is not considered in the building prototype, due to less
inhibiting the major cooling effects of the roof structure without the solar energy being absorbed by the roof surface or more internal heat
fiberglass. While in winter period with a much lower outdoor air tem­ being emitted out from the rooftop in the form of infrared radiation,
perature, the heat transmission from the interior to exterior roof surface buildings’ cooling load percentage decreases with the increase of the
works as a cooling effect for the roof structure, which is more obvious for surface solar reflectivity and thermal emissivity. While the heating load
the roof without fiberglass thermal insulation layer. Hence, in that case proportions exhibit the opposite trend. In comparison to variations of
the roof without the fiberglass will exhibit a lower interior roof surface the thermal emissivity, cooling and heating load amplitudes versus the
temperature during nighttime. solar reflectivity are quite higher, indicating the bigger impact of the
The results of annual energy savings in five typical cities for the surface solar reflectivity on buildings’ energy consumptions. More spe­
building prototype without the fiberglass thermal resistance layer are cifically, by the equal increase of both solar reflectivity Δρg and thermal
presented in Fig. 21. emissivity Δεg as indicated in Fig. 23, the typical cities with cooling
Under the circumstance without the fiberglass as the roof thermal loads dominant throughout the year like Hong Kong, Kunming and
resistance layer, the building’s cooling load savings and heating load Shanghai show more impacts from the increase of solar reflectivity than
penalties are both significantly enhanced due to the poorer roof heat that of thermal emissivity, whereas for the heating loads dominant city
insulation capacity. The achievable energy savings are correspondingly of Harbin under the severe cold climate zone, the effects of thermal
increased owing to the larger share of cooling load reductions in the emissivity overwhelmingly surpass the solar reflectivity since the in­
total energy consumptions. Among five cities, the energy saving rate crease of thermal emittance significantly aggrandizes the building’s heat

Fig. 21. Annual energy savings of the building prototype without the fiberglass thermal resistance layer between solar reflectivity of ρg = 0.2 and ρg = 0.95, and the
comparative results to buildings with the fiberglass layer. (a) Cooling and heating loads, and corresponding energy savings in buildings without the fiberglass. (b)
Energy saving rate for buildings without the fiberglass and its improvement to that with the fiberglass.

14
J. Chen and L. Lu Journal of Building Engineering 33 (2021) 101631

Fig. 22. Cooling and heating load percentage. (a) Load percentage versus solar reflectivity under thermal emissivity of 0.9. (b) Load percentage versus thermal
emissivity under solar reflectivity of 0.9.

ambient temperature range. Atmospheric window based selective

emitter though radiates the least radiative power still absorbs the
minimum atmospheric radiation, justifying its strength in achieving
the maximum cooling capacity under low surface temperature sce­
narios. While the transmittance based adaptive emitter can provide
in-between cooling capacities. Based on particular application sce­
narios, the three emitters can satisfy respective cooling
requirements.
3) Based on the developed building energy simulation model across
typical climatic regions in China, the exterior rooftop surface tem­
perature is significantly reduced under high solar reflectance with
the maximum cooler degree over 30 ◦ C in comparison to surfaces
with solar reflectivity of 0.2 as the baseline. The temperature
reduction is conducive to avoiding thermal shock to roof structure
and thus enhancing the lifespan of roofing systems. The roof bottom
temperature reduction can reach up to 13.8 ◦ C in Hong Kong for non-
Fig. 23. Comparisons of energy savings against variations of solar reflectivity
air-conditioned buildings to achieve indoor thermal comfort.
and thermal emissivity. Δρg = 0.4 means the difference of solar reflectivity
between ρg = 0.6 and ρg = 0.2; Δρg = 0.75 indicates the difference of solar
4) Annual energy savings benefited from the radiative roof cooling turn
reflectivity between ρg = 0.95 and ρg = 0.2; Δεg = 0.4 equals to the difference out to be positive for five typical cities, among which HSWW-based
of thermal emissivity between εg = 0.6 and εg = 0.2; Δεg = 0.75 indicates the Hong Kong and temperate Kunming exhibit the two most promi­
difference of thermal emissivity between εg = 0.95 and εg = 0.2. nent saving potentials with the saving rates of 7.1% and 9.3%
respectively. By contrast, for severe cold (Harbin), and cold (Beijing)
dissipation into external atmosphere. Thus more energy will be entailed climatic zones the energy savings are significantly suppressed but the
to offset the correspondingly increased heating loads, leading to nega­ cooling load saving still noticeably outperforms the heating load
tive energy savings. Beijing under the cold climate zone exhibits the penalty, leading to positive benefits.
similar degree of influence from both parameters. By and large, the in­ 5) The existence of roof fiberglass thermal insulation layer can effec­
crease in rooftop solar reflectivity will effectively step up energy saving tively restrain the heat transfer to reduce the roof bottom tempera­
potentials in buildings across all climatic regions, whilst the increase of ture. However, the temperature differences between roof structures
thermal emissivity turns out to have different effects based on buildings’ with or without the fiberglass turn out to be gradually unnoticeable
located atmospheric regions. under a higher solar reflective surface. The thick insulation layer can
be possibly removed from the thermal perspective for cost savings by
5. Conclusions applying radiative cooling surfaces.
6) The study sheds light on the fact that the application of radiative roof
1) The radiative roof cooling performance is dominated by components cooling by increasing rooftop solar reflectance combined with
directly interacting with roof surfaces including power radiated out appropriate thermal emittance will effectively improve the building
from the surface, absorbed incoming atmospheric radiation, incident energy efficiency across typical climatic regions in China, especially
solar power and non-radiative heat transfer between the surface and for those under cooling load dominant regions.
external environment. The atmospheric radiation is affected by local
precipitable water vapor, the increase of which will reduce the sky CRediT authorship contribution statement
transmittance, thus increasing the corresponding atmospheric radi­
ation. The increase of non-radiative heat coefficient can easily Jianheng Chen: Conceptualization, Data curation, Formal analysis,
reverse the cooling power from positive to negative, thereby signif­ Investigation, Methodology, Software, Validation, Visualization,
icantly undermining cooling potentials. Writing - original draft. Lin Lu: Funding acquisition, Project adminis­
2) Regarding three types of emitters, broadband emitter which is su­ tration, Methodology, Resources, Supervision, Writing - review &
perior in generating net cooling power exhibits both the highest editing.
radiative power and absorbed atmospheric radiation, the counter­
action of which undermines its passive cooling capacity under sub-

15
J. Chen and L. Lu Journal of Building Engineering 33 (2021) 101631

Declaration of competing interest [10] A.P. Raman, et al., Passive radiative cooling below ambient air temperature under
direct sunlight, Nature 515 (7528) (2014) 540–544.
[11] A.R. Gentle, G.B. Smith, A subambient open roof surface under the mid-summer
The authors declare that they have no known competing financial sun, Advanced Science 2 (9) (2015) 1500119.
interests or personal relationships that could have appeared to influence [12] S. Atiganyanun, et al., Effective radiative cooling by paint-format microsphere-
the work reported in this paper. based photonic random media, ACS Photonics 5 (4) (2018) 1181–1187.
[13] W. Li, et al., Photonic thermal management of coloured objects, Nat. Commun. 9
(1) (2018) 4240.
Acknowledgement [14] Y. Zhai, et al., Scalable-manufactured randomized glass-polymer hybrid
metamaterial for daytime radiative cooling, Science 355 (6329) (2017)
1062–1066.
The work is partially supported by Hong Kong Research Grant [15] J. Mandal, et al., Hierarchically porous polymer coatings for highly efficient
Council through General Research Fund (PolyU152184/17E). passive daytime radiative cooling, Science 362 (6412) (2018) 315–319.
[16] E.A. Goldstein, A.P. Raman, S. Fan, Sub-ambient non-evaporative fluid cooling
with the sky, Nature Energy 2 (9) (2017).
Appendix A. Supplementary data [17] D. Zhao, et al., Subambient cooling of water: toward real-world applications of
daytime radiative cooling, Joule 3 (1) (2019) 111–123.
Supplementary data to this article can be found online at https://doi. [18] H. Akbari, R. Levinson, Evolution of cool-roof standards in the US, Adv. Build.
Energy Res. 2 (1) (2008) 1–32.
org/10.1016/j.jobe.2020.101631. [19] W. Wang, et al., Performance assessment of a photonic radiative cooling system for
office buildings, Renew. Energy 118 (2018) 265–277.
Nomenclature [20] M. Kolokotroni, B.L. Gowreesunker, R. Giridharan, Cool roof technology in
London: an experimental and modelling study, Energy Build. 67 (2013) 658–667.
[21] D.D. Kolokotsa, et al., Cool roofs and cool pavements application in Acharnes,
c Speed of light in vacuum, 3.0 × 108 m/s Greece, Sustainable Cities and Society 37 (2018) 466–474.
h Planck’s constant = 6.62607004 × 10− 34 J s [22] J. Yang, et al., Green and cool roofs’ urban heat island mitigation potential in
hc Non-radiative heat transfer coefficient, W⋅m− 2⋅K− 1 tropical climate, Sol. Energy 173 (2018) 597–609.
[23] K.T. Zingre, M.P. Wan, X. Yang, A new RTTV (roof thermal transfer value)
IAM1.5 Radiation intensity of AM 1.5 spectrum, W⋅sr− 1 m− 3 calculation method for cool roofs, Energy 81 (2015) 222–232.
IBB Spectral radiance of a blackbody, W⋅sr− 1 m− 3 [24] J. Lei, et al., Cool colored coating and phase change materials as complementary
kB Boltzmann constant, 1.38064852 × 10− 23 J/K cooling strategies for building cooling load reduction in tropics, Appl. Energy 190
(2017) 57–63.
Patm: Atmospheric radiation absorbed by surface, W/m2 [25] E. Bozonnet, M. Doya, F. Allard, Cool roofs impact on building thermal response: a
Pnet: Net cooling power, W/m2 French case study, Energy Build. 43 (11) (2011) 3006–3012.
Pnonrad Non-radiative heat transfer between surface and [26] A.L. Pisello, F. Cotana, The thermal effect of an innovative cool roof on residential
buildings in Italy: results from two years of continuous monitoring, Energy Build.
environment, W/m2 69 (2014) 154–164.
Prad Radiative power emitted out from surface, W/m2 [27] H. Takebayashi, M. Moriyama, T. Sugihara, Study on the cool roof effect of
Psun Incident solar radiation absorbed by surface, W/m2 Japanese traditional tiled roof: numerical analysis of solar reflectance of
unevenness tiled surface and heat budget of typical tiled roof system, Energy Build.
t Atmospheric transmittance in zenith direction 55 (2012) 77–84.
T Surface temperature, ◦ C [28] M.C. Yew, et al., Experimental analysis on the active and passive cool roof systems
Tamb Ambient temperature, ◦ C for industrial buildings in Malaysia, Journal of Building Engineering 19 (2018)
134–141.
Teq Equilibrium surface temperature, ◦ C
[29] L.L. Han, S.Z. Ke, M.L. Chen, The application of container-type green roof in
Troof Rooftop surface temperature, ◦ C ancient architecture—examples of the green roofs of CPPCC and NPC buildings in
Tceiling Roof ceiling surface temperature, ◦ C huangpu district of shanghai, China Garden 31 (2015) 9–12.
εr Emissivity of surface [30] A. Synnefa, M. Santamouris, H. Akbari, Estimating the effect of using cool coatings
on energy loads and thermal comfort in residential buildings in various climatic
εa Emissivity of atmosphere conditions, Energy Build. 39 (11) (2007) 1167–1174.
εg Total thermal emissivity of roof surface [31] P.J. Rosado, et al., Measured temperature reductions and energy savings from a
ρg Total solar reflectivity of roof surface cool tile roof on a central California home, Energy Build. 80 (2014) 57–71.
[32] A.L. Pisello, State of the art on the development of cool coatings for buildings and
θ Zenith angle cities, Sol. Energy 144 (2017) 660–680.
λ Wavelength, m [33] M.C. Yew, et al., Integration of thermal insulation coating and moving-air-cavity in
MBE Mean bias error, ◦ C a cool roof system for attic temperature reduction, Energy Convers. Manag. 75
(2013) 241–248.
RMSE Root mean square error ◦ C [34] T. Xu, et al., Quantifying the direct benefits of cool roofs in an urban setting:
d Index of agreement reduced cooling energy use and lowered greenhouse gas emissions, Build. Environ.
48 (2012) 1–6.
[35] S. Boixo, et al., Potential energy savings from cool roofs in Spain and Andalusia,
References Energy 38 (1) (2012) 425–438.
[36] R. Levinson, H. Akbari, Potential benefits of cool roofs on commercial buildings:
[1] F. Cotana, et al., Albedo control as an effective strategy to tackle Global Warming: a conserving energy, saving money, and reducing emission of greenhouse gases and
case study, Appl. Energy 130 (2014) 641–647. air pollutants, Energy Efficiency 3 (1) (2010) 53–109.
[2] J.C. Lam, et al., Multiple regression models for energy use in air-conditioned office [37] Y. Gao, et al., Cool roofs in China: policy review, building simulations, and proof-
buildings in different climates, Energy Convers. Manag. 51 (12) (2010) 2692–2697. of-concept experiments, Energy Pol. 74 (2014) 190–214.
[3] Y. Jiang, Current trend of building energy use and major conservation issues in [38] J.C. Lam, et al., Building energy efficiency in different climates, Energy Convers.
China, Green Build 7 (2006) 10–15. Manag. 49 (8) (2008) 2354–2366.
[4] C.K. Leung, et al., Optical and thermal performance analysis of aerogel glazing [39] M.M. Hossain, M. Gu, Radiative cooling: principles, progress, and potentials, Adv.
technology in a commercial building of Hong Kong, Energy and Built Environment Sci. 3 (7) (2016) 1500360.
1 (2) (2020) 215–223. [40] M. Zeyghami, D.Y. Goswami, E. Stefanakos, A review of clear sky radiative cooling
[5] Y. Gao, et al., Thermal performance and energy savings of white and sedum-tray developments and applications in renewable power systems and passive building
garden roof: a case study in a Chongqing office building, Energy Build. 156 (2017) cooling, Sol. Energy Mater. Sol. Cell. 178 (2018) 115–128.
343–359. [41] B. Zhao, et al., Radiative cooling: a review of fundamentals, materials,
[6] C.-G. Granqvist, Radiative heating and cooling with spectrally selective surfaces, applications, and prospects, Appl. Energy 236 (2019) 489–513.
Appl. Optic. 20 (15) (1981) 2606–2615. [42] A. Berk, et al., MODTRAN5: 2006 update, in: Defense and Security Symposium,
[7] C.G. Granqvist, A. Hjortsberg, Radiative cooling to low temperatures: General vol. 6233, SPIE, 2006.
considerations and application to selectively emitting SiO films, J. Appl. Phys. 52 [43] Gemini Observatory, IR-Transmittance, Available from: www.gemini.edu/sciops/t
(6) (1981) 4205–4220. elescopes-and-sites/observing-condition-constraints/ir-transmission-spectra.
[8] T.S. Eriksson, C.G. Granqvist, Radiative cooling computed for model atmospheres, [44] Nrel-Astm, Reference air Mass 1.5, Spectra-ASTM G-173-03 2003; Available from:
Appl. Optic. 21 (23) (1982) 4381–4388. https://www.nrel.gov/grid/solar-resource/spectra-am1.5.html.
[9] H. Fang, et al., Performance evaluation of a metamaterial-based new cool roof [45] B. Zhao, et al., Performance evaluation of daytime radiative cooling under different
using improved Roof Thermal Transfer Value model, Appl. Energy 248 (2019) clear sky conditions, Appl. Therm. Eng. 155 (2019) 660–666.
589–599. [46] J. Khedari, et al., Field investigation of night radiation cooling under tropical
climate, Renew. Energy 20 (2) (2000) 183–193.

16
J. Chen and L. Lu Journal of Building Engineering 33 (2021) 101631

[47] B. Landro, P.G. McCormick, Effect of surface characteristics and atmospheric [53] L. Yang, J.C. Lam, C.L. Tsang, Energy performance of building envelopes in
conditions on radiative heat loss to a clear sky, Int. J. Heat Mass Tran. 23 (5) different climate zones in China, Appl. Energy 85 (9) (2008) 800–817.
(1980) 613–620. [54] EnergyPlus weather data, Available from: https://energyplus.net/weather, 2019
[48] A.W. Harrison, Effect of atmospheric humidity on radiation cooling, Sol. Energy 26 25 October.
(3) (1981) 243–247. [55] K.T. Zingre, et al., Modeling of cool roof heat transfer in tropical climate, Renew.
[49] X. Ao, et al., Preliminary experimental study of a specular and a diffuse surface for Energy 75 (2015) 210–223.
daytime radiative cooling, Sol. Energy Mater. Sol. Cell. 191 (2019) 290–296. [56] C.J. Willmott, S.M. Robeson, K. Matsuura, A refined index of model performance,
[50] J.-l. Kou, et al., Daytime radiative cooling using near-black infrared emitters, ACS Int. J. Climatol. 32 (13) (2012) 2088–2094.
Photonics 4 (3) (2017) 626–630. [57] Cool Rool Rating Council, Rated products directory, Available from: https://cool
[51] H. Bao, et al., Double-layer nanoparticle-based coatings for efficient terrestrial roofs.org/directory, 2019 12 December.
radiative cooling, Sol. Energy Mater. Sol. Cell. 168 (2017) 78–84. [58] Energy Star, Certified roof products, Available from: https://www.energystar.go
[52] Z. Huang, X. Ruan, Nanoparticle embedded double-layer coating for daytime v/products/building_products/roof_products, 2019 12 December.
radiative cooling, Int. J. Heat Mass Tran. 104 (2017) 890–896.

17