Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Article

Subambient Cooling of Water: Toward Real-


World Applications of Daytime Radiative
Cooling
Dongliang Zhao, Ablimit Aili,
Yao Zhai, ..., Gang Tan, Xiaobo
Yin, Ronggui Yang
gtan@uwyo.edu (G.T.)
xiaobo.yin@colorado.edu (X.Y.)
ronggui.yang@colorado.edu (R.Y.)

HIGHLIGHTS
10.6 C subambient cooling of
water around noon under direct
sunlight

Subambient cool-water
production at various constant
temperatures

The effect of weather conditions


on the performance of radiative
sky cooling

kW-scale radiative sky cooling


system to demonstrate scalability
of the technology

Radiative sky cooling of large thermal mass water to 10.6 C below the ambient
temperature is demonstrated around noon under direct sunlight by using polymer-
based low-cost radiative cooling metamaterial. Subambient cool-water
production at various constant temperatures is experimentally demonstrated. The
parasitic effects of different weather conditions (local wind speed, precipitable
water, and cloud cover) on the performance of sky radiative cooling have been
investigated, which is critical for developing operation strategies for real-world
applications.

Zhao et al., Joule 3, 111–123


January 16, 2019 ª 2018 Elsevier Inc.
https://doi.org/10.1016/j.joule.2018.10.006
Article
Subambient Cooling of Water: Toward
Real-World Applications
of Daytime Radiative Cooling
Dongliang Zhao,1 Ablimit Aili,1 Yao Zhai,1 Jiatao Lu,1 Dillon Kidd,1 Gang Tan,2,* Xiaobo Yin,1,3,*
and Ronggui Yang1,3,4,*

SUMMARY Context & Scale


Real-world applications of radiative sky cooling require thoughtful design of the Radiative sky cooling dumps
system, along with clear understanding of weather effects on system perfor- excessive heat to the low-
mance. This work explores application of radiative sky cooling based upon a temperature sky through infrared
low-cost radiative cooling metamaterial that can be scalably manufactured. A thermal radiation. Recent
radiative cooled-cold collection (RadiCold) module is developed to cool water advances in materials have led to
to 10.6 C below ambient at noon under stationary conditions. The effects of the breakthrough in daytime
different weather conditions (wind speed, precipitable water, and cloud cover) radiative cooling, with
on the performance of radiative cooling have been investigated. A kilowatt subambient temperatures
(kW)-scale RadiCold system with 13.5 m2 radiative cooling surface area is then achieved. In this work, we
built and demonstrated to provide a maximum cooling power of 1,296 W demonstrate for the first time the
at night, and an average cooling power of 607 W at noon (12–2 p.m.) under cooling of water (with large
952 W/m2 average solar irradiance at 26.5 L/(h,m2) volumetric flow rate. thermal mass) to 10.6 C below
A building-integrated RadiCold system is proposed to provide continuous ambient around noon under direct
day-and-night cooling. sunlight. The effects of different
operating temperatures and
weather conditions (local wind
INTRODUCTION
speed, precipitable water, and
By emitting infrared thermal radiation that is transparent through the atmospheric
cloud cover) on the performance
transmission window with characteristic wavelength at 8–13 mm to the cold uni-
of radiative sky cooling have been
verse,1 radiative sky cooling can cool an object on the earth by sending excessive
investigated, which are essential
heat to outer space without the need for electricity. This is significantly different
for real-world applications. A
from other cooling technologies that dump the waste heat into the surroundings
kW-scale radiative sky cooling
and is truly appealing ‘‘without costing the earth.’’ Numerous studies have been
system with 13.5 m2 radiative
carried out on nighttime radiative cooling in the past few decades.2–6 Very
surface area has been built to
recently, owing to the progress in nanophotonics and metamaterials, daytime radi-
demonstrate scalability of the
ative cooling to subambient temperatures have been demonstrated, which has
technology. As cooling demand
revived interest and attracted further research.7–18 However, due to the intrinsic
increases dramatically in the 21st
low energy density of radiative cooling (generally with a cooling flux in the order
century, this work paves the way
of 100 W/m2 at ambient temperature), most of the recent works have been
for promoting radiative sky
restricted to the demonstration of its potential by showing a temperature drop
cooling for energy saving, water
from the ambient temperature under the sun, but only for a small thermal mass.
saving, and more efficient power
By eliminating the thermal loss through a delicate vacuum with an extremely small
generation in the near future.
thermal mass, an average temperature drop of as much as 37 C has been
achieved.19 However, to have practical impact, objects with large thermal mass
need to be cooled down to target temperatures according to the application’s
need when utilizing radiative cooling. In real-world applications, the effectiveness
of radiative cooling can be affected by both operating strategies and weather con-
ditions.20 Weather effects such as local wind speed, precipitable water (humidity),

Joule 3, 111–123, January 16, 2019 ª 2018 Elsevier Inc. 111


and cloud cover can play important roles on the radiative cooling performance and
thus should be thoroughly investigated.

The applications of radiative sky cooling can be either via a passive way by simply
attaching a radiative cooling material to an object that needs to be cooled (e.g.,
cool roof21,22 and solar cell applications23,24), or in an active way by producing
cooled heat transfer fluid (e.g., water25) that can be coupled to other thermal sys-
tems (e.g., building air conditioner26 or power plant condensers27). Passive applica-
tions of radiative cooling have the advantage of system simplicity, but suffer from
possibly excessive cooling when cooling is not needed, especially during winter or
during the night.28 Active radiative cooling systems that generate cold heat transfer
fluids can be better controlled through circulation, and thus are more feasible for
those applications whereby cooling power needs to be regulated, and potentially
with greater efficiency gains. However, cooling of a heat transfer fluid poses a chal-
lenge due to the low energy density of radiative cooling along with significant para-
sitic thermal loss. By using a commercially available polyester film that has high
reflectance in the visible part of solar spectrum and near-blackbody emittance in
the infrared spectrum on top of a silver reflective surface, Goldstein et al.26 built a
circulated water-cooling system and showed a temperature drop between 3 C
and 5 C under a water flow rate of 0.2 L/(min,m2) for a total panel surface area of
0.74 m2, after a successful demonstration of subambient cooling of air during the
day using nanophotonic material.7

Very recently we reported a glass-polymer hybrid metamaterial thin film that can
generate 93 W/m2 radiative cooling power at noon under direct sunlight, and an
average of 110 W/m2 radiative cooling power for three consecutive days.8 This
‘‘designer metamaterial’’ consists of a transparent polymer over the whole solar spec-
trum, encapsulating randomly distributed silicon dioxide (SiO2) microspheres. The
metamaterial is extremely emissive across the entire atmospheric transmission win-
dow (infrared wavelength 8–13 mm) due to phonon-enhanced Fröhlich resonances
of the embedded microspheres. In this work, by applying the glass-polymer hybrid
metamaterial, we develop a radiative cooled-cold collection (RadiCold) module.
Here we demonstrate that water can be cooled to 10.6 C below ambient at noon un-
der the stationary conditions. The effects of radiative surface temperatures and
weather conditions including wind speed, precipitable water, and cloud cover, on
the cooling power of the RadiCold module have been studied and validated with
a theoretical model. Moreover, we demonstrated two operating strategies for
different application scenarios, controlled intermittent flow for producing water
with constant subambient temperatures and continuous flow for maximizing cooling
energy generation, respectively. We also demonstrated 72-hr continuous operation
of a kW-scale (13.5 m2 radiative surface area) RadiCold system consisting of multiple
RadiCold modules. The system can be potentially integrated with buildings to pro-
vide continuous day-and-night cooling. As cooling and air conditioning demand
1Department of Mechanical Engineering,
will increase dramatically in the 21st century, our work on studying the effects of oper-
University of Colorado, Boulder, CO 80309, USA
ating strategies and weather conditions on the radiative cooling performance can be 2Department of Civil and Architectural
essential for developing real-world applications of this appealing technology for en- Engineering, University of Wyoming, Laramie,
ergy saving, water saving, and more efficient power generation in the near future.29 WY 82071, USA
3Materials Science and Engineering Program,
University of Colorado, Boulder, CO 80309, USA
RESULTS AND DISCUSSION 4Lead Contact
Subambient Cooling of Water: Radiative Cooling Power versus Surface
*Correspondence: gtan@uwyo.edu (G.T.),
Temperature xiaobo.yin@colorado.edu (X.Y.),
Figure 1A shows a schematic cross-section of the RadiCold water-cooling module. It ronggui.yang@colorado.edu (R.Y.)
consists of a 10-mm-thick flat panel water container in the center, and 5.1-cm-thick https://doi.org/10.1016/j.joule.2018.10.006

112 Joule 3, 111–123, January 16, 2019


A

B C

Figure 1. The RadiCold Module and RadiCold Metafilm


(A) Schematic showing the thickness of each layer and heat transfer processes of the RadiCold
module under the sun. A 15-mm-thick polyethylene (PE) film (shown as solid yellow line) is placed
15 mm above the RadiCold metafilm to reduce convective thermal loss to the ambient. The module
has a 5.1-cm-thick polyisocyanurate rigid foam insulation board at the four sides and the bottom.
P rad is the infrared radiation power density from the RadiCold metafilm. P atm is the absorbed
atmospheric radiation power density on the RadiCold metafilm. Psolar is the absorbed solar
irradiation power density on the RadiCold metafilm. P conv+cond is the convective and conductive
heat fluxes between the RadiCold module and ambient air.
(B) Schematic of the bilayer RadiCold metafilm on top of the polycarbonate water container (size
not to scale).
(C) Photograph of the roll-to-roll (R2R) produced RadiCold metafilm without (left) and with (right)
silver layer.

polyisocyanurate board insulation at the four sides and the bottom, while the meta-
material thin film (hereafter named RadiCold metafilm) with a nominal thickness of
50 mm is laminated on the top surface of the water container using a 90-mm-thick
layer of pressure-sensitive adhesive. A 15-mm-thick air gap is created on top of
the RadiCold metafilm by using a 15-mm-thick polyethylene (PE) film to reduce
the convective thermal loss. Total radiative surface area of the module is 0.34 m2
(0.58 3 0.58 m). The water container, with multiple flat channels (channel cross-sec-
tion 20 3 9 mm) arranged in parallel inside, has a total volume of 2.72 L. The water
container is made of commercially available polycarbonate material with a wall
thickness of 0.5 mm. Here polycarbonate is used due to its long-term durability
for outdoor applications.30 The low cost of polycarbonate paneling is indeed very
appealing for large-scale deployment of radiative cooling systems in the future. Fig-
ures 1B and 1C show a schematic and a photograph of the roll-to-roll (R2R) manufac-
tured RadiCold metafilm.8 It consists of a 50-mm-thick polymer-glass bead hybrid
organic-inorganic film formed by the R2R extrusion, and a 200-nm-thick silver thin
film deposited by R2R physical vapor deposition. The RadiCold metafilm has an
average emissivity of 0.86 within the atmospheric transmission window (8–13 mm
wavelength), and a solar reflectivity of 0.95 over the wavelength of 0.3–2.5 mm
(see the blue curve in Figure 2B).

Joule 3, 111–123, January 16, 2019 113


Figure 2. Comparative Tests Showing RadiCold Module Performance
(A) Photograph of four water modules built for a comparative study on the rooftop of the Engineering Center at University of Colorado Boulder,
Colorado, USA.
(B) Measured wavelength-dependent emissivity of the RadiCold metafilm (blue curve), silver (yellow curve), and polycarbonate water container (red
curve) over a wavelength from 0.3 to 25 mm. The AM1.5 solar spectrum is plotted for visualization.
(C) Measured water temperatures in the four modules along with the ambient air temperature during the daytime in August 2017. Water temperature
inside the RadiCold module (blue solid curve) stays well below the ambient (black curve) throughout the day, with an average temperature difference of
10.6  C between 12:30 p.m. and 3 p.m.
(D) Measured water temperatures in the four modules along with the ambient air temperature during a night in August 2017. With the RadiCold
metafilm, water was cooled down to 11  C below ambient air.
(E) Comparison of experimentally measured net radiative cooling power (hollow dots) with the modeling results (curves) under clear sky conditions
for the RadiCold module, with measured 7.4- to 10.7-mm precipitable water at night and 12.5- to 17.0-mm precipitable water during the day. Here,
700 W/m 2 direct solar irradiance on the RadiCold metafilm is assumed for daytime performance modeling. The error bar in the reported cooling power
is based on the propagation of instrument uncertainty.

The energy balance of the RadiCold module is given by



cw mw + cp mp vT
= Prad Patm Psolar Pconv + cond ; (Equation 1)
A vt
where c and m are heat capacity and mass, and A is the radiative surface area
exposed to the sun and the sky. The subscripts w and p denote water and polycar-
bonate water container, respectively.

The net cooling power, which equals the rate of cold generation in water and poly-
carbonate container (left-hand side of Equation 1), can be expressed by the four
terms that denote the detailed heat transfer processes at the right-hand side of
Equation 1, as shown in Figure 1A. The thermal resistance for heat conduction
across the 0.5-mm-thick polycarbonate wall (0.0074 K/W) and thermal resistance
for natural convection across the 9-mm-thick water layer (0.03 K/W) are
much smaller than the thermal resistance for radiation on the module surface
(0.62 K/W). The ratio of thermal resistance for natural convection across the water
layer to the thermal resistance for radiation on the module surface is 0.05, suggest-
ing that water temperature inside the RadiCold module should be relatively uni-
form. Also, water temperature is assumed to be the same as the polycarbonate

114 Joule 3, 111–123, January 16, 2019


water container due to the low thermal conduction resistance across the wall.
Detailed modeling of the transient heat transfer processes is available in Supple-
mental Information 1.

Four identical water modules were fabricated for the comparative study (see Fig-
ure 2A). All water modules were placed on the rooftop with a 15 tilt angle to the
north, which allows easy filling of water and reduces incident solar irradiation on their
top surfaces. All modules are wrapped by highly reflective silver films except the top
surfaces. The shading between the modules is not a concern due to its flat shape
(68 3 68 3 8 cm) and being installed at the same height and orientation (Figure 2A).
The top surfaces of the four modules are at the same level and are in close proximity,
which makes the effect of wind on the four modules identical. Local wind speed and
relative humidity are recorded by a weather station (Ambient Weather WS-1002-
WIFI observer), which is installed at the same height but 2 m away from the water
modules. The temperatures of water in the modules and the ambient air were
measured by precalibrated K-type thermocouples (0.3 C accuracy). The thermocou-
ples that measure temperatures of the water modules are submerged in the water
inside the polycarbonate water container. The thermocouples that measure temper-
atures of ambient air were placed at sun-shielded locations where air can freely pass
by. Before the comparative test, another test was carried out to show that the four
water modules are truly identical by using the same surface material (i.e., RadiCold
metafilm). The thermocouple-measured ambient temperature is also compared with
the weather station-measured ambient temperature to demonstrate the accuracy of
measurement. See Supplemental Information 2 for thermocouple locations and
measurement details.

The top surfaces of the four water modules were then covered with materials of
different optical and thermal properties for the comparative study. These modules
are a RadiCold module as described above, a reflective module, a well-insulated mod-
ule, and a bare module, respectively. The RadiCold module is covered with the
RadiCold metafilm that is solar-reflective and infrared-emissive. The bare module
has no additional surface material on top of the polycarbonate water container, and
the polycarbonate material is directly exposed to the sun and the sky, which represents
those materials that are solar-absorptive and infrared-emissive (e.g., regular shingle
roof). The reflective module is covered by low-emissivity/absorptivity silver that is
highly reflective in the solar spectrum, which represents those materials that reduce
surface temperature through reflection (e.g., reflective cool roofs). The well-insulated
module is covered with a 2.5-cm-thick polyisocyanurate insulation board on the top
surface, which results in neither solar absorption nor radiative cooling to the sky, rep-
resenting an insulated confined space (e.g., well-insulated building attic).

Figure 2B shows the spectral absorptivity/emissivity of RadiCold metafilm along with


a silver film on a polymer (reflective module), both R2R manufactured in-house, and a
polycarbonate water container (bare module). Here a UV-visible near-infrared spec-
trophotometer is used to measure the solar reflectivity (0.3–2.5 mm) and a Fourier
transform infrared spectrometer is used to measure the emissivity at infrared regions
(2.5–25 mm), respectively. Figure 2C gives the measured average water tempera-
tures inside the four modules when exposed to the clear sky, along with the ambient
air temperature, in Boulder, Colorado, as one of the typical tests in August/
September 2017. For comparison, initial water temperatures were controlled to
be identical for all the modules. After the test started at 11 a.m., it was observed
that the water temperature of the RadiCold module experienced a drop at the
beginning from 11 a.m. to 12:30 p.m., after which the water temperature was kept

Joule 3, 111–123, January 16, 2019 115


at about 10.6 C below the ambient air between 12:30 p.m. and 3 p.m., even under
the highest direct solar irradiance of 710 W/m2. The temperature difference be-
tween ambient air and water increased further after 3 p.m. due to the decline of
the solar irradiance. For the reflective module, although the majority of the solar irra-
diance was reflected, the water temperature still went up and was the second highest
among the four modules. Water in the well-insulated module was also heated up and
reached a temperature slightly higher than ambient air in the afternoon. In contrast,
water in the bare module was heated up very quickly in the morning due to the ab-
sorption of solar irradiance, even though polycarbonate is a broadband emitter in
the infrared region (2.5–25 mm). Figure 2D shows the results of the radiative cooling
test during nighttime by using the same four modules. Before the experiment begin,
all modules were covered with 2.5-cm-thick thermal insulation boards to prohibit
radiative cooling effect. The insulation boards were removed and the experiment
started at midnight and the temperature of water in the RadiCold module dropped
down to approximately 11 C below the ambient temperature. The bare module
where the polycarbonate material has a high broadband infrared emissivity (see
red curve in Figure 2B) cooled down water to the second lowest, about 10.5 C below
the ambient. This comparison between a selective radiative cooling surface
(RadiCold module) and a broadband radiative cooling surface (bare module) indi-
cates that lower temperature can be achieved by using a wavelength-selective radi-
ative surface, although the rate of temperature drop could be slightly higher at the
beginning for the broadband radiative cooling surface. The radiative cooling effect
observed for the reflective module was negligible due to the low emissivity of the
silver surface. The temperature of the well-insulated module stayed highest
throughout the test. The performance of the RadiCold module over 72 continuous
hours is given in Supplemental Information 3.

The cooling power of radiative cooling to the sky is a function of radiative surface
temperature. A series of experiments at different surface temperatures of RadiCold
module were conducted under the clear sky conditions during both daytime and
nighttime (see Supplemental Information 4 for the measurement method). The pre-
cipitable water, which indicates the humidity level (see Equation 2), was between 7.4
and 10.7 mm for nighttime tests and 12.5 and 17.0 mm for daytime tests. Figure 2E
shows the net radiative cooling power, which equals the net cooling power when
Pconv+cond = 0 (see Equation 1), as a function of RadiCold metafilm surface temper-
ature. All tests were conducted under the condition that the RadiCold metafilm sur-
face temperature is close to the ambient air temperature to eliminate the thermal
loss to the ambient. Figure 2E shows that RadiCold modules at a higher surface tem-
perature have a higher net radiative cooling power for both daytime and nighttime.
At night, when RadiCold metafilm temperature is increased from 1 C to 20.5 C,
the net radiative cooling power increases from 61.2 to 101.4 W/m2. Essentially,
over 100 W/m2 net radiative cooling power can be expected with the RadiCold
metafilm for a temperature higher than 20 C when precipitable water is lower
than 10 mm. Daytime experiments were all performed around noon with a direct so-
lar irradiance on the RadiCold metafilm between 680 and 730 W/m2. When RadiCold
metafilm temperature increased from 10.9 C to 31.9 C, net radiative cooling power
increased from 39.7 C to 79.0 W/m2. Also shown in Figure 2E is the calculated
net radiative cooling power from the model (described in Supplemental Informa-
tion 1), which agrees well with the measurement results.

Weather Effects on Radiative Cooling


Radiative cooling performance in real-world applications, including both the subam-
bient temperature drop and the net cooling power, are strongly affected by the

116 Joule 3, 111–123, January 16, 2019


A B

Figure 3. Effect of Local Wind Speed on Radiative Cooling Performance


(A) A typical radiative cooling test showing the effect of wind. To better show the effect of wind, PE
film is not used on top of the RadiCold metafilm. Other influencing factors (e.g., precipitable water,
ambient temperature) are relatively stable during the test. Temperatures of ambient air (black short
dashed curve) and water (red solid curve) are plotted. The predicted transient water temperature
(blue dashed curve) in RadiCold module matches well with the measured data.
(B) Precipitable water (solid red curve) and wind speed (black dots) change during the test period.

weather conditions. For example, the absorbed power density of atmospheric radi-
ation (Patm) is affected by precipitable water (humidity) and cloud cover, the ab-
sorbed power density of solar irradiance (Psolar) is affected by both solar irradiation
and cloud cover, and convection and conduction heat fluxes (Pconv+cond) are affected
by both the local wind speed and the temperature difference between RadiCold
module surface and the ambient air. Here we present the study of the weather effects
on the radiative cooling performance of RadiCold metafilms arising from wind and
precipitable water. Cloud cover has a very complicated impact on the atmospheric
radiation due to the different types of clouds (e.g., cirrus, stratus), different height of
clouds, different constitution of clouds, and the constantly changing cloud behaviors
over time and space. Recognizing the importance of cloud impact on a radiative
cooling system for real-world operations, we present some preliminary results in
Supplemental Information 5 and 6 showing that the decrease in radiative cooling po-
wer could be proportional to the increase of cloud cover both at night and during the
day, when compared with clear sky.

Local wind speed affects convective heat transfer between the RadiCold metafilm
surface and the ambient air. The effect of wind on the RadiCold module can be either
beneficial or detrimental, depending on the temperature difference between
RadiCold module and ambient air. For subambient radiative cooling applications,
wind-induced thermal loss should be suppressed. Figure 3 gives a typical cooling
test of RadiCold module whereby the wind effect is obvious. The test was conducted
at night under clear sky conditions to avoid the impact of solar irradiance and clouds.
The ambient air temperature (between 9 C and 12 C) and precipitable water (be-
tween 4.5 and 6 mm) are relatively stable during the test period (highlighted in
gray), as shown in Figure 3B. Figure 3A shows that maximum subambient cooling
temperature difference between water and ambient air decreased from 7.0 C to
4.3 C (without a PE film on top of the RadiCold module), due to the change of
wind speed since it is much smaller at the beginning of the test compared with
that at the end. The evaluation of convective heat transfer coefficient between
RadiCold module and ambient air is given in Supplemental Information 1 and the
validation is provided in Supplemental Information 7.

Humidity affects the radiative cooling performance as it affects the atmospheric


emissivity, and thus the absorbed atmospheric radiation (Patm). It is usually

Joule 3, 111–123, January 16, 2019 117


Figure 4. Effect of Precipitable Water (Humidity) on Net Radiative Cooling Power under Clear
Sky Conditions
Modeling and experimental results are plotted as curves and dots, respectively. Daytime and
nighttime modeling results are plotted in solid yellow and dashed blue color. All tests were
conducted at relatively stable RadiCold metafilm temperatures (ambient air temperatures) of
15  C G 1.5  C with clear sky conditions. All daytime tests were conducted around noon with direct
solar irradiance on the RadiCold metafilm between 680 and 730 W/m 2 . The daytime modeling
results (dashed yellow curve) assume 700 W/m 2 direct solar irradiance on the RadiCold metafilm.
The error bar in the reported cooling power and precipitable water represents the propagation of
instrument uncertainty. The error bar in ambient temperature represents a conservative estimation
on ambient temperature change during the measurement period.

quantified as precipitable water (PW), which is defined as the depth of water in a col-
umn of the atmosphere if all the water in that column were precipitated as rain.31 The
increase in precipitable water results in an increased Patm and thus a reduced net
radiative cooling power. The precipitable water (in millimeters) at an altitude of
around 1,600 m (‘‘mile high city,’’ Boulder, Colorado) for clear sky conditions can
be calculated from experimentally measured relative humidity (RH) and ambient
temperature (Tamb) using the equation
 17:625T 
amb
3800 exp
Tamb + 243:04
PWz2:15RH 0:82; (Equation 2)
patm
where patm is the ambient pressure, Pa.

Figure 4 shows the effect of precipitable water on radiative cooling under the clear
sky conditions. The net radiative cooling power is measured in the same way as in
Figure 2E to rule out the impact of convective thermal loss to the ambient. When pre-
cipitable water changed from 7.4 to 17.1 mm at night, the net radiative cooling po-
wer dropped from 89.2 to 84.2 W/m2, and when precipitable water changed from

118 Joule 3, 111–123, January 16, 2019


A B

Figure 5. Cool-Water Production to Target Temperatures during the Day


(A) Two identical RadiCold modules were tested simultaneously with different target temperatures
at 24  C (red curve) and 26  C (green curve), respectively. Inlet water temperatures are controlled to
be close to ambient air temperature (black curve).
(B) Modeling results show cool-water production rate as a function of the targeted subambient
temperature drop at three different ambient temperatures. Solar irradiance is kept at 700 W/m 2
and wind speed is constant at 1 m/s.

4.1 to 13.9 mm during the day, net radiative cooling power dropped from 54.1 to
47.2 W/m2. The modeling results show that net radiative cooling power decreases
with the increase in precipitable water in a logarithmic relationship, which means
that the rate of decrease is faster when precipitable water is smaller.

Subambient Cool-Water Production at Constant Temperatures


Stationary tests as described above demonstrated the capability to generate sub-
ambient cool water using radiative cooling for both daytime and nighttime. Practical
use of radiative cooling could involve both controlled intermittent flow for producing
water with constant subambient temperatures and continuous flow for maximizing
cooling energy generation. Here, we first demonstrate subambient cool-water pro-
duction at target temperatures. Two identical RadiCold modules were tested simul-
taneously with different target temperatures in the daytime, 24 C (module 1) and
26 C (module 2), respectively. In real-world applications, the target temperature
can be determined by the specific application scenarios. Temperature of the supply
water to the RadiCold module tracks ambient air temperature, which mimics those
scenarios in which supply water is first cooled by an air-cooled heat exchanger, usu-
ally with a temperature slightly higher than the ambient. Water was stationary in the
modules for most of the time during the test. Once the water temperature in the
module reached the target temperature, the module was drained and refilled. Since
water stays stationary (i.e., controlled intermittent flow) in each module for most of
the cooling time, to obtain continuous production of cool water in real-world appli-
cations, many modules can be connected into a system32 and operated with a con-
trol strategy (see Supplemental Information 8). Compared with continuous flow of
water, this operation strategy can also save electricity for pumping the water.

Figure 5A shows that between 10:10 a.m. and 17:30 p.m., both modules can cool
water down to the specified target temperatures for multiple times (i.e., multiple
drain and refill). The rate of temperature drop at each time just after water refilling
is the greatest due to the highest water temperature and smallest subambient tem-
perature difference. The results show that the RadiCold module can produce decent
amount of 24 C water on a day that has maximum direct solar irradiance of 700 W/m2
and maximum ambient air temperature of 32 C. The total cool-water production was
32.1 L/m2 (equivalent water flow rate 4.4 L/(h,m2)) for targeted temperature of 24 C

Joule 3, 111–123, January 16, 2019 119


Figure 6. Radiative Cooling Test on a kW-Scale RadiCold System
(A) The kW-scale RadiCold system. It consists of ten modules with a total radiative surface area of
13.5 m 2 .
(B) Solar irradiance. The weather on July 1 and July 2 are mostly clear, while scattered clouds are
present on the afternoon of July 3.
(C) Inlet (black solid curve) and outlet (blue dashed curve) water temperatures. Before entering the
RadiCold system, inlet water went through an air-cooled heat exchanger to ensure its temperature
tracked the ambient temperature.
(D) The maximum cooling power at night during these three days is 96 W/m 2 , and the average
cooling power at noon time (12–2 p.m.) is 45 W/m 2 on July 1.

in module 1 and 64.1 L/m2 (equivalent water flow rate 8.7 L/(h,m2)) for targeted tem-
perature of 26 C in module 2. The average net cooling power was 31.9 W/m2 for
module 1 in comparison with 41.8 W/m2 for module 2, which was kept at a higher
target temperature. The difference indicates that for subambient radiative cooling
of water, the net cooling power is a function of both subambient temperature differ-
ence and the radiative surface temperature. A larger subambient temperature differ-
ence results in a larger thermal loss to the ambient and, therefore, a smaller net
cooling power. On the other hand, a smaller temperature difference might not
generate sufficient useful cooling utility for practical applications. There is appar-
ently a need to strategize the operation to balance the net cooling power and the
subambient temperature, which are dependent upon each other. Figure 5B shows
cool-water production rate as a function of the targeted subambient temperature
drop at three different ambient temperatures. Higher ambient temperature and
smaller targeted temperature drop result in higher cooling water production rate.

kW-Scale RadiCold System toward Real-World Applications


To demonstrate the scalability of radiative cooling, we built a kilowatt (kW)-scale
RadiCold system using our RadiCold metafilm on the roof of CU Engineering Center
in Boulder, Colorado. The system consists of ten modules, each has surface area
1.35 m2, giving a total radiative cooling surface area of 13.5 m2 (Figure 6A). Three
consecutive days of tests were carried out on July 1–3, 2018. Figure 6B shows that
continuously high solar irradiance was observed on July 1 and July 2, while scattered

120 Joule 3, 111–123, January 16, 2019


clouds on the afternoon of July 3 brought in lower solar irradiance. As mentioned
earlier, active use of radiative cooling could involve cool-water production with a
continuous flow. Water volumetric flow rate was 26.5 L/(h,m2) during the test. Inlet
and outlet water temperatures of the RadiCold system were monitored and are
plotted in Figure 6C. The inlet water temperature tracks ambient temperature
throughout the test. Net cooling power is calculated from the inlet/outlet tempera-
_
ture difference by using the equation q = c mDT
A , where A is 13.5 m . Figure 6D shows
2

that the net cooling power per square meter fluctuates between 40 and 100 W/m2
during the 3-day period. Maximum net cooling power at night is 96 W/m2
(1,296 W for the 13.5 m2 system) when the water was cooled to 3.1 C subambient
on average, and the average cooling power at noon (12–2 p.m.) is 45 W/m2
(607 W for the system) under an average of 952 W/m2 solar irradiance on July 1.
Under scattered cloud conditions on the afternoon of July 3, the system generated
50–60 W/m2 cooling power.

RadiCold System to Provide Continuous Day-and-Night Cooling for Buildings


As shown in Figure 6C, the outlet water temperature of the RadiCold system
operated under continuous-flow conditions changes significantly between day
and night. The cooled water can be used to directly cool the condenser to improve
the efficiency of an air conditioner during the day.26 However, considering that there
could be limited or even no cooling demand from buildings at night, this operation
strategy limits the effectiveness of the radiative cooling system. We address this
challenge by introducing scheduled operations and a cold storage unit (see Fig-
ure 6A on the right-hand side of the RadiCold modules) to maximize the use of
the RadiCold system. A building-integrated RadiCold system is proposed to sepa-
rate the use of daytime and nighttime radiative cooling (Figure S13A). During the
day, the subambient temperature heat transfer fluid generated, as demonstrated
in Figure 6C, is used to directly cool the condenser, while at night, the RadiCold sys-
tem stores the cooling energy in a storage unit, which is retrieved during the day to
reduce the cooling load on the air conditioner. The system efficiency can be signif-
icantly enhanced with day-and-night continuous cooling (see Supplemental Informa-
tion 10).

A modeling tool based on EnergyPlus33 and MATLAB has been developed for the
evaluation of electricity saving for cooling a 5,000 m2 floor area commercial office
building34 by employing the building-integrated RadiCold system (Supplemental In-
formation 10). The system is designed with a radiative surface area of 810 m2 (about
half of the building roof area), corresponding to an area ratio of 1:6.2 (radiative cool-
ing surface area to building total floor area). Annual performance of the RadiCold
system has been evaluated for three different locations in the United States
(Phoenix, Houston, and Miami) using TMY3 weather data. Results show that for
different locations, the RadiCold system could potentially save 64%–82% of the elec-
tricity consumption for cooling in winter (from November to February), and save
32%–45% of the electricity consumption for cooling in summer (from May to August).

In summary, by using the R2R manufactured solar-reflective and infrared-emissivity


RadiCold metafilm, we developed a RadiCold module and experimentally demon-
strated cooling of water to 10.6 C below ambient under the sun under stationary
conditions. Higher radiative surface temperature gives higher net radiative cooling
power during both daytime and nighttime, which provides a guideline for the design
of different radiative cooling systems that have different operating temperatures.
Continuous production of cool water at constant subambient temperatures has
been demonstrated. The comparison between different target temperatures (24 C

Joule 3, 111–123, January 16, 2019 121


and 26 C) indicates that for subambient radiative cooling, there exists a compromise
between the subambient temperature difference and the net cooling power, the two
performance indices of the technique. The correlation between cool-water produc-
tion rate (i.e., net cooling power) and subambient temperature drop at different
ambient temperatures has been confirmed. The effects of different weather condi-
tions (local wind speed, precipitable water, and cloud cover) on the performance
of radiative cooling have been investigated. The RadiCold module constructed in
this work uses various low-cost and commercially available polymer-based materials,
which is appealing for large-scale deployment of radiative cooling systems in the
near future. A kW-scale RadiCold system has been built to demonstrate the scalabil-
ity of our radiative sky cooling technology, which provides a maximum cooling
power of 1,296 W at night and an average cooling power of 607 W at noon
(12–2 p.m.) under 952 W/m2 average solar irradiance at a water volumetric flow
rate of 26.5 L/(h,m2). As cooling and air conditioning demand increases dramatically
in the 21st century,35 our work here paves the way for pursuing RadiCold technology
for real-world applications, which could result in energy saving, water saving, and
more efficient power generation.

EXPERIMENTAL PROCEDURES
Full experimental procedures are provided in the Supplemental Information.

SUPPLEMENTAL INFORMATION
Supplemental Information includes Supplemental Experimental Procedures and 14
figures and can be found with this article online at https://doi.org/10.1016/j.joule.
2018.10.006.

ACKNOWLEDGMENTS
The authors acknowledge the financial support of this work from the US Department
of Energy’s Advanced Research Projects Agency – Energy (ARPA-E) under contract
no. DE-AR0000580.

AUTHOR CONTRIBUTIONS
D.Z. and R.Y. conceived the concept of this work. D.Z. led all the efforts under the
supervision of R.Y., X.Y., and G.T. All authors participated in the construction, exper-
imentation, and data analysis. D.Z. drafted the paper. R.Y., D.Z., X.Y., and G.T. final-
ized the paper, while all co-authors participated in the discussion.

DECLARATION OF INTERESTS
The authors declare no competing interests.

Received: May 1, 2018


Revised: July 22, 2018
Accepted: October 3, 2018
Published: October 26, 2018

REFERENCES
1. Smith, G.B., and Granqvist, C.G. (2010). Green radiative cooling of selective surfaces. Sol. 4. Martin, M., and Berdahl, P. (1984).
Nanotechnology: Solutions for Sustainability Energy 17, 83–89. Characteristics of infrared sky radiation in the
and Energy in the Built Environment (CRC United States. Sol. Energy 33, 321–336.
Press). 3. Granqvist, C.G., and Hjortsberg, A. (1981).
Radiative cooling to low temperatures: general 5. Meir, M.G., Rekstad, J.B., and Løvvik, O.M.
considerations and application to selectively (2002). A study of a polymer-based
2. Catalanotti, S., Cuomo, V., Piro, G., Ruggi, D., emitting SiO films. J. Appl. Phys. 52, 4205– radiative cooling system. Sol. Energy 73,
Silvestrini, V., and Troise, G. (1975). The 4220. 403–417.

122 Joule 3, 111–123, January 16, 2019


6. Eicker, U., and Dalibard, A. (2011). 16. Fan, S. (2017). Thermal photonics and energy with the sky. Nat. Energy 2, https://doi.org/10.
Photovoltaic-thermal collectors for night applications. Joule 1, 264–273. 1038/nenergy.2017.143.
radiative cooling of buildings. Sol. Energy 85,
1322–1335. 17. Atiganyanun, S., Plumley, J.B., Han, S.J., Hsu, 27. Zeyghami, M., and Khalili, F. (2015).
K., Cytrynbaum, J., Peng, T.L., Han, S.M., and Performance improvement of dry cooled
7. Raman, A.P., Anoma, M.A., Zhu, L., Rephaeli, Han, S.E. (2018). Effective radiative cooling by advanced concentrating solar power plants
E., and Fan, S. (2014). Passive radiative cooling paint-format microsphere-based photonic using daytime radiative cooling. Energy
below ambient air temperature under direct random media. ACS Photonics 5, 1181–1187. Convers. Manag. 106, 10–20.
sunlight. Nature 515, 540–544.
18. Yang, P., Chen, C., and Zhang, Z.M. (2018).
8. Zhai, Y., Ma, Y., David, S.N., Zhao, D., Lou, R., A dual-layer structure with record-high solar 28. Pisello, A.L., and Cotana, F. (2014). The thermal
Tan, G., Yang, R., and Yin, X. (2017). Scalable- reflectance for daytime radiative cooling. Sol. effect of an innovative cool roof on residential
manufactured randomized glass-polymer Energy 169, 316–324. buildings in Italy: results from two years of
hybrid metamaterial for daytime radiative continuous monitoring. Energy Build. 69,
cooling. Science 355, 1062–1066. 19. Chen, Z., Zhu, L., Raman, A., and Fan, S. (2016). 154–164.
Radiative cooling to deep sub-freezing
9. Gentle, A.R., and Smith, G.B. (2015). A temperatures through a 24-h day-night cycle. 29. Stark, A.K., and Klausner, J.F. (2017). An R&D
subambient open roof surface under the mid- Nat. Commun. 7, 13729. strategy to decouple energy from water. Joule
summer sun. Adv. Sci. (Weinh). 2, 1500119. 1, 416–420.
20. Suichi, T., Ishikawa, A., Hayashi, Y., and Tsuruta,
10. Kou, J., Jurado, Z., Chen, Z., Fan, S., and K. (2018). Performance limit of daytime 30. Sethi, V.P., and Sharma, S.K. (2008). Survey and
Minnich, A.J. (2017). Daytime radiative cooling radiative cooling in warm humid environment. evaluation of heating technologies for
using near-black infrared emitters. ACS AIP Adv. 8, 055124. worldwide agricultural greenhouse
Photonics 4, 626–630. applications. Sol. Energy 82, 832–859.
21. Zinzi, M., and Agnoli, S. (2012). Cool and green
11. Zhang, K., Zhao, D., Yin, X., Yang, R., and Tan, roofs. An energy and comfort comparison
G. (2018). Energy saving and economic analysis 31. Duan, J., Bevis, M., Fang, P., Bock, Y., Chiswell,
between passive cooling and mitigation urban
of a new hybrid radiative cooling system for S., Businger, S., Rocken, C., Solheim, F., van
heat island techniques for residential buildings
Hove, T., Ware, R., et al. (1996). GPS
single-family houses in the USA. Appl. Energy in the Mediterranean region. Energy Build. 55,
224, 371–381. meteorology: direct estimation of the absolute
66–76.
value of precipitable water. J. Appl. Meteorol.
12. Bao, H., Yan, C., Wang, B., Fang, X., Zhao, C.Y., 22. Synnefa, A., and Santamouris, M. (2012). 35, 830–838.
and Ruan, X. (2017). Double-layer nanoparticle- Advances on technical, policy and market
based coatings for efficient terrestrial radiative aspects of cool roof technology in Europe: the 32. Zhang, K., Zhao, D., Zhai, Y., Yin, X., Yang, R.,
cooling. Sol. Energy Mater. Sol. Cells 168, Cool Roofs project. Energy Build. 55, 35–41. and Tan, G. (2017). Modelling study of the low-
78–84. pump-power demand constructal T-shaped
23. Zhu, L., Raman, A., Wang, K.X., Anoma, M.A., pipe network for a large scale radiative cooled-
13. Zou, C., Ren, G., Hossain, M.M., Nirantar, S., and Fan, S. (2014). Radiative cooling of solar cold storage system. Appl. Therm. Eng. 127,
Withayachumnankul, W., Ahmed, T., cells. Optica 1, 32–38. 1564–1573.
Bhaskaran, M., Sriram, S., Gu, M., and
Fumeaux, C. (2017). Metal-loaded dielectric 24. Li, W., Shi, Y., Chen, K., Zhu, L., and Fan, S. 33. EnergyPlus. https://energyplus.net/.
resonator metasurfaces for radiative cooling. (2017). A comprehensive photonic approach
Adv. Opt. Mater. 5, 1700460. for solar cell cooling. ACS Photonics 4,
34. 90.1 Prototype Building Models—Medium
774–782.
14. Huang, Z., and Ruan, X. (2017). Nanoparticle Office. Building Energy Codes Program.
embedded double-layer coating for daytime 25. Zhao, D., Martini, C.E., Jiang, S., Ma, Y., Zhai, https://www.energycodes.gov/901-prototype-
radiative cooling. Int. J. Heat Mass Transf. 104, Y., Tan, G., Yin, X., and Yang, R. (2017). building-models-medium-office.
890–896. Development of a single-phase thermosiphon
for cold collection and storage of radiative 35. Isaac, M., and van Vuuren, D.P. (2009).
15. Yang, P., Chen, C., and Zhang, Z.M. (2018). cooling. Appl. Energy 205, 1260–1269. Modeling global residential sector energy
A dual-layer structure with record-high solar demand for heating and air conditioning in the
reflectance for daytime radiative cooling. Sol. 26. Goldstein, E.A., Raman, A.P., and Fan, S. (2017). context of climate change. Energy Policy 37,
Energy 169, 316–324. Sub-ambient non-evaporative fluid cooling 507–521.

Joule 3, 111–123, January 16, 2019 123

You might also like