You are on page 1of 15

Energy 278 (2023) 127706

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Balancing energy harvesting and crop production in a nanofluid spectral


splitting covering for an active solar greenhouse
Yu Yuan a, b, c, 1, Yaning Ji b, c, 1, Wei Wang b, c, Dawei Shi d, Long Hai d, Qianlei Ma c, d,
Qichang Yang c, d, Yuming Xie a, Bin Li a, Gang Wu b, c, 1, *, Lingling Ma e, 1, **
a
Institute of Facility Agriculture, Guangdong Academy of Agricultural Sciences, Guangzhou, 510640, China
b
Institute of Environment and Sustainable Development in Agriculture, Chinese Academy of Agriculture Sciences, Beijing, 100081, China
c
Key Laboratory of Energy Conservation and Waste Management of Agricultural Structures, Ministry of Agriculture, Beijing, 100081, China
d
Institute of Urban Agriculture, Chinese Academy of Agricultural Sciences, Chengdu, 610000, China
e
Agricultural Genomics Institute at Shenzhen, Chinese Academy of Agriculture Sciences, Shenzhen, 518000, China

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

Handling Editor: Krzysztof (K.J.) Ptasinski This work represents a nanofluid spectral splitting (NSS) covering applied to the greenhouse roof to achieve the
co-production of crops and energy harvesting. NSS is a hollow structure composed of highly light-transmissive
Keywords: rigid materials with 10 mm thick 0.005 vol% ATO-WO3 nanofluids circulating in the hollow layer with a flow
Solar energy rate of 400 L/h. To study the growth response of plants in the NSS greenhouse (NSS-G), plant cultivation ex­
Plant
periments and photothermal tests were carried out. The energy analysis and performance estimation of NSS-G
Greenhouse
were proved. The results show that NSS-G can absorb 77.9% of the solar energy within 800–1500 nm wave­
Spectral splitting
Nanofluid lengths to generate heat. Compared with the conventional hollow covering greenhouse (Air-G), NSS-G can
reduce the temperature of indoor air and plant leaves by 6.3 ◦ C and 8.7 ◦ C, respectively. Meanwhile, 76.6% of
sunlight within 300–800 nm wavelengths is transmitted to plants for photosynthesis. The results of plant growth
experiments showed that, in comparison with Air-G, the photosynthetic rate in NSS-G reduced by 6.5%, but plant
dry weight was increased by 4.3% and there was no significant difference in biochemical component contents. It
proved that it is feasible to use NSS as the covering of the greenhouse roofs, and the total solar energy utilization
efficiency in NSS-G was more than 55%.

1. Introduction pollutants. Therefore, it requires the development of a highly efficient


and sustainable system to employ renewable energy. Solar energy can be
The Food and Agriculture Organization (FAO) estimates that food considered the most widely distributed and no geographically depen­
production needs to increase by 70% to feed the expected global pop­ dent renewable energy [6]. It is becoming an effective way to solve the
ulation of 9 billion in 2050 [1]. However, food insecurity may elevate energy crisis and environmental problems [7].
with global warming, disasters (like COVID-19), and wars [2]. Sunlight consists of ultraviolet, visible, and infrared spectra. Ac­
Compared with open-field farming, modern cultivation techniques in cording to its application, the solar spectrum can be mainly divided into
greenhouses have been proven to be an effective solution to rise pro­ plant active spectrum (PAS) (300–800 nm) and heat active spectrum
duction and overcome the problem of crop production caused by (HAS) (800–1500 nm) [8,9]. PAS is used in plant photosynthesis. The
extreme weather conditions [3]. Currently, the global area of the absorption peaks of chlorophyll a and chlorophyll b of plants lie on blue
greenhouse is larger than 3 600 000 ha worldwide [4]. However, the light within 400–500 nm wavelength and red light within 600–700 nm
thermal regulation of greenhouses requires higher energy demand to wavelength [10,11]. Blue light can improve the cotyledon unfolding and
maintain a controlled environment for plant growth [5]. Conventional expansion, and regulate flowering by the length of the day [12]. Red
fossil fuels always lead to large energy consumption and emission of light plays a critical role in mediating plant growth and development

* Corresponding author. Institute of Environment and Sustainable Development in Agriculture, Chinese Academy of Agriculture Sciences, Beijing, 100081, China.
** Corresponding author.
E-mail addresses: yuanyu_0530@163.com (Y. Yuan), wugang01@caas.cn (G. Wu), Lingling00@126.com (L. Ma).
1
These authors contributed equally to this work.

https://doi.org/10.1016/j.energy.2023.127706
Received 9 February 2023; Received in revised form 23 April 2023; Accepted 30 April 2023
Available online 12 May 2023
0360-5442/© 2023 Elsevier Ltd. All rights reserved.
Y. Yuan et al. Energy 278 (2023) 127706

Fig. 1. Diagram of a planting greenhouse using NSS as the roof.

Fig. 2. (A) TEM image of ATO-WO3 NPs. Insertion: volume of NPs diameter distribution and crystal structure of WO3 through XRD analysis. (b) Extinction coefficient
of de-ionized water, ATO-WO3 NFs with different volume concentrations of 0.0025 vol%, 0.005 vol%, and 0.01 vol%; Test optical path is 10 mm. Insertion: pho­
tographs of ATO-WO3 NFs with different volume fractions. (c) Solar spectrum distribution passed through NSS with different NFs layer thicknesses at 12:00 p.m. 10
mm-Water represents hollow PMMA covering filled with 10 mm water layer thickness. 5 mm, 10 mm, 15 mm, and 20 mm represent the NSS with 5 mm, 10 mm, 15
mm, and 20 mm 0.005 vol% ATO-WO3 NFs layer thickness. (d) Average transmittance in PAS and absorptance in HAS of NSS with 10 mm thick 0.005 vol% ATO-WO3
NFs layer at different flow rates at 10:00 a.m [9].

[13]. HAS is not suitable for plant growth, but it will cause indoor high to find a way to transmit the PAS light to plants inside the greenhouse
temperatures when entering the greenhouse [14]. It needs to be venti­ while absorbing the HAS radiation of solar energy in the covering layer.
lated and cooled at noon in winter, which causes energy loss. Mean­ Spectral beam splitting (SBS) can optimally split the solar spectrum
while, compulsory refrigeration measures are needed in summer, according to the needs [2,16]. SBS applied in the cover layer of the
increasing the cost of greenhouse energy [15]. Therefore, it is necessary greenhouse not only improves the solar utilization efficiency but also

2
Y. Yuan et al. Energy 278 (2023) 127706

Fig. 3. (A) Diagram of test greenhouse. (b) Optical properties of reflective film, 6 mm thick PMMA, and 0.005% ATO-WO3 NFs (10 mm optical length) for the test. (c)
Schematic diagram of the operation principle of NSS and measuring sensor layout. The surface temperature of the greenhouse wall and roof includes internal and
external surfaces.

realizes the cooling ability on sunny days. unknown. It prevents the application of NFs in practical greenhouses
Generally, the SBS technologies applied in the greenhouse include and needs to be studied clearly.
photothermal (PT) [9], and photovoltaic (PV) [17]. PV mainly used The main objective of the presented paper is to study the photo­
crystalline silicon cells, whose effective response wavelength is about thermal performance of the greenhouse using NFs as a roof, and the
400–1100 nm. However, the plant absorption is located at 300–800 nm, growth response of plants in this light and thermal environment. Out­
which causes competition between plants and PV [18]. Other research door experiments using 10 mm thick 0.005 vol% ATO-WO3 nanofluids
on some PV materials such as organic photovoltaics (OPVs) [19], circulating in the hollow layer with a flow rate of 400 L/h were con­
wavelength selective semitransparent polymer solar cells (ST-PSCs) ducted as a roof in a practical greenhouse, lettuce was grown in it and
[20], perovskite solar cells (PSCs) [21], and dichroitic film [22] had tested. Then, the numerical model is established to analyze the energy
attracted a large number of scientists. However, the photoelectric con­ flow of the greenhouse using NFs as the roof. This research aims to make
version efficiency of the above PV materials is still relatively low, and use of the NFs in the greenhouse and improve the utilization efficiency of
the transmitted sunlight is difficult to meet the needs of plants. Previous solar energy.
researches proved that PT receivers have higher photothermal conver­
sion efficiency compared with PV [23,24]. 2. Description of NSS greenhouse
PT includes the indirect absorption collector (IDAC) and the direct
absorption collector (DAC) according to different heat-collection The conceptual diagram of the NSS greenhouse is depicted in Fig. 1.
methods [25]. Previous studies have proved that the heat loss of DAC The greenhouse roof is covered by a nanofluid spectral splitting (NSS)
was lower than that of IDAC due to the lower temperature on the system. NSS is a hollow structure consisting of highly light-transmissive
heat-absorbing surface [9,25,26]. The DAC system absorbs solar radia­ rigid materials with nanofluids (NFs) flowing inside. The NFs are
tion directly by using a heat-collecting medium, which is not only a composed of nanoparticles (NPs) suspended in the base fluid to achieve
heat-absorbing material but also a heat transfer medium [9]. DAC based the desired spectrum partition: transmit the plant photosynthesis active
on nanofluids (NFs) has selective strong absorption and high trans­ spectrum (PAS: plant active spectrum, 300–800 nm) to plant growth,
missivity of sunlight [27]. It is feasible to use NFs as SBS in the green­ while absorbing the rest spectrum of sunlight (HAS: heat active spec­
house roof, and solar energy utilization efficiency is high [28–30]. trum, 800–1500 nm) to generate heat.
In our previous work [9], ATO-WO3/H2O NFs were synthesized and Through previous studies, ATO-WO3/H2O NFs were selected and
tested in outdoor experiments. It verified that 0.005 vol% ATO-WO3 NFs showed superior performance to transmit PAS and absorb HAS [9]. The
with 10 mm optical path-length had a transmissivity of 79.56% in PAS combined ATO-WO3 NPs are composed of 2.4 vol% ATO and 97.6 vol %
and an absorption of 75.78% in HAS. However, the light and thermal WO3, and the base fluids are deionized water. Fig. 2a shows TEM image
environment distribution in the greenhouse using NFs as the roof was of ATO-WO3 NPs. The average diameter of ATO-WO3 NPs is 42.18 nm,
not clear, and the growth response of plants in this environment was also and the crystal structure of WO3 is composed of a series of

3
Y. Yuan et al. Energy 278 (2023) 127706

perovskite-like [WO6] octahedral building blocks with connecting cor­ capacity of 80 L was regarded as a hydraulic close cycle connected with
ners in three-dimensional (see inserting in Fig. 2a). NSS. The experimental control was set with the daylight roof of the
The absorption coefficient a(λ) of NFs can be calculated by the Beer- greenhouse using a 10 mm thick hollow air layer (Air-G) instead of NFs.
Lambert Law [31]: The reflective film pasted on the internal greenhouse wall excluded the
( ) ( ) daylight roof to eliminate the marginal effect around the greenhouse on
1 1 1 I0 (λ)
a(λ) = ln = ln (1) plants. The reflectance of the reflective film used in this experiment is
l τ(λ) l I(λ)
shown in Fig. 3b. The greenhouse was kept closed during the test. A
heater was used to heat the greenhouse at night to maintain suitable
where λ is the wavelength, nm; l is the optical path-length (assumed to
environments for plants growing at 15–18 ◦ C during the experiment.
be 10 mm); τ(λ) is the optical transmittance of NFs (− ); I0(λ) is the
Lettuce (Lactuca sativa L. cv. “Tiberius”) was selected as experimental
radiant flux received by NFs (W⋅m− 2⋅nm− 1); I(λ) is the radiant flux
material. On February 6th, the seeds were sown in sponge cubes (2.5 ×
transmitted by NFs (W⋅m− 2⋅nm− 1). The UV–Vis–NIR spectrophotometer
2.5 × 2.5 cm) in a dark environment for 2 days. After the seeds germi­
recorded I0(λ) and I(λ) on a spectrophotometer, and the fluids were filled
nated, the seedlings were transferred to the light intensity of 200 μmol
in a high light transmittance quartz cuvette with an optical path-length
m− 2 s− 1 and the photoperiod of 16/8 h white LED in an environment-
of 10 mm. Fig. 2b shows the absorption coefficient of NFs at different
controlled plant factory. After 14 days of seedling growth, the seed­
volume concentrations of NFs. The crop growth factor was used to
lings were transplanted into pots (1.3 L, 135 mm, and 95 mm in top and
quantitatively assess the impact of the transparent devices on plant
bottom diameter, and 125 mm in height) containing quartz sand. There
growth [20,32]. The most suitable volume concentration of 0.005 vol%
were 20 plants for each treatment. Lettuce was cultured in Hoagland’s
NFs was determined by previous work [9].
solution containing 6 mmol L− 1 KNO3, 1 mmol L− 1 NH4H2PO4, 4 mmol
The NSS system was designed and fabricated for outdoor experi­
L− 1 Ca(NO3)2⋅4H2O, 2 mmol L− 1 MgSO4⋅7H2O, 46 μmol L− 1 H3BO3, 71
ments with 0.005 vol% ATO-WO3/H2O NFs [9]. Fig. 2c shows the solar
μmol L− 1 Fe-EDTA-Na2, 9.6 μmol L− 1 MnSO4⋅4H2O, 0.07 μmol L− 1
spectrum distribution of NSS with different NFs layer thicknesses at
(NH4)6Mo7O24⋅4H2O, 0.8 μmol L− 1 CuSO4⋅5H2O (Electro conductibility:
12:00 p.m. Fig. 2d shows the average transmittance in PAS and
1.63 mS cm− 1, pH: 5.70).
absorptance in HAS of NSS with 10 mm thick 0.005 vol% ATO-WO3 NFs
layer at different flow rates at 10:00 a.m. The average transmittance
3.1. Environmental testing methods
(AT) can be calculated from the following equation [33]:
∫ λmax
T(λ)⋅I0 (λ)dλ The solar irradiance on the internal and external surfaces of the
(2)
λ
AT = min∫ λmax greenhouse roof and the plant canopy level was recorded on a spec­
I (λ)dλ
λmin 0
trometer (AvaSolar-3). Indoor air temperature and humidity were
measured by T-type thermocouples. Outdoor air temperature, humidity,
where I0(λ) is the radiant flux received of the NSS surface
and velocity were monitored with one T-type thermocouple and wind-
(W⋅m− 2⋅nm− 1); T(λ) is the transmittance of the NSS; the subscript min
sonic placed 1.2 m above the open ground outside the greenhouse
and max represent the minimum and maximum wavelength (− ). For the
near the test site. The surface temperatures on the internal and external
PAS, λmin, and λmax are 300 nm and 800 nm, respectively. For the HAS,
of the greenhouse walls, NSS, and leaves were measured by T-type
λmin, and λmax are 800 nm and 1500 nm, respectively.
thermocouples. The T-type thermocouples with water-repellent surfaces
The average absorptance (AA) value can be calculated from the
were used to measure the soil temperatures. The sensor locations are
following equation:
shown in Fig. 3c. The same sensor locations were adopted for the con­
∫ λmax
(1 − T(λ) − R(λ))⋅I0 (λ)dλ trol. A flow meter was used to measure the volumetric flow rate (Vflow),
(3)
λ
AA = min ∫ λmax two PT 1000 type thermocouples were used to measure the inlet (Tinlet)
I (λ)dλ
λmin 0
and outlet NFs temperature (Toutlet) of the NSS system, and the pressure
of the inlet and outlet were measured by pressure meters. Four data
where R(λ) is the reflectance of the NSS. Please refer to Ref. [9] for the
loggers (JZYG-NHY5, China, accuracy: ±0.1 ◦ C) with 10-s scan intervals
detailed calculation processes and measurement methods (− ).
were used for data collection, and the averages per minute were recor­
The previous research work proved that 0.005 vol% ATO-WO3 NFs
ded. The infrared pictures were taken by a thermal imager (Testo 872,
with 10 mm optical path-length and 100 L/h flow rate had a superior
Testo SE & Co. KGaA, Germany, accuracy: ±2 ◦ C).
photothermal performance. For further details, such as the test process
and results, please refer to Ref. [9].
3.2. Plant growth, leaf photosynthesis, and biochemical components
measurements
3. Experimental setup

Six plants from each treatment were randomly taken at 18:00 after
The experiment (Fig. 3) was conducted for 20 days from February
20 days of growth, and the fresh and dry weights of plant organs (leaves,
22nd to March 13th, 2022 in Beijing (40.10◦ N, 116.92◦ E), to study the
roots, and stems) were determined. The shoot/root ratio was calculated
photo-thermal performance of NSS system, the distribution of light and
by the dry weight ratio of the shoot to root. Leaf area was measured with
thermal environment in the greenhouse using NSS as a roof (NSS-G), and
a leaf area meter (LI-3100C, Li-Cor Inc., Lincoln, USA). The specific leaf
the response of plants growth inside NSS-G. Fig. 3 shows the test site and
area was calculated by dividing the leaf area by the leaf dry weight.
experiment device. Two small greenhouses were built. The span, length,
Leaf photosynthesis measurements were taken one day before the
and ridge height of the greenhouse were 1.8 m, 1.2 m, and 1.44 m,
destructive measurements, which were performed on the upper-most
respectively (seen Fig. 3c). The daylight roof of the greenhouse was
fully expanded leaves, between 9:30 and 15:30. The measurements
south-oriented with an angle of 28◦ and 1.8 m2 area (1.2 × 1.5 m). The
were performed with the LI-6400XT photosynthesis system (Li-Cor
NSS was made of a hollow highly transparent rigid polymethyl meth­
Biosciences, Lincoln, Nebraska, USA) with the leaf chamber fluorometer
acrylate (PMMA) sheet with a thickness of 6 mm (the transmittance of 6
(Li-Cor Part No. 6400–40, 2 cm2). The starting photosynthetic photon
mm PMMA used in this experiment is shown in Fig. 3b). The gap be­
flux density (PPFD) was 0 μmol m− 2⋅s− 1, followed by 1600 μmol
tween the two PMMA sheets is 10 mm, and 0.005 vol% ATO-WO3 NFs
m− 2⋅s− 1. At each PPFD, the measurements were taken when the
were circulated in the hollow layer with the flow rate of 400 L/h (ac­
photosynthetic rate (Pn) reached a steady state (after about 5–10 min).
cording to the conversion the 0.3 m cross-section width flow rate of 100
Chlorophyll fluorescence imaging was used to measure the
L/h in previous test [9]), as shown in Fig. 3c. A heat storage tank with a
maximum quantum efficiency of photosystem II (Fv/Fm) by using an

4
Y. Yuan et al. Energy 278 (2023) 127706

Fig. 4. Schematic diagram of energy analysis of NSS-G.

IMAGING-PAM Chlorophyll Fluorometer (Walz, Effeltrich, Germany) at 0.048%, respectively.


9:00 and 12:00 after 15 days of lettuce growth. The uppermost fully
expanded leaves were randomly selected and dark-adapted for at least 4. Energy balance analysis and performance estimation of
30 min before measurement of minimal (Fo) and maximal (Fm) fluo­ system
rescence. The current fluorescence yield (F) and maximum light-adapted
fluorescence yield (Fm’) were measured under actinic illumination. The To quantify the energy flow of solar radiation within PAS and HAS in
parameters were calculated as follows: Fv/Fm = (Fm − Fo)/Fm. the NSS greenhouse (NSS-G), the energy balance was developed. The
On one day before destructive measurements, the uppermost fully model considers solar and thermal radiation, convective and conductive
expanded leaves were collected in vials, flash-frozen in liquid nitrogen, heat transfer between different components of NSS-G (Fig. 4). Based on
and transferred to a freezer (− 80 ◦ C) for storage. These samples were the greenhouse climate modeling study of B⋅H.E. Vanthoor [35], the
used to measure chlorophyll contents. Leaf pigments were extracted following assumptions are made:
with 95% ethanol and the absorbance was measured at 665, 649, and
470 nm using a UV–Vis spectrophotometer (UV-1800, Shimadzu, (1) The air in the greenhouse is considered to be uniform, which
Japan). The chlorophyll concentrations were calculated by using equa­ means that there are no spatial differences in temperature and
tions derived by Wellburn (1994). Dry leaf samples from destructive vapor pressures. Therefore, all the model fluxes were described
measurements were collected and powdered by using a grinder to per m2 of greenhouse floor.
determine leaf total nitrogen with an elemental analyzer (Vario PYRO (2) The air in the greenhouse neither absorbs nor emits radiant
cube, Isoprime, UK). energy.
(3) There is no evaporation from the soil.
3.3. Uncertainty analysis (4) The greenhouse has no ventilation during the experimental
period.
In the presented experiment system, the ranges, accuracy, and error
percentage of the instruments are given in Fig. 3. According to the un­
certainty theory, the relative uncertainty (UR(y)) can be calculated as 4.1. Energy balance of the canopy
[34]:
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
n ( ) The energy balance function for the plant canopy (Can) in the
∑ ∂ln y 2 2
UR (y) = U (xi ) (4) greenhouse is given as:
i=1
∂xi
dTCan ( )( )( )
CapCan = EPAS,Sun τPAS,NSS 1 − ρPAS,Can 1 − e− KPAS,Can ⋅LAI 1 − αPAS,Wall
where dt ( )( )( )
+EHAS,Sun τHAS,NSS 1 − ρHAS,Can 1 − e− KHAS,Can ⋅LAI 1 − αHAS,Wall
y = y(x1 , x2 , ⋯, xn ) (5) 6
− 2aCanAir LAI(TCan − TAir ) − 2.45 × 10 VECCanAir (VPCan − VPAir )
− RCanCov,in − RCanFlr − RCanSky
Therefore, the maximum uncertainties of transmittance and photo- (6)
thermal conversion efficiency calculated from Eq. (4) are 0.013% and

5
Y. Yuan et al. Energy 278 (2023) 127706

where Cap is the heat capacity of the object (J⋅K− 1⋅m− 2); T is the tem­ e. the internal cover layer RFlrCov,in, the sky RFlrSky (W⋅m− 2), calculated
perature of the object (◦ C); t is the time (s); EPAS,Sun is the solar direct with Eq. (7); HFlrSo is the conductive heat transfer between floor and soil
radiation intensity within PAS wavelength (W⋅m− 2); EHAS,Sun is the solar (W•m− 1), pleases refer to Ref. [38]. The conductive heat transfer fluxes
direct radiation intensity within HAS wavelength (W⋅m− 2); τ, ρ and α are from ‘i’ to ‘j’, Hij is described by Ref. [37]:
the transmittance, reflectance and absorptance of the objects, respec­ ( ) χ ij
Hij = Ai Ti − Tj (11)
tively (− ); K represents the extinction coefficient (− ); LAI is the leaf area δij
index (m2⋅m− 2); aCanAir is the convective heat exchange coefficient from
the canopy to the air inside greenhouse (W•m− 2•K− 1); 2.45 × 106 is the where χ ij is the conductive heat coefficient from object ‘i’ to ‘j’
latent heat of evaporation (J⋅kg− 1 water); VECCanAir is the vapor ex­ (W⋅K− 1•m− 1), δij is the distance from object ‘i’ to ‘j’ (m).
change coefficient between the canopy and air (kg⋅Pa⋅s− 1), which can be
calculated referring to Refs. [35,36]; VPCan and VPAir are the saturated 4.4. Energy balance of NSS
vapor pressure at canopy temperature and air in greenhouse (Pa); the
subscripts Can, Wall, NSS, Air, Sun, Cov,in, Flr, and Sky represent can­ NSS is applied to the greenhouse roof and made of a hollow highly
opy, greenhouse wall of cover except NSS roof, NSS for the greenhouse transparent rigid PMMA sheet with a thickness of 6 mm, through which
roof, air inside the greenhouse, sunlight, the internal cover layer of the NFs are circulated. Therefore, the analysis of the energy balance of NSS
greenhouse including the internal layer of NSS, the floor inside the is necessary. It is assumed that conductive heat transfer is the dominant
greenhouse, and the sky, respectively; R is the radiative heat transfer mode of energy transport between NFs with the internal and external
between the canopy and surrounding elements i.e. the internal cover layer of NSS. The temperature of the NFs (TNSS_NF), the external layer of
layer RCanCov,in, the floor RCanFlr, the sky RCanSky (W⋅m− 2). NSS (TNSS_e), and the internal layer of NSS (TNSS_in) are described by:
The radiative heat transfer equation from the surface ‘i’ to ‘j’, Rij is
dTNSS NF
described by Ref. [37]: CapNSS = αPAS,NSS NF τPAS,NSS e EPAS,Sun + HNSS inNSS NF
(12)
NF
dt
( ( )4 ) +RCanSky + αHAS,NSS NF τHAS,NSS e EHAS,Sun − HNSS NFNSS e + RFlrSky
Rij = Ai εi εj Fij σ (Ti + 273.15)4 − Tj + 273.15 (7)
dTNSS e ( )
CapNSS = αPAS,NSS e 1 − ρPAS,NSS e EPAS,Sun + HNSS NFNSS
where Ai is the surface of object ‘i’ per square meter greenhouse soil e
dt ( )
e
(13)
(m2⋅m− 2); εi and εj are the thermal infrared emission coefficients for +αHAS,NSS e 1 − ρHAS,NSS e EHAS,Sun − HNSS eOut − RNSS eSky
object ‘i’ and ‘j’, respectively; Fij is the view factor from object ‘i’ to ‘j’; σ
is the Stefan Boltzmann constant (W⋅m− 2⋅K− 4); Ti and Tj are the tem­ dTNSS in ( )
CapNSS = αPAS,NSS in 1 − ρPAS,NSS in τPAS,NSS e τPAS,NSS NF EPAS,Sun
peratures of object ‘i’ and ‘j’ respectively (◦ C).
in
dt ( )
+αHAS,NSS in 1 − ρHAS,NSS in τHAS,NSS e τHAS,NSS NF EHAS,Sun + RFlrNSS in
4.2. Energy balance of air inside the greenhouse +cHECin (TAir − TNSS in )1.33
ANSS in
+ RCanNSS in − HNSS inNSS NF
AFlr
ANSS in
The equation for energy balance for indoor air temperature is given +2.45 × 10 × 6.4 × 10 cHECin (TAir − TNSS in )0.33
6 − 9
(VPAir − VPNSS in )
AFlr
as:
(14)
dTAir
CapAir
dt
= 2aCanAir LAI(TCan − TAir ) − HECAirFlr (TAir − TFlr ) where the subscripts NSS_in, NSS_e and NSS_NF represent the internal
(8)
ACov in and external layer of NSS, and NFs circled inside NSS, respectively; R is
− cHECin (TAir − TCov in )1.33 − HAirOut
AFlr the radiative heat transfer between the internal layer of NSS and sur­
rounding elements i.e. the canopy RCanNSS_in, the floor RFlrNSS_in, RNSS_eSky
where HECAirFlr is the convective heat coefficient from the air to the floor is the radiative heat transfer from the external layer of NSS to the
inside the greenhouse referred to Ref. [35] (W•m− 2•K− 1); cHECin is the ambient (W⋅m− 2), can be calculated by Eq. (7); The NFs can absorb the
heat exchanges coefficient between the air inside the greenhouse and the long-wave radiation from the canopy RCanSky and floor RFlrSky to the sky;
internal cover layer of the greenhouse (W⋅m− 2⋅K− 1); HAirOut is the heat HNSS_inNSS_NF and HNSS_NFNSS_e are the conductive heat transfer between
exchanges from the air inside the greenhouse to outside, including the internal layer and NFs of NSS, and between NFs and the external
leakage and ventilation. The experimental greenhouses were closed and layer of NSS, respectively, which can be calculated by Eq. (11); 6.4 ×
the energy flux lost or gained by the greenhouse only due to leakage: 10− 9 is the conversion factor relating the heat exchange coefficient
HAirOut = ρAir cp,Air fLeakage (TAir − TOut ) (9) (W⋅m− 2⋅K− 1) to the vapor exchange coefficient (kg•m− 2•s− 1•Pa− 1);
cHECin is the heat exchange coefficient between the air inside greenhouse
where ρAir is the density of air (kg⋅m− 3); cp,Air is the specific heat ca­ to the internal layer of NSS (W⋅m− 2⋅K− 1); VPNSS_in is the saturated vapor
pacity of air (J⋅K− 1⋅kg− 1); fLeakage is the ventilation rate of the green­ pressure at the internal layer temperature of NSS (Pa); HNSS_eOut is the
house influenced by the greenhouse leakage rate which depends on wind conductive heat fluxes from the external layer of NSS to the outside
speed [35]; TOut is the temperature outside greenhouse (◦ C). environment, can be calculated by Ref. [37]:
ANSS e
4.3. Energy balance of floor inside the greenhouse HNSS eOut = (cHECOut 1 + cHECOut 2 ⋅ VWind cHECOut 3 )(TNSS e − TOut ) (15)
AFlr

According to the energy balance in the top soil layer, the function of where cHECOut_1, cHECOut_2, and cHECOut_3 are the convective heat ex­
the floor inside the greenhouse can be calculated by: change variables between the cover and outdoor air, which depend on
dTFlr the greenhouse shape (J⋅m− 3⋅K− 1); VWind is outside wind speed (m⋅s− 1);
CapFlr = αHAS,Flr τHAS,NSS τHAS,Can EHAS,Sun TNSS_e is the temperature of the external layer of NSS (◦ C).
( ) dt ( ) ( ) Note that this paper gives a brief introduction to the model used. For
+αPAS,Flr 1 − ρPAS,NSS τPAS,NSS 1 − ρPAS,Can e− KPAS,Can ⋅LAI 1 − αPAS,Wall EPAS,Sun
+HECAirFlr (TAir − TFlr ) − HFlrSo + RCanFlr − RFlrCov,in − RFlrSky more details, such as the model parameters and the physical meaning of
(10) the variables, please refer to Ref. [35].

where the subscript Flr represents the floor inside the greenhouse; R is
the radiative heat transfer between the floor and surrounding elements i.

6
Y. Yuan et al. Energy 278 (2023) 127706

Fig. 5. Flow chart of calculation.

4.5. Performance estimation


ELoss = ρPAS,NSS EPAS,Sun + ρHAS,NSS EHAS,Sun
( )( )( )( )( )
A nonlinear algebraic equation (Eq. (16)) describes the energy bal­ + 1 − ρPAS,NSS 1 − αPAS,NSS 1 − αPAS,Wall 1 − αPAS,Can 1 − αPAS,Flr EPAS,Sun
( )( )( )( )( )
ance of NSS-G, where the terms involved in the equation were changing + 1 − ρHAS,NSS 1 − αHAS,NSS 1 − αHAS,Wall 1 − αHAS,Can 1 − αHAS,Flr EHAS,Sun
concerning time (t) to satisfy the law of energy conservation [39]: (18)

ESun (t) = QPlant (t) + QAir (t) + QAbs,Flr (t) + QUse (t) + QLoss (t) + ELoss (t) (16) The flow of solar energy in NSS-G mainly includes the optical and
thermal flow in various parts of the greenhouse. The optical efficiency
where ESun(t) is the sunlight incident on the NSS surface, which consists (ηOpt) can be calculated as:
of PAS (EPAS,Sun) and HAS (EHAS,Sun) solar energy (W⋅m− 2); QPlant(t) is Eobject
the energy absorption for plant canopy calculated by Eq. (6) (W⋅m− 2); ηOpt = (19)
ESun
QAir(t) is the thermal energy exchange rate by the air inside greenhouse
calculated by Eq. (8) (W⋅m− 2); QAbs,Flr(t) is the thermal energy exchange where, Eobject includes EPAS, SunNSS, EHAS, SunNSS, EPAS, SunCan, EHAS, SunCan,
rate by the floor inside greenhouse calculated by Eq. (10) (W⋅m− 2); EPAS, SunFlr, EHAS, SunFlr, EPAS, SunLoss and EHAS, SunLoss.
QUse(t) is the heat production of NSS system calculated by Eq. (12) The thermal efficiency (ηth) can be calculated as:
(W⋅m− 2); QLoss(t) is the heat lost to the outside calculated by Eq. (17)
(W⋅m− 2); ELoss(t) is the sunlight lost to the outside calculated by Eq. (18) ηth =
Qobject
(20)
(W⋅m− 2). ESun
The heat loss from greenhouse components to the ambient (QLoss) is
where Qobject includes QUse, QAbs,Flr, and QLoss.
given as:
The root-mean square error (RMSE) is used to assess the discrepancy
QLoss = HAirOut + HCov eOut + RCov eSky (17) between the experimental and simulation results quantitatively. RMSE
can be expressed by Ref. [40]:
where HCov_eOut is the conductive heat fluxes from the external cover √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
∑n
layer of the greenhouse to the outside environment calculated by Eq. i=1 (measurement − prediction)
2
(21)
(15) (W⋅m− 2); RCov_eSky is the radiative heat transfer from the external
RMSE =
n
cover layer of the greenhouse to the sky calculated by Eq. (7) (W⋅m− 2).
The sunlight reflected from the greenhouse to outside (ELoss), where n denotes the total number of observations.
including the sunlight reflected on the NSS surface and various parts
inside the greenhouse, can be calculated as [35]: 4.6. Method for solving model

The thermal performance of the model was time-dependent, it can be


simulated by a dynamic simulation model. The solution method for

7
Y. Yuan et al. Energy 278 (2023) 127706

Fig. 6. (A) Outdoor weather parameters on February 27th, 2022. (b) Numerical and experimental air temperature inside greenhouse. NSS-G and Air-G represent
greenhouse using NSS and hollow as a roof, respectively.

HAS radiation of the solar energy. The average temperature of the air
inside NSS-G was 6.3 ◦ C lower than that inside Air-G. The cooling effect
is more significant in NSS-G when the solar flux intensity is higher.
Meanwhile, the temperature of the air inside NSS-G and Air-G was
higher than that outside because solar radiation entered the greenhouse,
and the indoor temperature raised in a closed state.

5.2. Photothermal conversion of NSS system

The photothermal conversion efficiency of the NSS system (η′ th)


obtained by the test can be calculated according to Eq. (22). After 6 h of
operation, the NFs temperature of the NSS system increased from 12 ◦ C
to 29.6 ◦ C, and the average η′ th of NSS system was 19.9% (Fig. 7). The
RMSE of the temperature of NFs and photothermal conversion efficiency
of NSS between the experimental and simulation results are 1.43 ◦ C and
1.39%. It indicates that the mathematical model is reliable.
In Fig. 7, the NFs temperature of the NSS system increased from
15.1 ◦ C to 27.8 ◦ C from 10:00 a.m. to 1:00 p.m. And the average η′ th of
the NSS system was 26.3%. However, the NFs temperature of the NSS
Fig. 7. Photothermal conversion efficiency and NFs temperature of system decreased 0.2 ◦ C from 2:40 p.m. to 3:00 p.m., and the average η′ th
NSS system. of the NSS system was − 4.8%. The larger difference between the tem­
perature of NFs and the outdoor air temperature is, the more heat the
solving the non-linear differential equations is solved in MATLAB, as NSS system losses to the outside environment, and the photothermal
shown in Fig. 5. The information on the design, and physical parameters conversion efficiency of the NSS system will also be reduced. Therefore,
of NSS-G and actual weather data were used as input data. The initial the operation time of the NSS system is determined to be from 10:00 a.
conditions were set to be consistent with the experimental condition. m. to 1:00 p.m. in the cold season according to the overall efficiency and
The overall objective of the model was established to analyze the energy economy of the system. Through the previous experimental results [9],
flow of the greenhouse using NFs as the roof. The time step used for the the NSS system was suitable to operate from 9:00 a.m. to 2:00 p.m. in the
mathematical model is △t = 300 s (5 min). When the solution is pro­ hot season.
duced, the computed values are taken as initial values for the next time
cP,NF ρNF Vflow (Toutlet − Tinlet )
step. (22)

ηth =
ESun ANSS
5. Results and discussion where ANSS is the solar receiving area of the NSS (m2); Vflow is the vol­
ume flow rate (L⋅h− 1); Tinlet and Toutlet are the inlet and outlet temper­
5.1. Experimental data and model validation ature of NFs (◦ C), respectively; cP,NF is the specific heat capacity of NFs
(J⋅kg− 1⋅K− 1); ρNF is the density of NFs (kg⋅m− 3).
Fig. 6a shows the solar radiation flux intensity on the NSS’s surface,
the outdoor temperature, and wind speed measured at 5 min intervals.
Both the predicted and measured temperature for the air inside NSS-G 5.3. Light distribution in greenhouse
are shown in Fig. 6b. The RMSE value of the temperature of the air in­
side NSS-G between the experimental and simulation results is 1.32 ◦ C. Sunlight is transmitted into the greenhouse through the covering and
The agreement is good with an absolute error of 4.7% between the affects the growth of plants. Fig. 8 shows the light distribution in the
measured and predicted values. It should be pointed out that the pre­ greenhouse under different treatments. Fig. 8a represents the flux den­
diction error is relatively large when the temperature of the air outside sity within PAS at different locations in the greenhouse. The average flux
the greenhouse is low. It was similar to the results of Ref. [41]. density in PAS on the canopy in NSS-G is 235.2 W m− 2, 18.2% lower
Fig. 6b shows the variation in the temperature of the air inside NSS-G than that in Air-G with a value of 278 W m− 2. The difference between
and Air-G. It indicates that the temperature of the air inside NSS-G was the flux density within PAS on the canopy in NSS-G and Air-G from
significantly lower than that inside Air-G because the NFs absorbed the 11:00 a.m. to 1:00 p.m. is relatively lower than at other moments. This is
because the optical path of sunlight at noon is smaller, which is

8
Y. Yuan et al. Energy 278 (2023) 127706

Fig. 8. Distribution of solar flux in PAS (a) and its Red/Blue ratio (b) on February 27th, 2022. Insertion: Spectrometer layout. Spectral distribution at different
locations (c) and in different directions on the canopy in NSS-G 0.5 m above the ground (d) at 12:00 p.m. (For interpretation of the references to colour in this figure
legend, the reader is referred to the Web version of this article.)

consistent with the results in Ref. [9]. 5.4. Growth analysis of lettuce
Fig. 8b shows the Red/Blue ratio at different locations in the
greenhouse. It can be seen that the Red/Blue ratio on the surface of the Fig. 9 shows no significant difference in the lettuce plants’ growth
greenhouse roof is about 1.7, which is consistent with the experimental inside NSS-G and Air-G. Under 20 consecutive days of operation, the
results in Ref. [42]. It further confirmed the accuracy and reliability of average growth of lettuces cultivated in NSS-G was 19.8 cm. NSS-G can
the instruments and test methods adopted in this experiment. In Fig. 8b, promote the production of indoor cultivated lettuce (Fig. 9a), and the
the Red/Blue ratio on the canopy in NSS-G is 0.15 lower than that in shoot fresh weight and dry weight were increased by 7.2% and 4.3% in
Air-G. It indicates that the red wavelength energy of sunlight was comparison with that in Air-G (Fig. 9b), respectively. Consistent with the
absorbed by NFs proved by Figs. 2c and 8c. The red light can enhance the increased biomass production, the plant leaf area was also increased in
photosynthesis of plants, promote the chlorophyll synthesis of leaves, NSS-G. The specific leaf area of lettuce growing in NSS-G was 9.2%
and increase the sugar content of plants [43,44]. higher than that in Air-G. The shoot/root ratio in NSS-G was 12.5%
Fig. 8c shows the spectral distribution at different locations in the higher than that in Air-G (Fig. 9b).
greenhouse at 12:00 p.m. When the solar light passed through NSS, Fig. 9c shows that the total chlorophyll, carotenoid, and soluble
78.3% of PAS light is transmitted into the greenhouse, while 85.4% of sugar content of leaves in NSS-G were lower than that in Air-G. Leaves in
HAS radiation is absorbed by NSS. As can be seen from Fig. 8c, the NSS-G showed a 6.5% decrease in Pn when absorbed PPFD was above
spectral distribution in PAS on the canopy in NSS-G is similar to that in 300 μmol m− 2⋅s− 1 (Fig. 9c). The light absorption of leaf showed the same
Air-G, especially in 300–500 nm wavelength. However, sunlight in the trend under the two treatments (Fig. 9d). It can be seen that the light
500–800 nm wavelength on the canopy in NSS-G is reduced by 16.4% absorption of leaf in NSS-G within 515–605 nm and 700–800 nm was
compared to that in Air-G, which is absorbed by NFs of NSS. And the relatively higher than that in Air-G, while the opposite results reversed
sunlight after 1400 nm wavelength in NSS-G almost disappeared and is in other wavelengths.
absorbed by NSS. The specific leaf area is the ratio of the leaf area to dry weight, which
Fig. 8d represents the non-uniform spectral distribution on the can­ means that the weaker the light is, the larger the specific leaf area is.
opy in different directions of NSS-G at 12:00 p.m. The shading on the Therefore, the specific leaf area in NSS-G is larger (Fig. 9b), indicating
south side and reflective film are the main reasons for the uneven dis­ that the light in NSS-G is weaker than that in Air-G (Fig. 8a). Meanwhile,
tribution of solar flux in the north-south direction inside the greenhouse. the lower chlorophyll content of the leaf is resulting in the reduction of
However, the distribution of the flux intensity is relatively uniform in the photosynthetic rate of plants (Fig. 9c). The Pn of lettuce growing in
the east-west direction. NSS-G was slightly decreased compared with Air-G, but the plant fresh
and dry weight did not reduce. That is, the use of NSS as the greenhouse
roof does not affect crop yield.
Fig. 10a and b shows the temperature distribution of lettuce leaves. It

9
Y. Yuan et al. Energy 278 (2023) 127706

Fig. 9. (A) Growth of lettuce on the 0th, and 20th day. Note: Day 0 and Day 20 represent the day transplanted the seedings into pots and lettuce harvest day,
respectively. (b) Growth indexes of lettuce growing in NSS-G and Air-G. (c) Response of steady-state net leaf photosynthesis (Pn) at different treatments. Insertion:
Biochemical component content of lettuce growing in NSS-G and Air-G. (d) Light absorption of the lettuce leaf growing in NSS-G and Air-G. Insertion: Average light
absorption of leaf within 300–800 nm. Means followed by different letters within one row differ significantly (P < 0.05) as established by the least significant
difference test.

can be seen that the average temperature of the lettuce leaf growing in production accounts for 25.2%, and the soil absorbed accounts for 4.3%.
NSS-G was 8.7 ◦ C lower than that in Air-G (Fig. 10a). The difference in 25.8% of solar energy is absorbed by plants, which contained 25.5% PAS
the leaf temperature between NSS-G with Air-G was higher when the and 0.3% HAS. Most of the rest of the solar energy entering into the
solar flux intensity increased. The thermal images of the temperature greenhouse was absorbed by the ground and converted into heat.
distribution on leaves in NSS-G were relatively uniform and maintained As can be seen from Fig. 11, the energy loss in NSS-G accounts for
at 25 ◦ C (Fig. 10b), which is the suitable temperature for plant growth. 44.7%, which included 5.6% for reflective optical loss and 39.1% for
The plant leaves can absorb solar radiation both in PAS and HAS. heat loss to the environment. The heat loss accounts for 87.5% of the
However, PAS, related to plant photosynthesis, 78.3% were transmitted total energy loss, due to the low outdoor temperature and high
into the plants. HAS caused a rise in leaf temperature, 85.4% was conductive heat coefficient of the covering material. However, in the
absorbed by NSS. Meanwhile, it can improve the water utilization effi­ previous experiments [9], the average solar energy utilization efficiency
ciency due to the transpiration rate decreased when the leaf temperature of NSS was up to 73.6% in summer. Therefore, the next step in
is decreased, and contribute to developing water-saving irrigation for improving the solar energy utilization efficiency of NSS-G is to provide
facility agriculture. high transparent covering insulation measures or materials, such as
When leaf temperature was raised, the maximum photosystem II aerogels [46].
efficiency (Fv/Fm) was reduced from 9:00 a.m. to 12:00 p.m., which The spectral energy of PAS and HAS accounts for 67.9% and 32.1%
indicated that the plants were subjected to inhibition. This reduction of outdoor solar spectra at 12:00 p.m., respectively. It can be seen from
was more pronounced in Air-G than that in NSS-G (Fig. 10c and d). Fv/ Fig. 11 that NSS-G can absorb 77.9% of the energy in HAS to generate
Fm of the plant in NSS-G was 0.021 lower than that in Air-G at 9:00 a.m. heat and transmit 76.6% of sunlight in PAS to be absorbed by the plant.
However, the Fv/Fm of the plant in NSS-G was only 0.001 lower than that 52% of PAS light was transmitted into the greenhouse, while only 13.7%
in Air-G at 12:00 p.m. (Fig. 10d). This indicated that lettuces growing in of PAS energy was absorbed by NSS, and 1.1% of PAS light was reflected
NSS-G were less inhibited compared with Air-G due to the lower leaf into the ambient (Fig. 11). However, 24.8% of HAS energy was absorbed
temperature. by NSS, while only 6.9% of HAS light transmitted into the greenhouse
and 0.2% of HAS light was reflected into the ambient.

5.5. Solar energy flow of NSS-G


5.6. Spectral splitting performance of NSS
According to the dynamic thermal model of the greenhouse in Sec­
tion 4, the solar energy flow in the NSS greenhouse at 12:00 p.m. was The spectral splitting performance of NSS is the focus of this study.
calculated, as shown in Fig. 11. The solar energy utilization efficiency in Solar spectra were split into PAS and HAS by NSS: PAS light was used to
NSS-G was 55.3%, in which the plants account for 25.8%, the heat participate in plant photosynthesis to synthesize carbohydrates, while

10
Y. Yuan et al. Energy 278 (2023) 127706

Fig. 10. (A) Leaf temperature under Air-G and NSS-G changes with time. (b) Thermal image of lettuce surface under NSS-G and Air-G at 1:00 p.m. (c) Image of
maximum quantum efficiency of photosystem II photochemistry (Fv/Fm) distribution and its value. (d) Average Fv/Fm under Air-G and NSS-G. Fv/Fm value of 0.8
indicates whether the plant is under duress. When Fv/Fm is higher than 0.8, the plant is not inhibited. On the contrary, it means that the plant is inhibited [45]. Error
bars show ± SE (n = 4). * and ** indicate significant differences at P < 0.05 and P < 0.01, respectively.

Fig. 11. Solar energy flow of NSS-G at 12:00 p.m. Subscript Trf represents the heat transfer of objects.

11
Y. Yuan et al. Energy 278 (2023) 127706

observed from Fig. 12 that the SOS of this work is 2.5 times higher than
that of the previous studies. Meanwhile, NSS used 0.005 vol% ATO-­
WO3/H2O NFs, the payback period and levelized cost of energy are only
0.76 years and 0.041 $/kWh within a 10-year life cycle [9].

6. Conclusion

This paper describes a nanofluid spectral splitting (NSS) covering


applied to the greenhouse roof to achieve the co-production of crops and
energy harvesting. Through experiments and analysis, the normal plant
growth and superior photothermal conversion performance of the NSS
greenhouse (NSS-G) were proved. The main conclusion of this study can
be summarized as follows:

(1) NSS can prevent the environment in the greenhouse from high
temperatures when the outdoor radiation is high. Compared with
the conventional hollow covering greenhouse (Air-G), NSS-G can
reduce indoor air and plant leaves temperature by 6.3 ◦ C and
8.7 ◦ C, respectively. The higher the solar radiance, the more HAS
energy was absorbed by the NSS system, and the more significant
the cooling capacity of NSS-G was.
(2) NSS reduces the sunlight entering the greenhouse. The average
Fig. 12. Compared proportion of PAS and HAS of transmitted sunlight and SOS flux density of PAS on the canopy in NSS-G was 18.2% lower than
of this study with Ref. [27,39,47]. Sunlight reached the surface of the green­ that of Air-G. The results of plant growth experiments showed
house roof (Solar flux on the roof) on the experimental day at 12:00 p.m. that compared with Air-G, the photosynthetic rate in NSS-G was
reduced by 6.5%, but plant dry weight was increased by 4.3%,
HAS radiation was absorbed by NSS to generate heat. To quantify the and there was no significant difference in biochemical component
spectral splitting ability of NSS, the spectral splitting coefficient (SSO) contents.
was defined as: (3) The total solar energy utilization efficiency in NSS-G was 55.3%,
∫ 800 including 25.8% for the plant, and 25.2% for heat production.
I (λ)τPAS (λ)dλ
300 PAS
SSO = ∫ 1500 (23) 39.1% of solar energy was lost to the outdoor environment due to
800
IHAS (λ)τHAS (λ)dλ the low ambient temperature. The total utilization efficiency will
be improved when the outdoor ambient temperature increases.
SSO can be understood as the ratio of PAS to HAS sunlight of the
(4) NSS has a strong spectral splitting ability. The spectral splitting
transmitted solar spectra through subjects. Fig. 12 represents the com­
coefficient of NSS is 2.5 times higher than that of the previous
parison of SOS between this work and published studies [27,39,47]. The
studies. NSS-G can absorb 77.9% of the solar energy within HAS
spectral energy of PAS and HAS accounts for 67.9% and 32.1% in sun­
to generate heat while transmitting 76.6% of sunlight within PAS
light on the surface of the greenhouse roof on an experimental day at
to plants. (see Table 1).
12:00 p.m., respectively. According to Eq. (23), the SSO of the sunlight
passed through the atmosphere was 2.1. However, when sunlight
Credit author statement
transmitted NSS, SSO was up to 9.6. PAS and HAS account for 90.6% and
9.4% of the transmitted sunlight through NSS.
Yu Yuan: Conceptualization, Methodology, Writing-original draft;
Chen et al. [47] and Shen et al. [27] reported that 0.01 vol% ATO
Yaning Ji: Visualization; Wei Wang: Investigation; Dawei Shi: Data
NFs and 0.025 vol% ATO plus 0.0001 vol% graphite NFs were used as
curation; Long Hai: Formal analysis; Qianlei Ma: Software; Qichang
spectrum filters to allow only visible light to enter buildings and absorb
Yang: Supervision, Visualization; Yuming Xie: Formal analysis; Bin Li:
other energy. Ma et al. [39] used spectral splitting film to transmit plant
Data curation; Gang Wu: Writing-review and editing, Validation;
active radiation while reflecting near-infrared radiation. It can be

Table 1
Constants used in solving thermal model.
Parameter Value Unite Parameter Value Unite Parameter Value Unite

CapCan 1200 J⋅K− 1⋅m− 2 εWall 0.091 – αPAS,PMMA 0.005 –


LAI 2.5 m2⋅m− 2 εCan 1 – τHAS,PMMA 0.865 –
KPAS,Can 0.7 – FCanNSS_in 0.882 – αHAS,PMMA 0.077 –
KHAS,Can 0.27 – FCanWall_in 0.241 – hPMMA 0.006 M
ρPAS,Can 0.07 – FNSS_eSky 1 – cp,NF 4200 J⋅K− 1⋅kg− 1
ρHAS,Can 0.35 – FFlrNSS_in 0.882 – ρNF 1000 kg⋅m− 3
αPAS,Can 0.767 – FWall_eSky 1 – hNF 0.01 M
8
αHAS,Can 0.087 – σ 5.670 × 10− W⋅m− 2⋅K− 4 ρFlr 1500 kg⋅m− 3
αPAS,Wall 0.079 – AFlr 2.16 m2 cp,Flr 1390 J⋅K− 1⋅kg− 1
αHAS,Wall 0.099 – AWall 6.9 m2 χFlr 1.12 W⋅m− 1⋅K− 1
cp,Air 1000 J⋅K− 1⋅kg− 1 ANSS 1.8 m2 cHECin 1.86 W⋅m− 2⋅K− 1
ρAir 1.2 kg⋅m− 3 cp,PMMA 1464 J⋅K− 1⋅kg− 1 cHECOut_1 2.8 W⋅m− 2⋅K− 1
aCanAir 5 W⋅m− 2⋅K− 1 ρPMMA 1200 kg⋅m− 3 cHECOut_2 1.2 J⋅m− 3⋅K− 1
εFlr 1 – χPMMA 0.2 W⋅m− 1⋅K− 1 cHECOut_3 1 J⋅m− 3⋅K− 1
εNSS 0.9 – τPAS,PMMA 0.934 –

Note: The optical properties varied with the incident angle on the subject’s surface. The values related to the optical properties τ, a and ρ in the table are taken when the
incident optical angle is 0.

12
Y. Yuan et al. Energy 278 (2023) 127706

Lingling Ma: Project administration, Writing-review and editing. Acknowledgment

Declaration of competing interest This work was supported by the Beijing Municipal Science and
Technology Project (Z211100004621002). The authors would also like
The authors declare that they have no known competing financial to thank the National Natural Science Foundation of China (51806244),
interests or personal relationships that could have appeared to influence Guangdong Basic and Applied Basic Research Foundation
the work reported in this paper. (2020A1515110677), the Project of Collaborative Innovation Center of
GDAAS-XTXM202201(XT202203), the key project of Youth Natural
Data availability Science Foundation (202202), the National Key Research and Devel­
opment Program, Ministry of Science and Technology of China
No data was used for the research described in the article. (2022YFB3604604).

Nomenclature

Symbols
A area [m2]
a convective heat exchange coefficient [− ]
AA average absorptance [− ]
AT average transmittance [− ]
cHEC heat exchanges coefficient [W⋅m− 2⋅K− 1]
Cap heat capacity of the object [J⋅K− 1⋅m− 2]
Cp specific heat capacity [ J⋅kg− 1⋅K− 1]
E solar direct radiation intensity [W⋅m− 2]
F view factor [− ]
fLeakage ventilation rate [− ]
H heat exchange [W⋅m− 2]
HEC convective heat coefficient [W•m− 2•K− 1]
I(λ) radiant flux [W⋅m− 2⋅nm− 2]
K extinction coefficient [− ]
LAI leaf area index [m2⋅m− 2]
Q energy [W⋅m− 2]
R thermal radiation [W⋅m− 2]
R(λ) reflectance [− ]
RMSE root-mean square error [− ]
SSO spectral splitting coefficient [− ]
T temperature [K]
Vflow volumetric flow rate [L⋅h]
VWind outside wind speed [m⋅s− 1]
VEC vapor exchange coefficient [kg⋅Pa⋅s− 1]
VP saturated vapor pressure [Pa]

Scripts
e external
in internal
inlet fluid inlet
Opt optical
outlet fluid outlet
sun sun
th thermal
total total
Trf heat transfer
Use useful energy
So soil
Wall Greenhouse wall

Greek symbols
α absorptance [− ]
χ conductive heat coefficient [ W⋅K− 1•m− 1]
ρ reflectance/density [kg•m− 3]
η efficiency [%]
λ wavelength [nm]
τ transmittance [− ]
σ Stefan Boltzmann constant [W⋅m− 2⋅K− 4]
δ distance [m]

13
Y. Yuan et al. Energy 278 (2023) 127706

Abbreviations
Abs Absorption
Air Air in greenhouse
BF Base fluid
Cov Cover
HAS Heat active spectrum, including 800–1500 nm wavelengths
Can Canopy
Flr Floor in greenhouse
Loss Heat loss
NF Nanofluid
NP Nanoparticle
Opt Optical
Out Outdoor air
PAR Photosynthetically active ration
NSS Nanofluid-based spectral splitter
PAS Plant active spectrum, including 300–800 nm wavelengths
PMMA Polymethyl methacrylate
PT Photothermal
PV Photovoltaic
SBS Spectral beam splitting

References [19] Emmott CJM, Röhr JA, Campoy-Quiles M, Kirchartz T, Urbina A, Ekins-Daukes NJ,
et al. Organic photovoltaic greenhouses: a unique application for semi-transparent
PV? Energy Environ Sci 2015;8(4):1317–28.
[1] Hollingsworth JA, Ravishankar E, O’Connor B, Johnson JX, DeCarolis JF.
[20] Shi H, Xia R, Zhang G, Yip H-L, Cao Y. Spectral engineering of semitransparent
Environmental and economic impacts of solar-powered integrated greenhouses.
polymer solar cells for greenhouse applications. Adv Energy Mater 2019;9(5).
J Ind Ecol 2019;24(1):234–47.
[21] Lamanna E, Matteocci F, Calabrò E, Serenelli L, Salza E, Martini L, et al.
[2] Sajid MU, Bicer Y. Comparative life cycle cost analysis of various solar energy-
Mechanically stacked, two-terminal graphene-based perovskite/silicon tandem
based integrated systems for self-sufficient greenhouses. Sustain Prod Consum
solar cell with efficiency over 26. Joule 2020;4(4):865–81.
2021;27:141–56.
[22] Liu W, Liu L, Guan C, Zhang F, Li M, Lv H, et al. A novel agricultural photovoltaic
[3] Gruda N, Tanny J. Protected crops recent advances, innovative technologies and
system based on solar spectrum separation. Sol Energy 2018;162:84–94.
future challenges. Conference Protected crops recent advances, innovative
[23] Crisostomo F, Hjerrild N, Mesgari S, Li Q, Taylor RA. A hybrid PV/T collector using
technologies and future challenges. International Society for Horticultural Science
spectrally selective absorbing nanofluids. Appl Energy 2017;193:1–14.
(ISHS), Leuven, Belgium, p. 271-278.
[24] An W, Chen L, Liu T, Qin Y. Enhanced solar distillation by nanofluid-based spectral
[4] Chang J, Wu X, Wang Y, Meyerson LA, Gu B, Min Y, et al. Does growing vegetables
splitting PV/T technique: preliminary experiment. Sol Energy 2018;176:146–56.
in plastic greenhouses enhance regional ecosystem services beyond the food
[25] Yuan Y, Wu G, Yang Q, Cheng R, Tong Y, Zhang Y, et al. Experimental and
supply? Front Ecol Environ 2013;11(1):43–9.
analytical optical-thermal performance of evacuated cylindrical tube receiver for
[5] Katzin D, Marcelis LFM, van Mourik S. Energy savings in greenhouses by transition
solar dish collector. Energy 2021:234.
from high-pressure sodium to LED lighting. Appl Energy 2021:281.
[26] Wu G, Yang Q, Zhang Y, Fang H, Feng C, Zheng H. Energy and optical analysis of
[6] Zhang H, Benoit H, Gauthier D, Degrève J, Baeyens J, López IP, et al. Particle
photovoltaic thermal integrated with rotary linear curved Fresnel lens inside a
circulation loops in solar energy capture and storage: gas–solid flow and heat
Chinese solar greenhouse. Energy 2020:197.
transfer considerations. Appl Energy 2016;161:206–24.
[27] Shen C, Lv G, Wei S, Zhang C, Ruan C. Investigating the performance of a novel
[7] An W, Zhang Y, Pang B, Wu J. Synergistic design of an integrated pv/distillation
solar lighting/heating system using spectrum-sensitive nanofluids. Appl Energy
solar system based on nanofluid spectral splitting technique. AIMS Energy 2021;9
2020:270.
(3):534–57.
[28] Sajid MU, Bicer Y. Performance assessment of spectrum selective nanofluid-based
[8] Hu X, Li Y, Tian J, Yang H, Cui H. Highly efficient full solar spectrum (UV-vis-NIR)
cooling for a self-sustaining greenhouse. Energy Technol 2020;9(1).
photocatalytic performance of Ag2S quantum dot/TiO2 nanobelt heterostructures.
[29] Said Z, Hachicha AA, Aberoumand S, Yousef BAA, Sayed ET, Bellos E. Recent
J Ind Eng Chem 2017;45:189–96.
advances on nanofluids for low to medium temperature solar collectors: energy,
[9] Yuan Y, Fang H, Wu G, Yang Q, Ma Q, Ji Y, et al. Experimental investigation of full
exergy, economic analysis and environmental impact. Prog Energy Combust Sci
solar spectrum utilization based on nanofluid spectral splitter for greenhouse
2021;84.
applications. Energy Convers Manag 2022:254.
[30] D. Feuermann RK, Zeroni M, Levi S, Gale J. Evaluation of a liquid radiation filter
[10] McCree KJ. The action spectrum, absorptance and quantum yield of photosynthesis
greenhouse in a desert environment. Transactions of the ASAE 1998;41(6):1781–8.
in crop plants. Agric Meteorol 1971;9:191–216.
[31] Han X, Guo Y, Wang Q, Phelan P. Optical characterization and durability of
[11] Chen M, Blankenship RE. Expanding the solar spectrum used by photosynthesis.
immersion cooling liquids for high concentration III-V photovoltaic systems. Sol
Trends Plant Sci 2011;16(8):427–31.
Energy Mater Sol Cell 2018;174:124–31.
[12] Giliberto L, Perrotta G, Pallara P, Weller JL, Fraser PD, Bramley PM, et al.
[32] La Notte L, Giordano L, Calabrò E, Bedini R, Colla G, Puglisi G, et al. Hybrid and
Manipulation of the blue light photoreceptor cryptochrome 2 in tomato affects
organic photovoltaics for greenhouse applications. Appl Energy 2020:278.
vegetative development, flowering time, and fruit antioxidant content. Plant
[33] Liu Y, Cheng P, Li T, Wang R, Li Y, Chang SY, et al. Unraveling sunlight by
Physiol (Sofia) 2005;137(1):199–208.
transparent organic semiconductors toward photovoltaic and photosynthesis. ACS
[13] Smith H. Phytochromes and light signal perception by plants - an emerging
Nano 2019;13(2):1071–7.
synthesis. Nature 2000;407(6804):585–91.
[34] Zhu ZY, Zheng HF, Wang QS, Chen MX, Li ZL, Zhang B. The study of a novel light
[14] Ravishankar E, Charles M, Xiong Y, Henry R, Swift J, Rech J, et al. Balancing crop
concentration and direct heating solar distillation device embedded underground.
production and energy harvesting in organic solar-powered greenhouses. Cell
Desalination 2018;447:102–19.
Reports Physical Science 2021;2(3).
[35] Vanthoor BHE, Stanghellini C, van Henten EJ, de Visser PHB. A methodology for
[15] Fernández EF, Villar-Fernández A, Montes-Romero J, Ruiz-Torres L, Rodrigo PM,
model-based greenhouse design: Part 1, a greenhouse climate model for a broad
Manzaneda AJ, et al. Global energy assessment of the potential of photovoltaics for
range of designs and climates. Biosyst Eng 2011;110(4):363–77.
greenhouse farming. Appl Energy 2022;309.
[36] Stanghellini C. Transpiration of greenhouse crops: an aid to climate management
[16] Sajid MU, Bicer Y. Experimental investigation of spectrum selective nanofluids for
[PhD thesis]. Wageningen: IMAG; 1987.
thermal management of greenhouses in hot arid climates. Appl Therm Eng 2023:
[37] De Zwart HF. Analyzing energy-saving options in greenhouse cultivation using a
223.
simulation model [PhD thesis]. Wageningen: Wageningen University; 1996.
[17] Mojiri A, Taylor R, Thomsen E, Rosengarten G. Spectral beam splitting for efficient
[38] Zhang G, Ding X, Li T, Pu W, Lou W, Hou J. Dynamic energy balance model of a
conversion of solar energy—a review. Renew Sustain Energy Rev 2013;28:654–63.
glass greenhouse: an experimental validation and solar energy analysis. Energy
[18] Mojiri A, Stanley C, Rodriguez-Sanchez D, Everett V, Blakers A, Rosengarten G.
2020:198.
A spectral-splitting PV–thermal volumetric solar receiver. Appl Energy 2016;169:
[39] Ma Q, Zhang Y, Wu G, Yang Q, Yuan Y, Cheng R, et al. Photovoltaic/spectrum
63–71.
performance analysis of a multifunctional solid spectral splitting covering for
passive solar greenhouse roof. Energy Convers Manag 2022:251.

14
Y. Yuan et al. Energy 278 (2023) 127706

[40] Hu M, Zhao B, Ao X, Ren X, Cao J, Wang Q, et al. Performance assessment of a [44] Aksenova NP, Konstantinova TN, Golyanovskaya SA, Schmulling T, Kossmann J,
trifunctional system integrating solar PV, solar thermal, and radiative sky cooling. Willmitzer L, et al. In vitro growth and tuber formation by transgenic potato plants
Appl Energy 2020:260. harboring rolC or rolB genes under control of the patatin promoter. Russ J Plant
[41] Su Y, Xu L, Goodman ED. Nearly dynamic programming NN-approximation–based Physl+ 1999;46(4):513–9.
optimal control for greenhouse climate: a simulation study. Optim Control Appl [45] Li T, Heuvelink E, Dueck TA, Janse J, Gort G, Marcelis LF. Enhancement of crop
Methods 2017;39(2):638–62. photosynthesis by diffuse light: quantifying the contributing factors. Ann Bot 2014;
[42] Liang Y, Zhang HH, Hu J, Zhai CY. Sunlight multiband photon flux density 114(1):145–56.
detection based on PFD calculation model. Int J Agric Biol Eng 2015;8(2):89–96. [46] Buratti C, Belloni E, Merli F, Zinzi M. Aerogel glazing systems for building
[43] Pennisi G, Blasioli S, Cellini A, Maia L, Crepaldi A, Braschi I, et al. Unraveling the applications: a review. Energy Build 2021:231.
role of red:blue LED lights on resource use efficiency and nutritional properties of [47] Chen N, Ma H, Li Y, Cheng J, Zhang C, Wu D, et al. Complementary optical
indoor grown sweet basil. Front Plant Sci 2019;10. absorption and enhanced solar thermal conversion of CuO-ATO nanofluids. Sol
Energy Mater Sol Cell 2017;162:83–92.

15

You might also like