Energy Conversion and Management 198 (2019) 111781

Contents lists available at ScienceDirect

Energy Conversion and Management

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

Coupled thermal-optical numerical modeling of PV/T module – Combining T

CFD approach and two-band radiation DO model
Seyed Reza Maadia,1, Meysam Khatibia,1, Ehsan Ebrahimnia-Bajestanb,c, , David Woodc
⁎

a
Department of Mechanical Engineering, Ferdowsi University of Mashhad, Mashhad, Iran
b
Department of Mechanical Engineering, Quchan University of Technology, Quchan, Iran
c
Department of Mechanical and Manufacturing Engineering, University of Calgary, 2500 University Dr. NW, Calgary, Alberta T2N 1N4, Canada

ARTICLE INFO ABSTRACT

Keywords: Photovoltaic thermal (PV/T) systems extract the waste heat and improve the overall energy efficiency of PV
Photovoltaic/thermal module (PV/T) modules. In this study, a 3D coupled thermal-optical model was developed, combining computational fluid
Discrete ordinate (DO) model dynamics and ray tracing to evaluate the performance of a glazed PV/T module. All layers of the module were
Two-band radiation simulated with their actual optical and thermophysical properties. To increase the accuracy of the model and
Greenhouse effect
capture the greenhouse effect in the glazed system, the optical characteristics of the layers were considered
Computational fluid dynamics (CFD)
radiation wavelength-dependent, using a two-band radiation model. The proposed model was validated using
the available measured data. Besides, a relatively simple and computationally cost-efficient 3D model compared
to the complete simulation of PV/T was introduced for the case of a straight tube configuration. The effects of
several parameters, including number of tubes, bonding width, and tube diameter on the performance of PV/T,
were also evaluated using the proposed model. The results indicated that considering the greenhouse effect in
the simulation has a significant effect on thermal efficiency, especially at lower solar irradiance, e.g., at solar
irradiance of 300 W/m2 applying the greenhouse effect increases the thermal efficiency by 12% compared to the
case without this effect. The electrical efficiency was not significantly altered by the greenhouse effect, although
a very slight reduction was observed. It was also concluded that increasing the tube number and bonding width
have a significant effect on the temperature distribution of the PV cells, and performance of the PV/T module.

1. Introduction electrical efficiency by 0.45%, 0.45%, and 0.25%, respectively [6].

Besides, the variation of PV module temperature over the day due to the
The Paris Climate Change Agreement aims to strengthen the changing solar irradiance, combined with spatial variations in cell
worldwide trend to lower greenhouse gas emissions, dealing with the temperature can lead to thermal stresses, which reduce the lifespan and
risk reduction of global warming. Accordingly, increasing the share of efficiency of PV modules [7]. Recently, Photovoltaic/Thermal (PV/T)
renewable energy, e.g., solar energy, and improving the efficiency of technology, which removes the heat from the PV module and simulta-
the energy systems have been proposed as the main strategies [1]. Solar neously uses this waste heat, has been attracted much attention. In fact,
photovoltaics (PV) is a commercially available technology to generate this technology is the combination of two well-established solar har-
electricity from sunlight [2]. The International Energy Agency (IEA) vesting technologies of PV modules and solar collectors, to improve the
envisions a global electricity share of 16% for PVs by 2050 [3]. overall energy efficiency of the system. Compared to PV modules, PV/T
One of the main issues in the widespread application of PV is its modules are in the first stage of commercialization, and countries such
relatively low energy conversion efficiency, which is only 4–23% [3,4] as Canada, US, UK, and Germany have seen attempts to incorporate this
depending on the type of solar cell. In these systems, a major part of the technology in buildings for heating purposes [8]. In these countries
absorbed solar radiation is converted to heat, which is responsible for around 80% of the total residential energy use is devoted to space and
the PV cell temperature rise, which reduces cell efficiency [5]. It has water heating [9,10]; therefore, this low carbon technology seems to be
been reported that a temperature rise of 1 °C in monocrystalline (c-Si), an innovative way to supply electricity and heat for the buildings. As
polycrystalline (pc-Si), and amorphous (a-Si) silicon cells reduces the suggested by Mellor et al. [11], further innovation, optimization, and

⁎
Corresponding author.
E-mail addresses: [email protected], [email protected], [email protected] (E. Ebrahimnia-Bajestan).
1
These authors have equal contribution in the paper.

https://doi.org/10.1016/j.enconman.2019.111781
Received 21 February 2019; Received in revised form 25 June 2019; Accepted 3 July 2019
Available online 18 July 2019
0196-8904/ © 2019 Published by Elsevier Ltd.
S.R. Maadi, et al. Energy Conversion and Management 198 (2019) 111781

Nomenclature Abbreviations

A area, m2 ARC antireflective coating

Cp specific heat capacity, J/kg K DO discrete ordinate
D diameter, m PCM phase change material
E power output of solar cell, W RTE radiative transfer equation
e power output of solar cell per unit area, W/m2 EVA ethylene-vinyl acetate
G solar irradiance, W/m2 PV/T photovoltaic/thermal
g gravity, m/s2 PV photovoltaic
h heat transfer coefficient, W/m2K
I radiation intensity, W/m2 Subscripts
K Thermal conductivity, W/mk
M mass, kg a absorption
N number of tubes, – abs absorber
n refractive index, – amb ambient
P pressure drop, pa b black body
Pa packing factor, – c collector
q Heat enegy per unit time per unit area, W/m2 cond conduction
T temperature, K conv convection
V velocity, m/s elec electrical
f fluid
Greek symbols g glass
in inlet
ref reference temperature coefficient, – inc incident radiation
β thermal expansion, – ref reference
η efficiency, – rad radiation
μ Viscosity, pa s r reflection
α Absorptance, – sun sun
ρ reflectance, – s solid
ε emissivity, – sky sky
τ transmittance, – surface surface
s scattering coefficient, – th thermal
σ absorption coefficient, 1/m tot total
phase function, – t transmission
'
solid angle, – out outlet
density, kg/m3 pv photovoltaic
shear stress, N/m3 wind wind
difference, – λ wavelength
W water

performance improvement are essential to increase the market poten- energy transferred from the sheet, as a fin, to the tube [19]. Yazdanifard
tial for PV/T technologies. This encourages researchers to study per- et al. [16] applied thermal-resistance-based modeling to a water-based
formance improvement techniques for PV/T modules including im- PV/T module. They found that the energy efficiency of the glazed
proved cooling system design, the use of nanofluids, phase-change system is more than that of unglazed one. They also reported that the
materials (PCM), glazing, and mass flow rate optimization [12,13]. effect of the studied parameters (e.g., solar irradiance, packing factor,
Numerical modeling is a practical and cost-efficient approach to Reynolds number, and geometrical specifications) on the module tem-
analyze and optimize the PV/T (or PV) modules in terms of thermal and perature is greater in the case of laminar cooling fluid flow compared to
optical characteristics of the elements of the system. It is well-estab- turbulent flow. Vera et al. [17] simulated a glazed and unglazed PV/T
lished that the efficiency of a PV cell depends on its temperature [14]. module by a thermal-resistance-based model to optimize the design
Also, for performance analysis of a PV/T module, it is essential to de- parameters including coolant mass-flow rate, collector length, packing
termine the amount of absorbed radiation which passes through the factor (the fraction of area in PV plate occupied by the solar cells) and
layers and transfers to the cooling fluid. These facts lead to a major air gap. Anderson et al. [20] performed a parametric study for im-
focus of researchers on the thermal modeling of the PV/T modules plementation of the PV/T systems into domestic and commercial
while few studies have incorporated optical models. In most of these buildings using a fin-based model. Generally, these thermal modeling
thermal models, the one-dimensional heat energy balance through the approaches are cost-efficient and fast methods for feasibility study of
module thickness is used to calculate the thermal loss, PV cell tem- systems in practical cases [21], time-dependent ambient conditions
perature, and the absorbed heat by the coolant section of the PV/T [18], and for performance optimization of the PV/T system [17] and
module [15–17]. These studies have usually employed thermal-re- the analysis of various cooling technologies [15].
sistance-based models for heat transfer assuming the absorbed solar Although the above-mentioned 1D thermal models are less time-
radiation as an internal heat source in the glass cover and/or PV cell consuming compared to the 2D and 3D models, the assumption of an
and uniform temperatures for each layer (the “isothermal-layer as- isothermal layer cannot be accurate in PV/T modules. This is because
sumption”) [18]. The other simple model is based on fin heat transfer the coolant flowing beneath the module can lead to a completely 3D
analysis — an analytical method for modeling the thermal performance heat transfer and non-uniform temperature distribution. This tem-
of sheet and tube solar collectors—calculating the amount of heat perature variation in PV cells can affect the efficiency of the module

2
S.R. Maadi, et al. Energy Conversion and Management 198 (2019) 111781

Table 1
Summary of 3D modeling of PV/T modules.
Ref. Simulated elements Modeling Summary and highlights

Optical Thermal

Zondag et al. [23] Glass cover; Air gap; EVA; Solar cell;
Absorber; Bonding; Tube
– ✓ • Developed a quasi 3D dynamic thermal model, assuming
constant temperature across the thickness of each layer,
solving by a Runge–Kutta procedure.
• Did not solve convective heat transfer equation inside air gap
and tubes; used the Nusselt number correlations in these parts.
• Showed that 1D analysis uses less time; 3D model is easily
adapted to other configurations and provides more detailed
information, e.g., collector temperature distribution.
• Concluded that hourly data are sufficiently accurate, and error
made by ignoring the dynamic effects is very small for the
calculation of daily performance
Siddiqui et al. [26] Glass cover; EVA; Solar cell; Tedlar; Channel; ✓ ✓ • Developed a 3D dynamic thermal model in ANSYS CFX.
Cooling fluid • Used a simple optical model to calculate the absorbed solar
radiation by PV cells. Assumed the optical properties were
wavelength independent.
• Did not model the greenhouse effect and ray tracing.
• Compared the performance of the PV and PV/T.
• Found that the simple 3D model is proper to predict the
temperature distribution of PV cells in PV/Ts with various
coolant passages.
Pierrick et al. [27] Glass cover; Air gap; EVA; Solar cell;
Channel; Cooling fluid
✓ ✓ • Developed a quasi 3D dynamic thermal model, assuming
constant temperature across the thickness of each layer,
solving the matrix of temperatures of different layers using the
Gauss Seidel method.
• Did not solve convective heat transfer equation inside air gap
and tubes; used Nusselt number correlations in these parts.
• Used a simple optical model to calculate the absorbed solar
radiation by glass and PV cells. Considered wavelength-
independent optical properties. Did not model the greenhouse
effect and ray tracing.
• Proposed a model for electrical power generation.
• Used a heat exchanger with a “fractal geometry” within the PV
module.
Guarracino et al. [28] Glass cover; Air gap; EVA; Solar cell; Tedlar;
Adhesive; Absorber; Tube; Cooling fluid
✓ ✓ • Developed a quasi 3D dynamic thermal model, assuming
constant temperature across the thickness of each layer.
• Did not solve convective heat transfer equation inside air gap
and tubes; used Nusselt number correlations in these parts.
• Used a simple optical model to calculate the absorbed solar
radiation by glass and PV cells. Considered wavelength-
independent optical properties. Did not model the greenhouse
effect and ray tracing.
• Showed that glazing the PV/T module increases the thermal
efficiency and slightly decreases the electrical efficiency.
Nasrin et al. [22] Glass cover; EVA; Solar cell; Tedlar;
Adhesive; Absorber; Channel; Cooling fluid
– ✓ • Developed a steady state 3D thermal model for various layers,
solving by COMSOL Multiphysics software
• Did not consider the air gap.
• Proposed new designs for thermal collector excluding the
absorber plate. Studied the effect of nanofluids as a cooling
fluid.
Hosseinzadeh et al. [29] EVA; Solar cell; Tedlar; Absorber; Tube;
Cooling fluid
– ✓ • Developed a steady state 3D thermal model for a section of the
module having one tube and various layers, solving by ANSYS
Fluent.
• Ignored the air gap and the direct simulation of the glass cover.
• Studied the effect of ZnO/water nanofluid.
Kazemian et al. [30] EVA; Solar cell; Tedlar; Absorber; Tube;
Cooling fluid; PCM
– ✓ • Developed a steady state 3D thermal model for a section of the
module having one tube and various layers, solving by ANSYS
Fluent.
• Ignored the air gap and the direct simulation of glass cover.
• Simulated melting process of PCM, using the enthalpy-porosity
technique.
Fayaz et al. [31] Glass cover; EVA; Solar cell; Tedlar;
Channels; Cooling fluid; PCM
– ✓ • Developed a steady state 3D thermal model for various layers,
solving by COMSOL Multiphysics software
• Ignored the air gap.
• Simulated melting process of PCM, using an enthalpy-based
method

3
S.R. Maadi, et al. Energy Conversion and Management 198 (2019) 111781

[22] and may result in the thermal stresses and aging issues. Further, described and the proposed mathematical modeling along with the
investigating new designs and configurations of thermal collectors in optical and thermophysical properties of different parts of the PV/T
PV/T modules is done more accurately in 3D models [22,23]. There are module is presented. The numerical solution of the governing equations
some studies on 2D [24,25] and 3D modeling of PV/T modules. A is also discussed.
summary of the available 3D simulations, along with the type of
modeling is provided in Table 1. Based on the literature review reported 2.1. Physical model
in Table 1, most of the 3D models have been focused on thermal
modeling of PV/T modules. It can also be seen that the available 3D Fig. 1 illustrates the modeled PV/T module whose outer surface is a
dynamic models, which have considered both optical and thermal glass layer. Below this, in order, is an air gap, Ethylene-vinyl acetate
modeling, are in fact, quasi 3D models. In these types of models, due to (EVA) and anti-reflective coating (ARC), and two layers of EVA, the
the small thickness of layers, the heat transfer equation across the poly-crystalline silicon cells (pc-Si), and Tedlar polyvinyl fluoride (Te-
thickness of each layer is not solved; i.e., it is assumed that the tem- dlar) underneath. The ARC layer improves the photovoltaic conversion
perature in each layer is constant. Therefore, these models are really 2D efficiency of the system, [41], while EVA and Tedlar layers also protect
thermal models. It should be mentioned that in the existing 3D dynamic PV cells, for example, from UV radiation, and humidity, [32,33]. All
modeling of PV/T modules the optical and thermal models are not these layers (shown as PV layers in Fig. 1) along with the glass cover,
coupled together, and the optical model predicts the energy absorbed air gap, absorber plate, and tubes are modeled in this study.
by a PV cell to use as an input parameter in the thermal model. The geometrical specifications, thermophysical properties along
Our literature survey reveals that most numerical studies on PV/T with operating conditions, and optical properties of the different com-
systems have focused only on the thermal aspect of the modules or used ponents are listed in Tables 2–4, in that order. The operational condi-
simplified optical characteristics for the layers (e.g., [32,33]). However, tions of a typical PV/T system presented in Table 3 are considered as
the dependence of the optical characteristics (i.e., reflectance, absorp- the base case in this study. Water is selected as the working fluid inside
tance, and transmittance) on the wavelength of the solar radiation has the tubes with the temperature dependent properties shown in Table 5.
not been accounted for in most studies [34]. Recently, Jia et al. [35]
also expressed that to develop novel PV/T systems, more effort is re-
2.2. Energy balance, heat transfer mechanisms and solar ray tracing in the
quired in accurate modeling of these systems, focusing more on the
PV/T module
radiation terms. Thus, to develop an accurate model for a PV/T (or PV)
module, the thermal and optical energy balance of the system should be
Fig. 2(a) and (b) depict the important heat transfer mechanisms in
considered simultaneously [36]. This means that both heat transfer and
the PV/T module and ray tracing for interaction between the glass
ray tracing considering wavelength-dependent optical properties need
cover and PV module (i.e., PV layers). In Fig. 2(a), G, and q represent
to be modeled especially in the case of glazed systems, in which the
the solar irradiance, and heat power per area, respectively. The thermal
received beam in a certain wavelength range is trapped between PV and
radiation is the energy emitted from the surface because of its tem-
its cover, causing a ‘greenhouse effect’. It is noteworthy that it has been
perature.
recommended to use glazing for PV/T modules [11,16], although most
Nearly all solar radiation has wavelengths in the range of 0.3–3 μm,
installed PV/T modules are unglazed [11]. The coupled thermal and
consisting of three wavelength bands: ultraviolet, visible, and IR [51].
optical modeling is also vital to design and optimize the selective layers
Glass is a semi-transparent material has different optical properties at
for a PV/T module [37].
different light wavelengths. It is almost opaque for the long wavelength
Recently, a few studies have simultaneously modeled the thermal
radiation and semi-transparent for the rest of the spectrum. Therefore,
and optical aspects of some solar systems, e.g., Fresnel collectors [38],
for the glazed PV/T module, short wavelengths of the solar radiation
solar-chimney power plants [39], and water evacuated tube solar col-
pass through the glass and reach the PV, converting to longer wave-
lectors [40] using 3D computational fluid dynamics (CFD) and ray
length; where part of these long wavelength waves cannot escape
tracing algorithms. However, there are no similar investigation in the
through the glass and are trapped inside the air gap (see Fig. 2(b)). This
PV or PV/T modules, which are more complex due to the presence of
phenomenon is called the ‘greenhouse effect’, which is considered in
different layers with small thicknesses.
the modeling of PV/T in this study.
Considering the above-mentioned challenges in the modeling of PV/
Most of the solar irradiance reaching the glass (Gsun in Fig. 2(a)) is
T systems, the present paper aims to construct a 3D model in-
transmitted (Gt,g) and the rest is reflected (Gr,g) or absorbed (Ga,g),
corporating solar ray tracing, heat transfer, and fluid flow in a glazed
depending on the glass optical properties. The effective glass optical
PV/T system. To the best of authors’ knowledge, this is the first time
properties are absorptance (α), reflectance (ρ), transmittance (τ), ab-
that such a model is studied. Accordingly, the proposed 3D model si-
sorption coefficient (σ), refractive index (n), and emissivity (ε). Since
mulates all layers of the PV/T module with all corresponding optical
the sky thermal radiation, (qrad,sky in Fig. 2(a)) is in the infrared band
and thermophysical properties as shown and explained later in Fig. 1.
(long wavelengths), it will not be transmitted through the glass.
The discrete ordinate (DO) method associated with solar ray tracing
Therefore, the total energy transfer from glass to the PV layers (see
algorithm is used to model the optical behaviour. Additionally, the
Fig. 2(a)) includes the transmitted solar irradiance (Gt,g) and the
greenhouse effect is considered by employing a two-band radiation DO
thermal radiation part from the interior surface of the glass (qrad,g),
model, considering the interchange of short and long wavelength ra-
diation in the reflection between the glass and the PV module. Besides,
a comparative study is conducted to examine the possibility of em-
ploying 3D simulation for only a small section of the PV/T module
instead of the whole module for the straight tube configuration. Finally,
the model is employed to investigate the effects of several parameters,
including number of tubes, bonding width between the tube and the
module, and tube diameter on the temperature distribution of PV cell
and PV/T performance.

2. Physical and mathematical modeling

In this section, the heat transfer mechanisms in a PV/T module are Fig. 1. Schematic of the modeled a) various layers and b) PV/T module.

4
S.R. Maadi, et al. Energy Conversion and Management 198 (2019) 111781

Table 2 Table 4
Geometrical specifications of different PV/T components used in this simulation Optical properties of different components for normal incidence light used in
[25,41–43]. this simulation [42,45–47].
Component Parameter Value Unit Refs. Component Parameter Value Unit Refs.

Glass cover Thickness 0.004 m [42] Glass cover Emissivity 0.88 [42]
Air gap Thickness 0.012 m – Absorption 26 for < 4.25 µm 1/m [42]
Encapsulated pc-Si Length 2 m [43] coefficient
Width 1 m [43] 1100 for 4.25 µm 1/m calculated
Thickness of pc-Si 0.0003 m [25] Refractive index 1.526 [46]
Thickness of EVA: 0.0005 m [41] Encapsulated pc-Si
Thickness of ARC: 8 × 10 8 m [41] pc-Si Emissivity 0.97 for < 4.25 µm [47]
Thickness of Tedlar: 0.0001 m [41] 0.85 for 4.25 µm
Absorber plate Thickness 0.0002 m [43] Effective 0.93 [33,42]
Tube Outlet diameter 0.00952 m [42] absorptance
Thickness 0.0009 m [42] Refractive index 3.69 [45]
Number 10 – [42] EVA Refractive index 1.45 [45]
Space between tubes 0.1 m [42] Tedlar Refractive index 1.45 [46]
Length 2 m [43] ARC (TiO2) Refractive index 2.3 [45]

Table 3
Material properties and operating conditions used in this simulation the incident wavelength. A part of the absorbed short wavelength ra-
[13,16,22,38,42–44]. diation is converted to electricity in PV cells, while the remainder is
Component/condition Parameter Value Unit Refs.
converted to heat. In addition, the heat transfer mechanisms, including
free convection, conduction, and thermal radiation exist within the air
Glass cover Thermal conductivity 1.5 W/mK [38] gap between surfaces.
Density 2650 kg/m3 [38] Considering Fig. 2(a) and (b), the energy losses of PV/T module
Specific heat capacity 786 J/kgK [38] include thermal and optical losses. The thermal losses occur through (i)
pc-Si layer (Poly-crystalline silicon) thermal radiation from glass to the sky (qrad,g-sky), and (ii) convection
Open circuit voltage 21.1 V [42] heat transfer to the ambient (Qconv, g-amb). While optical losses happen
Short circuit current 3.8 [42]
A
through (i) reflected solar irradiance from the glass surface to ambient
Max. power 17.1 W [42]
Thermal conductivity 100 W/mK [16]
(Gr , g ) , and (ii) some parts of reflections between glass and PV module,
Density 2330 kg/m3 [22] which escape from the glass cover to the ambient (G't , g , i.e., blue arrows
Specific heat capacity 700 J/kgK [22] in Fig. 2(b)).
Reference cell 0.173 – [42]
efficiency 2.3. Thermal-optical modeling
Reference temperature 0.0045 K 1 [42]
coefficient
Reference temperature 298 K [16] 2.3.1. Assumptions
Packing factor 0.804 – [42] In this modeling, the following assumptions have been made:
EVA
Thermal conductivity 0.35 W/mK [44] • The flow inside the tubes is laminar as assumed in various numerical
Density 960 kg/m3 [44] [22,25,29,52] and found in experimental studies [53]. Thus, the
Specific heat capacity 2090 J/kgK [44] Reynolds number is lower than 2300 in all cases.
ARC (TiO2) • The bottom of the module and tubes (illustrated in Figs. 1 and 2),
Thermal conductivity 8.4 W/mK [13] and also the sidewalls of the module are considered as adiabatic
Density 4157 kg/m3 [13] walls.
Specific heat capacity 692 J/kgK [13]
• The absorption coefficient of air is neglected compared to other
Tedlar parts.
Thermal conductivity
Density
0.2
1200
W/mK [44]
[44]
• The incident sunlight is unpolarized, and only the direct radiation is
kg/m3 considered [36,41].
Specific heat capacity 1250 J/kgK [44]
• The property of each solid component is temperature independent.
Absorber plate, bonding, and Thermal conductivity 387 W/mK [43] • Dust and partial shading on the module are negligible.
Tube (copper)
Density 8978 [43]
• The fluid flow is uniformly distributed among the tubes. In other
kg/m3 words, the flow rate per tube is the total flow rate divided by the
Specific heat capacity 381 J/kgK [43]
number of the tubes.
Operating conditions Water inlet
temperature
311 K –
• Gap walls and the pc-Si layer are assumed to be non-gray diffuse
surfaces. Noted that, the optical properties of a non-gray surface
Wind velocity 1 m/s –
depend on radiation wavelength, while they are wavelength-in-
Total mass flow rate 0.044 kg/s –
dependent for a gray surface.
•
Ambient temperature 303.15 K –
Solar irradiance 700 W/m2 – The contact between the PV layers and the absorber plate is con-
Pump efficiency 0.8 – [16] sidered to be a perfect thermal contact. In other words, the thickness
of the adhesive material between the Tedlar and the absorber plate
is ignored [43,54].
because of its temperature.
The transmitted irradiance (Gt,g) reaching the PV module (PV layers
• The incident solar radiation is assumed to be perpendicular to the
module surface.
in Fig. 2(a)) is absorbed, and/or reflected, then due to greenhouse effect
re-absorbed and/or reflected continuously (See Fig. 2(b)) depending on
• The analysis is quasi-steady. This assumption is reasonable for a
clear sunny day when the solar irradiation varies smoothly and

5
S.R. Maadi, et al. Energy Conversion and Management 198 (2019) 111781

Table 5
The correlations used for water properties as working coolant. The R2 values were calculated with respect to the available data [48], for the temperature range of
290–370 K.
Property Equation R2 Unit Ref.

Viscosity 1448.538 521926.58 0.9997 kg/m s [49]

µ w = 1.788 × 10 3 exp 1.704 +
TW 2
TW

Density = 2
0.003 × TW + 1.505 × TW + 816.781 0.9996 kg/m3 [50]
w

Specific heat capacity Cp, w = 3

0.0000463 × TW 2
+ 0.0552 × TW 20.86 × TW + 6719.637 0.9975 J/kg K [50]

Thermal conductivity kw = 2
0.000007843 × TW + 0.0062 × TW 0.54 0.9977 W/m K [50]

slowly, and the variation of the thermal inertia of the system is small quasi-steady. Noted that Zondag et al. [23] found that hourly data are
[55]. sufficiently accurate for the calculation of the daily performance of PV/
T modules (at a nearly clear sunny day), which strongly reduces the
Although the model developed in this study is steady state, it is processing time of the prediction.
easily possible to extend this simulation to a dynamic model, as long as
the solution convergences for each time step in ANSYS Fluent software.
On the other hand, the proposed steady-state model can predict the 2.3.2. Thermal modeling
daily performance of PV/T module using hourly climate data (e.g., solar Considering the above assumptions, the governing equations for the
irradiance and ambient temperature), which makes the present model heat transfer in this problem are as follows.
For natural convective/conduction heat transfer inside the air gap,

Fig. 2. Schematic of a) the incorporation of heat transfer mechanisms in a PV/T module and b) ray tracing for interaction between a glass cover and PV module (i.e.,
PV layers). (Figures are not in scale.)

6
S.R. Maadi, et al. Energy Conversion and Management 198 (2019) 111781

Fig. 2. (continued)

and also mixed convective heat transfer of water flowing through the each solid layer. To determine the Sh the Radiative Transfer Equation
tubes under the module: (RTE) must be solved with the other conservation equations of mass,
Continuity: momentum and energy, simultaneously, as discussed in Section 2.3.3.
In addition, the method of predicting Eelec in Eq. (5) will be discussed in
. ( f Vf ) = 0 (1) Section 2.3.4.

Momentum:
2.3.3. Electrical power modeling
. ( f Vf Vf ) = P+ + g The most common relation for the conversion of the absorbed
1 f (2)
sunlight to the electricity in the PV cells is expressed as Eq (6). In this
Energy: equation, the electrical efficiency reduces linearly with PV cells’ tem-
perature [28,31,56].
. ( f Cp, f Tf Vf ) = . (kf Tf ) (3)
Eelec = S ref (1 ref (Tpv Tref )) Pa (6)
In Eqs. (1) and (2) f , Vf , P , and 1 are density, velocity, pressure and
where, in order, ref , ref , and Tref, are reference cell efficiency, re-
shear stress, respectively and Cp, f , Tf , and kf in Eq. (3) are specific heat
ference temperature coefficient, and reference temperature. The re-
capacity, temperature, and thermal conductivity. Subscript f refers to
ference cell efficiency is an electrical efficiency of PV cell measured
the fluid (air or water). For water flowing through the tubes, the
under standard condition (solar irradiance of 1000 W/m2 and reference
thermophysical properties listed in Table 5 are used. For air inside the
temperature of Tref =298 K). Tpv denotes the average temperature of the
gap, all properties are considered constant ( µair = 1.7894 × 10 5kg/ms ,
PV cell computed from Eq. (5). Pa is the packing factor (i.e. the fraction
Cp, air = 1006.43J/kgK, and kair = 0.0242W/mK ) except the density in
of area in PV plate occupied by the solar cells) [56].
buoyancy term of the momentum equation which follows the Boussi-
In Eq. (6), S is the solar irradiance which is absorbed by the PV cells.
nesq approximation:
A common expression of the absorbed energy for the glazed PV cells is
( air amb ) g amb (T Tamb ) g (4) as follows [33]:

where amb is the density of air at the ambient temperature of Tamb (see S= g ¯ pv AGsun (7)
Table 3), and β is the thermal expansion coefficient.
Finally, pv , A, and Gsun are effective absorptance of PV cells , PV
For the conduction heat transfer in the solid parts of the module:
plate area and direct solar irradiance normal to the module reaching the
ks 2 (T )
s + e'elec + Sh = 0 (5) glass, respectively, as defined in Tables 3 and 4. The transmittance of
the glass for normal incident (as assumed in this study) is calculated as
In Eq. (5), = 1 for the PV cells (i.e., pc-Si layer) and zero other- [33,46]:
wise. The subscript s indicates the solid layers. Moreover, Sh is a heat
source which represents the amount of solar irradiance absorbed by = exp( L)(1 r) (8)

7
S.R. Maadi, et al. Energy Conversion and Management 198 (2019) 111781

where L and are the glass thickness and spectral absorption coeffi- pv, is the PV cells emissivity in each band, σ is the Stefan-Boltzmann
cient, respectively. more ever, r is the reflection of unpolarized radia- constant which is equal to 5.67 × 10 8W/m2K4 , and TPV is the tem-
tion on passing from the glass, due to the different refractive index of perature of the module [50].
two adjacent medium defined as follows [46]: The critical part of this optical modeling is the consideration of the
2 glass cover as a semi-transparent material. Fig. 2 shows two possible
ng nair radiation sources in the case of exterior semi-transparent walls: thermal
r=
ng + nair (9) radiation from surrounding zones and beam radiation from outside the
computational domain. For interior semi-transparent walls, the incident
where ng is the refractive index of the glass reported in Table 4, and radiation can transmit and possibly refract to the adjacent medium, be
nair = 1 is the refractive index of air. reflected or absorbed through the wall.
The simulation with the glass cover presents two important chal-
2.3.4. Optical modeling lenges, namely capturing the greenhouse effect and modeling the re-
In the present study, for solving the RTE, the DO radiation model is fraction. Here, it is essential to consider the non-gray model to include
employed. The full details of this method can be found in Ref. [57]. The the different behavior of the material in the different wavelength re-
DO model is used because of some advantages over alternative models gions. This model was applied by dividing the radiation spectrum into
such as: two wavelength bands (i.e., two-band model), which are 0–4.25 µ m and
greater than 4.25 μm for a typical glass [58]. The optical properties
• The ability to model semi-transparent materials (e.g., glass) remain constant for each band. Beer-Lambert’s law is used to determine
• Modeling non-gray radiation using a gray assumption for each band a dual-band absorption coefficient for glass with 4 mm thickness. For
• Applicable to a wide range of optical thickness wavelengths below 4.25 μm, the absorption coefficient is 26m−1 which
translates to 10% absorption by the glass cover. This range will be
The RTE for the spectral intensity I ( r , s ) at position r in the called “semi-transparent”. For wavelengths above 4.25 μm, the ab-
direction s is written as: sorption coefficient is 1100m−1 which means that around 99% ab-
sorption in this “opaque” band.
. (I ( r , s ) s ) + ( + s )I (r , s )
4
s ' ' 2.3.5. Boundary conditions
= n2Ib + I ( r , s ) ( s . s )d
4 (10) The boundary conditions of the tubes are mass flow inlet (i.e., mass
0
flow rate and inlet temperature), and outflow (i.e., zero gradients for all
where s' and s are scattering direction vector and scattering coeffi- variables at the outlet). The no-slip boundary condition is applied to all
cient, respectively. , n, Φ, and ' denote spectral absorption coeffi- walls exposed to the fluid flow. The boundary of the exterior surface of
cient, refractive index, phase function, and solid angle, respectively. Ib the glass cover is set as a combined convection and radiation boundary
is the blackbody intensity given by the Planck function. The first and conditions. The external radiation temperature is set equal to the sky
second terms at the left-hand side of Eq. (10) represent the radiation temperature, which is calculated as following [59]:
intensity change through the computational cell and radiation losses 1.5
Tsky = 0.0552Tamb (15)
due to the absorption and scattering in the media, respectively. The first
term at the right-hand side of Eq. (10) is the radiation absorbed due to The convective heat transfer coefficient of ambient air flowing over
emission and the other term represents the integration of the scattered the glass is calculated based on:
radiation in the domain of the scattering solid angle. The total intensity m
I ( r , s ) in each direction s at position r over the wavelength bands is 5.7 + 3.8Vwind if Vwind < 5 s
h wind =
computed using: 0.78
6.47 + V wind if Vwind 5s
m
(16)
I (r , s ) = I k (r , s ) k where Vwind is the local wind speed [25].
k (11)

It is noteworthy that to simulate the wavelength-dependent optical 2.3.6. Performance of PV/T module
properties, the non-gray DO model should be applied. For this purpose, Although cooling of a PV module increases the electrical power
based on the spectral properties of glass and encapsulated PV plate, the output, it is at the expense of pump power used for the coolant.
wavelength interval should be defined. The RTE is integrated over each Therefore, in this study, the effect of pump power has been taken into
considered wavelength interval. The optical properties for each band account in calculating the final electrical efficiency, as shown in Eq.
are assumed wavelength independent. The black body emission in each (19). Following [16], the pump power consumption, Ppump , for the
wavelength band per unit solid angle is written as: pumping of working fluid in tubes is calculated as:
T4 Nm p
Eb ( , T ) = [F (0 n 2T) F (0 n 1 T )] n2 Ppump =
(12) w pump (17)
where F (0 n T ) provides the Planck distribution which is the por- where N, pump , and Δp are the number of the tubes, pump efficiency,
tion of radiant energy emitted by a black body in a medium of refractive and the pressure loss in one tube, respectively. The required pressure
index n over the wavelength interval from 0 to λ at temperature (T). 2 drop in this equation is obtained through data post-processing on the
and 1 are the wavelength boundaries of each band. solved governing momentum equation for the working flow inside the
The radiation incident on the opaque PV surface and the net ra- tubes (Eq. (2)).
diative energy flux leaving the PV surface are computed as: Finally, to study the performance of the PV/T module, the thermal
efficiency ( th ) and electrical efficiency ( elec ) are defined as [16,60,61]:
qinc, PV , = Iin, s . n d (13)
mCp, w (TW , out TW , in )
=
and th
AGsun (18)
qout , PV , = (1 pv, ) qPV , inc, + pv, Eelec Ppump
=
[F (0 n 2T) F (0 n 1 T )] n2 Tpv 4
(14)
elec
AGsun (19)

8
S.R. Maadi, et al. Energy Conversion and Management 198 (2019) 111781

Tw,out and Tw,in represent the outlet and inlet water temperature 8 × N × N directions of the RTE equation are computed. Moreover,
flowing through the tubes, respectively. The heat capacity of water, Cp,w these control angles are then divided into pixelations. Therefore, divi-
depends on temperature as presented in Table 5. sion and pixel independency should be investigated in addition to the
grid study. Tables 7 and 8 show the output of some important variables
2.4. Numerical solution in different divisions and pixelations, respectively. The pixelations and
divisions do not have a significant effect on the outlet temperature and
The computational fluid dynamics (CFD) software ANSYS-Fluent pressure drop of the water in the tubes, as well as the PV module
18.1 was used to solve the governing equations. The pressure-based temperature, surface incident radiation, and energy flux. However, a
solver was used for this laminar model, and the coupling of pressure pixilation and a division of 3×3 are used due to the existing semi-
and velocity was by the SIMPLE scheme [62]. The “second-order up- transparent boundary in this simulation as recommended by ANSYS-
wind” was selected as a discretization scheme for the convection, dif- Fluent.
fusion, and radiation terms and also PERESTO scheme [63] was used
for pressure. The sun’s rays that enter the computational domain were 3. Validation
modeled by the ray tracing algorithm. By applying the solar load in
ANSYS-Fluent, the calculated heat by the ray tracing algorithm is The numerical results are validated by the comparison of the
coupled to the energy equation via a heat source term. Noted that this thermal and electrical efficiencies with the experimental data of a PV/T
model is only available for 3D simulation. The normalized convergence module, measured by Bhattarai et al. [42]. The operating conditions for
criterion for the continuity, momentum, and DO equations were set 10 6 this validation study are presented in Table 9. Note that, following the
and 10 8 was used for the energy equation. calculation of electrical efficiency done by Bhattarai et al., the effect of
pump power was ignored only for this validation study.
2.5. Grid, division, and pixel independency As shown in Fig. 4, reasonable agreement is observed between the
results of simulation and experimental data with maximum errors of
The 3D structured grids adopted inside the computational domain 3.7% and 5.1% for the electrical and thermal efficiencies, respectively.
with refined grids near the walls are presented in Fig. 3. Thanks to using The simplifying assumptions in the present numerical model and the
a pre-specified pattern to sort the nodes and elements and also lower uncertainties of experimental data may lead to the small discrepancies
cell numbers, structured grids require less computational memory and between the results of these studies. In conclusion, the proposed three-
CFD computation time compared to the alternative, unstructured grids. dimensional model is accurate enough to simulate the optical and
Moreover, structured grids are characteristically aligned in the flow thermal performance of a PV/T.
direction, resulting in more accurate prediction and better convergence
in CFD solution algorithms and solvers [64,65]. 4. Results and discussions
A grid study was done to evaluate the dependence of results on the
mesh for the conditions listed in Tables 2–4 which will be referred to as We consider, first, the effect of the greenhouse phenomenon in the
the “base-case”. For this purpose, the outlet temperature and pressure simulation of PV/T modules. Next, a method is proposed to reduce the
drop of the working fluid in the tube, as well as the PV module tem- computational cost of 3D simulation of whole PV/T module. Finally, the
perature, surface incident radiation (Eq. (13)), and net energy flux effects of several geometrical specifications on the performance of the
leaving the PV surface (Eq. (14)) were examined for various mesh sizes. PV/T and temperature distribution of the PV cells are studied. It should
The results of this grid study are presented in Table 6, where the be mentioned that all investigations are performed in the base-case
changes in pressure drop and surface incident radiation from Grid 1 to condition defined in Tables 2–4.
Grid 2 are about 5% and 3%, respectively; however, these variables
remain nearly constant from Grid 2 to Grid 4. The values of other 4.1. Effects of greenhouse phenomena
variables are almost the same for all grid systems. Grid 2, with the total
cell number of 1731600, was selected to balance computational cost As mentioned in Section 2.3.3, to take into account the greenhouse
and good accuracy. effect, the two-band radiation DO model and the non-gray model was
In solving the optical equations, a total of N × N directions are used to include the different behavior of the material in the different
solved for each band (ultraviolet region, visible, or IR band) in the wavelength bands. In this section, the importance of greenhouse effect
numerical calculations along with non-gray model. For 3D calculations is investigated. For this purpose, the results of the numerical solution

Fig. 3. The mesh used in the present simulation.

9
S.R. Maadi, et al. Energy Conversion and Management 198 (2019) 111781

Table 6
The grid study with divisions and pixelations of 1 × 1.
Grid system Cell number TW , out (K ) TPV (K ) p (Pa) qinc,PV (W /m2) qout, PV (W /m2)

Grid 1 750000 316.21 320.36 106.098 543.2 70.08

Grid 2 1731600 316.349 320.17 111.153 525.6 72.8
Grid 3 3000000 316.39 319.91 112.849 520.5 72.65
Grid 4 6240000 316.394 319.99 113.388 522.1 72.57

Table 7
The optical division study for the numerical model with 1,731,600 cells and
pixelations of 1 × 1.
Divisions TW , out (K ) TPV (K ) p (Pa) qinc,PV (W /m2) qout, PV (W /m2)

1× 1 316.39 320.33 111.31 524 72.1

3× 3 316.39 320.12 111.59 521.3 74.5
6× 6 316.39 320.12 111.59 521.1 74.6
9× 9 316.39 320.11 111.38 520 74.6

Table 8
The pixels study for the numerical model with 1,731,600 cells and 3 × 3 di-
visions.
Pixels TW , out (K ) TPV (K ) p (Pa) qinc,PV (W /m2) qout, PV (W /m2)

1× 1 316.39 320.44 111.35 520.9 74.8

Fig. 4. Validation of the PV/T module thermal and electrical efficiencies in
3× 3 316.39 320.42 111.58 521.5 74.6
numerical with the experimental results [42].
5× 5 316.39 320.43 111.49 521.3 74.5
10× 10 316.40 320.41 111.76 521.1 74.3
(where inlet set to Z = 0), with and without the greenhouse effect;
where the increment of the temperature can be observed in all positions
Table 9 of the PV layer due to the greenhouse effect. Therefore, the amount of
Parameters used for validation study [35].
heat energy transferred to the cooling system is higher in the presence
Component Parameter Value Unit of the greenhouse effect, which augments the temperature of cooling
fluid, as shown in Table 10. As a result, the thermal efficiency of the
Glass cover Area 2 m2
PV/T module considerably increases (see Table 11). However, con-
Thickness 0.004 m
Emissivity 0.88 sidering Eq. (6), the PV temperature increment in the presence of the
Absorption coefficient 26 1/m greenhouse effect, slightly reduces the electrical efficiency, as shown in
Refractive index 1.526 Table 11 and confirmed by the findings of other researchers, e.g. [28].
PV plate (Polycrystalline) Area of cell 1.607 m2 A comparative study was also performed in different solar irra-
Packing factor 0.804 –
diance to determine the influence of the greenhouse effect on the per-
Tube Outlet diameter 0.00952 m
Thickness 0.0009 m formance of the PV/T module. According to Table 11, the greenhouse
Number 10 – effect reduces the thermal losses by nearly 30%, 23%, and 20% for solar
Length 1.916 m irradiance of 300, 700, and 1100 W/m2, respectively. Similarly, the
Operating conditions thermal efficiency is substantially affected by the greenhouse effect at
Wind velocity 1 m/s
lower solar irradiance; while it has a slight effect on the electrical ef-
Ambient temperature 303.15 K
Solar irradiance 800 W/m2
ficiency.
Tilt angle 30 It can be concluded that applying the greenhouse effect in the
modeling of glazed PV/T modules is important, and essential especially
at lower solar irradiance. Note that the importance of the greenhouse
are compared to a case without modeling the greenhouse effect, con- effect has inspired researchers to use glazed PV/T modules [11].
ducted by considering the glass cover as a gray material.
Table 10 shows the greenhouse effect on the temperature of the 4.2. Simulation of a one-tube section instead of the whole module
glass cover, outlet temperature, PV cells temperature, surface incident
radiation, and total thermal losses of the module. In general, the Fig. 6 shows the temperature distribution of the PV cells in the 3D
greenhouse effect leads to the trapping of sunlight in the gap zone and simulation of the PV/T module. The coolant flowing through the tubes
consequently, reduction of the total thermal loss of the system, and decreases the temperature of the PV cells along the flow direction.
increasing the light incident to the PV surface. It can be seen in Fig. 6(b) also shows the temperature distribution across the width of the
Table 10, the radiation incident on the PV cells increases by 11.59%
when the greenhouse effect is considered. In consequence, the amount Table 10
of absorbed energy by the PV cells will increase, which in turn, leads to Importance of considering the greenhouse effect in the simulation of PV/T
a temperature rise in PV module. According to Table 10, the greenhouse module.
effect also causes a ∼1 K increase in the PV module temperature. To Simulation consideration TW , out (K ) TPV (K ) qinc,PV (W /m2) qloss, tot (W /m2)
better show the influence of the greenhouse effect on the temperature
of PV cells, the temperature distribution of PV cells in a selected section With greenhouse 315.67 320.26 552.89 160.06
having one tube is depicted in Fig. 5 at three lines along the module, Z Without greenhouse 315.17 319.31 495.47 207.37

10
S.R. Maadi, et al. Energy Conversion and Management 198 (2019) 111781

Fig. 7. Schematic of the proposed 3D one-tube PV/T section against the whole
PV/T module having 10 tubes.

Table 12
One-tube PV/T section compared to whole module with 10 tubes at solar ir-
radiance of 700 W/m2.
Simulation Case TW , out (K ) TPV (K ) p (Pa) qinc,PV (W /m2) qout, PV (W /m2)
Fig. 5. The temperature profile of PV cells at three positions, for a selected
One-tube model 315.67 320.25 758.06 552.89 43.53
section having one tube, in the absence and presence of the greenhouse effect
Whole module 315.67 320.24 758.08 552.85 43.50
under the solar irradiance of 700 W/m2. X and Y show the transverse and water
flow directions, respectively, (see Fig. 6). (The minimum temperature coincides
with the location of the tube).
the PV temperature immediately above the tubes and along the flow
direction. It can be seen that the temperature distribution of the PV
Table 11 layer is transversely periodic — with a small deviation due to the
Importance of the greenhouse effect consideration in the PV/T simulation at thermally insulating walls. This suggests simulating a section of the PV/
different solar irradiance. T module having only one tube instead of whole system to reduce the
Gsun status TPV (K ) qloss, tot th (%) elec (%) computational cost. In other words, this one-tube section is assumed to
(W/m2 (W/m2 be a part of a module that is infinite in the transverse direction (see
Fig. 7). The results of this one-tube simulation are compared with the
300 with green house 314.60 85.93 52.86 11.01
whole PV/T simulation (having 10 tubes) in Table 12.
without green house 313.82 123.35 40.92 11.05
700 with green house 320.26 160.06 61.42 10.71 A 3D one-tube PV/T section with periodic boundary conditions can
without green house 319.31 207.37 54.84 10.76 accurately represent a PV/T module. This finding is very significant as
1100 with green house 325.72 234.44 63.95 10.42 the required computational time for the 3D one-tube PV/T simulation is
without green house 324.59 291.92 58.79 10.48
nearly 2 h, while this time is around 13 h for the 3D simulation of the
whole module—using a desktop computer with an Intel Core (TM) i7-
7700HQ CPU and 16 GB RAM.
On the other hand, Fig. 6 shows maximum temperature differences
of 4 K and 12 K along the X (i.e., transverse) and Z (i.e., water flow)
directions, respectively. This temperature gradient may cause serious
thermal stresses and lifetime reduction for the module as well as re-
ducing the effectiveness of maximum power point tracking. Therefore,
the following sections explore means to reduce the temperature gra-
dient of the PV plate. In addition, reducing the PV plate temperature to
achieve higher electrical efficiency will be investigated.

4.3. Effect of number of tubes

As mentioned in the previous section, the heat transfer from the PV

layer is greater above the tubes, causing periodic variations in the
temperature profile in the X direction. Thus, it is expected that by in-
creasing the number of tubes (N), the temperature gradient in the X
direction of the PV plate reduces. For this purpose, N = 5, 10, 16, 20,
and 25 were investigated. Fig. 8(a–e) shows the temperature distribu-
tion across the width of the PV plate at three different sections (Z = 0.7,
1.1, and 1.5 m). The total mass flow rate is constant in all cases. It is
evident from these figures that increasing the number of tubes makes
the transverse temperature distribution of PV plate more uniform. Be-
sides, Table 13 shows the temperature rise of water
( Tw = TW , out TW , in ), total water pressure drop ( p), maximum tem-
perature difference of the PV module ( TPV , max ) , thermal and electrical
Fig. 6. (a) Temperature contour on the PV cells and (b) temperature distribu- efficiencies, and total bonding and tube mass (Mt). By increasing the
tion of the PV cells at three different positions. number of tubes from 5 to 25, TPV , max decreases by almost 74%;
however, the major reduction is experienced by increasing the number
module at different positions from the inlet. It is obvious that the of tubes from 5 to 10. This leads to a reduction in average temperature
temperature distribution across the PV layer is significantly non-uni- of the module, which in turn, increases the electrical efficiency (see Eqs.
form, indicating the importance of 3D modeling of PV/T modules and (6) and (19)). In addition, for a given total water mass flow rate, in-
weakness of 1D models. This non-uniformity of temperature is caused creasing N reduces the flow rate through each tube which reduces the
by the cooling fluid flowing through the tubes, leading to reduction of total pump power and increases the electrical efficiency (see Eqs. (17)
and (19)).

11
S.R. Maadi, et al. Energy Conversion and Management 198 (2019) 111781

Fig. 8. Variations of the transverse temperature distribution of the PV module for different number of tubes of (a) N = 5, (b) N = 10, (c) N = 16, (d) N = 20, and (e)
N = 25 at different positions.

Table 13
Effect of number of tubes on the performance of PV/T module.
N Re Tw (K) p (Pa) TPV ,max (K) th (%) elec (%) Mt (kg)

5 2130 4.4 828 23.44 57.72 10.06 3.065

10 (base case) 1065 4.67 758 10 61.42 10.71 6.130
16 666 5.26 729 7.2 69.1 10.79 9.808
20 532 5.31 718 6.48 69.83 10.83 12.260
25 426 5.36 709 6.06 70.54 10.86 15.325

Table 14
Effect of tube diameter on the performance of PV/T module.
Inner tube diameter Re TPV ,max (K) TPV ,x (Z = 1.9) th (%) elec (%)

0.4 D 2160 10.85 3.805 60.42 10.66

0.6 D 1440 10.67 3.71 61.01 10.69
0.8 D (base-case) 1065 10 3.69 61.42 10.71
Fig. 9. Schematic of the bonding between absorber and tube.

12
S.R. Maadi, et al. Energy Conversion and Management 198 (2019) 111781

transfer area between them increases and the thermal resistant reduces.
Thus, increasing L2 / L1 lowers the average temperature and makes the
temperature of the module more uniform. This achievement is very
important in terms of thermal stresses. The results show that, for
L2 / L1 > 0.6 , the changes in temperature distribution are not significant.
Table 15 shows that an increase in L2 / L1 reduces TPV , max , which in turn
gradually increases both thermal and electrical efficiencies. Never-
theless, a larger bonding width also increases significantly the mass of
the module.
According to the results of Sections 4.2, 4.3, and 4.4, for a given
total mass flow rate, the tube diameter has not a significant effect on the
performance and temperature distribution. Increasing N and bonding
Fig. 10. Variations of the temperature distribution in the width of the PV plate width have a significant effect on the temperature distribution of the PV
with the width of bonding at Z = 1.1 m. cells, which is important in terms of thermal stresses and life span.
These parameters also affect the electrical and thermal efficiencies of
Table 15 the PV/T module. However, increasing the module mass is one of the
Effect of bonding width on the performance of PV/T module. disadvantages of these methods.

L2 /L1 Tw (K) TPV ,max (K) th (%) elec (%) Mt (kg)

5. Conclusions
0.00952 (base case) 4.67 10 61.42 10.71 6.130
0.2 4.72 9.6 62.03 10.77 14.270 Employing PV/T technology not only increases the electrical effi-
0.4 4.79 7.9 62.96 10.82 30.540 ciency of a typical PV module but also is a smart way of using waste
0.6 4.82 6.9 63.35 10.85 46.810
energy – i.e., Heat. In this study, a 3D coupled thermal-optical model
0.8 4.84 6.04 63.54 10.86 63.080
1 4.86 5.9 63.86 10.87 83.460 using computational fluid dynamics and ray tracing was developed to
evaluate the performance of a glazed PV/T module. To improve the
accuracy of the model, optical and thermal losses, and the greenhouse
With an increase in N, the area of the contact bond to the absorber effect in the air gap between the glass cover and PV cell were con-
(see Fig. 2) and in turn, the rate of heat transfer to the water increases. sidered. Besides, all layers of the glazed PV/T module including pc-Si,
However, for a given total mass flow rate, a PV/T with more tubes EVA, ARC, and Tedlar, as well as other parts of PV/T module like glass
means lower mass flow rate in each tube, which can reduce the amount cover, air gap, absorber plate, and tube, along with all corresponded
of heat absorbed by each tube. These opposite behaviours lead to a optical and thermal properties were simulated. A validation study was
considerable increase in thermal efficiency from 10 to 16 tube-PV/T performed on the proposed thermal-optical model, where a reasonable
compared to larger numbers of tubes. Moreover, the results show that agreement was achieved. Key findings of this study are summarized as
the total thermal losses decrease by adding more tubes. For instance, follows:
the total thermal losses decrease 15.8%, 29.4%, 31.8%, 34% for a
collector with 10, 16, 20 and 25 tubes compared to a collector with 5 • The greenhouse effect in the air gap beneath the glass cover has a
tubes, respectively. On the other hand, Table 13 shows that increasing significant effect on thermal efficiency, especially at lower solar ir-
N causes a noticeable augmentation in the mass of PV/T module, which radiance. For instance, at solar irradiance of 300 W/m2 the green-
in turn can increase the cost of the system. Consequently, 16 would be a house effect increases the thermal efficiency by 12% compared to
suitable number of tubes for this PV/T module. the case where it is ignored. However, it does not have a consider-
able effect on the electrical efficiency, although a very slight re-
duction in the electrical efficiency was observed in the presence of
4.4. Effect of tube diameter
gthe reenhouse effect.
The effects of tube inner diameter on TPV , max , maximum transverse • For the PV/T module with straight tubes, an accurate and compu-
tationally cost-efficient 3D one-tube section using periodic boundary
temperature difference of the PV module ( TPV , x ) and performance of
conditions was shown to simulate the complete PV/T module ac-
the module are presented in Table 14. For this study, the outer diameter
curately.
•
of tube was kept constant at the outer diameter of tube in base case (D),
For a given total mass flow rate, there is an optimum number of
and the total mass flow rate is constant. Then, the tube thickness
tubes for the performance of the PV/T module. By increasing the
changes with variations of inner diameter. It is observed that increasing
number of tubes from 5 to 10, 16, 20 and 25, the overall thermal-
the tube diameter reduces TPV , max . It makes the temperature dis-
optical losses decrease by 15.8%, 29.4%, 31.8% and 34%, respec-
tribution of the module slightly more uniform. Consequently, for a
tively.
•
given total mass flow rate, the efficiencies of PV/T system gradually
It is found that increasing the number of tubes and bonding contact
increase with increasing the inner diameter.
area result in a considerable reduction in the temperature gradient
across the module, at the expense of additional mass.
4.5. Effect of bonding width
Declaration of Competing Interest
The effect of contact area between absorber and tubes is in-
vestigated in this section, while the other specifications in Tables 2–4 None.
remain unchanged. For the base case, the width of bonding is equiva-
lent to the tube outer diameter (D). The transverse variation of the PV References
module temperature at Z = 1.1 m is presented in Fig. 10 at different
bonding width ratio (L2/L1 as illustrated in Fig. 9). L2 / L1 = 0.0952 and [1] Commission EE. GREEN PAPER-a 2030 framework for climate and energy policies.
1 represent the base case and the case in which bonding covers the COM 2013;2013:169.
[2] Comello S, Reichelstein S, Sahoo A. The road ahead for solar PV power. Renew
entire bottom of the absorber layer, respectively. With an increase in Sustain Energy Rev 2018;92:744–56.
the contact area of the tube bond with the absorber plate, the heat

13
S.R. Maadi, et al. Energy Conversion and Management 198 (2019) 111781

[3] International Energy Agency (IEA). Technology Roadmap - Solar Photovoltaic [33] Aly SP, Ahzi S, Barth N, Abdallah A. Using energy balance method to study the
Energy, 2014. thermal behavior of PV panels under time-varying field conditions. Energy Convers
[4] de Wild-Scholten MM. Energy payback time and carbon footprint of commercial Manage 2018;175:246–62.
photovoltaic systems. Sol Energy Mater Sol Cells 2013;119:296–305. [34] Sahli M, Correia J, Ahzi S, Touchal S. Multi-physics modeling and simulation of heat
[5] Idoko L, Anaya-Lara O, McDonald A. Enhancing PV modules efficiency and power and electrical yield generation in photovoltaics. Sol Energy Mater Sol Cells
output using multi-concept cooling technique. Energy Rep 2018;4:357–69. 2018;180:358–72.
[6] Kalogirou SA, Tripanagnostopoulos Y. Hybrid PV/T solar systems for domestic hot [35] Jia Y, Alva G, Fang G. Development and applications of photovoltaic–thermal
water and electricity production. Energy Convers Manage 2006;47:3368–82. systems: a review. Renew Sustain Energy Rev 2019;102:249–65.
[7] Kolahan A, Maadi S, Passandideh Fard M, Sardarabadi M. Numerical and experi- [36] Abdolzadeh M, Zarei T. Optical and thermal modeling of a photovoltaic module and
mental investigations on the effect of adding nanoparticles on entropy generation in experimental evaluation of the modeling performance. Environ Prog Sustainable
PVT systems. 17th Conference On Fluid Dynamics. 2017. Energy 2017;36:277–93.
[8] Das D, Kalita P, Roy O. Flat plate hybrid photovoltaic-thermal (PV/T) system: a [37] Cox III C, Raghuraman P. Design considerations for flat-plate-photovoltaic/thermal
review on design and development. Renew Sustain Energy Rev 2018;84:111–30. collectors. Sol Energy 1985;35:227–41.
[9] Improving Energy Performance in Canada. Report to Parliament under the Energy [38] Moghimi M, Craig K, Meyer JP. A novel computational approach to combine the
Efficiency Act (2015-2016) Natural Resources Canada pp80, 2016. optical and thermal modelling of Linear Fresnel Collectors using the finite volume
[10] UK Department for Business EIS. Evidence Gathering – Low Carbon Heating method. Sol Energy 2015;116:407–27.
Technologies. London: Hybrid Solar Photovoltaic Thermal Panels; 2016. [39] Gholamalizadeh E, Kim M-H. CFD (computational fluid dynamics) analysis of a
[11] Mellor A, Alvarez DA, Guarracino I, Ramos A, Lacasta AR, Llin LF, et al. Roadmap solar-chimney power plant with inclined collector roof. Energy 2016;107:661–7.
for the next-generation of hybrid photovoltaic-thermal solar energy collectors. Sol [40] Essa MA, Mostafa NH. Theoretical and experimental study for temperature dis-
Energy 2018;174:386–98. tribution and flow profile in all water evacuated tube solar collector considering
[12] Chauhan A, Tyagi V, Anand S. Futuristic approach for thermal management in solar solar radiation boundary condition. Sol Energy 2017;142:267–77.
PV/thermal systems with possible applications. Energy Convers Manage [41] Lu ZH, Yao Q. Energy analysis of silicon solar cell modules based on an optical
2018;163:314–54. model for arbitrary layers. Solar Energy 2007;81:636–47.
[13] Yazdanifard F, Ameri M, Ebrahimnia-Bajestan E. Performance of nanofluid-based [42] Bhattarai S, Oh J-H, Euh S-H, Kafle GK, Kim DH. Simulation and model validation of
photovoltaic/thermal systems: a review. Renew Sustain Energy Rev sheet and tube type photovoltaic thermal solar system and conventional solar col-
2017;76:323–52. lecting system in transient states. Sol Energy Mater Sol Cells 2012;103:184–93.
[14] Yuan W, Ji J, Li Z, Zhou F, Ren X, Zhao X, et al. Comparison study of the perfor- [43] Khanjari Y, Pourfayaz F, Kasaeian AB. Numerical investigation on using of nano-
mance of two kinds of photovoltaic/thermal (PV/T) systems and a PV module at fluid in a water-cooled photovoltaic thermal system. Energy Convers Manage
high ambient temperature. Energy 2018;148:1153–61. 2016;122:263–78.
[15] Al-Waeli AH, Chaichan MT, Sopian K, Kazem HA, Mahood HB, Khadom AA. [44] Michael JJ, Selvarasan I, Goic R. Fabrication, experimental study and testing of a
Modeling and experimental validation of a PVT system using nanofluid coolant and novel photovoltaic module for photovoltaic thermal applications. Renewable
nano-PCM. Sol Energy 2019;177:178–91. Energy 2016;90:95–104.
[16] Yazdanifard F, Ebrahimnia-Bajestan E, Ameri M. Investigating the performance of a [45] Krauter S, Hanitsch R. Actual optical and thermal performance of PV-modules. Sol
water-based photovoltaic/thermal (PV/T) collector in laminar and turbulent flow Energy Mater Sol Cells 1996;41–42:557–74.
regime. Renewable Energy 2016;99:295–306. [46] Duffie JA, Beckman WA. Solar Engineering of Thermal Processes. John Wiley &
[17] Vera JT, Laukkanen T, Sirén K. Performance evaluation and multi-objective opti- Sons; 2013.
mization of hybrid photovoltaic–thermal collectors. Sol Energy 2014;102:223–33. [47] Guarracino Ilaria, Mellor Alexander, Ekins-Daukes Nicholas J, Markides Christos N.
[18] Kuo C-FJ, Lee Y-W, Umar ML, Yang P-C. Dynamic modeling, practical verification Dynamic coupled thermal-and-electrical modelling of sheet-and-tube hybrid pho-
and energy benefit analysis of a photovoltaic and thermal composite module tovoltaic/thermal (PVT) collectors. Appl Therm Eng 2016;101:778–95.
system. Energy Convers Manage 2017;154:470–81. [48] Bergman TL, Incropera FP, DeWitt DP, Lavine AS. Fundamentals of Heat and Mass
[19] Duffie JA, Beckman WA, Worek W. Solar engineering of Thermal Processes. Wiley Transfer. John Wiley & Sons; 2011.
Online Library; 2013. [49] White FM. White Frank M, editor. Fluid Mechanics. McGraw-hill; 2003.
[20] Anderson TN, Duke M, Morrison G, Carson JK. Performance of a building integrated [50] Ebrahimnia-Bajestan E, Moghadam MC, Niazmand H, Daungthongsuk W,
photovoltaic/thermal (BIPVT) solar collector. Sol Energy 2009;83:445–55. Wongwises S. Experimental and numerical investigation of nanofluids heat transfer
[21] Spertino F, D’angola A, Enescu D, Di Leo P, Fracastoro GV, Zaffina R. characteristics for application in solar heat exchangers. Int J Heat Mass Transf
Thermal–electrical model for energy estimation of a water cooled photovoltaic 2016;92:1041–52.
module. Sol Energy 2016;133:119–40. [51] Yunus AC. Heat Transfer: A Practical Approach. New York: MacGraw Hill; 2003.
[22] Nasrin R, Rahim NA, Fayaz H, Hasanuzzaman M. Water/MWCNT nanofluid based [52] Abdallah SR, Saidani-Scott H, Abdellatif OE. Performance analysis for hybrid PV/T
cooling system of PVT: experimental and numerical research. Renewable Energy system using low concentration MWCNT (water-based) nanofluid. Sol Energy
2018;121:286–300. 2019;181:108–15.
[23] Zondag HA, de Vries Dd, Van Helden W, van Zolingen RC, Van Steenhoven A. The [53] Nahar A, Hasanuzzaman M, Rahim N. Numerical and experimental investigation on
thermal and electrical yield of a PV-thermal collector. Sol Energy 2002;72:113–28. the performance of a photovoltaic thermal collector with parallel plate flow channel
[24] Modjinou M, Ji J, Yuan W, Zhou F, Holliday S, Waqas A, et al. Performance com- under different operating conditions in Malaysia. Sol Energy 2017;144:517–28.
parison of encapsulated PCM PV/T, microchannel heat pipe PV/T and conventional [54] Mousavi S, Kasaeian A, Shafii MB, Jahangir MH. Numerical investigation of the
PV/T systems. Energy 2019;166:1249–66. effects of a copper foam filled with phase change materials in a water-cooled
[25] Maadi SR, Kolahan A, Passandideh-Fard M, Sardarabadi M, Moloudi R. photovoltaic/thermal system. Energy Convers Manage 2018;163:187–95.
Characterization of PVT systems equipped with nanofluids-based collector from [55] Soriga I, Badescu V. Thermal inertia of flat-plate solar collectors in different ra-
entropy generation. Energy Convers Manage 2017;150:515–31. diative regimes. Energy Convers Manage 2016;111:27–37.
[26] Siddiqui MU, Arif AF, Kelley L, Dubowsky S. Three-dimensional thermal modeling [56] Skoplaki E, Palyvos J. On the temperature dependence of photovoltaic module
of a photovoltaic module under varying conditions. Sol Energy 2012;86:2620–31. electrical performance: a review of efficiency/power correlations. Sol Energy
[27] Pierrick H, Christophe M, Leon G, Patrick D. Dynamic numerical model of a high 2009;83:614–24.
efficiency PV–T collector integrated into a domestic hot water system. Sol Energy [57] Modest MF. Radiative Heat Transfer. Academic press; 2013.
2015;111:68–81. [58] Moghimi M, Craig K, Meyer JP. Optimization of a trapezoidal cavity absorber for
[28] Guarracino I, Mellor A, Ekins-Daukes NJ, Markides CN. Dynamic coupled thermal- the Linear Fresnel Reflector. Sol Energy 2015;119:343–61.
and-electrical modelling of sheet-and-tube hybrid photovoltaic/thermal (PVT) [59] Swinbank WC. Long-wave radiation from clear skies. Q J R Meteorolog Soc
collectors. Appl Therm Eng 2016;101:778–95. 1963;89:339–48.
[29] Hosseinzadeh Mohammad, Salari A, Sardarabadi M, Passandideh-Fard AM. [60] Yazdanpanahi J, Sarhaddi F, Adeli MM. Experimental investigation of exergy effi-
Optimization and parametric analysis of a nanofluid based photovoltaic thermal ciency of a solar photovoltaic thermal (PVT) water collector based on exergy losses.
system: 3D numerical model with experimental validation. Energy Convers Manage Sol Energy 2015;118:197–208.
2018;160:93–108. [61] Qiu Z, Zhao X, Li P, Zhang X, Ali S, Tan J. Theoretical investigation of the energy
[30] Kazemian A, Salari A, Hakkaki-Fard A, Ma T. Numerical investigation and para- performance of a novel MPCM (Microencapsulated Phase Change Material) slurry
metric analysis of a photovoltaic thermal system integrated with phase change based PV/T module. Energy 2015;87:686–98.
material. Appl Energy 2019;238:734–46. [62] Patankar S. Numerical Heat Transfer and Fluid Flow. CRC Press; 1980.
[31] Fayaz H, Rahim NA, Hasanuzzaman M, Rivai A, Nasrin R. Numerical and outdoor [63] Peyret R. Handbook of Computational Fluid Mechanics. Elsevier; 1996.
real time experimental investigation of performance of PCM based PVT system. Sol [64] Hoffmann KA, Chiang ST. Computational fluid dynamics volume I. Engineering
Energy 2019;179:135–50. Education System. 2000.
[32] Sakellariou E, Axaopoulos P. An experimentally validated, transient model for sheet [65] Versteeg HK, Malalasekera W. An Introduction to Computational Fluid Dynamics:
and tube PVT collector. Sol Energy 2018;174:709–18. The Finite Volume Method. Pearson education; 2007.

14