You are on page 1of 48

Journal Pre-proofs

Performance, Limits, and Thermal Stress Analysis of High Concentrator Mul-


tijunction Solar Cell under Passive Cooling Conditions

Essam M. Abo-Zahhad, Shinichi Ookawara, Ali Radwan, A.H. El-Shazly,


M.F. El-Kady, Mohamed F.C. Esmail

PII: S1359-4311(19)30741-0
DOI: https://doi.org/10.1016/j.applthermaleng.2019.114497
Reference: ATE 114497

To appear in: Applied Thermal Engineering

Received Date: 31 January 2019


Revised Date: 2 October 2019
Accepted Date: 4 October 2019

Please cite this article as: E.M. Abo-Zahhad, S. Ookawara, A. Radwan, A.H. El-Shazly, M.F. El-Kady, M.F.C.
Esmail, Performance, Limits, and Thermal Stress Analysis of High Concentrator Multijunction Solar Cell under
Passive Cooling Conditions, Applied Thermal Engineering (2019), doi: https://doi.org/10.1016/j.applthermaleng.
2019.114497

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version
will undergo additional copyediting, typesetting and review before it is published in its final form, but we are
providing this version to give early visibility of the article. Please note that, during the production process, errors
may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

© 2019 Published by Elsevier Ltd.


Performance, Limits, and Thermal Stress Analysis of High Concentrator
Multijunction Solar Cell under Passive Cooling Conditions

Essam M. Abo-Zahhad1,2,3,4, Shinichi Ookawara4, Ali Radwan5, A. H. El-Shazly1, M. F.


El-Kady1,6, and Mohamed F. C. Esmail2
1Chemicals and Petrochemicals Engineering Department, Egypt-Japan University of Science and
Technology, New Borg El-Arab City, Alexandria, Egypt
2Mechanical Power Engineering Department, Faculty of Energy Engineering, Aswan University,

Aswan, Egypt
3Department of Applied Mathematics to the Aerospace Engineering, School of Aerospace

Engineering, Universidad Politecnica de Madrid (UPM), Plaza Cardenal Cisneros 3,


28040 Madrid, Spain
4Department of Chemical Science and Engineering, Tokyo Institute of Technology, Tokyo 152-

8552, Japan
5Mechanical Power Engineering Department, Mansoura University, El-Mansoura 35516, Egypt
6Fabrication Technology Department, Advanced Technology and New Materials and Research

Institute (ATNMRI), the City of Scientific Research and Technological Applications,


Alexandria, Egypt

Abstract
Concentration of solar radiation onto the surface of triple-junction solar cells causes high
cell temperature and system failure. Recently, several cooling methods were proposed for these
systems. However, quantitative evaluation of the essential heat transfer coefficients to maintain
stable operation of these systems at different meteorological and operating conditions is not found
in the literature. Therefore, in this study, a comprehensive three-dimensional coupled thermal and
structural model is proposed for the latest triple-junction AZUR SPACE solar cell. The model is
used to investigate the performance of an HCPV system under different solar concentration ratios
(CRs), ambient temperature, direct solar irradiance, wind speed, backside heat transfer coefficient,
and copper-II substrate area ratios. In addition, a new structure of the solar cell is proposed by
modifying the typical solar cell assembly by changing the area of the rear copper layer. The results
indicate that by increasing the ambient temperature, CR and direct solar irradiance significantly
increase the predicted cell temperature at the same backside heat transfer coefficient. In addition,
increasing copper-II substrate area ratios significantly reduces the average cell temperature at the
same backside heat transfer coefficient and CR. At the highest backside heat transfer coefficient,
when the copper-II substrate area increased, the cell temperature decreased to a certain limit and
subsequently remained constant. Critical values of the highest backside heat transfer coefficient

1
were about 200, 600, 1000, and 1600 W/m2 K at CRs of 50, 500, 1000, and 1500 Suns,
respectively. In addition, at the highest backside heat transfer coefficient of 1600 W/m2 K, the
critical area ratio values were about 2, 3, 4, and 6 at CRs of 50, 500, 1000, and 1500 Suns,
respectively.

Keywords; Triple-junction solar cell, Passive cooling, Thermal stress, Concentrator photovoltaic

1. Introduction
Rapid evolution of mechanization and industrialization technologies lead to significant
growth in global energy consumption every year [1]. According to projections of the Annual
Energy Outlook between 2017 and 2050, energy consumption increases by about 0.4 % per year
[2]. Furthermore, the World Energy Council reported that the limited and unsustainably finite
fossil fuels provide about 86 % of the worldwide energy requirements [3]. Moreover, burning of
traditional fossil fuels for electricity, transportation, and heating is normally associated with
massive amounts of emission, which is the largest source of greenhouse gases [4]. Utilization of
fossil fuels for most of the human activities has increased greenhouse gas emissions over the last
150 years, causing an overall increase in earth's atmospheric temperature. Hence, the world is in
need of constant, reliable, sustainable, and clean energy resources. Solar energy is considered as
one of the most committed sources of clean, unlimited, and eco-friendly energy [5]. The earth
receives and absorbs about 3.85 million exajoules of energy from the sun annually [6,7].

Photovoltaic (PV) technology is the most attractive and widespread solar energy
technology, which can convert solar energy into electricity directly. However, solar energy
utilization for power generation is expensive. In high concentrator photovoltaic (HCPV) systems,
the area of expensive semiconductor solar cell materials is replaced by much cheaper concentrating
systems such as lenses. In addition, HCPV systems commonly consist of multijunction (MJ) solar
cells and optical components to focus solar irradiance onto these solar cells. HCPV systems can
work efficiently at concentration ratios (CRs) of more than 3000 Suns [8]. MJ cells are projected
to yield more electrical energy per unit area of cell than the conventional polycrystalline solar
cells. Recently, the electrical efficiency of these solar cells reached 46 % at CR = 500 Suns, and it
is expected to reach 50 % by the next decade.

2
Despite the enormous progress in MJ solar cells for compact HCPV arrangements, a major
portion of the received solar irradiance is converted into heat [9]. The conversion of solar energy
to heat within the cell body, especially under high CR values, causes remarkably high cell
temperatures, which significantly decrease cell efficiency and cause system failure. Nishioka et al.
[10] introduced a relation between triple-junction conversion efficiency and cell temperature
coefficient. They mentioned that the solar cell conversion efficiency decreased by 0.248 % per
degree increase in the cell temperature (Tcell) at CR = 1 Sun and by 0.098 % at CR = 200 Suns.
Open circuit voltage (Voc) of the cell decreased by 0.135 % per degree at CR = 500 Suns according
to the latest version of AZUR SPACE product data sheet [11]. The high CR value decreases the
proportion of Voc drop as the cell temperature increases [12].

Therefore, the design of a reliable and simple cooling system that offers low and uniform
cell temperature is desirable. The cooling system should be developed to withstand worst-case
scenarios with sufficient capacity and minimal power consumption [13]. Under CR of 500 Suns
with no applied cooling, the MJ cell temperature theoretically can exceed 1000 ℃ [14]. Mi et al.
[15] numerically predicted the temperature of a 3 × 3 mm2 uncooled MJ cell with CR = 400 Suns,
and they stated that the cell temperature increased to 1200 °C. Therefore, incorporating an effective
cooling technique for HCPV systems is a vital requirement. The cooling technology selection is
subject to wide range parameters such as available area for cooling, coolant flow rate, and heat
transfer coefficient [5]. Jakhar et al. [16] concluded that the cooling heat sink design must be
functional, flexible, and practically efficient during operation.

Cooling systems are commonly classified as passive and active. According to the literature,
many comprehensive studies have directly considered both passive and active cooling for HCPV
cooling. Recently, it was reported that passive cooling for a linear Fresnel lens concentrator was
insufficient for CR > 20 Suns. On the other hand, many reports [17] proved the ability of passive
cooling to handle heat of the HCPV arrangements effectively. Minano et al. [18] thermally
modeled passive cooling for a single cell under different high CR values. They reported that
passive cooling was more effective in maintaining Tcell at accepted levels, if the cell size was
reduced to less than 5 mm in diameter. Wang et al. [19] performed a series of computational fluid
dynamic simulations for passive cooling of HCPV solar cell modules under various weather
conditions. Their results showed that heat dissipation from the HCPV module highly depends on

3
wind velocity when it is less than 1 m/s. Furthermore, the cell maximum temperature had a linear
relationship with direct normal irradiance (DNI) and ambient temperature (Ta). To predict the
practical limit of passive cooling for HCPV modules, Gualdi et al. [20] developed a 3D thermal
model. They concluded that flat heat sink systems were suitable to maintain cell temperatures
below 80 °C under CR of a few thousand suns. Micheli et al. [21] investigated the performance of
plate micro-fins in natural convection conditions and found that micro-fins were able to enhance
the mass specific power by 50 % compared to a flat surface. Renzi et al. [22] experimentally
investigated the outdoor performance of solar cell under CR = 476 Suns. It was found that
aluminum flat plate temperatures ranged between 55 and 65 °C; however, no information on the
cell temperature was mentioned. Arak et al. [23] discussed a simple heat spreader for a single
concentrator cell and demonstrated that MJ cell efficiency loss due to cell temperature rise was 2
% without the support of heat sinks. Wang et al. [24] proposed several designs of passive and
active cooling approaches for HCPV systems. For passive cooling, they combined extended heat
pipes with annular or radial fins and concluded that passive cooling for HCPV arrangements is
promising for a heat dissipation requirement of less than 50 W (13.0 W/cm2). Chou et al. [25]
utilized the finite elements method to thermally model the performance of HCPV cooled with an
aluminum plate. They reported that the cell temperature was about 69 °C for a CR of 380 Suns.
Furthermore, Royne et al. [26] reported that when enough area is available underneath an HCPV
module for heat sink installation, passive cooling may be sufficient for a single MJ cell with CR
greater than 1000 Suns. Theristis et al. [27] concluded that a thermal resistance lower than 1.4
K/W is required to dissipate heat with passive cooling from a single HCPV cell (area = 1 cm2)
under a high CR of 500 Suns, especially at high Ta (higher than 40 °C).

In previous investigations, no quantitative estimation of the critical backside heat transfer


coefficient was performed to sustain a safe and efficient operation mode for the latest HCPV
systems under different operating and meteorological conditions. This parameter is essential
because it may be accomplished by using a passive cooling method. In this case, several merits
could be accomplished. Particularly, by using a light-weight passive heat sink, minimization of
consumed energy by tracking could be achieved [28]. In addition, the cooling system complexity
could be avoided. Therefore, compromise between the size and performance of a heat sink must
be comprehensively understood to limit the load on tracking systems, and simultaneously enhance
the electrical efficiency of HCPV systems.

4
However, in view of the HCPV complex structure, the dissimilar thermal expansion
constants of the board material and cell layer in an HCPV leads to thermal stress. Renno et al. [29]
concluded that a radical declination in the performance of HCPV systems occurs due to durable
thermal stress. On the other hand, when the cell operation temperature exceeds the permitted limit,
the formation of hot spots usually occurs. The heat flux at a hot spots can exceed 137 W/cm2,
which physically decline the PV cell lifespan due to thermal fatigue [26]. Many studies focused
on thermal stresses developed during thermal management processes for polycrystalline [30] and
organic PV cells [31]. However, very limited existing studies considered the structural analyses
for MJ cells. Cao et al. [32] provided outlines of the structural analyses of HCPV modules. They
considered extreme and nominal working situations of HCPV systems without considering the
cooling scheme.

Based on our literature survey, in the present research work, we aim to provide a complete
understanding of the essential cooling requirements for the latest HCPV systems under different
operating and meteorological conditions. Accordingly, the originality of the present study is based
on three key concepts. First, the critical convective heat transfer coefficients essential to keep the
latest concentrator triple-junction solar cell safely and effectively operational at different CR is
quantitatively estimated. Second, a new HCPV cell assembly is proposed. Finally, computational
coupling between thermal and structural analysis of the HCPV system under different operating
scenarios is studied. To cover these research gaps, a comprehensive three-dimensional (3D)
thermal model coupled with a thermomechanical model for HCPV was developed. The model was
used to study the performance of the most recent solar cell product, AZUR SPACE [11]. The effect
of different solar CRs, ambient temperature, direct solar irradiance, wind speed, backside heat
transfer coefficient, and copper-II substrate area on the cell temperature, cell temperature
uniformity, and thermal stress were computationally evaluated. The requirements to avoid thermal
degradation and maintain effective solar cell operation below the maximum operating temperature
were also evaluated.

2. Physical model
Physical assembly of the proposed HCPV system adopted in the current work is shown in
Fig. 1. High CR was attained by using an array of Fresnel lenses. As shown in the figure, Fresnel
lens was employed to focus solar light onto small areas of the MJ cell. The used triple-junction

5
cell has an area of 1 cm2 and was constructed on a direct bonded copper (DBC) board for current
handling and thermal management. The latest version of an AZUR SPACE typical
GaInP/GaInAs/Ge solar cell product is modeled in the present work. The cell manufacturer
mentioned that the recommended operating temperature ranges from 25 to 80 °C, and the extreme
working temperature (Tallowable ) must be less than 110 °C [33]. The used cell material comprised
of GaInP and GaInAs, which are responsible for electricity generation, on a Ge substrate.
Generally, for concentrator modules, the cell is attached with a board of ceramic and DBC
substrates. The investigated solar cell and board assembly sizes are presented in Table 1.

In the AZUR SPACE solar cell, the copper-II layer has similar dimensions as the copper-I
layer. To enhance the heat dissipation rate from the Ge layer, the area of copper-II, which was
subjected to back cooling, was increased. A new parameter called area ratio, Ar, which is the ratio
of the modified copper-II area to the area of conventional copper-II was defined. In the present
work, Ar was increased to 20. The proposed idea is very convenient in HCPV systems that use
point focus Fresnel lenses as concentrators. The solar cell has a small area and there is a large
space available for extending the copper-II area. In addition, using a thin and high conductive
material such as copper could enhance the system performance with little consumption of power
in tracking systems.

6
Tracking
Axis-2

Tracking Solar
Axis -1 irradiance
Fresnel
W Lens
N
S E

Concentrated
solar radiation

MJ Solar
cell
Germanium Ceramic

Copper I

Copper II

Fig. 1. Schematic representation of the proposed HCPV arrangement.

Table 1. Solar cell structure dimension values


Solar cell layer Thickness (mm) Length (mm) Width (mm)
Ge 0.19 10 10
Copper-I 0.25 24 19.5
Al2O3 ceramic 0.32 25.5 21
Copper-II 0.25 25 20.5

3. Numerical analysis
In this section, a detailed description of the thermal and structural numerical model is
presented. In Section 3.1, the thermal model governing equations along with the thermal boundary
conditions are presented. However, Section 3.2 presents detailed structural model governing
equations and the structural analysis boundary conditions.

7
3.1. HCPV system thermal modeling
The present model was developed to predict the performance of the HCPV system with 3D
temperature distribution of the solar cell under different operating scenarios. This 3D model is the
most appropriate for hot spot location assessment. In an HCPV, the cell gains energy by absorption
of solar irradiance. Generally, the absorbed solar radiation is divided into three main parts. The
first part is converted into electricity, which mainly depends on the electrical efficiency of the solar
cell because it is a function of Tcell and net concentrated solar irradiance. In the second part, the
foremost portion of the received solar energy is transformed into heat, which is dissipated through
the cooling heat sink. The maximum and minimum convective heat transfer coefficients offered
by the passive cooling heat sink employed in the current work were selected according to [27].
The last and smallest part of the solar radiation dissipates from the cell by combined convective
and radiative heat losses to the surrounding. This share is dependent on weather conditions such
as wind velocity, ambient temperature, top surface emissivity, and sky radiation temperature [34].

In the current developed 3D model, the following assumptions were considered:

1. The cell was modeled as one entity of Ge layer. The heat absorbed by this layer is used as
a heat source in the energy equation as studied by Chou et al. [25,27]. The typical material
properties of the HCPV are given in Table 2.
2. All the HCPV sides were assumed adiabatic because the thickness of each layer was very
small compared to the surface area.
3. Thermo-physical properties of the solid materials were assumed to be isotropic and
temperature independent.
4. The cell was not covered with glass or any other material [11].

8
Table 2. The thermo and optical properties of the current cell materials [33]
Layers
Property
Ge Cu Al2O3-Ceramic
Thermal conductivity, k, (W/m K) 60 400 30
Specific heat capacity, C, (J/kg K) 320 385 900
Density, ρ, (kg/m3) 5323 8700 3900
Emissivity (ε) 0.9 0.05 0.75

Considering multiple layers of the HCPV cell structure, the vector form of the heat
conduction for these layers can be represented as follows [35]:

∇.(𝑘𝑖 ∇𝑇𝑖) + 𝑞𝑖 = 0 (1)

where ki and Ti are the thermal conductivity and temperature of material i, respectively.
The heat generation source term q was added in the previous equation to consider solar irradiance
absorption in the layer. Therefore, the energy equation for the Ge layer is as follows:
∇.(𝑘𝐺𝑒 ∇𝑇𝐺𝑒) + 𝑞𝐺𝑒 = 0 (2)

The value of 𝑞𝐺𝑒 is principally dependent on the absorptivity (α) of the Ge layer [36]. In
the current work, the heat produced in the Ge layer is determined as

(1 ― ƞ𝑐𝑒𝑙𝑙) × 𝐷𝑁𝐼 × CR × 𝛼𝐺𝑒 × 𝐴𝐺𝑒


𝑞𝐺𝑒 = 𝑉𝐺𝑒
(3)

where ƞcell is the cell electrical efficiency, 𝐷𝑁𝐼 × CR represents the net concentrated solar
irradiance on the cell, and the concentrator optical efficiency (αGe) is the Ge absorptivity. A and V
are the top surface area and volume of the cell, respectively. An iterative technique was
implemented to determine the 𝑞𝐺𝑒 values in the cell accurately. The iterative technique was applied
because ƞ𝑐𝑒𝑙𝑙 is a function of Tcell, which is unknown. The equation governing the relation between

ƞ𝑐𝑒𝑙𝑙 and its Tcell is formulated as [37]


ƞ𝑐𝑒𝑙𝑙 = ƞ𝑅𝑒𝑓 ― [𝛽𝑡ℎ𝑒𝑟𝑚𝑎𝑙(𝑇𝐶𝑒𝑙𝑙 ― 𝑇𝑅𝑒𝑓)] (4)

where ƞ𝑅𝑒𝑓 is the cell electrical efficiency at the reference cell temperature TRef = 25 °C
[11], and βthermal is the cell temperature coefficient and is equal to 0.047 % for Ge [33]. Heat

9
generation in the other layers below the Ge layer was neglected due to the very low transmissivity
of the Ge layer.

3.1.1. Thermal boundary conditions


The upper surfaces of the Ge, copper-I, ceramic, and copper-II layers were subjected to
combined convection and radiation heat dissipation. At these boundaries, the convection heat
transfer coefficient by wind (ℎ𝑐𝑜𝑛𝑣,𝑤𝑖𝑛𝑑), ambient temperature (Ta), surface emissivity, and
external radiation temperature were carefully applied. Because of the very small thickness of these
layers, the sidewalls of the computational domain were assumed adiabatic. Moreover, backside
heat transfer coefficient (hback) with different values up to 1600 W/m2 K is defined at the lower
wall of the copper-II substrate as shown in Fig. 2 with free stream temperature equal to Ta.
Uniform solar Irradiance
Germanium Combined convection and
layer radiation heat loss
Copper I

Ceramic
Adiabatic
Y Sides

Modified x
Copper II Convection from passive cooling, hback

Fig. 2. The HCPV and heat sink assembly and boundary conditions.

Further, the mathematical expression of the used boundaries can be represented as the
following:

For top surfaces of the Ge layer,

∂𝑇𝐺𝑒
―𝑘𝐺𝑒 ( )∂𝑦
= 𝑞𝑟𝑎𝑑,𝐺𝑒→𝑆 + 𝑞𝑐𝑜𝑛𝑣,𝐺𝑒→𝑎. (5)

For the top surface of the copper-I layer,

∂𝑇𝑐𝑢
―𝑘𝑐𝑢 ( )
∂𝑦
= 𝑞𝑟𝑎𝑑,𝑐𝑢→𝑆 + 𝑞𝑐𝑜𝑛𝑣,𝑐𝑢→𝑎, (6)

10
where q rad,Ge→s and q conv,Ge→a are the radiative flux and convective heat flux from the cell
to the sky and the surrounding, respectively. The terms q rad,Ge→s and q conv,Ge→a are calculated using
the following equations [33,35]:

𝑞𝑐𝑜𝑛𝑣,𝐺𝑒→𝑎 = ℎ𝑐𝑜𝑛𝑣, 𝑤𝑖𝑛𝑑 (𝑇𝐺𝑒 ― 𝑇𝑎), (8)

ℎ𝑐𝑜𝑛𝑣, 𝑤𝑖𝑛𝑑 = 5.82 + 4.07𝑉𝑤𝑖𝑛𝑑, (9)

where Vwind is the wind velocity (m/s). The radiation heat dissipation from the cell layer as
an example was calculated as follows [33]:

𝑞𝑐𝑜𝑛𝑣,𝐺𝑒→𝑎 = 𝜎𝜀𝐺𝑒(𝑇4𝐺𝑒 ― 𝑇4𝑠 ), (10)

𝑇𝑠 = 0.0522𝑇1.5
𝑎 , (11)

where εGe and Ts are the Ge emissivity and sky temperature, respectively. To consider this
boundary in ANSYS, mixed boundaries were nominated on the boundary.

A thermal coupling boundary was applied between every two solid layers. Hence, no
further boundaries were required because the Fluent solver calculated the heat transfer from the
solution of the adjacent cells [38]. Thermally coupled boundary conditions between solid layers
were used. The conduction heat transfer happens at y-direction for all these interfaces. Therefore,
the boundary conditions at these interfaces can be described as follow:-

Between Ge and copper-I material,

∂𝑇𝐺𝑒 ∂𝑇𝑐𝑢
k𝐺𝑒 = k𝑐𝑢 𝑎𝑛𝑑 𝑇𝐺𝑒 = 𝑇𝑐𝑢 (12)
∂𝑦 ∂𝑦

Between copper-I and Al2O3-ceramics material,

∂𝑇𝑐𝑢 ∂𝑇𝑐𝑒
k𝑐𝑢 =k 𝑎𝑛𝑑 𝑇𝑐𝑢 = 𝑇𝑐𝑒 (13)
∂𝑦 𝑐𝑒 ∂𝑦

Between Al2O3-ceramics and bottom copper-II material,

∂𝑇𝑐𝑒 ∂𝑇𝑐𝑢,𝑏
k𝑐𝑒 =k 𝑎𝑛𝑑 𝑇𝑐𝑒 = 𝑇𝑐𝑢,𝑏 (14)
∂𝑦 𝑐𝑢,𝑏 ∂𝑦

11
At the backside of the solar cell, the backside heat transfer coefficient hback was defined.
The value of hback was changed according to the range of the heat transfer coefficient achieved by
passive cooling up to 1600 W/m2 K [27].

𝑞𝑐𝑜𝑛𝑣,𝑐𝑢→𝑎 = ℎ𝑏𝑎𝑐𝑘 (𝑇𝑐𝑢 ― 𝑇𝑎) (15)

3.2. Structural model and thermal stress analysis


Finite element analysis (FEA) code was applied using ANSYS APDL solver to investigate
the mechanical consequences of temperature gradient occurring during the cooling process and
estimate thermal stress inside a cell body. A 3D FEA thermo-mechanical model was developed.
The focus of this model is to estimate thermal stress within the cell body. Therefore, to cut down
the computational time, the model considered that the solar cell was attached to a copper-I substrate
only. After solving the thermal model, the calculated temperature distributions were used as
thermal load in the structural model under steady state conditions. The cell and copper-I
experienced significant temperature gradient. Therefore, the cell and copper-I thermal gradients
were imported to the structural model. The zero-stress state at the reference temperature (TRef = 25
°C) was assumed. The backside of copper-I was considered as fixed support, while all other faces
were set as load-free surfaces. The key material properties, thermal expansion coefficient, and
Young’s modulus were well defined. The thermal physics parameters of each layer are presented
in Table 3.
Table 3. Structural properties used in the structural model [39,40]
Layers
Property
Ge Copper
Young's modulus, E, GPa 102.7 128
Shear modulus, S, GPa 40.1 47.05
Bulk modulus, B, GPa 77.78 152.4
Poisson ratio, ν 0.28 0.34
Coefficient of thermal expansion, 𝛼, 1/℃ 6.1 × 10−6 16.9 × 10−6

3.2.1. Structural model governing equations


Because of the thermal-expansion mismatch between the cell layer and board components
made of different materials, thermal stress increased at the boundary of each layer. Hence, the first
step of structural investigation is importing temperature distribution from the thermal model,

12
which is defined as a free surface for all degrees of freedom. The backside of copper-I was
considered as fixed support, while all other surfaces were set as load-free boundaries. A simplified
structural model with two layers (Ge and copper-I) was established. Assuming that each layer
satisfies homogeneous, isotropic, and linear conditions, the Ge and copper-I layers were modeled
for structural analysis using the same geometry and mesh as the thermal model. In the present
study, the boundary conditions were calculated for the cell temperature exported from the thermal
model under an ambient temperature of 25 ℃ and assumed independent of temperature for
numerical solution simplicity.

Because the cell was considered as a constrained body, which cannot move with applied
thermal loading, the structural model can be developed relying on the FEA principle of virtual
work introduced by Yilbas et al. [36]. The virtual or very small change in the energy of internal
strain (δu) must be consistent to the corresponding variation in external work due to the applied
loads (δv). Because the cell was constrained in motion and by considering the strain energy due to
thermal stress resulting from the thermal load, the implementation of virtual work principle under
temperature body load only gives

{𝛿𝑢}𝑇 [𝐵]𝑇[𝐷][𝐵]𝑑𝑣{𝑢} = {𝛿𝑢}𝑇 [𝐵𝑇][𝐷]{𝜀𝑡ℎ}𝑑𝑣,


∫ 𝑣𝑜𝑙 ∫ 𝑣𝑜𝑙
(15)

where {𝛿𝑢}𝑇 is a set virtual arbitrary displacement vector for all the above terms, [B] is the
displacement gradient matrix, {𝜀𝑡ℎ} is the thermal strain vector, and [D] is the properties matrix of
the material [36].

[𝐾]{𝑢} = {𝐹𝑡ℎ} (16)

[K] is the element stiffness matrix and {Fth} is the element load vector. [K], {Fth}, and {𝜀𝑡ℎ}
are described by Eq. (17), (18), and (19), respectively [41,42]. Eq. (15) can be reduced to the
following:

[𝐵]𝑇[𝐷][𝐵]𝑑𝑣,
[𝐾] =
∫ 𝑣𝑜𝑙
(17)

{𝐹𝑡ℎ} = [𝐵𝑇][𝐷]{𝜀𝑡ℎ}𝑑𝑣,
∫ 𝑣𝑜𝑙
(18)

13
{𝜀𝑡ℎ} = {𝛼}∆𝑇. (19)

The thermal stress developing in the cell can be calculated considering Hook’s law as the
following equation [41,43]:

{𝜎} = [𝐷]{𝜀𝑒𝑙}, (20)

where {𝜎} and {𝜀𝑒𝑙} are the stress and elastic strain vectors, respectively. Because of strain

{𝜀} is equal to the sum of the elastic and thermal strains,

{𝜀} = {𝜀𝑒𝑙} + {𝜀𝑡ℎ}. (21)

The body cell was constrained; therefore, the total strain is equal to zero. Consequently,
the thermal stress can be calculated by rewriting Eq. (20) as follows:

{𝜎} = [𝐷]{𝜀𝑡ℎ}. (22)

3.3. Solution methodology for coupled thermal structural analysis


The solution procedure for coupled thermal and mechanical models is illustrated by the
flowchart in Fig. 3. The figure describes the overall simulation process of the current simulations.
The simulation procedure consists of four main steps. First, a 3D geometry was created using
design modular software with the dimensions listed in Table 1. Second, a mesh independence test
was performed using ANSYS meshing software. Third, for the best grid size, the thermal model
was iteratively solved using ANSYS Fluent software. Fourth, the thermal model was coupled with
the structural model by importing Tcell into mechanical static structure software. The material
mechanical properties were defined in the first step of the structural model.

As shown in Fig. 3, the thermal model runs iteratively. As the solar cell efficiency is a
function of the solar cell temperature, which is unknown, an assumed cell temperature of Tcell (s)
= 25 °C was used to start the simulation. At this value of temperature, cell efficiency was calculated
using equation (4). Then the heat generation in the cell layer was estimated using equation (3). At
that time the numerical solution of the energy equation for the whole domain was conducted until
the error in the energy equation become less than 10−8. From the energy equation solution, the new
value of the solar cell temperature Tcell (s+1) was obtained. The new value of the cell temperature
Tcell (s+1) was used to calculate the new value of cell efficiency using equation (4) and then the
new value of 𝑞𝐺𝑒 at Tcell (s+1) was obtained by equation (3). This iterative technique was repeated

14
until the maximum cell temperature difference between two consecutive iterations was less than
0.05 °C. At this condition, the temperature at each element of the computational domain was used
for thermal stress analysis. Fig. 3 (a) and (b) show the predicted Tcell after six iterations at an
ambient temperature of 25 °C and hback of 10 and 1600 W/m2 K, respectively. It is evident from
Fig. 3 that the iteration method is of utmost importance for the correct prediction of Tcell, especially
at high CR and low backside convective heat transfer coefficient. For instance, at CR of 1500
Suns, the difference between the Tcell predicted using the current iterative technique and the non-
iterative technique is about 4.5 and 161.6 ℃ for backside convective heat transfer coefficients of
1600 and 10 W/m2 K, respectively.

15
Fig. 3. Flowchart of the simulation work

16
Fig. 3. The iterative technique used for the accurate prediction of the cell temperature
with respect to iterations number at (a) hback = 10 and (b) 1600 W/m2 K.

17
3.4. Mesh independence test
Mesh independence test was performed to illustrate that the results were not dependent on
the number of elements, as shown in Fig. 4. This test was completed for each variation in the
computational domain dimensions. For instance, it is evident that increasing the number of
elements to more than 1.02 million slightly changes the predicted Tcell and cell uniformity index
ΔT. However, increasing the number of elements beyond this value consumes longer
computational time without significant increase in the solution accuracy.

Fig. 4. Grid independence test

4. Model validation
To validate the developed model, the predicted results were compared with the results of
[15,27] as shown in Fig. 6. Fig. 6 (a) shows the predicted Tcell for the uncooled case with the results
estimated by Min et al. [15]. In this part of the validation, a solar cell with an area of 3 × 3 mm2
under different solar CRs was numerically modeled with the same operating conditions as in [15].
It is evident that the current model accurately predicts Tcell, which is in good agreement with the
results of [15]. The maximum deviation between the predicted results and the results of [15] is
about 2.8 % at a CR of 1000 Suns. In the second set of validation, the predicted results were

18
compared with the results of Theristis et al. [27]. As shown in Fig. 5 (b), the boundary conditions
of the thermal model were definite for the HCPV structure. The variation between the two model
results is about 0.21 %. In this validation step, the backside heat transfer coefficient hback was kept
fixed at 1000 W/m2 K, while the ambient temperature was varied from 25 to 45 ℃. It is obvious
that the current model is in good agreement with the results of Theristis et al. [27] over a wide
range of ambient temperatures.

19
Fig. 5. Validation of the predicted results with the results of (a) Min et al. [15] and (b)
Theristis et al. [27].

The verification of this study model with the experimental and numerical data of Chow et
al.[44] was implemented as shown in Fig. 7. The comparison between the temperature values of
the 3J cell was estimated for the same cell size and structure of the previous validation set. Chow
et al. studied the effect of adding the thermal paste to the interface between the HCPV assembly
and the heat sink for different CR values under active cooling environment. The current model
calculations results agree well with experimental and mathematical results of Chow et al with a
minor average error less than 0.5%.

20
Fig.7 Verification of current model with experimental and mathematical results of chow et
al.[44] for the average temperature of the solar cell mounted on a receiver with/out thermal paste

5. Results and discussion


This section is organized into six subsections. The first subsection reveals the thermal
performance of the uncooled HCPV under different CR values. The second subsection investigates
the performance of HCPV systems under different ambient temperatures, wind speed, and DNI
values. In the third subsection, the performance of the HCPV was thermally estimated under
different values of the backside heat transfer coefficient. In the fourth subsection, the solar cell
structure was modified by extending the area of copper-II layer with different Ar values. In this
part, the thermal and electrical performances of the existing solar cell were determined at different
Ar and CR values. In the fifth subsection, static structural analysis of the HCPV system under
different operating scenarios was performed. Finally, the sixth subsection summarizes the current
research findings.

5.1. Uncooled solar cell performance


The operation of the MJ cell under the worst-case scenario occurs at no implemented
cooling method. To determine the cell behavior at no cooling, the cell temperature distribution is
presented in Fig. at different CR when the ambient temperature is 25 °C and wind speed is 1 m/s.
21
In this case, the top and backside of the solar cell assembly were assumed to lose heat by combined
convection and radiation heat loss calculated by Eq. (8) and (10), respectively. The temperature
contours are demonstrated with respect to width and length on a plane located at the top surface of
the solar cell.

As shown in Fig. , at the wind speed of 1 m/s with a backside convective heat transfer
coefficient of around 10 W/m2 K, the solar cell exhibited very high temperature operations above
the maximum permissible limit. Generally, increasing CR significantly increases the solar cell
temperature at the same DNI and weather conditions. Based on the predicted contours, the
maximum local cell temperature reached 271, 1028, 1359, and 1581 °C at the CR of 50, 500, 1000,
and 1500 Suns shown in Fig. 8 (a),( b), (c), and (d), respectively. One of the interesting findings
is that the solar cell temperature sharply increased by 757 ℃ while increasing the CR from 50 to
500 Suns. However, while increasing the CR from 1000 to 1500 Suns, the cell temperature
increased by around 222 ℃. This is attributed to the fact that at higher cell temperatures, i.e., CR,
heat loss by radiation from the solar cell surfaces becomes dominant compared to that at lower
values of cell temperature. This is because radiation heat loss is a function of the absolute cell
temperature raised to the fourth power. The predicted results are in good agreement with the results
of [33], which reported that the uncooled HCPV cell temperature reached 931 °C at a CR of 400
Suns. Fig. 8 shows that increasing the CR for uncooled HCPV increased the non-uniformity of
temperature. The temperature uniformity index (ΔT) is defined as the maximum temperature
difference on the solar cell layer (ΔT = Tmax − Tmin). Therefore, low ΔT leads to the best temperature
uniformity and is recommended to reduce thermal stress. The values of ΔT reached 1, 4, 12, and
18 at CRs of 50, 500, 1000, and 1500 Suns shown in Fig. 8 (a), (b), (c), and (d), respectively.

22
(a) (b)

(c) (d)
Fig. 8. Temperature contours (in Celsius) on the top surface of the uncooled cell when Ta = 25
℃, Vwind = 1 m/s, and CR is (a) 50, (b) 500, (c) 1000, and (d) 1500 Suns.

5.2. Effect of wind speed, ambient temperature, and DNI variation


In Fig. 9 (a), (b), and (c), the effect of wind speed, ambient temperature, and DNI variation,
respectively, on the HCPV thermal performance is displayed. In Fig. 9 (a), to include the variation
of wind speed, the ambient temperature and backside heat transfer coefficient hback are fixed to 25
℃ and 1600 W/m2 K, respectively. The value of hback is selected to mimic the maximum heat
transfer coefficient that can be attained by the passive cooling technique as recommended by
Mudawar [45]. Therefore, to include wind speed effect in the thermal model, the top convection
heat transfer coefficient is varied according to Eq. (8). Based on the predicted results, it is noticed

23
that increasing the wind speed from 0 to 15 m/s causes a very slight decrease in Tcell. This is
attributed to the fact that most of the heat generated in the solar cell is dissipated from the backside
of the solar cell with a high backside heat transfer coefficient. It is also concluded that the wind
speed effect is more pronounced at higher CR than that at lower CR values.
On the other hand, in Fig. 9 (b), the influence of Ta on the cell temperature is displayed at
a wind speed of 1 m/s and backside heat transfer coefficient of 1600 W/m2 K. It is very clear that
increasing the ambient temperature significantly increases the predicted cell temperature. The
increase in this case is pronounced because the ambient temperature affected both sides of the solar
cell. For instance, the heat dissipation from the top and backside is affected by the ambient
temperature. In addition, the sky temperature itself is dependent on the ambient temperature. For
instance, increasing the ambient temperature from 15 to 55 ℃ increases the cell temperature from
18 to 59 ℃ at a CR of 50 Suns and from 142 to 184 ℃ at a CR of 1500 Suns.

In Fig. 9 (c), variation of the cell temperature with DNI is displayed. The DNI values were
varied from 200 to 1000 W/m2. The simulation was performed at a wind speed of 1 m/s, ambient
temperature of 25 ℃, and hback of 1600 W/m2 K. The net concentrated solar irradiance in this case
is evaluated as the product of CR and DNI. Based on the predicted results, it is found that
increasing DNI linearly increases the cell temperature. In addition, increasing the CR also
increases the predicted cell temperature. This is because increasing DNI linearly increases the heat
generated in the Ge layer. For instance, increasing DNI from 200 to 1000 W/m2 increases the cell
temperature from 26 to 29 ℃ at a CR of 50 Suns and from 50 to 143 ℃ at a CR of 1500 Suns.

24
25
Fig. 9. Variation of the predicted solar cell temperature with (a) wind speed, (b) ambient
temperature, and (c) direct normal irradiance (DNI) at different CR values and hback of 1600
W/m2 K

5.3. Effect of backside heat transfer coefficient variation


Recently, several cooling techniques have been applied to limit cell temperature in HCPV
systems. In these methods, cooling is implemented by either natural or forced convection at the
backside of the solar cell. In this part of the simulation, the impact of hback on the cell thermal
performance at different CR values is investigated. The value of hback ranges from 10 to 1600 W/m2
K. Low values of hback can be attained by natural convection, while high values of hback can be
attained by forced convection at high wind speed or by using heat pipe for passive cooling.

Fig. 10 (a) and (b) show the impact of hback on the average cell temperature and solar cell
electrical efficiency, respectively, at different CR values from 50 to 1500 Suns. Fig. 10 (a) shows
that increasing hback significantly reduces the cell temperature, and with further increase in hback, a
slight reduction in the cell temperature was observed. At a low CR of 50 Suns, increasing hback
from 10 to 200 W/m2 K decreases the cell temperature from 220 to 48.7 ℃. A further increase in
hback from 200 to 1600 W/m2 K causes no reduction in the solar cell temperature. This trend was

26
also observed by the researchers in [46] and [47]. They provided three different explanations to
justify this trend. They reported that at a low hback, attained at a low coolant mass flow rate in active
cooling, the heat transfer mechanism between the backside of the solar cell and the cooling fluid
is dominated by convection, while at a high hback associated with high coolant mass flow rates, the
heat transfer mechanism is dominated by conduction within the thin boundary layer [46]. Another
point of view is that at a high hback, the heat extracted reaches a saturation level; therefore, the cell
temperature slightly decreases [47]. The value of hback at which the solar cell temperature reaches
a nearly saturated value is defined as critical backside heat transfer coefficient hback, cr. The value
of hback, cr is a function of CR at the same meteorological conditions. The values of hback, cr are about
200, 600, 1000, and 1600 W/m2 K at CRs of 50, 500, 1000, and 1500 Suns, respectively.

On the other hand, solar cell electrical efficiency variation with hback at different values of
CR is displayed in Fig. 10 (b). It is noticed that increasing CR decreases the solar cell electrical
efficiency at the same value of hback. This is attributed to the fact that, at a constant hback, increasing
CR results in a significant increase in the cell temperature, which resulted from an increase in the
received solar radiation density. It is also concluded that increasing hback enhances the solar cell
electrical efficiency until a constant value of hback, cr is reached, after which increasing hback slightly
enhances the solar cell electrical efficiency.

27
Fig. 10. Variation of (a) solar cell average temperature and (b) solar cell electrical efficiency
with the backside convective heat transfer coefficient at different CR values.

28
Fig. 11 illustrates the effect of varying CR on the electrical output and electrical efficiency
of a single HCPV at different hback values. It is evident that, as CR increased, the electrical output
and electrical efficiency increased. This trend was observed for all hback values. The electrical
output increased from about 5 to about 15 and 45 W by increasing the CR value for hback = 10 and
1600 W/m2 K, respectively. As shown in Fig. 10 (b), the electrical efficiency reduced as the CR
increased from 50 to 1500 Suns for all the hback values. This is mainly due to the rise in cell
temperature with an increase in CR.

Fig. 11. Variation of (a) electrical power and (b) electrical efficiency with CRs of 50,
500, 1000, and 1500 Suns.

29
Based on the results in Fig. 10, it is essential to estimate the value of hback, cr corresponding
to different CR values because increasing the value of hback to more than that of hback, cr results in
no enhancement in the cell temperature, cell electrical efficiency, and cell power. Hence, it is worth
mentioning that increasing hback to more than hback,cr increases the power consumption used in the
active cooling methods or the cooling system complexity. Consequently, the system net gained
electric power reduces, thereby increasing the system cost. The temperature contours at hback =
hback, cr are displayed in Fig. 12.

Fig. 12 shows the temperature contours of the top surface of the solar cell at CRs of 50,
500, 1000, and 1500 Suns. The maximum temperature on the solar cell surface reached 29, 66,
110, and 160 ℃ at the CR of 50, 500, 1000, and 1500 Suns, respectively, at an ambient temperature
of 25 ℃ and hback = hback,cr. Compared to the case of no cooling, mentioned in Fig. 8, it is noticeable
that increasing hback to hback,cr significantly reduced the cell temperature under CR = 50 and 500
Suns. In line with this, the maximum convective heat transfer coefficient of 1600 W/m2 K was not
sufficient to keep the solar cell average temperature within the recommended operating
temperature limit (25–80 ℃) for CR = 1000 and 1500 Suns, where the maximum temperature on
the solar cell surface reached 110 and 160 ℃ at the CR of 1000 and 1500 Suns, respectively. To
limit the temperature rise, another cooling technique must be implemented.

30
(a) (b)

(c) (d)
Fig. 12. Variation of average solar cell temperature contours when Ta = 25 °C, Vwind = 1 m/s, and
hback = hback,cr at (a) CR = 50, (b) 500, (c) 1000, and (d) 1500 Suns.

5.4. Effect of increasing the area of copper-II layer


In this section, increasing the copper-II area is proposed to decrease the cell temperature,
especially at high CRs of 1000 and 1500 Suns. We define the ratio of copper-II area to copper-I
area as Ar. To study the effect of varying Ar, additional simulations were conducted. In these
simulations, copper-II area was extended to provide better thermal management of the HCPV. This
idea is very convenient in the case of using a Fresnel lens concentrator with enough space between
the cells. In addition, the extension of the copper-II area by a very small thickness value could not
influence the power consumed in the tracking system.

31
Fig. 13 shows the solar cell average temperature variation with hback at different values of
Ar and CR. Some interesting findings can be concluded from this figure. First, it is noticed that
increasing hback decreases the average cell temperature as previously explained. Second, it is
predicted that increasing Ar significantly reduces the average cell temperature at the same hback and
CR. This is attributed to the fact that increasing the copper-II area enhanced the heat dissipation
rate from the solar cell. Third, increasing Ar reduces the value of hback, cr. This means that a lower
backside heat transfer coefficient for a large Ar can be used to attain the same cell temperature at
high hback when Ar is unity. Therefore, it is recommended to use heat spreaders combined with
active cooling techniques for cooling HCPV systems. Finally, it is noticed that increasing Ar from
the 1 to 19.5 at an hback of 1600 W/m2 K reduces the cell temperature from 110 to 78 ℃ and from
160 to 110 ℃ at CRs of 1000 and 1500 Suns, respectively.

32
(a) (b)

(c) (d)
Fig. 13. Variation of average cell temperature with hback for different Ar under (a) CR =
50, (b) 500, (c) 1000, and (d) 1500 Suns.

Fig. 14 (a) and (b) show the variation of the average cell temperature with Ar under
different CRs at the lowest and highest investigated values of hback, i.e., 10 and 1600 W/m2 K,
respectively. At the same CR, it is evident that the average cell temperature reduced as Ar
increased. On the other hand, at the highest hback of 1600 W/m2 K, as the Ar ratio increased, the cell
temperature decreased to a certain limit and subsequently remained constant. This means that no
reduction in the cell temperature can be attained with further increase in Ar beyond a certain value.
The value of Ar at which the cell temperature reached a constant value can be defined as the critical
area ratio Ar,cr. The value of Ar,cr is a function of hback and CR at the same ambient temperature and
wind speed. The solar cell temperature at the lowest hback of 10 W/m2 K was not suitable for solar
cell operation. Therefore, in Fig. 13 (b), at the highest hback of 1600 W/m2 K, Ar,cr values were

33
about 2, 3, 4, and 6 at CRs of 50, 500, 1000, and 1500 Suns, respectively. It is recommended to
use these values of Ar,cr to avoid solar cell over-sizing.

Fig. 14. Variation of the average cell temperature with Ar at different CR with (a) hback =
10 and (b) 1600 W/m2 K.

34
5.5. Thermal structural analysis of the uncooled solar cell
A structural model was used to examine thermal behavior based on the theory of material
mechanics and thermal stress analysis of a single HCPV solar cell. Fig. 15 (a), (b), (c), and (d)
show the thermal stress contours of the top surface of the cell under CR = 50, 500, 1000, and 1500
Suns with an hback of 1600 W/m2 K and Ar of unity. It is obvious that the highest and lowest thermal
stresses are located at the center and edges, respectively. In addition, the thermal stress increased
with increase in CR. This trend of the results is in agreement with the results of Xiaoyan and
Shuang [48] for a bonded structure under temperature variation. Because the authors reported that
thermal stress distribution of the bonded structure had a parabolic shape with respect to material
length, the maximum stress appeared at the midpoint and the minimum stress at the endpoints.

(a) (b)

(c) (d)
Fig. 15. Variation of the thermal stress contours in MPa of the top surface of the solar cell at hback
= 1600 W/m2 K and Ar = 1 at (a) CR = 50, (b) 500, (c) 1000, and (d) 1500 Suns.

35
Fig. 16. Thermal stress distribution at the top surface at Ar = 1 and hback = 1600 W/m2 K for (a)
the center and (b) diagonal lines.

36
For easy comparison, thermal stress values were calculated at different locations. The
thermal stress at the top surface of the solar cell is displayed in Fig. 16, while that at the interface
between the solar cell and the copper-I layer is displayed in Fig. 17. The thermal stress over (a)
the cell center and (b) diagonal lines is plotted in both the figures. For all the investigated cases,
the cell showed peak thermal stress at the body center. It is obvious from Fig. 16 that at hback =
1600 W/m2 K, the highest local thermal stress was located in the interface between the cell and
copper-I layer. This is due to the mismatch in the coefficient of thermal expansion because of the
presence of unlike materials within the interface. Furthermore, maximum thermal stress values,
i.e., 9.95 and 299 MPa for CR = 50 and 1500 Suns, respectively, were observed at the corners of
the cell at the interface between the cell and copper-I layer.

5.6. Findings and Summary

The main research findings and results are summarized in

Table 4. As the ambient temperature increased, the solar cell temperature and consequently
the electrical efficiency and power increased. In contrast, wind speed had a positive effect on the
electrical performance of the solar cell, i.e., as the wind speed increased, the cell temperature and
consequently the electrical efficiency and power output decreased, as summarized in

Table 4. The high values for G and CR had significant effect on the electrical performance
of the cell even when high cell temperature was observed. Furthermore, at the same condition,
increasing Ar reduced the average cell temperature to a certain limit, which subsequently remained
constant.

37
Fig. 17. Thermal stress distribution at the interface between the solar cell and the copper-I layer
for Ar = 1 and hback = 1600 W/m2 K for (a) the center and (b) diagonal lines.

38
Table 4. Main results and findings
HCPV system performance parameters for a
10 × 10 mm2 multijunction solar cell
Parameter Tcell ηe ΔT Pel
Values Constraints
(unit) (℃) (%) (℃) (W)
15 96 40.23 19 34.2 Vwind = 1 m/s,
35 116.4221 39.85 19.14 33.88 DNI = 1000 W/m2,
Ta CR = 1000 Suns,
Meteorological conditions

(℃) hback = 1600 W/m2


55 136.8814 39.48 19.25 33.55
K,
and Ar = 1
1 106.2 38.78 19.1 33 Ta = 25 ℃,
5 105.48 40.05 19.06 34.04 DNI= 1000 W/m2,
Vwind
CR = 1000 Suns,
(m/s)
15 103.77 40.07 19.02 34.06 hback = 1600 W/m2
K, and Ar = 1
200 40.9 40.92 3.74 34.78 Vwind = 1 m/s,
600 73.28 40.46 11.35 34.39 Ta = 25 ℃,
DNI
CR = 1000 Suns,
(W/m2)
1000 106.2 38.78 19.1 33 hback = 1600 W/m2
K, and Ar = 1
50 28.9 41.82 0.91 1.78 Vwind = 1 m/s,
Ta = 25 ℃,
CR 500 64 41.20 9.18 17.5 DNI = 1000 W/m2,
(Suns)
1500 153.25 35.57 30.15 45.35 hback = 1600 W/m2
Design conditions

K, and Ar = 1
10 1240.879 1 19.35 17.26 Vwind = 1 m/s,
800 171.37 39.07 20.1 33.21 Ta = 25 ℃,
hback
DNI= 1000 W/m2,
(W/m2 K)
1600 106.2 38.78 19.1 33 CR = 1000 Suns,
and Ar = 1
1 106.2 38.78 19.1 33 Ta = 25 ℃,
10 77.2 39.37 20.79 33.47 DNI = 1000 W/m2,
Ar CR = 1000 Suns,
20 77.2 39.37 20.78 33.47 hback = 1600 W/m2 K
and, Vwind = 1 m/s,

Conclusions
Comprehensive 3D thermal and structural models were coupled to investigate the
performance of an HCPV system under different CRs, ambient temperature, direct normal
irradiance, wind speed, backside heat transfer coefficient (hback), and copper-II substrate area
ratios. Based on the results, we conclude the following:

1- Increasing CR for an uncooled HCPV system significantly increases the non-uniformity


on the solar cell temperature at the same DNI and weather conditions.

39
2- Based on the predicted results, it is noticed that increasing the wind speed causes a very
slight decrease in the solar cell temperature. The wind speed effect is pronounced at higher
CR in comparison to that at lower CR values. Increasing the ambient temperature,
increasing CR, and increasing DNI significantly increase the predicted cell temperature.
3- For a cooled HCPV system, increasing hback significantly reduces the cell temperature, and
with further increase in hback, a slight reduction in the cell temperature is observed. On the
other hand, increasing hback enhances the solar cell electrical efficiency until it reaches a
constant value of hback,cr, after which increasing hback slightly enhances the solar cell
electrical efficiency. It is noticed that increasing CR decreases the solar cell electrical
efficiency and increases the electrical power output for the same value of hback..
4- Increasing Ar significantly reduces the average cell temperature at the same hback and CR
values. On the other hand, at the highest hback, as the Ar ratio increases, the cell temperature
decreases to a certain limit and subsequently remains constant.
5- Thermal structural analysis of the uncooled solar cell shows that the highest and lowest
thermal stress is located at the center and edges of the cell, respectively. In addition, thermal
stress increases with increase in CR.

Acknowledgments
Sincere gratitude and appreciation for the Egyptian Ministry of Higher
Education (MOHE), EJUST University, and JICA for providing the opportunity and offering the
financial support and computational tools for this manuscript. This work was supported by the
Science and Technology Development Fund (Project ID: 33515).

Nomenclature
A solar cell area or area (m2)
B bulk modulus (GPa)
CR concentration ratio (Suns)
DNI direct normal radiation (W/m2)
E Young's modulus (GPa)
G net concentrated solar irradiance (W/m2)
h convective heat transfer (W/m2 K)
k thermal conductivity (W/m K)

40
q heat generation per unit volume (W/m3)
S shear modulus (GPa)
T temperature (°C)
V solar cell volume (m3)
Greek symbols
α thermal expansion coefficient or absorptivity
β temperature thermal coefficient
Δ difference
δ thickness (m)
ε thermal strain (mm/mm)
η efficiency (%)
σ thermal stress (Pa) or Stefan–Boltzmann constant 5.67 × 10−8 (W/ (m2 K4))
ѵ Poisson ratio
Subscript
a ambient
conv convection
cr critical
cu copper layer
el electrical
Ge germanium layer
i layer
r ratio
rad radiation
Ref reference
th thermal
Abbreviations
CR concentration ratio

DNI direct normal irradiance


FEA finite element analysis
HCPV high concentrator photovoltaic
MJ multijunction

41
PV photovoltaic
References
[1] Aman MM, Solangi KH, Hossain MS, Badarudin A, Jasmon GB, Mokhlis H, et al. A review
of Safety, Health and Environmental (SHE) issues of solar energy system. Renew Sustain
Energy Rev 2015;41:1190–204. doi:10.1016/j.rser.2014.08.086.

[2] Outlook AE. AEO2018 Early Release Overview. Energy 2018;2018:1–74. doi:26.05.2014.

[3] World Energy Council. World Energy Resources. vol. 2007. 2016.
doi:http://www.worldenergy.org/wp-
content/uploads/2013/09/Complete_WER_2013_Survey.pdf.

[4] Global Greenhouse Gas Reference Network. Trends in Atmospheric Carbon Dioxide. Earth
Syst Res Lab Glob Monit Div 2017.

[5] Ju X, Xu C, Hu Y, Han X, Wei G, Du X. A review on the development of


photovoltaic/concentrated solar power (PV-CSP) hybrid systems. Sol Energy Mater Sol
Cells 2017;161:305–27. doi:https://doi.org/10.1016/j.solmat.2016.12.004.

[6] Drennen TE. Renewable Energy: Sources for Fuels and Electricity. J Environ Qual
1994;23:622. doi:10.2134/jeq1994.00472425002300030032x.

[7] Daneshazarian R, Cuce E, Cuce PM, Sher F. Concentrating photovoltaic thermal (CPVT)
collectors and systems: Theory, performance assessment and applications. Renew Sustain
Energy Rev 2018;81:473–92. doi:10.1016/j.rser.2017.08.013.

[8] Shanks K, Ferrer-Rodriguez JP, Fernández EF, Almonacid F, Pérez-Higueras P,


Senthilarasu S, et al. A >3000 suns high concentrator photovoltaic design based on multiple
Fresnel lens primaries focusing to one central solar cell. Sol Energy 2018;169:457–67.
doi:https://doi.org/10.1016/j.solener.2018.05.016.

[9] Dong J, Zhuang X, Xu X, Miao Z, Xu B. Numerical analysis of a multi-channel active


cooling system for densely packed concentrating photovoltaic cells 2018;161:172–81.
doi:10.1016/j.enconman.2018.01.081.

[10] Nishioka K, Takamoto T, Agui T, Kaneiwa M, Uraoka Y, Fuyuki T. Annual output


estimation of concentrator photovoltaic systems using high-efficiency InGaP/InGaAs/Ge
42
triple-junction solar cells based on experimental solar cell’s characteristics and field-test
meteorological data. Sol Energy Mater Sol Cells 2006;90:57–67.
doi:10.1016/j.solmat.2005.01.011.

[11] 3C44A – with 10x10mm2Concentrator Triple Junction Solar cell. .www.azurspace.com


2016:1–6.

[12] Luque A, Hegedus S. Handbook of Photovoltaic Science and Engineering. 2011.


doi:10.1002/9780470974704.

[13] Royne A, Dey CJ, Mills DR, Royane A, Dey CJ, Milis DR, et al. Cooling of photovoltaic
cells under concentrated illumination: a critical review. Sol Energy Mater Sol Cells
2005;86:451–83. doi:10.1016/j.solmat.2004.09.003.

[14] Cotal H, Frost J. Heat transfer modeling of concentrator multijunction solar cell assemblies
using finite difference techniques. 2010 35th IEEE Photovolt. Spec. Conf., 2010, p.
000213–8. doi:10.1109/PVSC.2010.5614514.

[15] Min C, Nuofu C, Xiaoli Y, Yu W, Yiming B, Xingwang Z. Thermal analysis and test for
single concentrator solar cells. J Semicond 2009;30:044011. doi:10.1088/1674-
4926/30/4/044011.

[16] Jakhar S, Soni MS, Gakkhar N. Historical and recent development of concentrating
photovoltaic cooling technologies. Renew Sustain Energy Rev 2016;60:41–59.
doi:10.1016/j.rser.2016.01.083.

[17] Micheli L, Reddy KS, Mallick TK. Thermal effectiveness and mass usage of horizontal
micro-fins under natural convection. Appl Therm Eng 2016.
doi:10.1016/j.applthermaleng.2015.09.042.

[18] Minano JC, Gonzalez JC, Zanesco I. Flat high concentration devices. Proc. 1994 IEEE 1st
World Conf. Photovolt. Energy Convers. - WCPEC (A Jt. Conf. PVSC, PVSEC PSEC),
vol. 1, 1994, p. 1123–6 vol.1. doi:10.1109/WCPEC.1994.520159.

[19] Wang YN, Lin TT, Leong JC, Hsu YT, Yeh CP, Lee PH, et al. Numerical investigation of
high-concentration photovoltaic module heat dissipation. Renew Energy 2013;50:20–6.
doi:10.1016/j.renene.2012.06.016.

43
[20] Gualdi F, Arenas O, Vossier A, Dollet A, Aimez V, Ar??s R. Determining passive cooling
limits in CPV using an analytical thermal model. AIP Conf. Proc., vol. 1556, 2013, p. 10–
3. doi:10.1063/1.4822187.

[21] Micheli L, Reddy KS, Mallick TK. Plate micro-fins in natural convection: An opportunity
for passive concentrating photovoltaic cooling. Energy Procedia, vol. 82, 2015, p. 301–8.
doi:10.1016/j.egypro.2015.12.037.

[22] Renzi M, Egidi L, Comodi G. Performance analysis of two 3.5 kWp CPV systems under
real operating conditions. Appl Energy 2015;160:687–96.
doi:10.1016/j.apenergy.2015.08.096.

[23] Araki K, Uozumi H, Yamaguchi M. A simple passive cooling structure and its heat analysis
for 500 X concentrator PV module. Conf Rec Twenty-Ninth IEEE Photovolt Spec Conf
2002 2002:2–5. doi:10.1109/PVSC.2002.1190913.

[24] Wang S, Shi J, Chen H-H, Schafer SR, Munir M, Stecker G, et al. Cooling design and
evaluation for photovoltaic cells within constrained space in a CPV/CSP hybrid solar
system. Appl Therm Eng 2017;110:369–81.
doi:https://doi.org/10.1016/j.applthermaleng.2016.08.196.

[25] Chou T-L, Shih Z-H, Hong H-F, Han C-N, Chiang K-N. Thermal Performance Assessment
and Validation of High-Concentration Photovoltaic Solar Cell Module. IEEE Trans
Components, Packag Manuf Technol 2012;2:578–86. doi:10.1109/TCPMT.2011.2181165.

[26] Royne A, Dey CJ, Mills DR. Cooling of photovoltaic cells under concentrated illumination:
A critical review. Sol Energy Mater Sol Cells 2005;86:451–83.
doi:10.1016/j.solmat.2004.09.003.

[27] Theristis M, O’Donovan TS. Electrical-thermal analysis of III-V triple-junction solar cells
under variable spectra and ambient temperatures. Sol Energy 2015;118:533–46.
doi:10.1016/j.solener.2015.06.003.

[28] Micheli L, Fernández EF, Almonacid F, Mallick TK, Smestad GP. Performance, limits and
economic perspectives for passive cooling of High Concentrator Photovoltaics. Sol Energy
Mater Sol Cells 2016;153:164–78. doi:10.1016/j.solmat.2016.04.016.

44
[29] Renno C, Landi G, Petito F, Neitzert HC. Influence of a degraded triple-junction solar cell
on the CPV system performances. Energy Convers Manag 2018;160:326–40.
doi:10.1016/j.enconman.2018.01.026.

[30] Amalu EH, Hughes DJ, Nabhani F, Winter J. Thermo-mechanical deformation degradation
of crystalline silicon photovoltaic (c-Si PV) module in operation. Eng Fail Anal
2018;84:229–46. doi:10.1016/j.engfailanal.2017.11.009.

[31] Smirnova O, Fend T, Capuano R, Feckler G, Schwarzbözl P, Sutter F. Determination of


critical thermal loads in ceramic high concentration solar receivers. Sol Energy Mater Sol
Cells 2018;176:196–203. doi:10.1016/j.solmat.2017.11.033.

[32] Cao M, Butler S, Benoit JT, Jiang Y, Radhakrishnan R, Chen Y, et al. Thermal Stress
Analysis/Life Prediction of Concentrating Photovoltaic Module. J Sol Energy Eng
2008;130:21011–9.

[33] Aldossary A, Mahmoud S, Al-Dadah R. Technical feasibility study of passive and active
cooling for concentrator PV in harsh environment. Appl Therm Eng 2016;100:490–500.
doi:10.1016/j.applthermaleng.2016.02.023.

[34] Zhou J, Yi Q, Wang Y, Ye Z. Temperature distribution of photovoltaic module based on


finite element simulation. Sol Energy 2015;111:97–103.
doi:10.1016/j.solener.2014.10.040.

[35] Siddiqui MU, Arif a. FM. Electrical, thermal and structural performance of a cooled PV
module: Transient analysis using a multiphysics model. Appl Energy 2013;112:300–12.
doi:10.1016/j.apenergy.2013.06.030.

[36] Hasan O, Arif AFM, Siddiqui MU. Finite Element Modeling, Analysis, and Life Prediction
of Photovoltaic Modules. J Sol Energy Eng 2013;136:021022. doi:10.1115/1.4026037.

[37] Evans DL. Simplified method for predicting photovoltaic array output. Sol Energy
1981;27:555–60. doi:10.1016/0038-092X(81)90051-7.

[38] ANSYS FLUENT 14.5 Theory Guide http: //www.afs.enea.it /project/ neptunius/
docs/fluent/html/th/node322.htm n.d.

45
[39] Souare PM, Sylvestre J. Thermal and mechanical analysis of a concentrated photovoltaic
module with integrated secondary optics. AIP Conf Proc 2016;1766:90008.
doi:10.1063/1.4962114.

[40] Wang H, Wang A, Yang H, Huang J. Study on the thermal stress distribution of crystalline
silicon solar cells in BIPV. Energy Procedia, 2016. doi:10.1016/j.egypro.2016.06.019.

[41] Hetnarski RB, Eslami MR. Thermal stresses - Advanced theory and applications. Solid
Mech Its Appl 2009;158:1–591. doi:10.1007/978-1-4020-9247-3_1.

[42] Yilbas BS, Arif AFM, Abdul Aleem BJ. Laser welding of low carbon steel and thermal
stress analysis. Opt Laser Technol 2010;42:760–8. doi:10.1016/j.optlastec.2009.11.024.

[43] Callister WDJ. Fundamentals of Materials Science and Engineering. 2001.


doi:10.1017/CBO9781107415324.004.

[44] Chow S, Valdivia CE, Wheeldon JF, Ares R, Arenas OJ, Aimez V, et al. Thermal test and
simulation of alumina receiver with high efficiency multi-junction solar cell for
concentrator systems. Photonics North 2010, 2010. doi:10.1117/12.872894.

[45] Mudawar I. Assessment of high-heat-flux thermal management schemes. IEEE Trans


Components Packag Technol 2001;24:122–41. doi:10.1109/6144.926375.

[46] Xu Z, Kleinstreuer C. Computational Analysis of Nanofluid Cooling of High Concentration


Photovoltaic Cells. J Therm Sci Eng Appl 2014;6:031009. doi:10.1115/1.4026355.

[47] Du B, Hu E, Kolhe M. Performance analysis of water cooled concentrated photovoltaic


(CPV) system. Renew Sustain Energy Rev 2012;16:6732–6.
doi:10.1016/j.rser.2012.09.007.

[48] Xiaoyan W, Shuang Z. Analytical model of thermal stress distribution of bonded structure
under temperature field. Int J Adhes Adhes 2011. doi:10.1016/j.ijadhadh.2011.03.005.

46
Highlights:

1. Thermal and structural analysis of high concentrator photovoltaics were studied.


2. The influence of meteorological conditions on the cell performance was evaluated.
3. The critical backside heat transfer coefficient was obtained for various conditions.
4. Increasing the back-copper area significantly enhances cell performance.
5. Direct normal irradiance has a substantial effect on solar cell temperature.

47

You might also like