You are on page 1of 30

Solar Energy 264 (2023) 112025

Contents lists available at ScienceDirect

Solar Energy
journal homepage: www.elsevier.com/locate/solener

Optimizing performance of water-cooled photovoltaic-thermal modules: A


3D numerical approach
Koorosh Khosravi a, Hayder I. Mohammed b, Jasim M. Mahdi c, Mahyar Silakhori d,
Mohammadreza Ebrahimnataj Tiji e, Arash Kazemian f, Tao Ma f, *, Pouyan Talebizadehsardari g, *
a
Department of Mechanical, Aerospace and Nuclear Engineering, Rensselaer Polytechnic Institute, 110 8th street, Troy, NY 12180, USA
b
Department of Physics, College of Education, University of Garmian, Kurdistan, Kalar, Iraq
c
Department of Energy Engineering, University of Baghdad, Baghdad 10071, Iraq
d
School of Electrical and Mechanical Engineering, University of Adelaide, Adelaide, Australia
e
Department of Mechanical Engineering, Qom University of Technology, Qom, Iran
f
School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai, China
g
Power Electronics, Machines and Control (PEMC) Research Group, University of Nottingham, Nottingham, UK

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

Keywords: To evaluate and improve the efficiency of photovoltaic solar modules connected with linear pipes for water
Response surface method (RSM) supply, a three-dimensional numerical simulation is created and simulated via commercial software (Ansys-
Optimization Fluent). The optimization utilizes the principles of the 1st and 2nd laws of thermodynamics by employing the
Photovoltaic/thermal (PVT)system
Response Surface Method (RSM). Various design parameters, including the coolant inlet velocity, tube diameter,
Energy and exergy analysis
panel dimensions, and solar radiation intensity, are systematically varied to investigate their impacts on ener­
getic and exergitic efficiencies and destroyed exergy. The relationship between the design parameters and the
system responses is validated through the development of a predictive model. Both single and multi-objective
optimizations are performed using the predictive model to optimize the thermal and electrical productivity
under different scenarios. The findings indicate the significance of the thermal exergy effectiveness, as evidenced
by its low P-value for all solar system responses, indicating its crucial role in the predictive model. For single-
objective optimization, the desirability is equal to 1 in cases where only heat transfer efficiency, whole energy
effectiveness, or thermal exergy efficiency is maximized or only destroyed exergy is minimized. The improve­
ments in energy and exergy efficiencies range from 3.55% to 69.13%, with the amount of destroyed exergy
reduced by 81.47% compared to the base case. For multi-objective optimization, desirability values exceeding
0.829 and 0.655 are obtained for single and multi-objective scenarios, respectively, indicating that the expected
performance is within desirable limits. The findings provide valuable insights for designing high-efficiency
photovoltaic/thermal systems and addressing their challenges and limitations.

to fossil fuels in the long term, given its abundance and even distribution
1. Introduction worldwide [4]. Photovoltaics (PV), which transforms sunlight directly
into electricity, is the most widespread form of harnessing solar energy
As mounting concerns over climate change and environmental due to its affordability and technical maturity [5]. Based on a recent
degradation persist, primarily due to harsh fossil fuel consumption and study by the International Renewable Energy Agency, it has been found
directly impacting human life, reliance on fossil fuels to meet energy that over 98% of solar-based electricity is derived from PV systems [6,7].
demand continues to grow worldwide [1,2]. Therefore, efforts must be However, there are still some issues to be resolved with current tech­
made to develop more efficient approaches for harvesting energy from nology [8]. For example, the commercially available PV modules on the
sustainable supplies such as solar, wind, and hydropower [3]. Among market seldom exceed 20% efficiency [9,10]. This implies that 80% of
these, solar energy is considered the most promising and likely successor all inward sunlight is converted to heat, dissipating to nearby

* Corresponding authors.
E-mail addresses: khosrk@rpi.edu (K. Khosravi), hayder.i.mohammad@garmian.edu.krd (H.I. Mohammed), jasim@siu.edu (J.M. Mahdi), mahyar.silakhori@
adelaide.edu.au (M. Silakhori), ebrahimnataj.m@qut.ac.ir (M. Ebrahimnataj Tiji), kazemian_arash@yahoo.com (A. Kazemian), tao.ma@sjtu.edu.cn (T. Ma),
pouyan.talebizadehsardari2@nottingham.ac.uk (P. Talebizadehsardari).

https://doi.org/10.1016/j.solener.2023.112025
Received 23 May 2023; Received in revised form 17 August 2023; Accepted 9 September 2023
Available online 25 September 2023
0038-092X/© 2023 International Solar Energy Society. Published by Elsevier Ltd. All rights reserved.
K. Khosravi et al. Solar Energy 264 (2023) 112025

Nomenclature μ Viscosity (Pa.s)


ρ Density (kg.m− 3)
A Area (m2) σ sb Stefan Boltzmann constant (W.m-2⋅K− 4)
aii Quadratic gradient factor
Cp Specific heat (J.kg− 1.j-1) Subscript
d Desirability a Ambient
D Collector tube diameter (m) AP Absorber plate
Ex Exergy (W) c Convection
g Gravity acceleration (m.s− 2) e Electrical
h Heat transfer coefficient (W.m-2.K− 1) in Inlet
k Thermal conductivity (W.m− 1.K− 1) ex Exergy
ṁ Mass flow rate (kg.s− 1) out outlet
p Pressure (Pa) PV Photovoltaic panel
q Heat flux (W.m− 2) ref Reference
r Radius (m) s Solid
Qs Solar radiation intensity th Thermal
T Temperature (K) w Wind
u Velocity (m/s) Abbreviation
( )
V Volume m3 ACO Ant Colony Optimization

V Velocity vector (m.s− 1) GA Genetic Algorithm
HTF Heat transfer fluid
Greek PCM Phase change material
α Absorptivity PSO Particle Swarm Optimization
β Temperature coefficient (K− 1) PVT Photovoltaic/thermal
ε Emissivity RSM Response Surface Methodology
η Efficiency (%)

Fig. 1. The schematic of the water-cooled PVT system.

components, leading to PV overheating. This effect drives the PV tem­ resulting in increased conversion efficiency and a longer lifespan for the
perature to surge over 80 ◦ C in some regions, whereas the PV efficiency PV cells. Furthermore, PVT systems can provide the benefit of thermal
is subject to about a 0.5 % drop for every degree above the standard and electrical yield, resulting in greater overall efficiency than PV sys­
operating temperature of 25 ◦ C [11]. In this context, photovoltaic tems used alone.
thermal (PVT) is an emerging technology combining solar PV and solar PVT systems are typically categorized by the type of heat transfer
thermal to produce electricity and useful heat from PV systems simul­ fluid (HTF) employed and can primarily be classified as air-based or
taneously [12]. In these systems, a thermal collector is connected un­ water-based. The HTF may later be used for space heating or to supply
derneath the panel (PV) to remove excess heat from PV components, domestic hot water after it has sufficiently cooled the PV cells [13,14].

2
K. Khosravi et al. Solar Energy 264 (2023) 112025

While air-type systems present lower risks in building-integrated PVT


structures, particularly in terms of freezing in chilly weathers and po­
tential damage to roofs from leaks, water-type PVT systems offer a
notable advantage due to their higher specific heat and thermal con­
ductivity [15]. This implies that water-type PVTs have a better potential
for thermal management on the PV side as they can remove larger
amounts of heat at relatively faster removal rates. However, many
limitations act as barriers to the widespread application of water-type
PVT systems. One of them is that water still has poor heat diffusivity
compared to other HTFs, which means that the thermal output of the
system would not be sufficiently hot to be useful for heating applica­
tions. Therefore, design and operational constraints should be carefully
considered to achieve ideal electrical and thermal yields within a
particular temperature range [16]. Further thermal performance
improvement may be accomplished by using forced convection tech­
niques, such as pumps and blowers, in conjunction with incorporating
heat transfer enhancers, such as expanded fins and nanoparticles.
Therefore, research efforts have been devoted to finding ways to
improve the thermal functionality of standard PVT systems by making
modifications to the design layouts (e.g., channel configuration, channel
size, and the number of passes), employing forced convection means (e.
g., pumps and blowers), and incorporating heat transfer enhancers such
as extended fins and flow baffles. As water-based PVT is the primary
focus of this research, the following text is intended to provide a concise
review of the literature on that topic.
Bahaidarah et al. [17] tested a water-cooled PVT system and

Table 1
Absorber plate and PV panel properties [44].
PV panel Absorber plate

Length = 2 m Length = 2 m
Fig. 2. Schematic of the helical wavy tube PVT system: (a) front view (b) Width = 1 m Width = 1 m
top view. ηref = 12%, Tref = 25 C

Material = copper
εPV = 0.88 εAP = 0.05
1 αAP = 0.95
β = 0.0045
K

Fig. 3. A schematic depicting the system’s overall mode of operation for building applications.

3
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 4. The generated mesh selected after the mesh analysis.

water jet-impingement cooling equipment installed on the PV backside.


Table 2 The results reveal an improvement in the PVT power outputs of about
m 62% compared to the conventional PV module under a solar irradiance
The grid independency for case 2 for pure water with uin = 0.08 in.
s of 1000 W/m2, a flow rate of 0.167 kg/s, and an ambient temperature of
Grid number Outlet temperature (K) PV temperature (K) about 30 ◦ C. Wu et al. [19] investigated the thermal characteristics of a
525,000 308.389 309.759 water-cooled PVT system by positioning the cooling channel above or
970,000 308.391 309.771 below the PV module. Results showed that convective heat transfer is
1,925,000 308.392 309.774 better with a relatively higher electric yield when the cooling channel is
at the bottom PV surface. Xiao et al. [20] analyzed the performance of
stepped PVT with solar still marking to improve PV’s cooling rate and
freshwater production. The study reported a 16% rise in thermal output
and a 52% increase in the daily production of freshwater. However, the
electrical output remains almost unchanged. Hissouf et al. [21] tested
three different tube layouts for distributing the coolant fluids in a PVT. It
was found that, compared to the square-tube and circular-tube layouts,
the half-tube design provides 1.17% and 2.6% higher maximum effi­
ciency, respectively. Chen et al. [22] showed that about 70% of solar
energy could be turned into usable energy by coupling a PVT collector
with a desalination system to produce freshwater and electrical output
simultaneously. Ould-Lahoucine et al. [23] found that using a PVT
system with either a rectangular channel or a sheet-and-tube circulating
a water-TiO2 nanofluid coolant can significantly improve PV cell cool­
ing and result in a temperature drop of 19.4 ◦ C and 11.48 ◦ C, respec­
tively, from an average surface temperature of 64.6 ◦ C. Recently,
Shojaeefard et al. [24] tested a cooling system made up of an array of
copper pipes attached to the back of PV module at different inlet water
temperatures (ranging from 19.5 to 61 ◦ C). Their results showed that the
cooling water system can effectively reduce the PV surface temperature,
dropping it from 58.5 ◦ C to 33.5 ◦ C at 9:00 a.m. and from 71 ◦ C to 68 ◦ C
at 1:30p.m.
Finding an optimal engineering design across various dependent
Fig. 5. The validation study of the PVT unit comparted with Qinghua Yu variables through numerical simulations or experimental tests can often
et al. [45]. be time-consuming and costly. This process necessitates using statistical
and mathematical methods to model and analyze problems in which
reported a 9% increase in electrical efficiency compared to a standalone multiple independent variables influence a dependent variable or
PV system. It was found that the PVT system captured nearly 750 W of response. In the context of PVT systems, the independent variables
energy under 900 W/m2 irradiance, surpassing the standalone PV sys­ might include incoming solar radiation, ambient temperature, wind
tem’s 190 W production. Hasan et al. [18] tested a PVT system with speed, and various characteristics of the PVT system, such as its size,

4
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 6. Methodology flowchart for the modeling, simulation, prediction, and optimization.

Table 3
Presenting an example of the results of this work.
Case 21 Diameter Velocity Width Length Qs ṁ (kg/s) Tout Tin
(mm) (m/s) (mm) (mm) (W/m2) (K) (K)
12 0.08 200 2000 1000 0.009 307.92 298.15

Electrical Thermal Overall Power Electrical Thermal Overall Destroyed TPV Absorber
power Power Efficiency exergy exergy Exergy Exergy (W) Heat (W)
efficiency Efficiency efficiency efficiency efficiency
11.10 62.14 73.24 11.92 1.59 13.51 322.14 314.87 367.72

materials used, and design. The “response” typically measures the sys­ techniques have been developed, including the Genetic Algorithm (GA),
tem’s performance, like its thermal efficiency, electrical efficiency, or Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO),
overall energy output. Over the last three decades, several optimization and Response Surface Method (RSM). These techniques can provide a

Fig. 7. The contour profile of the temperature distribution on the PV panel for the data is shown in Table 3.

5
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 8. The contour profile of the temperature distribution on the PV panel for (a) inlet velocity = 0.04 m/s and (b) inlet velocity = 0.12 m/s.

swift response, making them a practical and cost-efficient tool for system’s output power by about 48% under optimal conditions. Another
evaluating a system’s performance under multiple objective variables. study conducted by Gelis et al. [29] employed the RSM to explore the
Among these, RSM is the most widely used in various experimental and impact of nanofluid on the electrical and thermal efficiency of PVT
computational studies because it is precise, quick to compute, and easy systems. They suspended SiO2 nanoparticles in deionized water as a base
to use [25,26]. For instance, Hatami and Jing [27] reported that the coolant and tested flow rate, nanoparticle concentration, and solar ra­
RSM was a very effective and readily adaptable optimization method­ diation as independent variables. Results showed that solar radiation,
ology for determining which wavy profile had the most desirable nanoparticle concentration, and nanofluid flow rate increase electrical
operating characteristics for a wavy solar collector incorporating a efficiency. The thermal efficiency of the PVT system increases with
nanofluid. volumetric concentration and decreases with solar radiation and nano­
In recent years, a few studies have applied RSM analysis to water- fluid flowrate.
cooled PVT systems to improve their energetic conversion efficiency. The primary objective of the present study is to conduct a compre­
These studies aim to identify the optimal values of various governing hensive analysis using the RSM technique to analyze the impact of a
parameters, such as flow rate, temperature, and irradiance, that can range of operational parameters on the performance of water-cooled
maximize the system’s energy output. One study by Javidan and Mog­ PVT systems. This is accomplished by developing a dynamic three-
hadam [28] applied RSM to optimize a water-cooled PVT design dimensional (3D) simulation model that optimizes both energy and
featuring a jet impingement system. The results of the RSM analysis exergy efficiencies of a water-based PVT system based on RSM-guided
showed that the optimal operating parameters significantly boost elec­ multi-objective optimization. The study attempts to bridge a critical
trical power output and moderate PV module temperature, ensuring gap in the existing literature by conducting an extensive energy and
temperature uniformity across PV cells. Applying these parameters exergy analysis with an optimized configuration for the water-cooling
reduced the PV surface temperature from 64 ◦ C to 34 ◦ C, increasing the channel to achieve high electrical and thermal outputs. To the

6
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 9. Effect of the inlet velocity on the (a) thermal and electrical energy efficiencies, (b) thermal and electrical exergy efficiencies, (c) overall energy and exergy
efficiencies, and (d) thermal and electrical exergies.

authors’ knowledge, no previous study has integrated 3D numerical for cooling purposes. All surfaces are considered insulated except the
simulation with RSM-based multi-objective optimization to evaluate the upper layer, which absorbs solar energy from the sun. Due to the sym­
thermal characteristics of water-cooled PVT systems. The present study metrical condition, only 1/5 of the whole system is considered for the
employs RSM as a predictive tool to analyze the system’s performance computational domain [30,31]. The length of the panel is considered in
based on five independent parameters: the flow rate of the coolant, the the range of 1500 mm to 2500 mm. The absorber plate absorbs the solar
size of the coolant channel, the panel’s dimensions, and the level of solar thermal energy depending on the absorber’s solar radiation intensity
radiation. The dependent variables, including thermal and electrical and absorptivity. It should be mentioned that a real PV panel consists of
energy efficiencies, exergetic efficiencies, and overall efficiency, are multilayer materials including glass, PV panel, EVA, Tedlar, etc. which
evaluated using RSM. Various scenarios for single and multi-objective are effective based on conduction heat transfer; however, based on the
optimizations are considered to propose an efficient design for PVT aim of this paper and also due to prevent extra complexity of the
systems. In essence, this study enriches the field of water-based PVT problem, it was assumed that perfect conduction heat transfer is
systems by providing an in-depth optimization process with energy and assumed for different layers of the PV above the absorber. This process
exergy analysis. has been widely used in the literature to simply the problem [32,33].
Fig. 3 describes the system operation schematic for building appli­
2. Numerical modeling cations. Alongside electricity generated by the PV module, the system
utilizes the heat removed from the PV module to heat water through an
This study investigates a PVT system integrated with cooling tubes. auxiliary heater. This allows the water to attain the necessary temper­
Fig. 1 shows the proposed PVT system, including the PV panel, the ature for utilization within the building’s heating system. The water can
absorber plate and the cooling pipe. The boundary conditions are also be also used for producing hot water in residential and industrial
displayed in Fig. 1. Water is considered as the HTF to regulate the buildings in a close loop system to prevent any dissipation.
temperature of the PV panel. The PV cell is located on the upper surface, The PVT model used in the present analysis is simplified by the use of
the absorber plate is in the middle, and the cooling tubes are placed at the following assumptions [34]:
the bottom of the module. As illustrated in Fig. 1, the boundary condi­
tion of the upper surface is the incoming solar irradiance and heat losses • The fluid flow is incompressible [35],
through convection and radiation mechanisms. The PV-T system • The problem is steady-state,
examined in this study has a fixed-tilt angle. It is considered that the • The flow regimes are laminar [36],
solar radiation is deemed perpendicular to the upper surface of the PV • Insulation is thought to exist between the tube’s exterior surface and
panel. the plate’s bottom which acts as the absorber.
The schematic of the present module is shown in Fig. 2. As seen in • Combined effects of convective and radiative heat losses are
Fig. 2a, the PV system benefits from five straight tubes under the panel considered [37].

7
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 10. The contour profile of the temperature distribution on the PV panel for (a) HTF tube diameter = 8 mm and (b) HTF tube diameter = 16 mm.

• The absorber plate and the PV panel have perfect contact [38]. →
Hence, it is simplified that their temperatures are equal. ∇⋅ V = 0 (1)

(→ ) ∂p ( )
Despite the time-varying nature of solar radiation, the assumption of ρ V .∇ui = − + μ ∇2 ui − ρgi (2)
steady-state conditions is academically justified for several reasons. By ∂xi
assuming steady-state behavior, the analysis can focus on the average (→ )
performance of the system, simplifying the mathematical model and ρCp V .∇T = k∇2 T (3)
reducing computational complexity. Time-averaging solar radiation
data allows for meaningful insights into the system’s thermal and elec­ For the solid part, the energy equation is given as [41]:
trical performance while accounting for typical operating conditions. 2
k∇ T = 0 (4)
Although the assumption may overlook transient effects, future studies
could explore the impact of time-varying solar radiation for a more The upper surface of the system absorbs solar energy from the sun. A
comprehensive understanding of the system’s behavior. Moreover, the fraction of this energy is absorbed, transferred to the HTF, and a fraction
velocity in this study ranges from 0.04 m/s to 0.12 m/s and thus the is dissipated due to convection and radiation losses. The net heat
Reynolds number in all considered cases is below 2300, indicating absorbed which is absorbed by the PV cell is calculated as [42]:
laminar flow. Also, to simplify and avoid complexity, a solar radiation ( )
qAp = Qs (αPV − ηe ) − εPV σ sb T PV 4 − T a 4 − hc (T PV − T a ) (5)
value is considered constant in all the simulation to have a steady state
condition. This auusmption has been widely used in the literature where σsb = 5.76 × 10− 8 mW2 K [36] and hc = 3ua +2.8 [43].
[6,32,33,39].
All of the components having touch with the thermal insulation are
The governing equations and boundary conditions are listed below:
assumed adiabatic. Both the absorber plate and the PV panel sides have
For steady-state laminar fluid flow, the continuity, momentum, and
two surfaces regarded to be symmetrical. A velocity inlet and pressure
energy equations are presented as follows [40]:
outlet are considered at the fluid’s intake and outflow. In every scenario,

8
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 11. The effect of the HTF tube diameter on the (a) thermal and electrical energy efficiencies, (b) thermal and electrical exergy efficiencies, (c) overall energy and
exergy efficiencies, and (d) thermal and electrical exergies.

the temperature measured at the HTF’s input is called the ambient


Exdestroyed = Exsolar − [Exelectrical + Exth ] (12)
temperature (298.15 K). The no-slip boundary condition is imposed
along the tube walls. Table 1 displays the properties of the absorber The term in the bracket is useful exergy. The exergy efficiency is then
plate and PV panel. The electrical efficiency (ηref ), is calculated at the given as follows [52]:
reference temperature when the PV temperature equals the reference
Exelectrical + Exth
value. Note that the material of the pipe is also copper. ηII = (13)
Exsolar
The electrical, thermal, and solar exergy are calculated as follows
2.1. Energy and exergy evaluation
[53]:
( )
According to the resources analysis, the thermoelectric performance T0
Exth = ṁf luid Cp,f luid (T out − T in )⋅ 1 − (14)
of the PV panel is defined as [45,46]: Tm

qAp dA
(6) T out − T in
A
ηI,th = Ap
Qs AAp Tm = (15)
ln TTout
in

ηI,electrical = ηpv (7) where [54,55]:

where ηpv and ηI are the panel efficiency and the first law efficiency Exelectrical = ηpv Qs Apv (16)
calculated as [47,48]:
The amount of solar exergy gained by the PV components is calcu­
( ( ))
ηpv = ηref 1 − β T pv − T ref (8) lated as [53]:
( ( )4 )
4 Ta 1 Ta
ηI = ηoverall = ηI,th + ηI,electrical (9) Exsolar = Qs AAP 1 − + (17)
3 T sun 3 T sun
The implementation of the second law of thermodynamics is
required for the exergy evaluation. While the energy evaluation pro­ where Tsun is 5774 K.
vides a measurement of the amount of energy, the exergy evaluation It should be noted that, in this study, energy and exergy analyses
reveals the quality of the energy. The exergy for the PVT assembly can be have been exclusively considered for the PV-T component alone.
estimated as [49,50]:
Exfluid,in + Exsolar = Exfluid,out + Exelectrical + Exdestroyed (10)
2.2. Grid independency and model validation
Exth = Exfluid,out − Exfluid,in (11)
After creating geometry using ANSYS Design Modeler (version
Eq. (10) can be written as [49,51]: 2022), the ANSYS Mesher is employed to generate the mesh. The

9
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 12. The contour profile of the temperature distribution on the PV panel for (a) width of the panel = 150 mm, and (b) width of the panel = 200 mm.

SIMPLE algorithm is implemented to obtain the pressure field, and the a study where they employed a straight tube under the PV module to
convective transport terms in the governing equations of conservation thermally manage the working temperature, enhance electrical power
are discretized using the Quick scheme. The convergence conditions for output, and harness thermal power from the PV system. They utilized a
the three conservation equations of mass, momentum, and energy are microencapsulated slurry with a water-based fluid for this purpose. The
set to be 10-6, 10-6, and 10-8, respectively. The Design Expert software module’s dimensions in their research were 2000 mm in length and
package (version 13) is also used for optimization purposes. 1000 mm in width, with 1640 mm covered by the panel. The cooling
To achieve the structured mesh, the geometry is divided into blocks tubes had a length of 2000 mm and a diameter of 8 mm. The coolant
and then meshed using the swipe element method. Grids with various inlet temperature and irradiation intensity were set at 298.15 K and
numbers of mesh structures are explored to guarantee grid indepen­ 1000 W/m2, respectively. Fig. 5 illustrates the variation in PV temper­
dency. Fig. 4 shows the generated mesh for case 2. This figure shows that ature in relation to the water inlet velocity for both present study and the
smaller meshing is implemented near the solid–fluid interfaces due to model presented by Qinghua Yu et al. [45]. The results demonstrate
the large velocity and temperature gradients. The results of grid inde­ almost a similar trend with a satisfactory level of accuracy. The differ­
pendency for case 3 are summarized in Table 2. The total number of ence and error between the validation model and our results are within
nodes is varied from 525,000 to 1,925,000. Outlet and PV temperatures an acceptable range of 2.5% (1 ◦ C). This discrepancy could be attributed
are picked to assess the grid independency. It is detected that the results to variations in grid numbers, post-processing, computational errors and
are almost similar for all the studied cases. The temperature is virtually simplifying assumptions.
no different by increasing the number of mesh elements above 970,000 It should be noted that the evaluating and enhancing the efficiency of
grids. As such, this number of mesh elements is chosen for future studies. photovoltaic solar modules connected with linear tubes for water supply
The current model’s predictions are compared with earlier published has some limitations, including a simplified model, a limited parameter
data on laminar flow to validate the developed numerical model. The range, and potential discrepancies between predicted and actual results
numerical predictions of Qinghua Yu et al. [45] are used as a benchmark which are the reasons for the difference between the numerical model
for validating the laminar flow model. Qinghua Yu et al. [45] conducted presented in this study and the actual results.

10
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 13. The effect of the width of the panel on the (a) thermal and electrical energy efficiencies, (b) thermal and electrical exergy efficiencies, (c) overall energy and
exergy efficiencies, and (d) thermal and electrical exergies.

3. Response surface method competence values, electrical and thermal exergy efficiency, the whole
energy and exergy efficiencies, and destroyed exergy as a function of
Optimization of the water-based PVT unit based on energy, exergy, control factors (i.e. dimeter of the tube, velocity, width and length of
and exergy destruction, utilizing reaction surface methodology and panel and solar radiation intensity).
considering the geometrical parameters and solar radiation intensity Accordingly, the second-order polynomial function is utilized, which
simultaneously, has not been carried out so far. Accordingly, the major owns few benefits such as: enhancing the expectation authenticity of the
purpose of this study to resolve the stated weaknesses can be summed up prediction method, limiting the technique divergent, and nominating
as follow: the ideal approach when finite numbers of control characteristics are
examined. This formula can be highlighted as follows [58]:
• To predict thermal/electrical energy and exergy and destroyed

n n ∑
∑ n
exergy. y = a0 + aii x2i + aii xi xj , i < j (18)
• To obtain the interaction effects of design factors and their responses. i=1 i=1 j=1

• To employ several operational scenarios for the unit in charge to


optimize design variables by using single and multi-objective where y is the method’s response, and n demonstrates the number of
schemes. characteristics. A0 is a constant factor, ai is the linear-gradient factor of
ith factors, aii is the quadratic gradient factor of the ith coefficient, and aij
Fig. 6 provides a concise flowchart to help understand the simulation is the reaction of ith and jth coefficients. x is the configuration variable.
and optimization processes. The RSM model proposes various configuration plans to describe the
coefficient level, where the Central Composite Design (CCD) is utilized
in this work [59]. Three levels of CCD are used to run the computational
3.1. Predictive model method for four various factors, involving the diameter of the tube (D),
velocity (m/s), the width of the panel (W), the length of the panel (L),
Unlike the traditional optimization techniques, which do not and the solar radiation intensity (Qs) as independent parameters.
consider the characteristics of communications and the responses [32], Consequently, the total number of simulations required for RSM (N) can
the RSM technique can efficiently react to this interaction, reducing the be found as [58,60]:
requested experiment number for the current work and spawning more
efficient numerical and practical processes [56]. The RSM was primarily N = 2k + 2k + nc (19)
used in practical work, but it has recently been used in computational
which k and nc describe the control operators and central configuration
studies. It numerically converts to a unique technique because of its
points, respectively.
significant advantages, such as accelerating the computational process
and predicting nonlinear ideal formulae [57]. This method illustrates
the connection between the design factor and the responses. However, a
link can be happened for every seven responses of electro-thermal power

11
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 14. The contour profile of the temperature distribution on the PV panel for (a) length of the panel = 1500 mm, and (b) length of the panel = 2500 mm.

3.2. Analysis of variance (ANOVA) ∑n


(yi − ỹi )2
R2 = 1 − ∑i=1 (22)
The ANOVA test is utilized to estimate the relevance and significance
n 2
i=1 (yi − yi )
of the predictive model. F and P-values and design coefficients are used
for the current resolve to assess the method’s competence [61]. The P Radj2 = (1 − R2 )
n− 1
(23)
and F-value present the importance of the gradient methods and the n− p− 1
influence of the factors. “F-values” less than 0.05 indicate that the
∑n 2
technique is essential. Likewise, P-values characterize the importance of yi − ̃
yi
i=1 (1− hii )
every term in the analytical method associated with other factors, where R 2
pred = 1 − ∑n (24)
i=1 (yi − yi )2
[62]:

1∑ n where n − p − 1 is the error of the freedom degrees, n − 1 is the total


yi = y (20) corrected freedom degrees, and hii is the ith diagonal element of the H =
n i=1 i
X(XT X)− 1 XT .
and
y = Xa
̃ (21) 3.3. Optimization
y represents the observations vector (n × 1), X shows the levels of the
To improve the stated configuration coefficients of the method,
independent variables (n × p), and a is the regression coefficients vector
multi-response optimization is used, employing Derringer′s suitability
(p × 1). Regarding the method suitability, resolve factors, including R2 ,
function coupled with RSM [64]. The interest function modifies each
R2 adj , and R2 pred are the significant factors, differing in the range of
reaction to a dimensionless individual interest scale (di), varying be­
0–1, where the value of 1 is the ideal fitness. The following equation
tween 0 as an unattractive value and 1, showing all responses within a
describes its formula [63].
desirable limit. Three potential plans for y can be studied depending on

12
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 15. The effect of the length of the panel on the (a) thermal and electrical energy efficiencies, (b) thermal and electrical exergy efficiencies, (c) overall energy and
exergy efficiencies, and (d) thermal and electrical exergies.

the optimization principle. To maximize the attractiveness, the √ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅


following formula is utilized [59]: D= m
d1 d2 ...dm (28)
⎧ ⎫

⎪ 0 if yi ≤ Low ⎪
⎪ where m describes the number of responses.

⎪ ⎪

⎨ y − Low ⎬
(25)
i
di = (
⎪ High − Low
) if Low ≤ yi ≤ High
⎪ 4. Results and discussions

⎪ ⎪


⎩ ⎪

1 if y ≥ High
i
In this section, the impact of various factors will be investigated.
Low and high indicate the appropriate boundaries of the worst and Next, a predictive model will be developed, and the effect of developing
best responses. If the minimum total interest is the last target, the variables using obtained predictive model will be presented. Finally, the
following formula can be employed [59]: optimum values for the design variables will be obtained.
⎧ ⎫

⎪ 1 if yi ≤ Low ⎪
⎪ 4.1. Effect of design parameters

⎪ ⎪

⎨ High − y ⎬
di = ( i
) if Low ≤ yi ≤ High (26)



High − Low ⎪

⎪ A Parametric study is performed in this section by simulating a
⎪ ⎪
⎩ 0 if y ≥ High
i
⎭ photovoltaic thermal hybrid solar collector (PVT) with a straight pipe
under the panel for cooling to examine the effect of various parameters.
When the target (TG) is in a specific desirable range, di can be An example of the results is represented as data in Table 3 for case 21,
calculated as [59]: which is introduced later in the optimization section. The influential
⎧ ⎫ independent parameters are the tube diameter (12 mm), the fluid flow
⎪ 0 if yi ⩽Low ⎪





⎪ velocity (0.08 m/s), and the dimensions of the panel (1000 mm × 2000
⎪ ⎪



⎪(
yi − Low s
) if Low⩽yi ⩽TG ⎪


⎪ mm – as discussed, for simplicity and due to the symmetrical boundary
⎨ ⎬
TG − Low condition, a one-fifth of the panel’s width is simulated which is equal to
di = ( )t (27)

⎪ High − yi ⎪ 200 mm), and the inlet temperature of the fluid (298.15 K). These pa­

⎪ if TG⩽yi ⩽High ⎪⎪



⎪ High − Low ⎪

⎪ rameters produce overall energetic and exergetic efficiencies of 73.24 %
⎪ ⎪

⎩ 1 if y ⩾High ⎭
⎪ and 13.51 %, respectively. The temperature of the PV reaches and sta­
i
bles at 314.87 K with the absorber heat rejection of 367.72 W. These
parameters and results show the system operation procedure.
where the overall desirability function (D) calculates the weighted
Fig. 7 shows the temperature distribution of the PV module. The
geometric average of di, which incorporates the emphasis on the lower
figure shows a lower temperature at the beginning of the panel, which
or higher boundaries through weighted range values (0 to 10) assigned
increases over the panel into the outlet since the temperature difference
to s and t in the above formula [64]:
between the panel and fluid is higher at the beginning, resulting in a
higher heat rejection from the panel to the fluid. The temperature shows

13
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 16. The contour -profile of the temperature distribution on the PV panel for (a) solar radiation intensity = 700 W/m2 and (b) solar radiation intensity = 1000
W/m2.

higher values at the edges of the panel, dropping the temperature from heat transfer between the HTF and the PV panel, resulting in increased
the border to the center (the place of contact with the HTF tube). At the thermal power and reduced PV temperature, thereby leading to greater
end of the panel, the temperature at the edge is around 47 ◦ C and 42 ◦ C electrical power production. The values presented in Fig. 9a show that
at the center. increasing the HTF velocity from 0.04 to 0.12 m/s results in a 2% in­
crease in electrical power and a 10% increase in thermal power,
4.1.1. Effect of the inlet velocity resulting in an overall power efficiency improvement of 8.7%. The in­
Fig. 8 shows the contours of the temperature distribution of the PV crease in electrical exergy efficiency from 12.1% to 12.3% when the HTF
panel for the velocities of 0.04 and 0.12 m/s. With the fluid’s slower inlet velocity increases from 0.04 to 0.12 m/s indicates that the system’s
movement, the panel releases less thermal energy to the fluid through performance exceeds that of the reverse condition. However, the ther­
the wall of the tube, and then the HTF has a higher temperature at the mal exergy efficiency decreases by 0.4 W as the flow velocity increases
outlet section. The temperature of the PV panel is relatively high from 0.04 to 0.12 m/s.
because of the less temperature difference between the panel and the Consequently, the exergy efficiency decreases from 13.6% to 13.3%
HTF (Fig. 8a). However, in the case of the 0.12 m/s (Fig. 8b), the fluid (Fig. 9c). Fig. 9d illustrates both the electrical and thermal exergies at
gains more heat from the panel due to the higher temperature difference different HTF velocities. Additionally, increasing the HTF velocity from
between the HTF and the panel, causing a colder surface of the PV panel. 0.04 to 0.12 m/s decreases the maximum thermal work required to bring
The average temperature registered of the HTF for both cases (0.04 and the system back to equilibrium from 3.7 to 2.5 W. However, it increases
0.12 m/s) is 41 ◦ C and 34 ◦ C, respectively. the maximum electrical work required by 1.9%, from 31.42 to 32.03 W,
The heat transfer fluid’s (HTF) velocity plays a significant role in the due to the changes in velocity. These observations are supported by the
efficiency of the photovoltaic/thermal (PVT) system. Increasing the HTF data presented in the figures and provide insights into the performance
velocity in the tube leads to improvements in both electrical and thermal characteristics of the PVT system under varying HTF velocities.
power efficiencies. This is because the flow rate or velocity enhances

14
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 17. The effect of the solar radiation intensity on (a) thermal and electrical energy efficiencies, (b) thermal and electrical exergy efficiencies, (c) overall energy
and exergy efficiencies, and (d) thermal and electrical exergies.

Table 4
Level of configuration factors for the predictive model.
Design parameters Definition Low or − α (-1) central High or +α (+1)

D (mm) Diameter 8 12 16
u(m/s) Inlet velocity 0.04 0.08 0.12
W (mm) Width 150 200 250
L (mm) Length 1500 2000 2500
( )
Qs W/m2 Solar radiation intensity 400 600 1000

4.1.2. Effect of diameter velocity increases the fluid’s flow rate, resulting in a higher heat transfer
Fig. 10 shows the temperature of the PV panel for the cases of 8 mm rate. Increasing the diameter from 8 to 16 mm enhances the electrical
and 16 mm HTF tube diameters. Fig. 10a clearly indicates that the HTF and thermal energy efficiencies by 4.6% and 12.6%, prospectively.
at the lower diameter gains less heat from the panel due to the lower Fig. 11b indicates that the electrical and thermal exergy efficiencies are
fluid flow rate of the HTF. This causes a higher panel temperature at an directly and inversely proportional to the diameter of the HTF tube. The
8 mm tube diameter, which is higher at the edge of the panel and drops electrical exergy efficiency increases from 11.8% to 12% and then to
toward the center. With a larger diameter, the flow rate of the fluid is 12.2%, and the thermal exergy efficiency drops from 2.2% to 1.6% and
higher, which absorbs more heat from the panel, causing a lower then to 1.3% when the diameter expands from 8 to 12 and then to 16
average temperature on the panel’s surface (Fig. 10b). The average mm. The overall energy and exergy efficiencies with different diameters
temperature of the fluid in the case of 8 mm reaches 42 ◦ C, and this value are shown in Fig. 11c. The system’s overall energy efficiency increases
drops from 16 mm to 36 ◦ C. from 70.1% to 78.1% when the diameter changes from 8 to 16 mm due
The diameter of the HTF tube affects the system’s performance, as it to the increasing flow rate of the HTF passing the tube, which gains more
influences the thermal behavior via the Reynolds number or flow rate. heat from the panel. Increasing the diameter causes a drop in the overall
At a specific velocity, the flow rate of a fluid is directly proportional to exergy efficiency from 13.9% to 13.4%. Increasing the flow rate of the
the diameter of the tube. Increasing the tube diameter increases the flow HTF due to increasing the diameter of the tube causes an improvement
rate while decreasing the tube diameter decreases the flow rate. This in the electrical exergy by 3.7% when the diameter increases from 8 to
relationship is based on continuity, which ensures mass conservation 16 mm; however, the thermal exergy drops from 4.6 to 2.7 W.
along the flow path. Fig. 11a shows the system’s electrical and thermal
energy efficiencies at the velocity of 0.04 m/s. Increasing the diameter is 4.1.3. Effect of width
directly proportional to the electrical and thermal energy efficiencies Enlarging the panel’s width will increase the panel’s surface tem­
since enlarging the tube’s cross-section area with a constant HTF perature as the HTF’s flow rate is constant (Fig. 12). The heating surface

15
K. Khosravi et al. Solar Energy 264 (2023) 112025

Table 5
Design matrix required for RSM.
cases Design parameters Responses
Diameter Inlet Width Length Solar Electrical Thermal Overall Electrical Thermal Overall Destroyed
(mm) velocity (mm) (mm) radiation energy energy energy exergy exergy exergy exergy (W)
(m/s) intensity efficiency efficiency efficiency efficiency efficiency efficiency
(W/m^2) (%) (%) (%) (%) (%) (%)

1 8 0.04 250 1500 400 11.37 49.80 61.18 12.21 1.03 13.25 121.17
2 8 0.08 200 1500 700 11.26 59.20 70.46 12.09 1.33 13.42 169.29
3 8 0.12 150 1500 400 11.71 65.40 77.12 12.58 0.57 13.15 72.78
4 16 0.12 150 1500 400 11.82 70.03 81.85 12.69 0.33 13.03 72.89
5 8 0.04 150 1500 400 11.58 59.25 70.83 12.43 0.90 13.33 72.63
6 8 0.12 250 1500 400 11.54 57.73 69.27 12.40 0.73 13.13 121.34
7 8 0.12 150 1500 1000 11.29 65.50 76.79 12.12 1.42 13.54 181.15
8 8 0.04 150 1500 1000 10.95 59.19 70.14 11.75 2.18 13.94 180.31
9 16 0.12 250 1500 400 11.69 64.13 75.82 12.55 0.46 13.01 121.50
10 16 0.04 150 1500 400 11.74 66.63 78.37 12.61 0.51 13.12 72.81
11 16 0.12 150 1500 1000 11.55 70.17 81.71 12.40 0.84 13.23 181.78
12 16 0.04 250 1500 1000 10.95 59.36 70.31 11.76 1.64 13.41 302.37
13 8 0.12 250 1500 1000 10.86 57.64 68.50 11.67 1.79 13.45 302.21
14 16 0.04 250 1500 400 11.58 59.39 70.97 12.44 0.66 13.10 121.37
15 16 0.12 250 1500 1000 11.21 64.21 75.43 12.04 1.15 13.20 303.10
16 16 0.04 150 1500 1000 11.35 66.75 78.10 12.19 1.27 13.46 181.30
17 8 0.04 250 1500 1000 10.44 49.41 59.85 11.21 2.46 13.67 301.45
18 12 0.08 200 2000 700 11.37 62.12 73.49 12.21 1.12 13.33 225.96
19 12 0.04 200 2000 700 11.22 58.18 69.40 12.05 1.42 13.47 225.59
20 12 0.12 200 2000 700 11.44 63.96 75.41 12.29 0.97 13.26 226.14
21 12 0.08 200 2000 1000 11.10 62.14 73.24 11.92 1.59 13.51 322.14
22 12 0.04 200 2000 1000 10.89 58.14 69.03 11.69 2.01 13.70 321.43
23 12 0.12 200 2000 1000 11.20 64.00 75.20 12.03 1.39 13.42 322.49
24 12 0.08 150 2000 700 11.52 65.98 77.50 12.37 0.95 13.33 169.48
25 12 0.04 150 2000 700 11.39 62.68 74.08 12.24 1.25 13.49 169.17
26 12 0.12 150 2000 700 11.58 67.50 79.08 12.44 0.82 13.26 169.62
27 12 0.08 250 2000 700 11.23 58.41 69.64 12.06 1.23 13.29 282.60
28 12 0.04 250 2000 700 11.07 54.02 65.09 11.89 1.52 13.40 282.22
29 12 0.12 250 2000 700 11.31 60.49 71.80 12.15 1.08 13.23 282.80
30 12 0.08 250 2000 400 11.56 58.44 70.00 12.42 0.71 13.12 161.79
31 12 0.04 250 2000 400 11.47 54.13 65.59 12.31 0.88 13.19 161.66
32 12 0.12 250 2000 400 11.61 60.49 72.10 12.46 0.62 13.08 161.86
33 12 0.08 150 2000 1000 11.31 66.04 77.35 12.15 1.36 13.51 241.61
34 12 0.08 150 2500 1000 11.26 65.09 76.35 12.10 1.48 13.58 301.77
35 8 0.04 150 2500 1000 10.71 54.64 65.35 11.50 2.69 14.19 299.62
36 16 0.12 250 2500 400 11.64 62.16 73.80 12.50 0.55 13.06 202.39
37 8 0.04 250 2500 400 11.26 44.54 55.80 12.09 1.22 13.31 201.79
38 8 0.12 150 2500 400 11.66 63.00 74.66 12.52 0.70 13.23 121.20
39 16 0.04 150 2500 400 11.69 64.51 76.20 12.56 0.63 13.19 121.26
40 16 0.12 150 2500 1000 11.47 68.76 80.23 12.31 1.03 13.34 302.60
41 16 0.12 150 2500 400 11.79 68.63 80.42 12.66 0.41 13.07 121.42
42 8 0.04 250 2500 1000 10.10 42.49 52.59 10.85 2.64 13.49 503.45
43 16 0.04 150 2500 1000 11.24 64.59 75.83 12.07 1.55 13.62 301.63
44 8 0.12 150 2500 1000 11.15 63.05 74.20 11.98 1.74 13.71 301.30
45 8 0.12 250 2500 1000 10.69 54.36 65.06 11.48 2.10 13.58 502.94
46 16 0.04 250 2500 1000 10.80 56.42 67.22 11.60 1.93 13.53 503.24
47 8 0.04 150 2500 400 11.48 54.87 66.35 12.33 1.12 13.46 120.88
48 16 0.12 250 2500 1000 11.10 62.19 73.30 11.93 1.38 13.31 504.53
49 16 0.04 250 2500 400 11.52 56.55 68.07 12.37 0.79 13.16 202.15
50 8 0.12 250 2500 400 11.48 54.57 66.04 12.32 0.87 13.19 202.08

area is reduced by utilizing the narrower width of 150 mm, as depicted the panel, which reduces the system’s efficiency. Electrical and thermal
in Fig. 12a. Consequently, the HTF positioned at the panel’s center can energy efficiencies are reduced by 3.7% and 11.2% when the width of
gain more thermal energy from the panel’s surface. As a result, the the panel changes from 150 to 250 mm. The electrical and thermal
panel’s average temperature is lower compared to the scenario where a exergy efficiencies at various widths of the panel are shown in Fig. 13b.
broader PV panel (200 mm) is employed, as illustrated in Fig. 12b. With The thermal exergy efficiency is proportional to the width of the panel
the broader width of the panel, the surface gains more heat from the sun, due to gaining more heat at the surface. Unlike wise, the electrical
which increases its temperature, and because of locating the HTF tube at exergy efficiency is inversely proportional to the width of the panel. The
the center of the board, the temperature of the edge part remains high as thermal exergy efficiency increases from 1.36% to 1.79%, while the
the HTF cannot gain heat ideally from the edge of the panel causing a electrical exergy efficiency reduces from 12.15% to 11.75% when the
higher average temperature of the panel surface. The average temper­ width enlarges from 150 to 250 mm. Fig. 13c shows that the overall
atures recorded for the PV panel are 36 ◦ C and 42 ◦ C for the cases using energy efficiency drops with expanding the width of the panel due to the
150 mm and 250 mm panel width, respectively. high useless energy, which is presented as the heat on the surface.
Fig. 13a shows that the thermal and electrical energy efficiencies However, the overall exergy efficiency increases with increasing the
reduce with increasing the panel width. Increasing the width of the PV width of the panel’s surface. The overall energy efficiency drops by
panel enlarges the radiation absorber area and makes the panel warmer. 10.3%, and the overall exergy efficiency increases by 0.002%, which
Using a specific tube diameter limits the heat absorption process from consider a negligible value when the width of the panel enlarges from

16
K. Khosravi et al. Solar Energy 264 (2023) 112025

Table 6
Coefficient of each term in the predictive equations.
Electrical Energy (Thermal Energy (Overall Energy Electrical Exergy 1/Sqrt (Thermal Overall Exergy Ln (Destroyed
Efficiency Efficiency)^2.5 Efficiency)^3 Efficiency Exergy Efficiency) Efficiency Exergy)

Factor Coefficient Estimate

Intercept 11.38 30468.04 3.972E + 05 12.22 0.9490 13.35 5.42


A-Diameter 0.1519 4518.59 61380.40 0.1632 0.1231 − 0.1181 0.0014
B-Inlet Velocity 0.1179 3527.47 47948.94 0.1266 0.0964 − 0.0971 0.0011
C-Width − 0.1551 − 4633.62 − 63087.41 − 0.1665 − 0.0578 − 0.0470 0.2560
D-Length − 0.0568 − 1660.58 − 22584.66 − 0.0611 − 0.0473 0.0423 0.2549
E-Solar Radiation − 0.2990 − 31.17 − 4755.03 − 0.3211 − 0.2231 0.1814 0.4560
Intensity
AB − 0.0347 − 723.04 − 9028.44 − 0.0372 0.0120 – –
AC 0.0198 +183.78 1449.16 0.0212 − 0.0218 0.0367 − 0.0004
AD 0.0174 349.09 4329.88 0.0187 − 0.0068
AE 0.0656 88.72 1528.16 0.0704 − 0.0310 − 0.0365 0.0004
BC 0.0150 132.73 1002.25 0.0161 − 0.0172 0.0287 − 0.0003
BD 0.0147 288.90 3583.85 0.0158 − 0.0058 – –
BE 0.0523 77.91 1294.25 0.0562 − 0.0245 − 0.0282 0.0003
CD − 0.0092 − 78.39 – − 0.0099 0.0084 – –
CE − 0.0660 − 84.40 − 1439.47 − 0.0708 0.0146 − 0.0262 0.0004
DE − 0.0247 – – − 0.0266 0.0127 – –
A2 − 0.0413 − 1018.36 − 19310.88 − 0.0443 – – –
B2 − 0.0466 − 1190.15 − 15461.47 − 0.0500 − 0.0236 – − 0.0006
C2 – 305.64 4438.80 – 0.0160 – − 0.0319
D2 – – 6215.59 – – – − 0.0325
E2 – – – – 0.0719 – − 0.1014

Fig. 18. The graphs demonstrate the approach’s suitability for the PVT structure, showing the differences between (a) predicted and actual values, (b) ordinary
probability, and (c) inner studentized residuals and expected data.

150 to 250 mm respectively. Both electrical and thermal exergies in­ temperature difference between the panel surface and the Heat Transfer
crease with the panel’s width. The absolute thermal exergy increases Fluid (HTF) in the tube decreases compared to using a longer panel
due to delivering more heat to the HTF, which is used in the electrical (Fig. 14b), leading to more effective surface cooling. The average tem­
section resulting in higher electrical exergy. The thermal and electrical peratures registered on the panel surface for the cases of 1500 mm and
exergies increase by 120% and 61% when the width enlarges from 150 2500 mm are 36 ◦ C and 41 ◦ C, respectively.
to 250 mm. Fig. 15 shows the effect of the panel length on the general system
operation. Fig. 15a presents that both electrical and thermal energy ef­
4.1.4. Effect of length ficiencies decrease with increasing the panel length. The efficiencies
Fig. 14a and b show the contour of the temperature distribution of drop due to higher temperature at the panel’s surface, caused by the
the panel’s surface for the panel’s length of 1500 mm and 2500 mm, large solar collector surface area. The thermal and electrical energy ef­
respectively. By employing a shorter panel length (Fig. 14a), the average ficiencies drop by 3.4% and 1.07% when the panel length increases from

17
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 18. (continued).

1500 mm to 2500 mm. The electrical and thermal exergy efficiencies are to higher resistive losses along the longer interconnections and busbars
affected by the length of the panel, as shown in Fig. 15b. The figure within the panel. These observations emphasize the importance of
shows that the relationship between the length of the panel and thermal considering panel length in optimizing both thermal and electrical
and electrical exergy efficiencies can be summarized as follows. The performance in photovoltaic systems.
thermal exergy efficiency increases from 1.45% to 1.74% as the panel Further research is needed to explore these relationships and develop
length increases from 1500 mm to 2500 mm, indicating improved heat strategies to enhance overall system efficiency. Fig. 15c shows the effect
transfer characteristics. In contrast, the electrical exergy efficiency de­ of the panel length on the overall energy and exergy efficiencies. The
creases from 12.1% to 12% with the increase in panel length, attributed overall energy efficiency reduces with increasing the panel length as the

18
K. Khosravi et al.
Table 7
The analysis of variance (ANOVA).
Electrical energy Thermal energy efficiency Overall energy efficiency Electrical exergy Thermal exergy efficiency Overall exergy efficiency Destroyed exergy (W)
efficiency (%) (%) (%) efficiency (%) (%) (%)
F-value p-value F-value p-value F-value p-value F-value p-value F-value p-value F-value p-value F-value p-value

model 439.97 < 0.0001 2506.82 < 0.0001 2745.51 < 0.0001 439.97 < 0.0001 3664.82 < 0.0001 63.20 < 0.0001 1.669E + 06 < 0.0001
A-Diameter 926.02 < 0.0001 12694.93 < 0.0001 13965.77 < 0.0001 926.02 < 0.0001 10596.17 < 0.0001 113.04 < 0.0001 114.20 < 0.0001
B-Inlet Velocity 713.54 < 0.0001 9945.18 < 0.0001 10955.37 < 0.0001 713.54 < 0.0001 8314.94 < 0.0001 97.07 < 0.0001 99.18 < 0.0001
C-Width 1234.10 < 0.0001 17161.47 < 0.0001 18621.43 < 0.0001 1234.10 < 0.0001 2997.60 < 0.0001 23.00 < 0.0001 5.105E + 06 < 0.0001
D-Length 133.52 < 0.0001 1757.25 < 0.0001 1890.74 < 0.0001 133.52 < 0.0001 1606.50 < 0.0001 14.92 0.0004 4.099E + 06 < 0.0001
E-Solar Radiation Intensity 4252.99 < 0.0001 0.7019 0.4084 95.31 < 0.0001 4252.99 < 0.0001 39761.39 < 0.0001 317.72 < 0.0001 1.464E + 07 < 0.0001
AB 47.11 < 0.0001 319.16 < 0.0001 296.68 < 0.0001 47.11 < 0.0001 98.31 < 0.0001
AC 15.34 0.0004 20.62 < 0.0001 7.64 0.0094 15.34 0.0004 323.29 < 0.0001 10.62 0.0023 10.94 0.0022
AD 12.14 0.0015 75.77 < 0.0001 69.50 < 0.0001 12.14 0.0015 32.01 < 0.0001
AE 168.62 < 0.0001 4.81 0.0358 8.50 0.0064 168.62 < 0.0001 655.30 < 0.0001 10.47 0.0025 10.85 0.0023
BC 10.37 0.0029 12.70 0.0012 4.32 0.0458 10.37 0.0029 239.86 < 0.0001 7.65 0.0086 7.88 0.0081
BD 8.53 0.0064 50.95 < 0.0001 46.75 < 0.0001 8.53 0.0064 23.02 < 0.0001
19

BE 120.43 < 0.0001 4.16 0.0498 6.84 0.0135 120.43 < 0.0001 460.43 < 0.0001 7.04 0.0115 7.29 0.0106
CD 3.39 0.0750 3.84 0.0587 3.39 0.0750 49.81 < 0.0001
CE 187.76 < 0.0001 4.70 0.0377 8.09 0.0077 187.76 < 0.0001 6.16 0.0175 8.41 0.0064
DE 24.58 < 0.0001 24.58 < 0.0001 113.22 < 0.0001
A2 17.27 0.0002 139.96 < 0.0001 31.56 < 0.0001 17.27 0.0002
B2 13.05 0.0010 131.89 < 0.0001 123.01 < 0.0001 13.05 0.0010 64.94 < 0.0001 3.50 0.0698
C2 8.02 0.0079 9.69 0.0039 25.91 < 0.0001 8274.41 < 0.0001
D2 3.40 0.0745 10585.62 < 0.0001
E2 696.38 < 0.0001 88223.09 < 0.0001
R2 0.9957 0.9992 0.9993 0.9957 0.9995 0.9419
1.0000
2
Adjusted R 0.9935 0.9989 0.9990 0.9935 0.9993 0.9270
1.0000
Predicted R2 0.9864 0.9979 0.9980 0.9864 0.9987 0.8900
1.0000
Adeq Precision 92.1768 212.4407 224.5618 92.1768 261.6992 35.4745
4855.4543

Solar Energy 264 (2023) 112025


K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 19. Interaction impact of the design factors on electrical energy efficiency.

panel’s total area increases, which could absorb more heat from solar exergy increases by 59.5% and 23.3% when the length of the panel
radiation. The total exergy efficiency increases with increasing the changes from 1500 mm to 2500 mm.
length of the panel, whereas by changing the length of the panel from
1500 mm to 2500 mm, the total energy and exergy efficiencies change 4.1.5. Effect of solar radiation intensity
from 76.5% to 74.2% and 13.6% to 13.7%, respectively. Fig. 15d shows Studying the effect of solar radiation is the main factor for any
that increasing the length of the panel at a constant width result in investigation related to the PV panel. Different radiation intensities (400
higher absolute thermal and electrical exergies. The reason is that – 1000 W/m2) are used in the model to analyze their effectiveness.
increasing the panel’s length increases the surface’s thermal energy, Fig. 16 shows the contour of the temperature distribution on the panel’s
which is then released to the HTF tube. The thermal and electrical surface. Increasing solar radiation increases the temperature of the

20
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 20. Interaction impact of the design factors on thermal energy efficiency.

surface, which is demonstrated by the comparison between Fig. 16a and 0.65% to 1.59% when the solar power intensity rises from 400 to 1000
Fig. 16b. The average temperatures of the panel surface are 35 ◦ C and W/m2. Considering thermal and electrical energy and exergy efficiencies
43 ◦ C in the cases of 400 and 1000 W/m2, respectively. (Fig. 17c), the total energy efficiency drops by 0.86%, and the total
Fig. 17a shows the solar power intensity’s impact on electrical and exergy efficiency increases by 3.4% when the solar radiation increases
thermal energy efficiencies. The electrical and thermal energy effi­ from 400 to 1000 W/m2. The thermal and electrical exergies increase
ciencies are inversely and directly proportional to the solar power in­ with the solar radiation intensity because of the warmer panel surface
tensity, respectively. Increasing the solar intensity means delivering caused by the higher incident radiation (Fig. 17d). The thermal exergy
more heat to the panel surface, causing an increase in the surface tem­ increases from 0.96 W to 5.91 W, and the electrical exergy rises from
perature, which also affects energy efficiency. The electrical energy ef­ 18.9 W to 44.4 W when the radiation intensity increases from 400 to
ficiency drops by 6.4%, and the thermal energy efficiency increases by 1000 W/m2.
0.07%, which is negligible when the solar radiation intensity increases
from 400 to 1000 W/m2. The change in solar radiation also affects the
thermal and electrical exergy efficiencies (Fig. 17b). The thermal exergy 4.2. Developing a predictive model
increases because of delivering more heat from the warmer surface in
the case of 1000 W/m2 to the HTF; however, the electrical exergy ef­ The depended-on factors and their range must be specified to find a
ficiency drops for the same reason. The electrical exergy efficiency drops predictive analytical method and optimize the solar thermal unit. This
from 12.7% to 11.9%, and the thermal exergy efficiency increases from study uses the Response surface methodology (RSM) to achieve a correct
model matrix. Then, the numerical simulation is performed, and the

21
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 21. Interaction impact of the design factors on overall energy efficiency.

predictive method is developed by employing RSM. Furthermore, the parameter [41]. The studied levels of configuration factors for CCD are
manner of the operation fluid can be assessed, and the impact of various listed in Table 4, showing the range of each parameter.
model factors on the unit properties can be described. After all, applying In the RSM, once knowing the levels of each configuration’s rule, one
the single and multi-objective optimization, established by choosing is required to choose a proper configuration matrix. The matrix tech­
multiple scenarios, is performed toward the best configuration using the nique is achieved by creating professional software, shown in Table 5.
predictive model. The HTF unit with higher energy, exergy, and lower entropy production
is more desirable. Consequently, several factors are considered to
4.2.1. Design of numerical simulation configure an effective HTF, including electrical and thermal energy and
Numerous parameters influence the thermal performance of the unit; exergy efficiencies, overall energy and exergy efficiencies, and electrical
which are essentially challenging to identify a relation that includes all and thermal exergies. CFD simulation is ruled to compute the responses,
these factors. Therefore, selecting the key factors to decrease the as explained previously, built on the matrix configuration found by CCD
computational cost is crucial. Five design factors of the working fluid (Table 5). All the design parameters and the corresponding responses are
(HTF) are chosen for this purpose, i.e., velocity, the diameter of the HTF listed in Table 6.
tube, dimensions of the solar panel, and radiation intensity. Previous
works frequently apply some techniques, such as Taguchi [39] and RSM 4.2.2. Statistical and diagnostic analysis
[40,41], to create the tests. In this study, RSM is used due to the possi­ To assess the method’s capability statistically, various analytical
bility of considering the collaboration impacts of design factors. One of plots for every response are provided in Fig. 18. RSM presents various
the most critical types of RSM is central composite design (CCD) analytical contours for verifying the competence of the prediction
(considered in this work due to its accuracy of the predictive model method. These graphs are categorized into two groups: those built on the
[65]). The CCD method involves central locations that are improved Hat matrix and residuals. The Design Expert software package computes
with a set of “star locations” and “factorial points” that are utilized for these graphs. Fig. 18a shows the assessment of the predicted results
curvature approximation [66]. Thus, each configuration factor’s three found by RSM compared to the findings found by the numerical simu­
stages (lower, central, and higher levels) are necessary for each design lations. The term “actual data” in this figure refers to the data obtained

22
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 22. Interaction impact of the design factors on electrical exergy efficiency.

through numerical simulations. It is important to note that the accuracy predicted data. More details and explanations about these analytical
of the numerical simulation has been validated through experimental graphs are described in [30].
work conducted by Qinghua Yu et al. [45], and further details can be Moreover, the appropriateness and statistical significance of the
found in Section 3. Fig. 18b illustrates the typical residuals against inside suggested model are assessed via ANOVA analysis. This method uses
studentized residuals. The continuation of the linear connection in calculation indicators (P-value and F-value) to analyze the method’s
graphs 12a and 12b indicates an acceptable linear fitness between the validity. These indicators are utilized to identify the importance of the
achieved data by the predictive method and those found by ANSYS applied regression methods and the impacts of combining the parame­
Fluent. Fig. 18c reveals studentized residuals compared with the pre­ ters. The importance of the technique is shown when the F-values are
dicted data, in which the random distribution of residuals in this graph less than 0.05. Likewise, P-values describe the meaning, representing the
indicates that the discrepancy of the outcomes is steady for any values of every term regarding the others. A model with a large F-value

23
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 23. Interaction impact of the control factors on thermal exergy efficiency.

(greater than 4) and less P-value (smaller than 0.05) is considered a indicators.
reliable predictive model [67]. As previously stated, from the perspec­
tive of the method fitness, predicting coefficients R2, R2adj, and R2pred are 4.2.3. Assessment of the predictive method
three significant values, shifting in the range between 0 and 1, where the The RSM method can be applied to find the predictive method based
quality of the fitness increases when those values approach 1. It should on the responses. The predictive method reveals the connection between
be noted that the Backward selection is used to remove the P-values the used configuration factors in Table 5 and the reactions. Furthermore,
which are greater than 0.1. a relationship can be driven for each response set, including the elec­
The outcome of the analysis of the ANOVA is presented in Table 7. It trical and thermal efficiency values, electrical and thermal exergies, and
is shown that the method is numerically satisfied based on various destroyed exergy as a function of the design parameters (i.e., operation

24
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 24. Interaction impact of the control factors on overall exergy efficiency.

Fig. 25. Interaction impact of the control factors on destroyed exergy.

fluid (HTF) velocity, diameter of the HTF tube, dimensions of the solar quadratic model cannot accurately predict the responses. Therefore, the
panel, and the radiation intensity). As previously stated, a second-order Box-Cox transformation increases the accuracy of the predictive model.
polynomial equation commonly utilized in RSM is applied to state the A power transformation with coefficients of 2.5 and 3 is selected for the
predictive method. The factors of the predictive method, those related to thermal energy efficiency and overall energy efficiency, respectively. An
the correlation, are listed in Table 6 for every response. Due to a rapid inverse square root and natural log transformation are also chosen for
and abrupt enhancement of some responses in the design factors, the thermal and destroyed exergy efficiency, respectively. The units of the

25
K. Khosravi et al. Solar Energy 264 (2023) 112025

design factors and their responses are the same as those in Table 5. To

Desirability
examine the numerical significance of the fitted method, the technique
of discrepancy (ANOVA) is applied [68]. ANOVA is used to evaluate the

0.994

0.995

0.829
quality among various methods by evaluating the variations among sets

1
1

1
relative to the variation within sets (random error). This assessment can
calculate the significance of the regression method and the arrangement
between the parameter configurations by determining the P-value,
Predicted Response which ranged between 0 and 1 for every term of the fitted method (listed
in Table 6). The importance of the P-value of every term is stated as
statistically significant (when P-value is less than 0.05) and little static
importance (when the P-value is greater than 0.1).
2.69
11.81
70.25
81.85
12.68

13.99
72.62
4.2.4. Researching the effect of developing variables
After estimating the predicted values (Table 6) for each response, it is
essential to show the impact of design factors on each response indi­
Solar radiation intensity (W/m^2)

vidually. Fig. 19 reveals the difference of Electrical energy efficiency


based on the terms of (a) D-u, (b) D-W, (c) D-L, (d) D-Qs, (e) u-W, (f) u-L,
(g) u-Qs, (h) W-L, (i) W-Qs and (j) L-Qs. Furthermore, according to
Fig. 19 (a), it can be concluded that increasing the diameter and inlet
velocity can improve the electrical energy effectiveness because of the
increase in the Reynolds number and, consequently, increase in heat
transfer coefficient and decreases in temperature of PV cell as discussed
549.52

994.63

comprehensively in the first of the discussion. The same reason is valid


1000

1000
400

400

400

for the panel’s solar radiation intensity, width, and length. With the
increase in solar radiation intensity, size, and width of the PV panel, the
temperature of the cell increases, resulting in a reduction in electrical
energy efficiency.
Length (mm)

Fig. 20 illustrates the changing in thermal energy efficiency based on


2367.15

the terms (a) D-u, (b) D-W, (c) D-L, (d) D-Qs, (e) u-W, (f) u-L, (g) u-Qs,
1500
1500
1500
1500

2500
1500

(h) W-L and (i) W-Qs. These figures show that an increase in the diam­
eter, inlet velocity, the width of the panel, and solar radiation intensity
can improve the thermal energy, which is perfectly compatible with the
actual result in the previous section. It is worth mentioning that
Width (mm)

increasing the length of the panel has a negative effect on the thermal
189.82

efficiency of the panel, as shown in the first part and Fig. 20.
150
150
150
150

150
150

Fig. 21 shows the impact of design parameters on the overall energy


efficiency, which is the sum of thermal and electrical energy efficiencies.
This figure illustrates the relationship between overall thermal energy
response based on the terms (a) D-u, (b) D-W,(c) D-L, (d) D-Qs, (e) u-W,
Inlet velocity (m/s)

(f) u-L, (g) u-Qs, and (h) W-Qs which are in compatible with the first
section of the discussion on the evaluation of overall energy efficiency.
The interactions between the electrical exergy and design parame­
ters are shown in Fig. 22. Since electrical energy and exergy have the
0.087

0.088
0.12
0.12

0.04
0.04
0.04

same value, the same trend can be seen in the contours below based on
the terms of (a) D-u, (b) D-W, (c) D-L, (d) D-Qs, (e) u-W, (f) u-L, (g) u-Qs,
Optimal control factors

(h) W-L, (i) W-Qs and (j) L-Qs compared with Fig. 19.
Moreover, based on Fig. 23, with increasing the diameter, and the
Diameter (mm)

inlet velocity, the thermal exergy efficiency of the system decreases due
to enhancement in thermal entropy generation in the system, as shown
14.054

15.74
14.06

previously in sections 6.1 and 6.2. Besides, by increasing the length and
8.1

8.6
The single-objective designing factor’s optimal range.

16

the width of the panel, as well as solar radiation intensity, the thermal
8

exergy efficiency of the system increases. These changes can see in


Fig. 23 based on the terms (a) D-u, (b) D-W,(c) D-L, (d) D-Qs, (e) u-W, (f)
u-L, (g) u-Qs, (h) W-L, (i) W-Qs and (j) L-Qs for the variation of thermal
To maximize only electrical energy efficiency

To maximize only electrical exergy efficiency


To maximize only thermal energy efficiency

To maximize only thermal exergy efficiency


To maximize only Overall energy efficiency

To maximize only Overall exergy efficiency

exergy efficiency.
The overall exergy efficiency is the sum of thermal and electrical
To minimize only destroyed exergy

exergy efficiencies is shown in Fig. 24, presenting the relationship be­


tween the overall thermal energy response based on the terms of (a) D-
W, (b) D-Qs, (c) u-W, (d) u-Qs, and (e) W-Qs. The interactions between
exergy destruction and design parameters are also illustrated in Fig. 25.
Goals of optimization

These changes can see based on the terms (a) D-W, (b) D-Qs,(c) u-W, (d)
u-Qs, and (e) W-Qs. It should be noted that the Backward selection is
used to remove the P values which are greater than 0.1.
Table 8

4.2.5. Optimization
To find the optimal estimation of the configuration parameters for

26
K. Khosravi et al. Solar Energy 264 (2023) 112025

Fig. 26. Comparison between the based case (Table 3) and the improved case (Table 8) for single- objective scenario.

the solar thermal system, two typical optimization theories (single and efficiency and exergy, and reducing the destroyed exergy). Regarding
multiple objective optimizations) are performed utilizing the RSM multi-objective optimization, the optimization aims to improve elec­
model. Special intentions are identified to suggest the best values of trical and thermal energy efficiency, concurrently enhance electrical
ruled reasons for various terms related to single-objective optimization and thermal exergy efficiencies, and reduce the destroyed exergy.
(enhancing the electrical exergy efficiency, improving the thermal The desirability function is specified to demonstrate whether the

Table 9
The multi-objective designing factor’s optimal range.
Optimal control factors
Goals of optimization Diameter Inlet velocity Width Length Solar radiation Predicted Desirability
(mm) (m/s) (mm) (mm) intensity (W/m^2) Response

To maximize electrical and thermal energy efficiency 16 0.104 150 1500 400 11.78 0.982
69.83
To maximize electrical and thermal exergy efficiency 10.34 0.04 150 2500 599.96 12.194 0.755
1.372
To maximize Overall energy and exergy efficiency 10.022 0.064 150 1500 1000 75.75 0.655
13.72
To maximize electrical and overall energy efficiency 15.82 0.106 150 1500 400 11.785 0.987
81.73
To maximize thermal and overall energy efficiency 16 0.12 150 1500 843.77 70.17 1
81.85
To maximize electrical and overall exergy efficiency 8.94 0.063 150 1500 1000 11.96 0.92
13.77
To maximize thermal and overall exergy efficiency 8 0.04 150 2500 1000 2.67 0.91
13.92
To maximize electrical, thermal, and overall energy 16 0.112 150 1500 400 11.77 0.987
efficiency 69.93
81.7
To maximize electrical, thermal, and overall exergy 8.6 0.041 150 1500 1000 11.79 0.7
efficiency 2.09
13.88
To maximize electrical, thermal, and overall energy 12.79 0.081 150 1500 1000 11.41 0.72
efficiency and 67.84
To maximize electrical, thermal, and overall exergy 79.25
efficiency 12.26
1.16
13.42
To maximize electrical, thermal, and overall energy 12.54 0.082 150 1500 981.57 11.41 0.69
efficiency and 67.69
To maximize electrical, thermal, and overall exergy 79.1
efficiency and minimize destroyed exergy 12.26
1.16
13.42
181.26

27
K. Khosravi et al. Solar Energy 264 (2023) 112025

Table 10 property is supposed to be optimized without considering the other


Sensitivity analysis on design parameters. properties. The ideal values of rule parameters for the aims defined are
Design parameters Definition Change ranges listed in Table 8. This table describes that the desirability of all the pa­
rameters is close to or equal to 1, which reveals that the expected re­
D(mm) Diameter 4–20
u(m/s) Inlet velocity 0.02–0.2 actions are within the appropriate limits.
W(mm) width 100–300 Afterward, to better understand the optimization process, comparing
L(mm)
( )
Length 1000–3000 the optimum values of the responses in the single-objective optimization
Qs W/m2 Solar radiation intensity 200–1200 (Table 8) with the base case (Table 3) is necessary. It should be noted
that the base case in not the worst case, and the parameters selected for
the base case are based on the average values in the range defined for
outcome results are optimized, which designates values between 0 and
each parameter or based on the usage in the literature. The comparison
1. It reflects the desirable ranges for each response. Approaching the
is shown in Fig. 26. It is shown that the system’s efficiency is improved
value of 0 refers to the undesirable obtained value, and by close to 1, the
significantly for each objective scenario. These improvement values for
model’s desirability increases [69]. The numerical optimization locates
electrical and thermal energy efficiencies, overall energy efficiency, and
a position where the desirability function is maximized. A goal’s attri­
electrical and thermal exergy efficiencies are 6.4%, 13.05%, 11.75%,
butes can be changed by changing its weight or priority. All objectives
6.37%, 69 0.13%, and 3.55%, respectively. The destroyed exergy re­
come together into one popularity function for a variety of responses and
duces by 81.47% compared to the base case.
circumstances. It needs to be mentioned that ’Don’t get diverted’ by
constantly aiming for a high desirability rating. The degree to which the
4.2.7. Multi-objective optimization
lower and higher limits are set in relation to the real optimum de­
The multi-objective optimization is employed by applying RSM to
termines the value. Instead of aiming to reach a desirability value of 1,
find the ideal rule parameters for various aims, which consist of con­
optimization seeks to identify a good set of conditions that will satisfy all
current electrical and thermal energy efficiencies, enhancing both
the objectives. Desirability is merely a mathematical technique for
electrical and thermal exergy efficiencies concurrently, and reducing the
determining the ideal [70].
destroyed exergy. It is presumed that the whole reactions are similarly
valuable and can be considered in different situations [22]. The fitness
4.2.6. Single-objective optimization
attitude is utilized to achieve the ideal values of rule parameters. This
The single-objective optimization is approved to describe the ideal
technique combines different reactions and parameters and aims to
operating and geometrical requirements of the solar unit when just a
develop a desirability function. The ideal values of ruled parameters for
single objective function is chosen for optimization. Several objectives
every aim of optimization are presented in Table 9. The Desirability
are defined to provide a variety of selections for improving the electrical,
values imply that the expected reactions are within the standard and
thermal energy, and exergy efficiencies and reducing the destroyed
acceptable limits.
exergy. Towards achieving a single-objective optimization, only one

Table 11
The optimum value of design parameters is based on sensitivity analysis (multi-objective).
Optimal control factors
Goals of optimization Diameter Inlet velocity Width Length Solar radiation Predicted Desirability
(mm) (m/s) (mm) (mm) intensity (W/m2) Response

To maximize electrical and thermal energy efficiency 16.2 0.08 100 1240 578.78 11.82 1
72.17
To maximize electrical and thermal exergy efficiency 9.84 0.2 100 3000 1148 12.21 0.79
1.64
To maximize Overall energy and exergy efficiency 8.95 0.055 100 1000 1200 80.43 0.91
14.07
To maximize electrical and overall energy efficiency 14.4 0.1 108.4 2422.46 212.36 11.94 1
82.58
To maximize thermal and overall energy efficiency 15.3 0.12 141.78 1099.7 253.78 70.78 1
83.65
To maximize electrical and overall exergy efficiency 8.48 0.6 100 1000 1200 12.1 0.78
14.07
To maximize thermal and overall exergy efficiency 4 0.051 169.64 2834.45 1131.91 3.31 1
14.22
To maximize electrical, thermal, and overall energy 19.73 0.099 120.65 2736 212.2 11.83 1
efficiency 70.69
82.52
To maximize electrical, thermal, and overall exergy 8 0.02 100 1000 1092 11.88 0.8
efficiency 2.18
14.06
To maximize electrical, thermal, and overall energy 11.1 0.057 100 1000 1200 11.42 0.85
efficiency and 70.18
To maximize electrical, thermal, and overall exergy 81.6
efficiency 12.27
1.23
13.5
To maximize electrical, thermal, and overall energy 10.41 0.065 100 1000 1200 11.42 0.85
efficiency and 70.2
To maximize electrical, thermal, and overall exergy 81.62
efficiency and minimize destroyed exergy 12.27
1.22
13.4
101.12

28
K. Khosravi et al. Solar Energy 264 (2023) 112025

4.2.8. Sensitive analysis of design parameters design factors and performance criteria. The findings highlight the
Based on the provided previous methods, the RSM method can importance of optimizing photovoltaic/thermal systems to achieve high
expand the range of the design parameters to configure a new broader efficiency and address the challenges and limitations associated with
model for optimization purposes. For the new range, the diameter of the these systems. Future directions for research in photovoltaic solar
HTF tube is considered in the range of 4 to 20 mm; the inlet velocity modules connected with linear pipes for water supply include experi­
values are between 0.02 and 0.2 m/s, the width and the length of the mental validation, dynamic analysis, advanced optimization techniques,
panel are 100 to 300 mm and 1000 to 3000 mm, respectively, and the design optimization under uncertainty, consideration of real-world
solar radiation intensity is between 200 and 1200 W/m2. All the values constraints, and comparative analysis. These directions aim to
are also presented in Table 10. enhance reliability, address practical challenges, and provide valuable
The multi-objective optimization is employed on these parameters insights for designing high-efficiency photovoltaic/thermal systems in
by using RSM to locate the optimized regulation parameters for different real-world applications.
goals. The ideal values of ruled parameters for every aim defined for the
optimization are presented in Table 11. The optimal predicted responses
to maximize the electrical and thermal energy efficiencies performed Declaration of Competing Interest
resulted in 11.82 and 72.17, respectively, with the desirability of 1.
Likewise, the optimal electrical and thermal exergy efficiencies are The authors declare that they have no known competing financial
12.21 and 1.64, respectively, with a desirability of 0.79. To maximize interests or personal relationships that could have appeared to influence
electrical, thermal, and overall energy efficiency and to maximize the work reported in this paper.
electrical, thermal, and overall exergy efficiency and minimize
destroyed exergy, their predicted values are 11.42%, 70.2%, 81.62%, References
12.27%, 1.22%, 13.4 and 101.2 W. These values are performed when the
diameter of the tube, the velocity of the HTF, the width and the length of [1] Hasan, H.A., Togun, H., Abed, A.M., Mohammed, H.I., 2023. Experimental study
on improving the thermal efficiency using fin array with different slant angles in
the solar panel, and the radiation intensity are 10.41 mm, 0.065 m/s,
the finned plate solar air heater FPSAH. Int. J. Thermophys. 44, 71.
100 mm,1000 mm, and 1200 W/m2, respectively. The desirability [2] IEA. Electricity market report. IEA, Paris2021.
values imply that the expected reactions are within the standard and [3] Ma X, Wan Y, Wang Y, Dong X, Shi S, Liang J, et al. Multi-Parameter Practical
Stability Region Analysis of Wind Power System Based on Limit Cycle Amplitude
acceptable limits.
Tracing. IEEE Trans. Energy Convers. 2023.
[4] Zhang, W., Zheng, Z., Liu, H., 2021. Droop control method to achieve maximum
5. Conclusion power output of photovoltaic for parallel inverter system. CSEE J. Power Energy
Syst. 8, 1636–1645.
[5] Gu, Q., Li, S., Gong, W., Ning, B., Hu, C., Liao, Z., 2023. L-SHADE with parameter
This study focused on developing a predictive model to optimize the decomposition for photovoltaic modules parameter identification under different
performance of a hybrid photovoltaic-thermal (PVT) collector with a temperature and irradiance. Appl. Soft Comput. 143.
straight tube designed for cooling purposes. The primary objective was [6] Mahdi, J.M., Mohammed, H.I., Talebizadehsardari, P., 2021. A new approach for
employing multiple PCMs in the passive thermal management of photovoltaic
to enhance the collector’s performance by employing a response surface modules. Sol. Energy 222, 160–174.
model to predict its thermoelectric behavior based on five key design [7] IRENA I. Global renewables outlook: Energy transformation 2050. International
factors. These factors included the inlet velocity of the heat transfer fluid Renewable Energy Agency Abu Dhabi. 2020.
[8] Huang, N., Zhao, X., Guo, Y.u., Cai, G., Wang, R., 2023. Distribution network
(HTF), the tube diameter, the panel’s width and length, and the intensity expansion planning considering a distributed hydrogen-thermal storage system
of solar radiation. The effects of these factors on the thermal and elec­ based on photovoltaic development of the Whole County of China. Energy 278.
trical energy and exergy efficiencies, as well as the destroyed exergy, [9] Gorjian, S., Shukla, A., 2020. Photovoltaic solar energy conversion: technologies,
applications and environmental impacts. Academic Press.
were assessed, and a predictive model was developed for optimization. [10] Liao, Q., Li, S., Xi, F., Tong, Z., Chen, X., Wan, X., Ma, W., Deng, R., 2023. High-
The following points summarize the conclusions without mentioning performance silicon carbon anodes based on value-added recycling strategy of end-
specific numerical results: of-life photovoltaic modules. Energy 281, 128345.
[11] Ma, T., Yang, H., Zhang, Y., Lu, L., Wang, X., 2015. Using phase change materials
in photovoltaic systems for thermal regulation and electrical efficiency
• Thermal exergy efficiency plays a crucial role in determining the improvement: a review and outlook. Renew. Sustain. Energy Rev. 43, 1273–1284.
overall performance of the photovoltaic/thermal (PVT) system, as [12] Lv, S., Zhang, B., Ji, Y., Ren, J., Yang, J., Lai, Y., Chang, Z., 2023. Comprehensive
research on a high performance solar and radiative cooling driving thermoelectric
evidenced by a significant P-value of less than 0.0001 for all system
generator system with concentration for passive power generation. Energy 275.
responses analyzed. [13] Rashid, F.L., Kaood, A., Al-Obaidi, M.A., Mohammed, H.I., Alsarayreh, A.A., Al-
• The study achieved improvements in electrical and thermal energy Muhsen, N.F.O., Abbas, A.S., Zubo, R.H.A., Mohammad, A.T., Alsadaie, S.,
efficiency, overall energy efficiency, and electrical and thermal Sowgath, M.T., Abd-Alhameed, R., Mujtaba, I.M., 2023. A Review of the
Configurations, Capabilities, and Cutting-Edge Options for Multistage Solar Stills in
exergy efficiency, indicating the effectiveness of the optimization Water Desalination. Designs 7 (3), 67.
process. [14] Kalkan, C., Ezan, M.A., Duquette, J., Yilmaz Balaman, Ş., Yilanci, A., 2019.
• The multi-objective optimization scenarios resulted in high desir­ Numerical study on photovoltaic/thermal systems with extended surfaces. Int. J.
Energy Res. 43 (10), 5213–5229.
ability values, indicating desirable outcomes for maximizing elec­ [15] Candanedo, L.M., Candanedo, J.A., O’Brien, W., Chen, Y., 2010. Transient and
trical and thermal energy efficiencies and electrical and thermal steady state models for open-loop air-based BIPV/T systems. ASHRAE Trans. 116,
exergy efficiencies. 600.
[16] Al-Waeli, A.H., Kazem, H.A., Chaichan, M.T., Sopian, K., 2019. Photovoltaic/
• The dimension of the HTF tube diameter, inlet velocity, width and thermal (PV/T) systems: principles, design, and applications. Springer Nature.
length of the panel, and solar radiation intensity were critical pa­ [17] Bahaidarah, H., Subhan, A., Gandhidasan, P., Rehman, S., 2013. Performance
rameters in achieving optimal system performance. evaluation of a PV (photovoltaic) module by back surface water cooling for hot
climatic conditions. Energy 59, 445–453.
• The specific values of the parameters that yielded the best results for [18] Hasan, H.A., Sopian, K., Jaaz, A.H., Al-Shamani, A.N., 2017. Experimental
maximizing electrical and thermal energy efficiencies and electrical investigation of jet array nanofluids impingement in photovoltaic/thermal
and thermal exergy efficiencies are reported. collector. Sol. Energy 144, 321–334.
[19] Wu, S.-Y., Chen, C., Xiao, L., 2018. Heat transfer characteristics and performance
• All single and multi-objective scenarios demonstrated desirability
evaluation of water-cooled PV/T system with cooling channel above PV panel.
values above the desirable limits, indicating successful attainment of Renew. Energy 125, 936–946.
expected performance. [20] Xiao, L., Shi, R., Wu, S.-Y., Chen, Z.-L., 2019. Performance study on a photovoltaic
thermal (PV/T) stepped solar still with a bottom channel. Desalination 471.
[21] Hissouf, M., Feddaoui, M., Najim, M., Charef, A., 2020. Performance of a
Generally, this study comprehensively optimized a PVT hybrid solar photovoltaic-thermal solar collector using two types of working fluids at different
collector with a straight tube for cooling purposes, considering various fluid channels geometry. Renew. Energy 162, 1723–1734.

29
K. Khosravi et al. Solar Energy 264 (2023) 112025

[22] Chen, Q., Burhan, M., Akhtar, F.H., Ybyraiymkul, D., Shahzad, M.W., Li, Y., Ng, K. coolant in photovoltaic thermal systems (PVT) from energy, exergy and entropy
C., 2021. A decentralized water/electricity cogeneration system integrating generation viewpoints. Energy 162, 210–223.
concentrated photovoltaic/thermal collectors and vacuum multi-effect membrane [47] Ji, J., Lu, J.-P., Chow, T.-T., He, W., Pei, G., 2007. A sensitivity study of a hybrid
distillation. Energy 230. photovoltaic/thermal water-heating system with natural circulation. Appl. Energy
[23] Ould-Lahoucine, C., Ramdani, H., Zied, D., 2021. Energy and exergy performances 84 (2), 222–237.
of a TiO2-water nanofluid-based hybrid photovoltaic/thermal collector and a [48] Li, M., Zhong, D., Ma, T., Kazemian, A., Gu, W., 2020. Photovoltaic thermal module
proposed new method to determine the optimal height of the rectangular cooling and solar thermal collector connected in series: Energy and exergy analysis. Energ.
channel. Sol. Energy 221, 292–306. Conver. Manage. 206.
[24] Shojaeefard, M.H., Sakran, N.B., Sharfabadi, M.M., Hussein, O.A., Mohammed, H. [49] Demirel, Y., 2013. Thermodynamic analysis. Arab. J. Sci. Eng. 38 (2), 221–249.
A., 2023. Experimental and Numerical Investigation of the Effect of Water Cooling [50] Kazemian, A., Hosseinzadeh, M., Sardarabadi, M., Passandideh-Fard, M., 2018.
on the Temperature Distribution of Photovoltaic Modules Using Copper Pipes. Effect of glass cover and working fluid on the performance of photovoltaic thermal
Energies 16 (10), 4102. (PVT) system: An experimental study. Sol. Energy 173, 1002–1010.
[25] Lohrasbi, S., Gorji-Bandpy, M., Ganji, D.D., 2017. Thermal penetration depth [51] Kazemian, A., Salari, A., Ma, T., Lu, H., 2022. Application of hybrid nanofluids in a
enhancement in latent heat thermal energy storage system in the presence of heat novel combined photovoltaic/thermal and solar collector system. Sol. Energy 239,
pipe based on both charging and discharging processes. Energ. Conver. Manage. 102–116.
148, 646–667. [52] Sardarabadi, M., Hosseinzadeh, M., Kazemian, A., Passandideh-Fard, M., 2017.
[26] Lohrasbi, S., Bandpy, M.G., Ganji, D.D., 2016. Response surface method Experimental investigation of the effects of using metal-oxides/water nanofluids on
optimization of V-shaped fin assisted latent heat thermal energy storage system a photovoltaic thermal system (PVT) from energy and exergy viewpoints. Energy
during discharging process. Alex. Eng. J. 55 (3), 2065–2076. 138, 682–695.
[27] Hatami, M., Jing, D., 2017. Evaluation of wavy direct absorption solar collector [53] Wu, S.-Y., Guo, F.-H., Xiao, L., 2015. A Review on the Methodology for Calculating
(DASC) performance using different nanofluids. J. Mol. Liq. 229, 203–211. Heat and Exergy Losses of a Conventional Solar PV/T System. Int. J. Green Energy
[28] Javidan, M., Moghadam, A.J., 2021. Experimental investigation on thermal 12 (4), 379–397.
management of a photovoltaic module using water-jet impingement cooling. [54] Kazemian, A., Taheri, A., Sardarabadi, A., Ma, T., Passandideh-Fard, M., Peng, J.,
Energ. Conver. Manage. 228. 2020. Energy, exergy and environmental analysis of glazed and unglazed PVT
[29] Gelis, K., Naci Celik, A., Ozbek, K., Ozyurt, O., 2022. Experimental investigation system integrated with phase change material: An experimental approach. Sol.
into efficiency of SiO2/water-based nanofluids in photovoltaic thermal systems Energy 201, 178–189.
using response surface methodology. Sol. Energy 235, 229–241. [55] Ma, T., Kazemian, A., Habibollahzade, A., Salari, A., Gu, W., Peng, J., 2022.
[30] Ekramian, E., Etemad, S.G., Haghshenasfard, M., 2014. Numerical analysis of heat A comparative study on bifacial photovoltaic/thermal modules with various
transfer performance of flat plate solar collectors. J. Fluid Flow Heat Mass Transfer cooling methods. Energ. Conver. Manage. 263.
(JFFHMT) 1, 38–42. [56] Darvishmotevalli, M., Zarei, A., Moradnia, M., Noorisepehr, M., Mohammadi, H.,
[31] Kazemian, A., Khatibi, M., Ma, T., Peng, J., Hongxing, Y., 2023. A thermal 2019. Optimization of saline wastewater treatment using electrochemical
performance-enhancing strategy of photovoltaic thermal systems by applying oxidation process: prediction by RSM method. MethodsX. 6, 1101–1113.
surface area partially covered by solar cells. Appl. Energy 329. [57] Zhou, J., Hatami, M., Song, D., Jing, D., 2016. Design of microchannel heat sink
[32] Eisapour, A.H., Eisapour, M., Hosseini, M.J., Shafaghat, A.H., Talebizadeh with wavy channel and its time-efficient optimization with combined RSM and
Sardari, P., Ranjbar, A.A., 2021. Toward a highly efficient photovoltaic thermal FVM methods. Int. J. Heat Mass Transf. 103, 715–724.
module: Energy and exergy analysis. Renew. Energy 169, 1351–1372. [58] Basati, Y., Mohammadipour, O.R., Niazmand, H., 2019. Numerical investigation
[33] Eisapour, M., Eisapour, A.H., Hosseini, M.J., Talebizadehsardari, P., 2020. Exergy and simultaneous optimization of geometry and zeta-potential in electroosmotic
and energy analysis of wavy tubes photovoltaic-thermal systems using mixing flows. Int. J. Heat Mass Transf. 133, 786–799.
microencapsulated PCM nano-slurry coolant fluid. Appl. Energy 266. [59] Bezerra, M.A., Santelli, R.E., Oliveira, E.P., Villar, L.S., Escaleira, L.A., 2008.
[34] Fontenault BJ, Gutierrez-Miravete E. Modeling a combined photovoltaic-thermal Response surface methodology (RSM) as a tool for optimization in analytical
solar panel. Proc 2012 COMSOL Conf Boston2012. chemistry. Talanta 76 (5), 965–977.
[35] Ma, T., Li, M., Kazemian, A., 2020. Photovoltaic thermal module and solar thermal [60] Kazemian, A., basati, Y., Khatibi, M., Ma, T., 2021. Performance prediction and
collector connected in series to produce electricity and high-grade heat optimization of a photovoltaic thermal system integrated with phase change
simultaneously. Appl. Energy 261. material using response surface method. J. Clean. Prod. 290.
[36] Salari, A., Kazemian, A., Ma, T., Hakkaki-Fard, A., Peng, J., 2020. Nanofluid based [61] Mäkelä, M., 2017. Experimental design and response surface methodology in
photovoltaic thermal systems integrated with phase change materials: Numerical energy applications: A tutorial review. Energ. Conver. Manage. 151, 630–640.
simulation and thermodynamic analysis. Energ. Conver. Manage. 205. [62] Kazemian, A., Khatibi, M., Reza Maadi, S., Ma, T., 2021. Performance optimization
[37] Kolahan, A., Maadi, S.R., Kazemian, A., Schenone, C., Ma, T., 2020. Semi-3D of a nanofluid-based photovoltaic thermal system integrated with nano-enhanced
transient simulation of a nanofluid-base photovoltaic thermal system integrated phase change material. Appl. Energy 295.
with a thermoelectric generator. Energ. Conver. Manage. 220. [63] Emamjomeh, M.M., Jamali, H.A., Moradnia, M., 2017. Optimization of nitrate
[38] Kazemian, A., Parcheforosh, A., Salari, A., Ma, T., 2021. Optimization of a novel removal efficiency and energy consumption using a batch monopolar
photovoltaic thermal module in series with a solar collector using Taguchi based electrocoagulation: prediction by RSM method. J. Environ. Eng. 143, 04017022.
grey relational analysis. Sol. Energy 215, 492–507. [64] Derringer, G., Suich, R., 1980. Simultaneous optimization of several response
[39] Karaaslan, I., Menlik, T., 2021. Numerical study of a photovoltaic thermal (PV/T) variables. J. Qual. Technol. 12 (4), 214–219.
system using mono and hybrid nanofluid. Sol. Energy 224, 1260–1270. [65] Danmaliki, G.I., Saleh, T.A., Shamsuddeen, A.A., 2017. Response surface
[40] Liaw, K.L., Kurnia, J.C., Sasmito, A.P., 2021. Turbulent convective heat transfer in methodology optimization of adsorptive desulfurization on nickel/activated
helical tube with twisted tape insert. Int. J. Heat Mass Transf. 169. carbon. Chem. Eng. J. 313, 993–1003.
[41] Kazemian, A., Salari, A., Hakkaki-Fard, A., Ma, T., 2019. Numerical investigation [66] Hanrahan, G., Lu, K., 2006. Application of factorial and response surface
and parametric analysis of a photovoltaic thermal system integrated with phase methodology in modern experimental design and optimization. Crit. Rev. Anal.
change material. Appl. Energy 238, 734–746. Chem. 36 (3-4), 141–151.
[42] Du, Y., Fell, C.J., Duck, B., Chen, D., Liffman, K., Zhang, Y., Gu, M., Zhu, Y., 2016. [67] Brown, S.D., Tauler, R., 2020. Comprehensive Chemometrics: Chemical and
Evaluation of photovoltaic panel temperature in realistic scenarios. Energ. Conver. Biochemical Data Analysis. Elsevier Science.
Manage. 108, 60–67. [68] Armstrong, R.A., Eperjesi, F., Gilmartin, B., 2002. The application of analysis of
[43] Kumar, S., Mullick, S.C., 2010. Wind heat transfer coefficient in solar collectors in variance (ANOVA) to different experimental designs in optometry. Ophthalmic
outdoor conditions. Sol. Energy 84 (6), 956–963. Physiol. Opt. 22 (3), 248–256.
[44] Bhattarai, S., Oh, J.-H., Euh, S.-H., Kafle, G.K., Kim, D.H., 2012. Simulation and [69] Raeisian, L., Niazmand, H., Ebrahimnia-Bajestan, E., Werle, P., 2019. Thermal
model validation of sheet and tube type photovoltaic thermal solar system and management of a distribution transformer: An optimization study of the cooling
conventional solar collecting system in transient states. Sol. Energy Mater. Sol. system using CFD and response surface methodology. Int. J. Electr. Power Energy
Cells 103, 184–193. Syst. 104, 443–455.
[45] Yu, Q., Romagnoli, A., Yang, R., Xie, D., Liu, C., Ding, Y., Li, Y., 2019. Numerical [70] Singh, S., Dhiman, P., 2014. Thermal and thermohydraulic performance evaluation
study on energy and exergy performances of a microencapsulated phase change of a novel type double pass packed bed solar air heater under external recycle using
material slurry based photovoltaic/thermal module. Energ. Conver. Manage. 183, an analytical and RSM (response surface methodology) combined approach.
708–720. Energy 72, 344–359.
[46] Kazemian, A., Hosseinzadeh, M., Sardarabadi, M., Passandideh-Fard, M., 2018.
Experimental study of using both ethylene glycol and phase change material as

30

You might also like