Applied Thermal Engineering 224 (2023) 119997

Contents lists available at ScienceDirect

Applied Thermal Engineering

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

Research Paper

A bidirectional solar thermoelectric generator combining heat storage for

daytime and nighttime power generation
Francisco J. Montero a, b, *, Ravita Lamba c, *, Alfonso Ortega d, Wolfram Jahn a,
Wei-Hsin Chen e, f, g, Amador M. Guzmán h
a
Departamento de Ingeniería Mecánica y Metalúrgica, Escuela de Ingeniería, Pontificia Universidad Católica de Chile, Vicuña Mackenna 4860, Santiago de Chile
7820436, Chile
b
Departamento de Ciencias de la Energía y Mecánica, Universidad de las Fuerzas Armadas ESPE, P.O. Box 171-5-231B, Sangolquí, Ecuador
c
Department of Electrical Engineering, Malaviya National Institute of Technology Jaipur, JLN Marg, Malaviya Nagar, Jaipur 302017, India
d
Laboratory for Advanced Thermal and Fluid Systems, Department of Mechanical Engineering, Villanova University, Philadelphia, USA
e
Department of Aeronautics and Astronautics, National Cheng Kung University, Tainan 701, Taiwan
f
Research Center for Smart Sustainable Circular Economy, Tunghai University, Taichung 407, Taiwan
g
Department of Mechanical Engineering, National Chin-Yi University of Technology, Taichung 411, Taiwan
h
Department of Industrial Engineering, Diego Portales University, Vergara 210, Santiago de Chile, Chile

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

Keywords: A solar thermoelectric generator (STEG) is a promising technology for harvesting solar energy for standalone
Solar thermoelectric generator applications. However, the STEG cannot generate electricity during nighttime due to unavailability of solar
Latent heat energy. The efficiency of thermoelectric generator (TEG) is also low that limits its application areas. This low
Cooling system
efficiency can be improved by partially utilizing the waste heat from its cold side using phase change materials
Thermal management
Day-night operation
(PCMs). Further, the STEG systems operating during day and nighttime are not proposed so far. Therefore, an
Desert locations experimental test rig of a bidirectional (operative in day and night both) STEG coupled with latent heat storage
and cooling system (LHSCS) has been developed in this paper. The LHSCS acts as a sink by storing waste heat
from the TEG cold side in a phase change material during the daytime and regulates its temperature effectively.
During nighttime, LHSCS acts as a heat source for TEG power generation. This proposed bidirectional model aims
to provide non-intermittent electricity generation for 24 h. An experimental setup was tested under laboratory
conditions and adjusted using a numerical model previously developed in COMSOL Multiphysics software. Once
both models are mutually adjusted and verified, the proven numerical model is used to simulate the prototype in
the environmental conditions of the Atacama Desert in Chile. The transient effects of solar radiation, ambient
temperature and wind speed of the selected location on the hot and cold side temperatures, voltage, power
output and efficiency of STEG have been analyzed. A maximum temperature difference of 120 ◦ C is obtained
between the TEG hot and cold sides. The experimental results showed TEG efficiency of 5%. The system
generated average annual electricity of 5735 Wh. The STEG generated around 0.6 % of the total electricity
during the night in the Atacama Desert location. The levelized cost of energy and storage have also been
calculated for the proposed system and compared with PV and STEG systems. The LCOE and LCOS of the pro­
posed system are 8850 and 566 USD/MWh respectively. The proposed configuration may provide a reference
study for design and development of an efficient and cost-effective STEG coupled LHSCS system.

advantages over conventional energy sources, mainly in that it is clean,

pollution-free, and inexhaustible. Consequently, it is deemed to be the
1. Introduction future of global electricity production. Among renewables, solar energy
is a vital technology for achieving future sustainability worldwide. It
The environmental concerns of global warming and continuously provides a promising solution to environmental and energy issues [1].
increasing energy demand for socioeconomic developments led to the Solar photovoltaic (PV) and thermal technologies are the two main
adoption of renewable energy sources. Renewable energy has several

* Corresponding authors.
E-mail addresses: [email protected] (F.J. Montero), [email protected] (R. Lamba).

https://doi.org/10.1016/j.applthermaleng.2023.119997
Received 13 July 2022; Received in revised form 7 December 2022; Accepted 3 January 2023
Available online 6 January 2023
1359-4311/© 2023 Elsevier Ltd. All rights reserved.
F.J. Montero et al. Applied Thermal Engineering 224 (2023) 119997

Nomenclature W TEG power output (W)

ZT Dimensionless figure of merit
As TEG absorptance area (m2)
Copt Optical concentration Greek Symbols
Cp Specific heat capacity (J/K) α Mass fraction
h Heat transfer coefficient (W/m2 K) ρ Fluid density (kg/m3)
I TEG electrical current (A) ηopt Optical efficiency (%)
Im Matched current (A) θ PCM solid fraction
k Thermal conductivity (W/m K) μ Dynamic viscosity (N s/m2)
Qloss Heat losses (W/m2) ηTEG TEG efficiency (%)
Qsun Solar irradiation (W/m2) Abbreviations
Qconc Concentrated heat flux (W/m2) Bi2Te3 Bismuth Telluride
Qteg Heat directed to TEG (W/m2) DAQ Data acquisition system
Rm Matched resistance (Ω) LHS Latent heat storage
Ri TEG internal resistance (Ω) LHSCS Latent heat storage and cooling system
S Seebeck coefficient (V/K) PCM Phase change material
Tpcm PCM temperature (K) PV Photovoltaic system
Tcold TEG cold side temperature (K) STEG Solar thermoelectric generator
Thot TEG hot side temperature (K) TEC Thermoelectric cooler
u Fluid velocity (m/s) TEG Thermoelectric generator
Vm Matched voltage (V) TIM Thermal interface material
VOC Open circuit voltage (V)

paths to harvesting solar energy. Solar PV systems use only the visible to store a large amount of residual heat during the PCM melting. A PCM
part of the solar spectrum for electricity generation; the rest is wasted in storage system includes at least three components [10]: the PCM, where
heat. Solar thermal systems use the entire solar spectrum with advanced the energy will be stored in the form of latent heat, a container, and a
coating materials. The efficiency of PV systems has been improving in contact surface with the heat source. One of the most widely used phase
recent years. However, for modeling the photovoltaic systems, a con­ change materials is paraffin, in which the heat of fusion has an average
version efficiency of around 20% for crystalline silicon cells is value of 35–40 kcal/kg. Paraffin’s other advantages are that it is
considered. chemically inert, non-toxic, and available at a reasonably low cost [11].
The thermoelectric generator (TEG) is an alternative to harvesting An LHS system can also be coupled to a STEG system to manage the cold
solar energy. It is a device that converts a temperature gradient into side temperature and maintain high-temperature differences between
electricity. Solar energy can be used as a heat source to produce a the hot and cold sides. Therefore, an LHS system can cool the TEG’s cold
temperature gradient across the TEG system and this solar powered TEG side and thus store the waste from the TEG [12]. An LHS system that
is called as a solar thermoelectric generator (STEG). It consists of a TEG serves this dual purpose is named as a latent heat storage and cooling
module and a solar collector or selective solar absorber that convert system (LHSCS). It is also necessary to optimize the thermal and
solar energy into heat. In most cases, a cooling system is required to geometrical parameters of the LHS container. The PCM container’s size
maintain the cold side of the TEG at a relatively lower temperature. A and shape and the PCM’s melting time must be carefully considered for
passive cooling system is always recommended. It boosts the conversion the daily insolation of a particular geographic location [13]. The cy­
efficiency of STEG , thus allowing for its techno-economic feasibility. lindrical shell and tube-type configurations have been widely used for
The efficiency of STEG systems can be boosted by exploring different design of the heat storage container. Inclusion of fins also boosts the heat
materials and system configurations. Despite being a promising tech­ transfer within the PCM. The fins in the heat sink also increase the
nology that envisions a conversion efficiency of over 15% [2], currently, melting speed and decrease melting time during the charging and dis­
most researchers report an efficiency of around 5 % [3–6], and the charging process.
highest reported efficiency of a STEG with solar concentration reaches An experimental investigation of the TEG system coupled with an
around 10 % [7]. The low conversion efficiency of STEG makes it LHS system has been carried out by Jaworski et al. [14]. The authors
possible to store waste heat in the form of sensible, latent, or thermo­ used the LHS system to cool the cold side of the TEG. Selvam et al. [15]
chemical heat. A higher temperature at the hot side of the STEG can be conducted a computational study of an encapsulated heat sink-PCM to
obtained with the help of a concentrator. A concentrated STEG system manage the TEG cold side temperature. The temperature of both sides of
normally operates above 150 ◦ C that further demands proper heat the TEG has been controlled, resulting in an increased TEG conversion
dissipation at its cold side [8]. Some cooling systems, including PCM and efficiency of around 33%. Karthick et al. [16] experimentally analyzed
heat pipes, have been developed for effective heat removal [9]. Heat the power generation of a reversible TEG-coupled PCM system. It has
pipes also play a crucial role in the heat transfer between the TEG and been observed that, under laboratory conditions, the TEG system could
PCM systems. However, heat is first transferred to the PCM in such generate an additional amount of electricity using the heat stored in the
configurations and subsequently expelled into the environment. Such a PCM. Lamba et al. [17] developed and optimized a concentrating STEG
system does not consider storage or subsequent use of the rejected heat, thermodynamic model in MATLAB. They also analyzed the effects of the
focusing only on keeping the cold side temperature of the TEG at around number of thermocouples, electrical current, concentration ratio, and
78 ◦ C. load resistance ratio on the system’s power output, energy, and exergy
Like other solar-based systems, STEG also requires an energy storage efficiencies. Luo et al. [18] used paraffin/expanded graphite PCM
system that makes it possible to generate electricity during nighttime. A coupled to a STEG for thermal management by enhancing the paraffin’s
latent heat storage (LHS) based system has higher density and lower thermal conductivity and anti-leakage ability. Bhuiya et al. [19] pro­
temperature variations than the sensible and thermochemical heat posed a method of improving the energy efficiency of thermoelectric
storage based systems. LHS systems use a phase change material (PCM) cooler (TEC) by minimizing its hot side temperature and by adding PCM

2
F.J. Montero et al. Applied Thermal Engineering 224 (2023) 119997

in the heat sink using ANSYS. The effects of phase change material providing a new way of thinking for the design and application of the
volume, heat transfer coefficient and heat sink geometries on the per­ STEG coupled LHSCS and have practical guidance significance for the
formance of the TEC has been analysed. Wang et al. [20] experimented optimization and improvements of the experiments. Also, the heat re­
with the TEG-integrated PCM system thermal management. Shittu et al. covery using LHSCS may also results in cost and energy savings, better
[21] analyzed the use of PCM to reduce the effect of transient and non- reliability and less CO2 emissions. Economic analysis has also been
uniform heat flux by integrating it into the top surface of the concen­ carried out in brief. In short, the novelty of the proposed study consists
trated STEG. Huang et al. [22] used a TEG-integrated PCM system in of the following points:
automotive applications to reduce the exhaust gas temperature varia­
tions on the TEG system. They achieved around 1% improvements in the • A novel bidirectional STEG coupled heat storage system is proposed
TEG’s open-circuit voltage and power output. Alireza et al. [23] devel­ for 24 h power generation
oped the numerical model of a TEG-integrated PCM system in Multi­ • LHSCS is a passive method for cooling the cold side of STEG during
physics simulation to optimize the system’s performance using the daytime and hence, maximizing the temperature difference between
Taguchi method. The authors compared the system’s electrical perfor­ the hot and cold sides of the STEG.
mance with the experimental results of similar systems in the literature. • LHSCS acts as a heat sink, i.e. connected to cold side of STEG during
Escobar et al. [24] investigated a TEG- thermosyphon hybrid system daytime and the same system acts as a heat source, i.e. connected to
with low operating temperature. TEG is used for electricity generation, hot side of STEG during nighttime
and the thermosyphon is used to maintain a constant temperature at the • The large ambient temperature variations between day and night in
TEG cold side and recover heat from the condenser zone. Shittu et al. desert locations results in heat storage in LHSCS during daytime and
[25] studied the effects of non-uniform and transient heat flux on the power generation using this stored heat during the nighttime
STEG system performance by developing its detailed three-dimensional • The impact of harsh environmental conditions on the operation of
model in COMSOL Multiphysics software. Phase change material (PCM) solar energy conversion systems and the effect of LHSCS insulation
reduces the non-uniform and transient heat flux effects on STEG per­ thickness in controlling the heat losses that affect the TEG cold side
formance. Wang et al. [26] performed the thermal management of the temperature and stored energy are analyzed.
TEG systems by adding PCM to the hot side of the TEG and compared its • LCOE and LCOS of proposed system are compared with the other
performance with the conventional TEG system without considering solar energy conversion systems.
PCM. The authors studied the effects of PCM parameters on the TEG
performance under different pulse transient thermal boundary condi­ This paper is outlined as follows: Section 2 describes the methodol­
tions. Thus, most of these studies reflect that the use of PCM is beneficial ogy including the physical configuration and operation of the proposed
for a TEG in providing enhanced performance. Almost all these studies system, its experimental setup and uncertainty analysis, In Section 3, the
used PCM has a heat sink in regulating the cold side temperature of the formulation including governing equations of the STEG and LHSCS
TEG. systems are developed and then numerical simulations are performed
The literature review indicates that in TEG integrated PCM systems, and system validation is carried out. Section 4 discusses the results and
the waste energy stored in the PCM at the TEG cold side was rejected to discussion regarding the performance of the proposed system in terms of
the outside environment. Studies related to the utilization of the rejected temperature difference, voltage and power output and efficiency for
heat for any heating or cooling applications are not found in the liter­ different operating conditions. Lastly, conclusions regarding perfor­
ature. Also, the studies reporting the performance study of the TEG- mance and application are drawn and summarized in Section 5.
integrated PCM system for the high solar irradiation and high ambient
temperature variation environmental conditions are very few. 2. Methodology
Comprehensive studies of TEG-PCM systems considering both theoret­
ical and experimental analyses are still lacking. Therefore, it is indis­ 2.1. Physical configuration and operation
pensable to address these issues by considering the need for continuous
energy supply using integrated systems. Hence, this study proposes an A two-dimensional (2-D) schematic diagram of the proposed STEG-
experimental setup of a hybrid system consisting of a STEG coupled coupled LHSCS is shown in Fig. 1. It consists of two subsystems. One
latent heat storage and cooling system to effectively operate the TEG is a solar thermoelectric generator system (STEG), and the other is a
during day and nighttime with the advantages of ease of manufacturing latent heat storage and cooling system (LHSCS). The STEG system in­
and simple structure. The proposed system passively cools the TEG cold cludes a Bismuth Telluride (Bi2Te3) thermoelectric module and an op­
side during daytime and maximizes the temperature difference between tical concentrator consisting of a Fresnel lens collector. The TEG cold
the hot and cold sides of the TEG. Moreover, the LHSCS system simul­ side is fixed to the LHSCS. The LHSCS includes an insulated acrylic
taneously stores the TEG residual heat in a phase change material (PCM) plastic container filled with PCM and an embedded copper heat sink,
during the day, which can then be used for electricity generation at which includes a solid bar with radial fins. The cylindrical container has
night. The transient study is also performed and the electrical perfor­ top and bottom covers made of acrylic plastic. Organic paraffin is used as
mance of the STEG is examined. The importance of this study lies in PCM due to its lower ambient impact as its components can quickly
providing evidence that the proposed concept of an LHSCS can operate degrade in a lower time compared to inorganic paraffin (petroleum-
as a cooling system and heat reservoir in actual environmental condi­ based), which are limited in temperature range, containment methods,
tions of the geographic location of the Atacama Desert. This study allows and thermal cycles. Organic paraffins have the characteristics of high
us to verify the impact of harsh environmental conditions including the stability, low flammability, low toxicity, and highly sustainable perfor­
solar radiation, ambient temperature and wind velocity on the operation mance [27]. The optimum PCM melting temperature was obtained from
of solar energy conversion systems. In other words, the large ambient the geometric mean between ambient and heat source temperatures
temperature differences between day and night in a desert location can using the analysis presented by De Lucia and Bejan [28]. The TEG cold
help the proposed system generate and store energy during the night. side projected temperatures (around 80 ◦ C) and the average ambient
Additionally, this analysis shows the effect of insulation thickness of the temperature of the Atacama desert during the hours of solar insolation
LHSCS to control the heat losses that will affect the TEG cold side (around 25 ◦ C) were used to obtain an optimum casting temperature of
temperature and the stored energy. It has been verified that the pro­ around 48 ◦ C. Therefore, Puretemp 48X paraffin is chosen as PCM. The
posed STEG coupled LHSCS system can enhance the electricity genera­ amount of Puretemp 48X included in the computer model for the
tion duration of the STEG system, thus allowing it to operate without described dimensions was 6 kg.
solar irradiation during nighttime. The objective of this study lies in The working principle of the proposed system can be described as

3
F.J. Montero et al. Applied Thermal Engineering 224 (2023) 119997

Fig. 1. Schematic of Solar Thermoelectric Generator Coupled Latent Heat Storage and Cooling System, presented previously by Montero et al. [29].

follows: the TEG hot side receives a concentrated heat flux (Qconc) to load resistance. The application of transistors to commute different
maintain a higher temperature at the TEG hot side (Thot) surface. The signals is widely used in various electronic systems and integrated cir­
solar radiation flux (Qsun) is focused on using the optical concentrator. It cuits. The design of the described electronic system is not a part of this
is found that a Fresnel lens with a 40x concentration can achieve around analysis. The methodology applied in this analysis is shown in Fig. 2.
200 ◦ C temperature at the TEG hot side. The LHSCS is integrated with
the TEG cold side to maintain its temperature low by storing the waste
heat. A part of the total concentrated heat flux (Qconc) is rejected to the 2.2. Experimental set-up
environment as convective and radiative heat losses, and the remaining
heat is transferred to the TEG hot side (Qteg) by conduction. Qteg is The experimental set-up and schematic of the STEG-coupled LHSCS
partially used to generate electricity using the thermoelectric effects, are shown in Fig. 3 (a) and (b) respectively. The setup includes the heat
and the remaining heat is rejected from the TEG cold side to the LHSCS. concentrator, ice bath, zone box, DAQ system, and power supply. This
This residual heat, received from the TEG cold side, is conducted to the setup is developed to determine the feasibility and confirm the previous
PCM during its charging process. The PCM starts to melt and, thus, stores computational simulation model results [29]. The authors take the ge­
that heat as latent heat of fusion. The PCM temperature (Tpcm) increases ometry, dimensions, and materials from a previous numerical study
while charging, and it starts to melt when its melting temperature is [29]. Cartridge heaters are used to replicate the concentrator which is
achieved. The melting occurs in all compartments of the PCM container, used to simulate the concentrated solar radiation on the TEG hot side
including radial fins. Thus, the PCM maintains a relatively lower tem­ under the laboratory environment. The aluminum heat concentrator
perature at the TEG cold side (Tcold) because it absorbs the heat released consists of three 96 W each cartridge heaters to control the heat flux over
from the TEG cold side. The reverse process occurs in the PCM during the TEG system. The concentrator is insulated with glass wool to reduce
the PCM discharging, when its solidification starts indicating that the the convective heat losses to the environment.
operation of LHSCS is reversed compared to the charging cycle. During Additionally, the glass wool is covered with aluminum foil to reduce
the discharging period, the heat flows from the PCM into the TEG cold radiative heat losses to the ambient. The cartridge heaters of the heat
side. The charging process in the LHSCS is coupled to the available concentrator are connected to a variable electrical power supply to
diurnal hours of the solar radiation at a specific geographic location control the heat flux on the concentrator. The TEG module (model TE-
using the correct amount of the PCM volume and exact LHSCS container MOD-5W5V-40S) has dimensions of 40-by-40-by-5 mm and is manu­
geometry. The discharging process of PCM is coupled with the off- factured by TEGPro. The TEG module is placed between the concen­
sunshine hours. During this period, the stored heat is released back to trator and LHSCS using a thermal interface material (TIM) to reduce the
TEG’s cold side to produce electricity. Since the ambient temperature is contact resistances between the components.
much lower, the TEG hot side becomes the cold side in this period. The The experimental test rig uses a data acquisition system to record
heat flow direction reverses during the discharging process, thus signals from the various temperature sensors. The TEG module is
inducing the electrons to flow in the opposite direction inside the TEG. coupled to a data acquisition system composed of a NI Compact DAQ –
Therefore, a transistor circuit can manage the change of poles of the 9174, a NI 9207 voltage module, and a LabView program that stores the
thermoelectric module during the discharging process. It is used as a voltage data. In addition, two temperature modules NI 9213, are con­
commutator to interchange the connections of the TEG module and the nected to the NI Compact DAQ – 9174 to acquire the temperature data
captured by the thermocouples installed at different positions. The

4
F.J. Montero et al. Applied Thermal Engineering 224 (2023) 119997

Fig. 2. STEG coupled LHSCS system methodology, including verification of the results in the numerical and experimental analysis.

positions of thermocouples used in this test rig are presented in Fig. 4. fins with the copper bar and the TEG cold side to the copper bars. The
Three type-T thermocouples are situated on the 2nd, 4th, and 6th MGCHEMICALS 8329TCM adhesive joins the aluminum heat concen­
compartments of the LHSCS, as shown in Fig. 3. All the thermocouple trator with the TEG hot side. The components and the materials used in
connections are isolated inside the zone box and then connected to the the model are shown in Fig. 5.
DAQ system. The data acquisition system includes a zone box and an ice
bath. All thermocouples are first connected inside the zone box and then 2.2.1. Uncertainty analysis
connected to the data acquisition system (DAQs) to avoid temperature Type-T and type-K thermocouples are used to measure the temper­
differences in the connections and introduce more uncertainty. The ice atures at different positions of the experimental test rig. Four Type-T
bath helps maintain a reference temperature relative to the ice point. In intrinsic thermocouples are present in each radial fin. Three type- T
this way, the DAQ module only measures the millivolt signal from each thermocouples, separated by 5 mm, are located near the axial copper bar
thermocouple. The reference signal, measured from the ice point ther­ to quantify the heat flux transferred by conduction to each fin. The
mocouple, is used to verify the temperature measurement of each fourth thermocouple is placed at the periphery of each fin to quantify the
thermocouple using a seventh-order polynomial fit provided by the heat flux transferred by convection to the PCM. Six type-T thermocou­
OMEGA Engineering Temperature Measurement Handbook for the type ples are located in the PCM to measure the temperature variations inside
K and T thermocouple pairs. The various properties of the materials used the PCM and verify the phase and the melting front at a specific time.
in the experimental test rig are described in Table 1. Type-K thermocouples measure the temperature on the heat concen­
In the LHSCS, two types of materials are present. One is low thermal trator, container wall, TEG hot and cold sides, and ambient temperature.
conductivity materials (insulation materials), and the other is high The thermocouples work in their measure range to ensure the tem­
thermal conductivity materials (heat transfer enhancers). The acrylic perature data reliability, type-T for 0 to 100 ◦ C and type-K for 0 to
PCM container and its side covers have low thermal conductivity, and 220 ◦ C. The reported uncertainty (δXi, N) for the used thermocouples is
the thermal conductivity of the copper heat sink consisting of copper calculated as [30]:
bar, copper plate, and copper fins is high. As observed from Table 1, the { }1/2
thermally conductive adhesive paste used to join different components δXi,N = B2i,0 + δX 2i,1 (1)
in the system has relatively low thermal conductivity compared to the
heat sink material (copper). The ARCTIC MX-4 paste joins the copper

5
F.J. Montero et al. Applied Thermal Engineering 224 (2023) 119997

Fig. 3. (a) Experimental set-up and (b) Schematic of the STEG coupled LHSCS prototype (the location of type-T probes are indicated in the white label).

where Bi,0 is the probably fixed error band of the instruments, and δXi,1 is TEG. The heat flux concentrated using an optical concentrator (Qconc) at
the first-order uncertainty associated with the experimental prototype. the TEG hot side is given as [6]:
The first term on the right-hand side of Eq. (1) is calculated from the
Qconc = Copt ηopt Qsun (2)
average error band of the measurements made against the calibration
instrument. Temperature measurements are compared with those taken
where Copt is the optical concentration factor, ηopt is the optical
by a calibrated thermistor of the Laboratory for Advanced Thermal and
concentrator efficiency, and Qsun is the inlet solar irradiation. From the
Fluid Systems, Department of Mechanical Engineering, Villanova Uni­
concentrated heat flux (Qconc), Qloss is wasted in the form of radiative and
versity, Philadelphia, USA. The second term is the standard deviation of
convective heat losses from the TEG hot side to the ambient. The
the measurements when a previously set value is measured. After cali­
remaining heat flux, used for electricity generation from the TEG mod­
bration, the uncertainty of temperature measurements is estimated to be
ule (Qteg), is given as [6]:
± 0.5 ◦ C and ± 1.1 ◦ C for the type-T and type-K thermocouples,
respectively. The observed uncertainty represents 0.5% of the mea­ Qteg = Qconc − Qloss (3)
surement made, which is the value reported in this analysis. The un­
Qteg is partially converted into electricity (W), and the remaining heat
certainty analysis is included in a supplementary file.
(Qres) is rejected from the TEG cold side. The net energy balance is given
as [6]:
3. Formulation
Qteg = W + Qres (4)
3.1. Governing equations of STEG
The TEG efficiency (ηTEG) is dependent on the Seebeck coefficient (S)
and the semiconductor material’s dimensionless figure of merit (ZT).
The STEG comprises an optical concentrator, heat absorber, and

6
F.J. Montero et al. Applied Thermal Engineering 224 (2023) 119997

Fig. 4. Thermocouples position in the experimental test rig.

internal resistance (Ri) and the TEG absorptance area (As). The STEG
Table 1
conversion efficiency is given as [6]:
Properties of the various materials used in the experimental setup.
Material Cp @25 ◦ C (J/g K) ρ (kg/ k @25 ◦ C I[S(Th − Tc ) − IRi ]
ηSTEG = (6)
m3) (W/m K) Qconc As
Glass wool 0.43 96 0.16
Aluminum 0.89 2700 237
Copper 0.38 8960 400
3.2. Governing equations of LHSCS
Acrylic plastic 0.63 1190 0.17
ARCTIC MX-4 Thermal – 2500 8.5 The heat transfer in the STEG-coupled LHSCS consists of the
Compound Paste conductive heat transfer in solids and the conductive and convective
MG CHEMICALS 8329TCM 2300 1.4
heat transfer in the PCM in the LHSCS. Fourieŕs law and the energy
–
Thermally conductive
epoxy adhesive conservation equation are used to govern the heat transfer, which are
Cartridge heaters Cylindrical electrical given as [32]:
resistances of 96 W each,
regulated by a power ∂T
ρC p + ρCp μ∇T + ∇⋅q = Q (7)
source. ∂t
Thermocouples Type K, standard error
0.75%. q = − k∇T (8)
Type T, standard error
0.75%. where ρ is the fluid density, Cp is the constant pressure specific heat
capacity, T is the fluid temperature, and q is the heat flux. The laminar
The maximum conversion efficiency (ηmaxTEG) of the TEG is the product flow is dominated by the continuity and Navier Stokes equations,
of the Carnot efficiency and TEG efficiency (ηTEG), which is given as assuming a Newtonian and incompressible fluid flow, which are
[31]: expressed as [32]:
( ) √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ∂u ( ( ))
ηmaxTEG = h
T − Tc 1 + ZTm − 1
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ Tc (5) ρ + ρ(u∇)u = ∇ − pI + μ ∇u + (∇u)T +F (9)
Th 1 + ZTm + Th
∂t

ρ∇u = 0 (10)
where Tc and Th are the TEG cold and hot side temperatures, and Tm is
the average temperature. where μ is the dynamic viscosity, u is the fluid velocity, I is the identity
Like the TEG efficiency, the STEG conversion efficiency depends on matrix, and F represents the buoyancy forces.
the TEG hot and cold sides temperature difference (Th-Tc), concentrated The natural convection inside the PCM has been modeled using a
heat flux (Qconc), Seebeck coefficient (S), electrical current (I), TEG buoyancy force (F) in laminar flow, following a Boussinesq approach.

7
F.J. Montero et al. Applied Thermal Engineering 224 (2023) 119997

Fig. 5. Materials of different components of STEG coupled LHSCS experimental test rig.

This means that the variation in the PCM density induced by the tem­ 3.3. Numerical method and validation
perature variation is taken into account for the force term only (despite
the assumption of incompressible flow). The action of the gravity force The STEG-coupled LHSCS system, described by Eqs. (7) to (14), was
inside the PCM finally causes the upward flow of the liquid PCM in the previously simulated in the COMSOL Multiphysics software as 2D and
compartments of LHSCS. The material properties are computed as a time-dependent axisymmetric model by Montero et al. [29]. This model
function of the temperature (T) using the phase change phenomenon comprises a non-isothermal flow Multiphysics for simulating the fluid
defined by Kim and Kaviany [33], which are defined as: flow and heat transfer coupling. Multiphysics interfacing is the conju­
gate heat transfer-laminar flow. The concentrated solar radiation for the
ρ = θρphase1 + (1 − θ)ρphase2 (11)
selected geographical zone is the heat flux at the TEG hot side in this
model. Alumina and Bi2Te3 are collectively modeled as the TEG, and the
1( ) ∂αm
Cp = θρphase1 Cp,phase1 + (1 − θ)ρphase2 Cp,phase2 + L (12) PCM is modeled as a fixed grid. Alumina and Bi2Te3 are considered
ρ ∂T
temperature-dependent, and their thermal resistances are chosen from
k = θkphase1 + (1 − θ)kphase2 (13) the COMSOL library. The experimental results have confirmed the cor­
rected height of the Bi2Te3 component by comparing the temperature
1 (1 − θ)ρphase2 − θρphase1 achieved on the TEG cold side. The TEG geometry is cylindrical and
αm = × (14) modeled as a two-dimensional axisymmetric model. A TEG circular
2 (1 − θ)ρphase2 + θρphase1
transversal area, equal to the TEG quadratic transversal area, has been
These material properties equations include the mass fraction (α), assumed to simplify the 3D model and does not imply significant errors
constant pressure heat capacity (Cp), PCM solid fraction (θ), thermal compared to the experimental results. The dimensional and physical
conductivity (k), density (ρ), and the subscripts phase1 and phase2 refer models of the system are depicted in Fig. 6.
to solid and liquid phases respectively. The LHSCS computational study consists of a time-dependent model
for 86400 s (24 h). COMSOL Multiphysics software uses a Finite Element

Fig. 6. Components and dimensions (mm) of the computational model of the proposed system, previously presented by Montero et al. [29].

8
F.J. Montero et al. Applied Thermal Engineering 224 (2023) 119997

Method (FEM) solver. The software will automatically adjust the step experimentally measured temperatures on the container walls. The
size based on the current error estimates concerning the defined toler­ following empirical correlations have been used to approximate the
ances. The implicit differential equations derived from finite element natural convection heat transfer coefficient in the container covers and
discretization in COMSOL Multiphysics are solved with predefined walls [35].
precision requirements, subject to user-supplied absolute and relative ( )1/4
tolerances. Software pre-defined relative tolerance of 0.005 and abso­ h = 1.32
ΔT
(15)
lute tolerance of 0.001 are used in this analysis. A step size adjustment D
(time convergence) is performed automatically if the predetermined ( )1/4
step does not meet the estimated error. In this case, the step size is h = 1.42
ΔT
(16)
reduced. It has been observed that the solving steps reach a minimum of D
0.01 s and a maximum of 100 s, according to the computational solving Using the container dimensions presented in Fig. 5, the minimum,
equations effort. The mesh independence for this numerical analysis is maximum and average natural convection heat transfer coefficients are
given in a previous study by the authors [29]. found to be 2, 4, and 3 W/m2 ◦ C, respectively. The opaque acrylic wall
emissivity is set at 0.8. The mesh for the numerical model was previously
4. Results and discussions analyzed [29]. The proposed system computational simulation considers
a time-dependent analysis with a one-second step and a time interval of
The results of the numerical model of STEG-coupled LHSCS, previ­ 0–86400 s. The LHSCS geometry is the same for both the experimental
ously developed by the authors in [29], have been verified by devel­ and computational models.
oping the experimental test rig in the Advanced Thermal and Fluid The experimental test rig includes the TEG hot side temperature
Systems laboratory at Villanova University. The proposed system control after applying the electrical power of 96 W (24 V*4 A) on the
maintains the TEG cold side temperature at relatively lower values by three cartridge heaters of the heat concentrator. The electrical power
utilizing its waste heat effectively. The system also produces electricity was applied to the cartridge heaters until the hot side reached a tem­
using the heat stored in the PCM. The proposed system has been simu­ perature of around 220 ◦ C; after that, the power supply was set off. The
lated for the Atacama Desert environmental conditions, and the proto­ TEG hot side is thermally connected to the heat concentrator; therefore,
type was previously adjusted under laboratory conditions. The geometry its temperature follows the variation of the heat concentrator tempera­
of the LHSCS has been optimized for obtaining a significant temperature ture, as shown in Fig. 7 (a). Further, the TEG cold side temperature
difference of above 100 ◦ C between the TEG hot and cold sides. The follows the TEG hot side temperature variation till 16.5 h, and after that,
LHSCS geometry to achieve this temperature difference is presented in the cold side temperature becomes higher than that of the hot side since
Fig. 6. Five fins are found to be sufficient for a low temperature at TEG the stored heat maintains the temperature of the TEG cold side. The
cold side, using an iterative numerical simulation method [29]. Six kg of average relative errors for the complete day simulations between the
Puretemp 48X (organic paraffin) was used as the PCM. The manufac­ experimental and computational models for the TEG hot side and TEG
turer provides its thermophysical properties, which are given in Table 2 cold side temperatures are 0.37% ±0.29% and 4.28% ±2.67%, respec­
[34], and are guaranteed by DSC and other tests [27]. The weight of the tively. These results show the correct fit between the experimental and
copper heat sink was 2.5 kg. Other material physical properties in the computational models by adjusting the TEG equivalent thermal resis­
numerical simulations have been taken from the COMSOL software. tance and the heat losses to the environment. A maximum temperature
As presented in the previously analyzed computational model [29], difference of 140 ◦ C (Th = 220 ◦ C and Tc = 80 ◦ C) is obtained between
the TEG module, heat sink, and container are simulated as solids, and the TEG hot and cold sides under laboratory conditions, as shown in
the PCM is simulated as a fluid. A phase change temperature difference Fig. 7 (a). For 17 to 24 h, the TEG cold side temperature is greater than
of 1 K is set between the PCM solid and liquid phases. Initially, all the that of the hot side, which indicates the reversed heat flux on the TEG
components are set at the reference temperature of 25 ◦ C. The TEG hot- and, therefore, the reversion of the electron flux, which generates a
side temperature of the computational model is set equal to the TEG hot- negative open-circuit voltage in the TEG. This change in the electron
side temperature measured on the experimental test rig. The experi­ flux direction can be controlled using a transistor circuit to generate
mental temperature variations throughout the day are controlled by positive voltage.
varying electrical power to the cartridge heaters. The TEG cold side Fig. 8 shows the variation of the TEG open-circuit voltage that fol­
temperature is also measured during the experimental test rig. The TEG lows the same trend of the TEG hot and cold side temperature differ­
cold side temperature obtained in the computational model was ences as shown in Fig. 7. The maximum open-circuit voltage measured
adjusted to the experimental results through the variation of the height under laboratory conditions is around 5.8 V, which is in line with the
of the Bi2Te3. The geometric and thermal properties of various compo­ data presented by the manufacturer [36] and observing a relative error
nents of the TEG are given in Table 3. of around 3% as compared with the theoretical datasheet voltage. The
The computational model also considers the heat losses by convec­ TEG open-circuit voltage becomes negative during the inversion of the
tion and radiation from the top and bottom covers and the container TEG temperatures at 16.5 h, as shown in Fig. 6 (b). During the day, the
walls to the environment. The convective heat transfer is calculated for hot side temperature is higher than the cold side, and the waste heat
an average laboratory environmental temperature of 20 ◦ C and the from TEG cold side is stored in the PCM. When the solar radiation in­
tensity decreases in the evening, the TEG hot side temperature starts
Table 2 decreasing, but the TEG cold side temperature is maintained at the PCM
Puretemp 48X physical properties (Puretemp LLC) [34]. melting temperature. The hot side temperature decreases at a higher
rate following the ambient temperature variations at night. Because the
Parameter Value
TEG hot side is in direct contact with the outside environment, at t =
PCM 48X 16.5 h, both TEG hot side and TEG cold side temperatures become equal,
Solid Thermal Conductivity (W/m◦ C) 0.25
Solid Specific Heat (J/g◦ C) 2.10
and, after that, the TEG hot side temperature becomes lower than that of
Solid Density (g/ml) 0.90 the cold side. Therefore, the inversion of heat transfer flux is observed
Melting point (◦ C) 48 from the TEG cold side to the TEG hot side. The inversion in the TEG heat
Liquid Thermal Conductivity (W/m◦ C) 0.15 flux also results in the electron flux inversion inside the thermoelectric
Liquid Specific Heat (J/g◦ C) 2.27
legs. The negative open-circuit voltage obtained during the inversion
Liquid Density (g/ml) 0.82
Heat Storage Capacity (J/g) 230 period is around − 0.075 V. This shows the possibility of generating

9
F.J. Montero et al. Applied Thermal Engineering 224 (2023) 119997

Table 3
Geometric and thermal properties of TEG components.
Material Height (m)*10-3 Area (m2)*10-3 Average Thermal Conductivity (W/m K) Average Thermal Resistance (K/W)

Alumina 1.60 1.600 36.000 0.031

Tin 0.43 0.560 64.000 0.012
Air 3.00 1.130 0.024 110.619
Insulation 3.00 0.002 0.080 18750.000
Bi2Te3 2.54 0.467 1.960 2.775
Bi2Te3 equivalent 8.52 1.600 1.960 2.720

Fig. 7. Transient variations of TEG hot and cold temperatures of both computational and experimental models for (a) the entire period and (b) for the reverse heat
flux period.

10
F.J. Montero et al. Applied Thermal Engineering 224 (2023) 119997

Fig. 8. Variation of open-circuit voltage produced by the TEG during the simulation under laboratory conditions.

surplus electricity by utilizing the heat stored in the PCM. experimental and computational models for the different PCM com­
Further, improvements in the TEG hot and cold side temperature partments temperatures is shown in Table 4. It is observed from Table 4
difference are necessary to enhance electricity generation during the and Fig. 9 that relative differences of over 5% are observed between the
inversion period. Moreover, it is essential to have a lower ambient computational and experimental PCM temperatures. The first explana­
temperature to get the inversion in the TEG hot and cold sides. In the tion for this difference is that the computational model only includes
experimental test, the environment temperature was controlled at heat transfer by conduction inside the PCM. The convection heat
around 20 ◦ C for 24 h period, which caused a low voltage generated transfer contributes significantly after the melting of the PCM. There­
during the inversion heat flux period. The average relative errors for the fore, the variations between the computational and experimental PCM
reverse heat flux period between the experimental and computational temperatures are more significant during the PCM melting periods. The
models for the TEG hot and cold side temperatures are 0.49% ±0.28% second explanation is that the computational model does not include the
and 1.36% ±1.12%. The air velocity in the laboratory was lower than contact resistances between the copper fins and the copper bar. Since the
0.1 m/s causing only natural convection heat losses to the surroundings thermal conductivity of the thermal adhesive pastes is around 40 times
at the container walls and covers. lower than the copper thermal conductivity, as observed in Table 2. It
The PCM temperatures are also measured in the experimental test implies a significant thermal resistance between the copper bar and
rig. The experimentally measured PCM temperatures are compared to copper fins that has not been included in the computational simulations
PCM temperatures obtained from the computational model for the 2nd, but affects the heat conducted to the PCM. These factors will be
4th, and 6th compartments, as shown in Fig. 9 (a)-(c), respectively. The considered for future experimental setup models to reduce the contact
positions and captions of all thermocouples are shown in Fig. 4. The resistances between the heat sink components. Finally, as the thermal
PCM temperature has been measured in each compartment at two conductivity paste used between the TEG module and the heat sink has a
different positions. For example, since the 2nd compartment is closest to higher conductivity than the TEG module, it implies a lower impact on
the TEG cold side and the 6th compartment is farthest, the PCM tem­ the temperatures obtained through the computational simulations, as
perature is highest in the 2nd compartment and then decreases with an observed in Fig. 7.
increase in the distance from the TEG cold side lowest PCM temperature The computational and experimental temperatures at the middle of
in the 6th compartment. The initial experimental temperature drop the PCM acrylic container in the 2nd, 4th, and 6th compartments are
observed in Fig. 9 (b) and (c) was due to the difference between the shown in Fig. 10. Three type-T probes are installed at the center of the
initial set temperature (25 ◦ C) and the laboratory room temperature 2nd, 4th, and 6th PCM compartments, as shown in Fig. 4. The temper­
(20 ◦ C). It is observed that the initial temperature in the middle and atures in the computational model at these three positions were
bottom of the LHSCS container was closer to the room temperature. compared to the experimentally measured temperatures. The average
The average relative error for the entire period between the relative error for the complete day simulation between the experimental
and computational models for the temperature of acrylic container in
the central part of the 2nd, 4th, and 6th compartments of the storage
Table 4
system are 3.04% ±1.50%, 1.83% ±1.39%, and 3.24% ±2.01%,
Average relative error compared to experimental measurements for selected
respectively. These lower differences between the computational and
PCM thermocouple positions.
experimental wall temperatures ensure that the convective and radiative
Compartment/ thermocouple Average relative Standard deviation of coefficients are well selected for adjusting the computational model. The
position error (%) relative error (%)
observed elimination of temperature peaks, appearance of valleys, or
Second/Position-1 (PCM21c) 7.28 2.76 damping in the temperature curves are due to several reasons. The
Second/Position-2 (PCM22c) 4.23 3.14
elimination of the temperature peak observed in the last period of
Fourth/Position-1 (PCM41c) 11.15 4.17
Fourth/Position-2 (PCM42c) 5.30 4.79 temperature rise on the TEG hot side in Fig. 7 is mainly due to the
Sixth/Position-1 (PCM61c) 3.71 1.68 geometric adaptation carried out in the computational model, which
Sixth/Position-2 (PCM62c) 3.52 2.13 consists of the change from a square section to a circular section of the

11
F.J. Montero et al. Applied Thermal Engineering 224 (2023) 119997

Fig. 9. Transient variations of the PCM temperatures of both computational and experimental models at two positions in the (a) 2nd, (b) 4th, and (c) 6th
compartment of the LHSCS.

TEG. This change is necessary to simplify the geometric model used in surface and, therefore, the elimination of temperature peaks.
computational simulation because it allows using an axisymmetric On the other hand, when simulating a 2D model, the heat flow is
model. The change from a square section to a circular section supposes, simplified to two axes, which entails minor differences with the volu­
among other things, an elimination of the vertices and lateral faces in the metric heat transfer verified in the experimental model. The appearance
TEG. The last results in a more homogeneous temperature profile on the of valleys or buffers is mainly observed in Figs. 7 and 9, showing the

Fig. 10. Variation of computational and experimental temperatures for the acrylic container in the 2nd, 4th, and 6th compartments of the LHSCS.

12
F.J. Montero et al. Applied Thermal Engineering 224 (2023) 119997

paraffin temperature inside the container. For this case, it must be annual average direct normal solar radiation is around 860 W/m2. The
considered that the compared temperatures correspond to the compu­ annual average temperature is 13 ◦ C. The significant temperature vari­
tational model that only includes conduction heat transfer. The pre­ ations during the entire day, having temperatures over 20 ◦ C with higher
ceding leads to the linearity of the temperature, typical of the phase solar radiation and under 10 ◦ C at night, are shown in Fig. 11. These
change physics at the constant temperature, being observed in the lower ambient temperatures during the night and higher environmental
melting period of the paraffin. Valleys are also seen in the experimental temperatures and solar radiation throughout the morning for the desert
results at the beginning of the paraffin temperature measurements in locations collectively contribute to the improved operation of the pro­
Figs. 9 and 10. The latter may be related to the procedure for reducing posed system. The latter improves the heat transfer between the system
the uncertainty in the measurements carried out in the laboratory. The and its surroundings, thus enhancing electricity generation. The
measurements use a reference temperature of a cold bath in which a computational model includes wind velocity to calculate the heat lost to
thermocouple was approximately at 0 ◦ C. This reference requires sta­ the environment through convection from the outer walls and the TEG
bilization at the beginning of the experiment measurements as the cold module. The annual average wind velocity is 6 m/s, as shown in Fig. 11.
bath contains a water–ice mixing in an insulated container. The The proposed computational model was simulated each month for
observed valleys in the temperature curves result from a rapid start of solar radiation for a whole day after fixing the environmental condi­
the temperature acquisition data before considering a waiting time until tions. The STEG system is exposed during the morning and afternoon to
the whole set of the water-ice mixture reaches a homogeneous tem­ the average monthly solar radiation for each hour, and during the night,
perature. This anomaly is corrected approximately 10 min after the the TEG system is still generating electricity by utilizing the heat stored
experiment starts. It is considered a non-determining error since it in the PCM. The simulations start at 6:00 AM and extend for 24 h for the
represents approximately 1% of the experimentation time. Nevertheless, computational model. A direct solar radiation concentration of 40x has
it should be considered that ice is added to the cold bath during the been used in the computational model. In the experimental model, this
experimentation time to ensure the reference temperature is 0 ◦ C. concentration is achieved using a Fresnel lens, but it is included as heat
Once the computational model was set up and adjusted to the supplied directly to the TEG hot side in the computational model. The
experimental configuration, this adjusted model was used to simulate variation of the TEG cold and hot side temperatures and the corre­
the proposed system under the environment and solar radiation of a sponding open-circuit voltages for each season are shown in Fig. 12 (a)-
particular geographic region. Moreover, an alternate method for (d). The results are presented in four graphs, one for each season. In
generating higher electricity at night is to select a place with a low Chile, summer corresponds to the months between December and
temperature during the night and early morning hours. Therefore, desert February, fall corresponds to the months between March and May,
sites are the most suitable locations allowing enhanced electricity gen­ winter corresponds to June through August, and spring corresponds to
eration using the proposed system. Therefore, a desert location with September through November. The observed peak is the consequence of
extreme temperature variations during morning and night and higher the lowest wind velocity (≈1 m/s) reported for the studied location. The
solar radiation levels was chosen. The main reason for choosing the latter results in lower heat convection losses to the ambient and higher
desert locations is that these zones perfectly matched the proposed temperatures at the TEG hot side. During the summer months, the TEG
model’s operational conditions. Like other desert areas, the Atacama hot side temperatures reach around 200 ◦ C during the daytime. During
Desert also has two favorable environmental conditions for STEG the sunset, the TEG temperature continuously decreases to around 40 ◦ C
operation and LHSCS. Firstly, it has a higher annual solar irradiation and and approximately 20 ◦ C at the end of the day. The TEG cold side follows
a quasi-total lack of clouds throughout the day. Secondly, the significant the TEG hot side temperature pattern. It reaches a maximum tempera­
variations of the environmental temperatures during the day and night ture of around 80 ◦ C and then decreases to 48 ◦ C around sunset, which is
result in temperatures close to 25 ◦ C at noon and 5 ◦ C at night. For this the PCM melting temperature. This temperature is maintained for
study, the Atacama Desert in Chile has been considered the experiment’s around 5 h and decreases to around 30 ◦ C. Thus, the TEG hot and cold
potential site. The environmental data provided by the Chilean gov­ sides maintain a difference of around 10 ◦ C for the entire inversion
ernment [37] is selected for this study. The environment data consists of period, which results in around 10 h of extra electricity generated by the
monthly average direct solar radiation, ambient temperature, and wind TEG.
velocity for each hour, as shown in Fig. 11. In this selected location, the The results reflect that the STEG-coupled LHSCS meets its objectives

Fig. 11. Hourly variation of monthly average direct solar radiation, ambient temperature, and wind speed [37].

13
F.J. Montero et al. Applied Thermal Engineering 224 (2023) 119997

Fig. 12. TEG hot side (Th) and cold side (Tc) temperature and corresponding open circuit voltage for (a) December (Summer), (b) May (Fall), (c) July (Winter), and
(d) October (Spring) under solar radiation and environmental conditions of the Atacama Desert.

of maintaining a more significant temperature difference between the follows the temperature difference observed for each month, the
TEG hot and cold sides and generating more electricity at night. How­ reversed period has more extension for the winter and fall seasons than
ever, the temperature difference is still low during the reverse period, the summer and spring seasons. The open-circuit voltage of 0.2 V, noted
considering that the environmental temperature during the night in the during the inversion period for the evaluated months, is quasi-constant,
summer months is around 8 ◦ C. Increasing the temperature difference and a low variation is observed in this value.
during the inversion period may be possible by adding a more efficient Similarly, in the winter months, the TEG hot side reaches a maximum
heat sink at the TEG hot side to reach the environment temperature temperature of 200 ◦ C, but it is only maintained for about 15 min. The
during the night. Systems like radiative cooling or enhanced convection TEG cold side reaches around 70 ◦ C. After the inversion time, it
cooling can help achieve this objective, but such a system should not permanently decreases to about 15 ◦ C at the end of the day. A temper­
block solar radiation during the daytime charging period. The open- ature difference of about 10 ◦ C is maintained during the inversion period
circuit voltage curve follows the temperature differences between hot for about 9 h. During the spring and summer months, the TEG hot side
and cold TEG sides. The peak open voltage observed in Fig. 12 (a)-(d) temperature is maintained at almost 200 ◦ C for about 5 h. The TEG cold
results from a high temperature on the hot side caused by the low air side temperature is maintained at 70 ◦ C for about 4 h. After the reversion
velocity and a low temperature on the cold TEG side. period, the TEG cold side temperature decreases to the PCM melting
Additionally, the calculated TEG voltage turns negative after a spe­ temperature of 48 ◦ C. In the reverse period, the temperature difference is
cific time. It is a consequence of the change in the heat flux direction in maintained at around 10 ◦ C. Thus, the TEG cold and hot sides reduce
the TEG system during nighttime due to the absence of solar radiation, their temperatures to around 20 ◦ C at the end of the day. The peak open-
the lower environment temperature, and the higher temperature of the circuit voltages for the winter and spring seasons were 4.8 V and 5.4 V.
LHSCS. For example, the peak of open-circuit voltage for summer was The open-circuit voltage generated by the STEG system is evaluated
around 6 V during the charging time. However, the negative voltage using a TEG temperature difference ΔT = TH-TC. This polynomial is
reaches around 0.2 V. In the inversion period, the heat flux inversion evaluated for the temperature difference obtained in the simulation of
time varies according to the monthly solar radiation levels and envi­ the STEG coupled LHSCS for each average monthly environment con­
ronmental conditions. dition and solar irradiations of the Atacama Desert.
In the fall months, dawn is around 8:00 h, meaning the TEG hot side The monthly energy produced by the proposed system under the
temperature increases immediately and reaches about 200 ◦ C around environmental conditions of the Atacama Desert has also been evalu­
noon. The charging period decreases to around 11:00 h and ends before ated. Eqs. (15), (16), and (17) were obtained by analyzing the temper­
the inversion of the TEG hot and cold sides temperatures around the ature curves presented by the TEG manufacturer. It is observed that the
sunset hours. It means that the inversion period extends for about 12 h. obtained values are consistent with the datasheet values published by
The temperature difference between the TEG hot and cold sides remains the TEG manufacturer [36]. The maximum matched voltage and current
at 10 ◦ C for around 8 h. This temperature difference reaches 15 ◦ C at the of 3 V and 0.6 A was calculated using the datasheet of the selected TEG
end of the day. In the simulation of a fall day, the peak open-circuit module and is considered for calculating the electricity generation. The
voltage reaches around 5.2 V. The negative voltage was maintained at maximum conversion efficiency corresponding to the temperature dif­
around 0.2 V for the entire inversion period. As the open-circuit voltage ference (145 ◦ C) obtained in the STEG-coupled LHSCS prototype is

14
F.J. Montero et al. Applied Thermal Engineering 224 (2023) 119997

Fig. 13. Transient variation of measured (a) open-circuit voltage, matched voltage, and resistance and (b) matched power and current of TEG.

around 5%. Fig. 13 (a) shows the transient variation of open-circuit 0.0012 W is obtained in the absence of the external heat source.
voltage, matched voltage, and resistance of the TEG module TE-MOD- The proposed system’s monthly average electricity generated
5W5V-40S. Three periods are observed during the measurements. The throughout the year is depicted in Fig. 14. The monthly average elec­
first period is the application of the heat flux on the TEG hot side. In this tricity produced by the TEG system is 477.92 Wh. Out of it, 475.29 Wh
period, the open-circuit voltage (VOC) reaches around 6 V following the and 2.63 Wh are generated during the day and the night, respectively.
temperature increase on the TEG hot side. The second period corre­ Electricity generation during the night is 0.6% of the total electricity
sponds to the absence of heat flux at the TEG hot side (the heat source is generation. The annual electricity generation using the STEG-coupled
turned off), and VOC reaches 0 V due to a decrease in temperature at LHSCS is 5735 Wh. Though electricity generation by the TEG at night
around 16 h. After this period, the heat stored in the PCM allows a is low, it establishes the viability of the latent heat storage and cooling
minimum inversed heat flux on the TEG, causing a minimum VOC of system for managing the TEG cold side temperature and generating
around 0.12 V. The last period generated a negative voltage caused by more electricity at night.
the inverse heat flux in the TEG module. However, only the magnitude The transient variation of energy efficiencies of TEG calculated for
of the generated VOC in the inversed heat flux period is shown in Fig. 13 the daytime and nighttime are shown in Fig. 15 (a) and (b), respectively.
(a). It also shows the matched voltage (Vm) calculated in the TEG module The maximum daytime energy efficiency reaches around 4.4% during
using the data presented in the TE-MOD-5W5V-40S data sheet [36]. A the presence of the external heat source. In contrast, the maximum
polynomial approximation for Vm is generated using the temperature nighttime energy efficiency reaches around 0.11%. Fig. 16 shows the
difference across the TEG hot and cold sides (Th - Tc), which is presented variation of the match current and power with the voltage for the TE-
in Eq. (15), where x is the temperature difference between the TEG hot MOD-5W5V-40S thermoelectric module. It is observed that the power
and cold sides (ΔT). output increases with the voltage, which is evident, and there is an
optimum voltage at which the TEG should be operated to get the
Vm = − 0.00000003x3 − 0.00003241x2 + 0.02732069x − 0.01757096 (15)
maximum current.
The Vm reaches around 3 V while applying the external heat flux on Fig. 17 (a) shows the variation of the matched voltage, current, and
the TEG hot side. When the heat flux is obtained from the LHSCS, Vm power output, and Fig. 17 (b) shows the variation of energy efficiency
reaches around 0.06 V. Since the ambient temperature is controlled at with the temperature difference between the hot and cold sides of the
around 21 ◦ C, the temperature difference (Th - Tc) was low during the TEG. It has been observed that the output voltage and power increase
absence of an external heat source. In this period, Vm represents around with an increase in temperature difference because of the output
0.2% of the Vm measured in the presence of the external heat source. The voltage, and thus, the output power are directly proportional to the
matched resistance (Rm) and current (Im), as shown in Fig. 13 (a) and (b), temperature difference. It is observed that for the most significant
respectively, are also calculated for this experimental setup. The poly­ temperature difference (145 ◦ C), the maximum TEG power output is
nomial approximations for both matched current and resistance are around 1.75 W. Further, due to the minimum temperature difference
presented in Eq. (16) and (17), respectively, where x is the temperature obtained in the absence of the input heat source, the TEG power output
difference between the TEG hot and cold side (Th - Tc) and y is the TEG in this period is 0.0012 W for a temperature difference of 3.4 ◦ C. As
hot side. shown in Fig. 17 (b), the TEG energy efficiency reaches a maximum of
around 4.5%, corresponding to the optimum temperature difference
Im = − 0.00000011x3 + 0.00000360x2 + 0.00564739x − 0.00270758 (16) achieved in this study. Further, it has been found that although the
power output increases with an increase in temperature difference, the
Rm = − 0.00000002y3 + 0.00000564y2 + 0.0084752y + 3.43415667 (17) energy efficiency does not always increase with an increase in temper­
The matched current reaches around 0.55 A and 0.01 A for the period ature difference. Thus, there is an optimum temperature difference at
with and without external heat flux. In the case of matched resistance, maximum energy efficiency.
the minimum value corresponds to 3.66 Ω. It is increased according to Fig. 18 shows the variation of the TEG energy efficiency with the
the TEG hot side temperature variation, reaching a maximum of 5.36 Ω input solar energy. The increase in the TEG conversion efficiency is
when the maximum temperature difference is achieved (Th – Tc ≈ observed according to the concentrated solar radiation (Qconc) received
145 ◦ C). The calculated matched power for the entire period is also on the TEG hot side. The conversion efficiency reaches the maximum
presented in Fig. 13 (b). The power is calculated using the match current value when the TEG receives the maximum direct normal irradiation
and voltage values obtained previously. The maximum matched power (around mid-day). The main advantage of the proposed passive cooling
obtained is around 1.75 W. This value is obtained at the end of the system is that it does not need electricity consumption during operation
charging process when the maximum temperature difference is observed time. Also, the LHSCS could be improved to allow the TEG system to
between the TEG hot and cold sides. The maximum matched power of generate more electricity during the night with a deep analysis of the

15
F.J. Montero et al. Applied Thermal Engineering 224 (2023) 119997

Fig. 14. Annual TEG electricity generation for the Atacama Desert during the night, morning, and the entire day.

Fig. 15. Transient variation of (a) nighttime and (b) daytime energy efficiency of STEG coupled LHSCS.

Fig. 16. Variation of matched power output and current of TEG with voltage.

16
F.J. Montero et al. Applied Thermal Engineering 224 (2023) 119997

Fig. 17. Variation of (a) matched power output, voltage, and current and (b) efficiency of TEG with temperature difference.

Fig. 18. TEG conversion efficiency variation with the concentrated solar radiation heat at the TEG hot side.

geometry and PCM selection, which will be presented in future analysis. In terms of technical feasibility, the STEG, coupled with the con­
On the one hand, the possibility of generating electricity using the version efficiency of commercial TEG modules, initially limits the LHSCS
heat stored in the phase change material increases when the proposed system. Although such systems’ energy conversion has increased, it
system is operated under the climate conditions of desert locations. The barely reaches 10% when used in solar energy conversion systems.
heat storage system design also allows it to operate as a temperature- Therefore, a holistic study is necessary that groups the development of
controlling system for the TEG cold side during the daytime. Since this new thermoelectric materials, new configurations of generators and
study demonstrates the STEG-coupled LHSCS system feasibility with systems, and new systems for managing and controlling the heat flow in
high solar radiation levels in a desert location, the cooling system is TEG systems. The preceding would have a significant impact on eco­
expected to fulfill its objective in any geographic location. On the other nomic feasibility. With low conversion efficiency, STEG systems have
hand, the capacity of the heat storage system for electricity generation limited competitiveness compared to other more developed technolo­
during nighttime in locations with relatively higher temperatures would gies such as PV or solar thermal, except in applications where the heat
be limited by the minimum ambient temperature reached during the resource is very abundant or in residual heat applications. However,
night period. The average nocturnal environmental temperature for the STEG systems could also complement the full use of solar radiation. It
studied desert location is 8 ◦ C. Although this temperature is lower than
that observed in the daytime, it does not differ excessively from the
Table 5
nighttime temperature in other geographic locations. In geographic lo­
Installation cost for some PV and STEG systems.
cations where the night temperature is close to the temperature of the
heat reservoir, a different heat transfer system could be implemented, Reference System Installation cost (USD/kW)

such as radiative cooling or an enhanced convective heat transfer sys­ Statista (2021), [39] PV 883
tem, to increase the heat flow through the thermoelectric generation CNE (2020), [40] PV 669–993
Narducci et al. (2018), [41] STEG 1954
system. Another alternative would be to use the storing heat in other
Habibollahzade et al. (2018), [42] STEG 1500
applications such as space conditioning, food drying, or applications Rodrigo et al. (2019), [43] STEG 510
that require low-temperature heat. Montero et al. (2021), [29] STEG 1500

17
F.J. Montero et al. Applied Thermal Engineering 224 (2023) 119997

has also been verified that the TEG systems can efficiently evacuate the Table 6
remaining heat from the PV systems, allowing them to maintain their LCOE and LCOS for some PV and STEG systems.
controlled operating temperature. Thus, maintaining or increasing the Reference System LCOE (USD/MWh)
conversion efficiency from solar to electric energy conversion in the
Lazard (2020), [44] PV 29–227
hybrid system improves its economic feasibility [38]. Tables 5 and 6 Lozano (2011), [45] STEG + Geothermal 8404
show economic parameters, including the installation cost, levelized Lazard (2020), [44] PV + BESS LCOS: 406–506
cost of energy (LCOE), and levelized cost of storage (LCOS) for PV and Chen & Soderberg (2013), [46] STEG 236
STEG systems and the proposed system [29]. Montero et al. (2021), [29] STEG + LHSCS 8850
LCOS: 566
Verifying the techno-economic feasibility, the STEG systems are not
yet competitive with PV solar energy conversion systems in terms of
conversion efficiency and costs. Therefore, research studies such as the • The temperature difference of 120 ◦ C is obtained between the hot
one carried out in this analysis seek to resolve the limitations of these and cold sides of TEG using 6 kg of PCM and 2.5 kg of copper heat
type of systems, including questions as to how to use the remaining heat sink.
from the conversion process and how to keep the temperature of the TEG • The proposed system outperforms in the geographical locations of
cold side controlled without the need for additional electricity using the desert areas because of higher solar radiations and maximum
passive storage and cooling systems. Further, the scalability of this ambient temperature differences during day and night.
system to large scale requires several TEG modules connected thermally • The STEG generated around 0.6% of the total electricity generation
in parallel to cover a large area over the LHSCS that can generate power during off-sunshine hours in the Atacama Desert.
at larger scale and effectively utilize the waste heat from the TEG cold • The proposed system generated average monthly and annual elec­
side to enhance the overall performance of this proposed system. This tricity of 478 Wh and 5735 Wh, respectively, because of the high and
STEG coupled LHSCS system will boost the energy generation and a quasi-constant annual solar irradiation in the Atacama Desert.
better use of the phase change material. • The LCOE and LCOS of the proposed system are 8850 and 566 USD/
During the solar radiation period (daytime), the PCM melting tem­ MWh respectively.
perature can control the temperature difference between the TEG hot • Natural convection plays a crucial role in PCM melting; therefore, it
and cold sides. If a PCM with a high melting temperature is used, it must be verified in the design of a PCM container.
reduces the TEG conversion efficiency during solar radiation hours. On
the other hand, if PCM with a low melting temperature is used, then it The effect of natural convection in PCM on the TEG cold side tem­
would reduce the conversion efficiency during nighttime. In other perature and the STEG performance is not studied in depth and will be
words, there exists a compromise between the TEG cold side tempera­ analyzed in future studies. Further, it is anticipated that a third system
ture operation and the PCM melting temperature that requires all-day can be coupled to take advantage of the remaining heat from the TEG
temperature coupling between the desired temperatures and the envi­ hot and cold sides during the daytime and nighttime for heating, drying,
ronment temperature of the system location. As explained previously, or other applications requiring low-temperature heat. Another proposal
the LHSCS system meets the objective of controlling the temperature of is to include more thermoelectric modules in the periphery of the storage
the TEG cold side. However, it has limitations of low efficiency of heat system to generate additional electricity.
storage system for electricity generation during nighttime. The PCM
melting temperature governs the TEG cold side temperature. Because Declaration of Competing Interest
the temperature difference would be given between the PCM tempera­
ture and that of the environment, considering the respective heat losses The authors declare that they have no known competing financial
to the environment, the LHSCS system would not allow a large amount interests or personal relationships that could have appeared to influence
of energy to be generated in by TEG at night. In this study, around 0.6% the work reported in this paper.
of the total electricity was generated during the nighttime. This
nocturnal generation was limited by the morning’s operating tempera­ Data availability
ture and the geographical location’s environmental conditions.
No data was used for the research described in the article.
5. Conclusions
References
In this study, the experimental test rig of STEG-coupled LHSCS is
developed, and the results of its computational model built in COMSOL [1] E.B. Alibakhsh Kasaeian, A. Shamaeizadeh, Christos Tzivanidis, Solar-driven poly-
generation systems: Recent progress and outlook, Appl. Energy 264 (2020),
Multiphysics software have been validated using the experimental
114764.
setup. An experimental setup is built using the container geometry of the [2] M.L. Olsen, E.L. Warren, P.A. Parilla, E.S. Toberer, C.E. Kennedy, G.J. Snyder, S.
numerical model and under laboratory environment-controlled condi­ A. Firdosy, B. Nesmith, A. Zakutayev, A. Goodrich, C.S. Turchi, J. Netter, M.
tions. The effect of ambient temperature, solar radiation and wind speed H. Gray, P.F. Ndione, R. Tirawat, L.L. Baranowski, A. Gray, D.S. Ginley, A High-
temperature, High-efficiency Solar Thermoelectric Generator Prototype, Energy
of the Atacama Desert on the performance of the proposed system are Procedia 49 (2014) 1460–1469, https://doi.org/10.1016/j.egypro.2014.03.155.
analyzed. The economic analysis has also been carried out in brief. The [3] Kumar R, Maduabuchi C, Lamba R, Vashishtha M, Upadhyaya S. Transient
experimental model has been used to confirm and rectify the results optimization of a segmented variable area leg geometry-based solar thermoelectric
generator. In2021 IEEE Green Energy and Smart Systems Conference (IGESSC)
obtained in the numerical approach. Moreover, the possibility of tem­ 2021 Nov 1 (pp. 1–6). IEEE.
perature control of the TEG cold side and enhancement of electricity [4] G. Chen, Theoretical efficiency of solar thermoelectric energy generators, J. Appl.
production by the STEG using the coupled LHSCS has been discussed. Phys. 109 (10) (2011) 104908.
[5] C.C. Maduabuchi, R. Lamba, M. Eke, E. Ejiogu, Multi-dimensional optimization of a
The main finds of this work are as follows: concentrated solar thermoelectric generator, International Journal of Energy
Research 46 (5) (2022 Apr) 5896–5907.
• The maximum hot and cold side temperatures of TEG are 200 ◦ C and [6] B. Lorenzi, G. Chen, Theoretical efficiency of hybrid solar thermoelectric-
photovoltaic generators, J. Appl. Phys. 124 (2) (2018) 024501.
80 ◦ C respectively. [7] D. Kraemer, Q. Jie, K. Mcenaney, F. Cao, W. Liu, L.A. Weinstein, J. Loomis, Z. Ren,
• Using a commercial TEG with a peak efficiency of 6%, an experi­ G. Chen, Concentrating solar thermoelectric generators with a peak efficiency of
mental TEG efficiency of 5% is obtained. 7.4%, Nat. Energy 1 (11) (2016) https://doi.org/10.1038/nenergy.2016.153.

18
F.J. Montero et al. Applied Thermal Engineering 224 (2023) 119997

[8] L. Tan, B. Singh, A. Date, A. Akbarzadeh, Sustainable Thermoelectric Power System [26] J. Wang, X. Song, Q. Ni, X. Li, Q. Meng, Experimental investigation on the
Using Concentrated Solar Energy and Latent Heat Storage, IEEE Int. Conf. Power influence of phase change material on the output performance of thermoelectric
Energy, no. December, pp. 2–5, 2012. generator, Renew. Energy 177 (2021) 884–894.
[9] L. Tan, R. Singh, A. Date, A. Akbarzadeh, Thermal performance of two-phase closed [27] Puretemp. (n.d.). Frequently asked questions about phase change materials.
thermosyphon in application of concentrated thermoelectric power generator using Retrieved September 5, 2022, from https://puretemp.com/?p=330#panel8.
phase change material thermal storage, Front. Heat Pipes, 2012. [28] M. De Lucia, A. Bejan, Thermodynamics of Energy Storage by Melting Due to
[10] A. Abhat, Low temperature latent heat thermal energy storage: Heat storage Conduction or Natural Convection, J. Sol. Energy Eng. 112 (2) (1990) 110–116,
materials, Sol. Energy (1983). https://doi.org/10.1115/1.2929642.
[11] I. Dincer, M. Rosen, Thermal Energy Storage Systems and Applications, 2011. [29] F.J. Montero, R. Lamba, A. Ortega, W. Jahn, A.M. Guzmán, A novel 24-h day-night
[12] M.E. Kiziroglou, S.W. Wright, T.T. Toh, P.D. Mitcheson, T. Becker, E.M. Yeatman, operational solar thermoelectric generator using phase change materials, J. Clean.
Design and fabrication of heat storage thermoelectric harvesting devices, IEEE Prod. 10 (296) (2021 May), 126553.
Trans. Ind. Electron. 61 (1) (2014) 302–309. [30] R.J. Moffat, Describing the uncertainties in experimental results, Exp. Therm Fluid
[13] F. Agyenim, N. Hewitt, P. Eames, M. Smyth, A review of materials, heat transfer Sci. 1 (1) (1988) 3–17.
and phase change problem formulation for latent heat thermal energy storage [31] G.D. Mahan, Inhomogeneous Thermoelectrics, J. Appl. Phys. 70 (8) (1991)
systems (LHTESS), Renew. Sustain. Energy Rev. 14 (2) (2010) 615–628. 4551–4554, https://doi.org/10.1063/1.349091.
[14] M. Jaworski, M. Bednarczyk, M. Czachor, Experimental investigation of [32] R.L. Panton, Incompressible Flow, second ed., John Wiley & Sons Inc., 1996.
thermoelectric generator (TEG) with PCM module, Appl. Therm. Eng. 96 (2016) [33] C.-J. Kim, M. Kaviany, A numerical method for phase-change problems, Int. J. Heat
527–533. Mass Transf. 33 (12) (1990) 2721–2734.
[15] C. Selvam, S. Manikandan, N.V. Krishna, R. Lamba, S.C. Kaushik, O. Mahian, [34] Puretemp. (n.d.). Puretemp 48X. Retrieved June 3, 2020, from https://puretemp.
Enhanced thermal performance of a thermoelectric generator with phase change octochemstore.com/shop/puretemp-48x/.
materials, Int. Commun. Heat Mass Transf. 114 (2020), 104561. [35] S.W. Churchill, H.H.S. Chu, Correlating Equations for Laminar and Turbulent Free
[16] K. Karthick, S. Suresh, G.C. Joy, R. Dhanuskodi, Experimental investigation of solar Convection from a Horizontal Cylinder, Int. J. Heat Mass Transfer 18 (1975) 1049.
reversible power generation in Thermoelectric Generator (TEG) using thermal [36] TEGpro. n.d. “TEGpro 5W TEG Module.” Retrieved June 5, 2020 (https://www.
energy storage, Energy Sustain. Dev. 48 (2019) 107–114. tegmart.com/thermoelectric-modules/5w-5v-40m-teg-module).
[17] R. Lamba, S. Manikandan, S.C. Kaushik, Performance analysis and optimization of [37] Universidad de Chile (2020). Explorador Solar. http://ernc.dgf.uchile.cl:48080/
concentrating solar thermoelectric generator, J. Electron. Mater. 47 (9) (2018 Sep) inicio.
5310–5320. [38] F.J. Montero, R. Kumar, R. Lamba, R.A. Escobar, M. Vashishtha, S. Upadhyaya, A.
[18] X. Luo, Q. Guo, Z. Tao, Y. Liang, Z. Liu, Modified phase change materials used for M. Guzmán, Hybrid photovoltaic-thermoelectric system: Economic feasibility
thermal management of a novel solar thermoelectric generator, Energy Convers. analysis in the Atacama Desert, Chile, Energy 239 (2022) 122058.
Manag. 208 (Mar. 2020), 112459. [39] Statista (2021). Average installed cost for solar photovoltaics worldwide from 2010
[19] R. Bhuiya, N. Shah, D. Arora, N.V. Krishna, S. Manikandan, C. Selvam, R. Lamba, to 2020. https://www.statista.com/statistics/809796/global-solar-power-
Thermal management of phase change material integrated thermoelectric cooler installation-cost-per-kilowatt/.
with different heat sink geometries, Journal of Energy Storage. 1 (51) (2022 Jul), [40] CNE (2020). Informe de costos de tecnologías de generación. https://www.cne.cl/
104304. wp-content/uploads/2020/03/ICTG-Marzo-2020.pdf.
[20] Y. Wang, Y. Peng, K. Guo, X. Zheng, J. Darkwa, H. Zhong, Experimental [41] D. Narducci, P. Bermel, B. Lorenzi, N. Wang, K. Yazawa, Hybrid Solar Harvesters:
investigation on performance improvement of thermoelectric generator based on Technological Challenges, Economic Issues, and Perspectives, in: Hybrid and Fully
phase change materials and heat transfer enhancement, Energy 229 (Aug. 2021), Thermoelectric Solar Harvesting, Springer International Publishing, 2018,
120676. pp. 137–151, https://doi.org/10.1007/978-3-319-76427-6_8.
[21] S. Shittu, et al., Transient and non-uniform heat flux effect on solar thermoelectric [42] A. Habibollahzade, E. Gholamian, P. Ahmadi, A. Behzadi, Multi-criteria
generator with phase change material, Appl. Therm. Eng. 173 (Jun. 2020), optimization of an integrated energy system with thermoelectric generator,
115206. parabolic trough solar collector and electrolysis for hydrogen production, Int. J.
[22] K. Huang, Y. Yan, G. Wang, B. Li, Improving transient performance of Hydrogen Energy 43 (31) (2018) 14140–14157, https://doi.org/10.1016/J.
thermoelectric generator by integrating phase change material, Energy 219 (Mar. IJHYDENE.2018.05.143.
2021), 119648. [43] P.M. Rodrigo, A. Valera, E.F. Fernández, F.M. Almonacid, Performance and
[23] A. Rezania, S.A. Atouei, L. Rosendahl, Critical parameters in integration of economic limits of passively cooled hybrid thermoelectric generator-concentrator
thermoelectric generators and phase change materials by numerical and Taguchi photovoltaic modules, Appl. Energy 238 (2019) 1150–1162, https://doi.org/
methods, Mater. Today Energy 1 (16) (2020 Jun), 100376. 10.1016/J.APENERGY.2019.01.132.
[24] P.V. Escobar, D.I. Oyarzun, A. Arias, A.M. Guzmán, Experimental study of a hybrid [44] Lazard, Levelized Cost of Energy and Levelized Cost of Storage 2020, 2020,
solar thermoelectric generator energy conversion system, Energy Convers. Manage. https://www.lazard.com/perspective/levelized-cost-of-energy-and-levelized-cost-
15 (238) (2021 Jun), 113997. of-storage-2020/.
[25] S. Shittu, G. Li, Q. Xuan, X. Xiao, X. Zhao, X. Ma, Y.G. Akhlaghi, Transient and non- [45] A. Lozano, Analysis of a Novel Thermoelectric Generator in the Built Environment,
uniform heat flux effect on solar thermoelectric generator with phase change 2011¸ https://repositories.lib.utexas.edu/handle/2152/ETD-UT-2011-08-4131.
material, Appl. Therm. Eng. 5 (173) (2020 Jun), 115206. [46] G. Chen, R. Soderberg, Concentrated Solar Thermoelectric Power, 2013, https://
www.osti.gov/servlets/purl/1191490.

19