You are on page 1of 9

Available online at www.sciencedirect.

com

ScienceDirect
Solar Energy 132 (2016) 386–394
www.elsevier.com/locate/solener

Modeling of flat-plate solar thermoelectric generators


for space applications
Lei Liu ⇑, Xue Sen Lu, Mao Lei Shi, Ya Kun Ma, Jian Ying Shi
College of Electronic and Information Engineering, Hebei University, Baoding 071002, PR China

Received 13 December 2015; received in revised form 13 March 2016; accepted 14 March 2016

Communicated by: Associate Editor Michael Epstein

Abstract

A model for a flat-plate solar thermoelectric generator (STEG) with sandwich-like structure is established in this paper. The influence
of the thermal concentration ratio, length of the thermoelectric legs and other geometrical factors on the performance of the STEG are
investigated by numerical calculations. The results indicate that the maximum conversion efficiency and output power per unit mass can
reach 5.5% and 6.5 W/kg, respectively, for Bi2Te3 based STEGs working in the Earth’s orbit. For PbTe based STEGs working in the
orbits of Venus and Mercury, the maximum conversion efficiency can reach 4.7% and 5.8%, respectively. It is also shown that, under
high intensity radiation, the conversion efficiency of a flat-plate STEG can be maintained at a relatively stable level by adjusting the
device geometry, which is precisely the advantage of STEG when compared to photovoltaic devices.
Ó 2016 Elsevier Ltd. All rights reserved.

Keywords: Space energy supply; Thermoelectric conversion; Solar energy; Thermal concentration

1. Introduction still the problems faced by RTGs. In this scenario, a solar


thermoelectric generator (STEG) may become a better
Photovoltaic technology and radioisotope thermoelec- choice in order to supply a spacecraft with energy (Karni,
tric generators (RTGs) have been the main ways for 2011; Deng and Liu, 2009; Macdonald et al., 2010).
supplying energy to spacecrafts over the past few decades Thermoelectric energy generators convert thermal energy
and to date they have provided sufficient power to support directly into electrical power via the Seebeck effect, which
space exploration. However the performance of photo- is a solid state solar-heat to electricity conversion process.
voltaic cells would be severely compromised at the elevated There are many advantages of STEGs and these include
temperatures close to the sun (Landis et al., 2005). In radiation resistance, high temperature resistance, environmen-
addition, particle radiation and the high temperature will tal friendliness, low cost and long operational life. Although
cause irreversible damage to the structure of the photo- they did not attract much attention for a long time due to
voltaic cells, which would ultimately reduce the lifetime their low conversion efficiency, recently the use of STEGs
of the spacecraft. Although the RTGs have no the above has become feasible and hence has been the subject of many
mentioned disadvantages of photovoltaic cells, the research studies (Su et al., 2014; Dehghan et al., 2015; Bjork
shortage of radioisotopes material and security issue are and Nielsen, 2015; Dallan et al., 2015; Chavez-Urbiola et al.,
2012; Chen et al., 2014; Cai et al., 2015).
⇑ Corresponding author. Tel.: +86 312 5079656. Temperature differences determine the energy conver-
E-mail address: beimingzy@126.com (L. Liu). sion efficiency of the thermoelectric generator which is also

http://dx.doi.org/10.1016/j.solener.2016.03.028
0038-092X/Ó 2016 Elsevier Ltd. All rights reserved.
L. Liu et al. / Solar Energy 132 (2016) 386–394 387

a Carnot heat engine. So a large temperature difference (Baranowski et al., 2012). Kraemer’s modeling yielded a
between the hot-end and cold-end of the thermoelectric predicted efficiency of 5% for Bi2Te3 based terrestrial
legs is the key of improving the performance of STEGs STEGs (Kraemer et al., 2012), which is higher than the
in both outer space and ground applications. Concentrat- result of our preliminary work (Liu et al., 2013). However,
ing sunlight through a lens or mirror is a common method it should be noted that most studies mentioned above were
to achieve a large temperature difference between the ends based upon specific temperature differences (DT) or specific
of the thermoelectric legs. Telkes achieved a temperature cold-end temperatures (Tc) of the thermoelectric legs. In
difference of 247 °C with a conversion efficiency of 3.35% fact, the temperature difference (DT) between the hot and
by concentrating sunlight through a 50 lens (Telkes, cold end of the thermoelectric legs should be a dependent
1954). Li and his coauthors built a STEG system based variable under certain operating conditions, which could
on silver antimony lead telluride alloys, which could be used to calculate the conversion efficiency of STEGs.
achieve a conversion efficiency of 5% with a 55 Fresnel This paper presents an analysis of the potential perfor-
lens concentrator and water cooling (Li et al., 2010). mance of flat-plate solar thermoelectric generators for
Amatya also reported a STEG system using n-type ErAs: outer space applications. The influence of geometrical
(InGaAs)1x(InAlAs)x and p-type (AgSbTe)x(PbSnTe)1x structure on the temperature difference (DT) between hot
that reached a conversion efficiency of 5.6% under a and cold ends of thermoelectric legs is investigated by finite
120 reflex mirror (Amatya and Ram, 2010). However, element analysis based upon temperature dependent trans-
in order to ensure continuous operation it is necessary to port properties of thermoelectric materials. The conversion
add a solar tracking system for light concentration, and efficiency, maximum output power and maximum output
this makes STEGs more complicated and costly. Another power per unit mass of STEGs are further calculated under
way to achieve a large temperature difference between the the influence of solar radiation at 1AU (Earth’s orbit),
ends of thermoelectric legs is by using a flat-plate to achieve 0.72AU (Venus’ orbit) and 0.39AU (Mercury’s orbit),
thermal concentration, which is more suitable for space respectively. Although we primarily consider the working
applications because of its simple structure. For example, performance of STEGs in an outer space environment, this
the thermal concentration via a flat-plate has been used model also applies to terrestrial operation conditions.
in Fuschillo’s and Scherrer’s work to build STEGs that
supply energy to space based technologies (Fuschillo 2. Model and calculation method
et al., 1966; Eggleston and Fuschillo, 1965; Scherrer
et al., 2003). In 2011, Kraemer carried out experimental A unit cell of flat-plate STEG is primarily composed of
research upon Bi2Te3 based STEGs with a flat-plate and three parts: a flat-plate thermal absorber, a pair of p-type
achieved a conversion efficiency of 5.2% under a 196 times and n-type thermoelectric legs and a heat sink forming a
(1.5 kW/m2) thermal concentration (Kraemer et al., 2011). sandwich-like structure, as shown in Fig. 1. We use Aa to
Unfortunately Kraemer’s result was obtained when the represent the area of the square absorber. Aa = a2, where
cold side was fixed at 20 °C, which is difficult to maintain a is the side length of the absorber. L is the length of
in practical applications. thermoelectric legs, b is the cross-sectional side-length of
Solar thermoelectric generators have also attracted a the p and n legs and therefore At (At = b2) represents
range of theoretical and design studies. Gibson analyzed the cross-sectional area of the legs. The area of the heat
the influence of material, radiation intensity, heat sink sink is designed to be equal to the flat-plate thermal absor-
area, contact resistance and other factors on conversion ber so that it would also serve as a mechanical support. The
efficiency in STEGs, and discussed the advantages of ratio of area of the thermal absorber to the cross-sectional
segmented STEG (Gibson and Fuschillo, 1965). Chen’s
calculation suggested that the optimal working tempera-
ture of STEGs with an optical concentrator is dependent Incident solar radiation
on the nondimensional figure-of-merit (ZT) and the opti-
cal properties of the system, but independent of the Thermal absorber
device geometry (Chen, 2011). However, Chen’s result b
was based on a constant property model (CPM) in which L Thermoelectric legs
Radiation loss
the transport properties of the thermoelectric elements
were assumed to be constant with temperature. In contrast Heat sink
to Gibson’s flat-plate thermal concentrator, McEnaney’s
a
design employed an optical concentrator in segmented
and cascaded STEGs and was able to show a conversion Fig. 1. Schematic of the unit cell in a sandwich-like STEG, which consists
efficiency of 15.9% for cascade device based upon Bi2Te3 of a flat-plate thermal absorber, a pair of p-type and n-type thermoelectric
legs and a heat sink. The thermal absorber absorbs the solar radiation and
and Skutterudite (McEnaney et al., 2011). Baranowski also
thermally concentrates the heat onto the thermoelectric legs. The heat is
carried out an analysis and showed that the conversion conducted along the thermoelectric legs and is finally transferred to the
efficiency of STEGs made from today’s materials could heat sink. In this process, the energy is simultaneously dissipated to the
reach 14.1% using an ideal optical concentration system surrounding environment in the form of radiation.
388 L. Liu et al. / Solar Energy 132 (2016) 386–394

area of one pair of p–n thermoelectric legs is defined as the where q is the density, c is the specific heat, k is the thermal
area thermal concentration ratio Cth (Cth = Aa/2At), which conductivity, q is the strength of heat source or sink, T is
combined with the length L of the thermoelectric legs are the temperature. The subscript i refers to the ith object.
considered to be the major structural parameters for The radiative flux at a point on the surface of the ith
flat-plate STEGs. The thermoelectric materials of the device element Xi is given by (Lobo and Emery, 1995)
N Z 
legs should be chosen according to the working environment
X cos hi cos hk
and therefore Bi2Te3 would make an excellent choice for the qr ¼ i rT i  i
4
b J k dCk ð3Þ
Earth’s orbit environment for its high ZT values at the k¼1 Ck pr2
temperature near 100 °C, but PbTe which own high ZT val-
ues between the temperature of 300 and 400 °C (Fuschillo where Ck is the boundary of the element Xk, e is the
et al., 1966; Rowe, 1995; Fuschillo and Gibson, 1966) may hemispherical total emissivity, r is the Stefan–Boltzmann
be a superior candidate for that of Venus or Mercury on constant, hi is the angle between the normal ni at the radi-
which higher solar intensity results in higher hot-end tem- ation point on Ci, and hk is the corresponding angle for the
peratures (Th) of the thermoelectric legs. normal nk at the radiation point on Ck, r is the distance
As incoming solar radiation is converted into heat between the radiation points of Ci and Ck, and Jk is the
energy by the flat-plate thermal absorber, heat is conducted radiation flow density. More details of finite element anal-
along the absorber to the thermoelectric legs and finally is ysis on the heat conduction coupled with radiation could
transferred to the heat sink. In this process, a thermal be found elsewhere (Bathe, 1996; Lobo and Emery, 1995;
electromotive force forms between the hot and cold ends Comini et al., 1993; Aziz and Lunardini, 1995; Daurelle
of the thermoelectric legs. If there is a load accessing the et al., 1994). The absorption coefficient of the thermal
device, there will be an output power in the circuit. Since absorber was assumed to be 0.9, which is slightly lower
there is no air convection in outer space, heat conduction than that of commercially available solar thermal absorber.
and radiation are the major factors affecting the thermal The average emission coefficients of the absorber, thermo-
distribution of the STEGs for space applications. In that electric legs and heat sink were 0.1, 0.4, and 0.9 respec-
case, the absorber which converts solar radiation into tively. The boundary temperature of the finite element
thermal energy should be a wavelength-selective surface analysis was set to be the cosmic background radiation
which has a high absorption to solar radiation but a low temperature of 3 K. The thermal conductivities, Seebeck
emittance at its operational temperature. In contrast, the coefficients and conductivities of Bi2Te3 and PbTe used in
surface radiation emitted by the heat sink should be as high the numerical calculation are shown in Fig. 2 (Rowe,
as possible in order to obtain a large temperature difference 1995, 2006; Fuschillo and Gibson, 1966; Cheng et al.,
between the ends of the thermoelectric legs. Therefore a 2015; Yan et al., 2010; McEnaney, 2010; Pei et al., 2011;
copper sheet with a wavelength-selective surface and an Snyder and Toberer, 2008).
anodized aluminum plate with high surface emission After obtaining the thermal distribution from the finite
coefficient would make excellent candidates for the thermal element analysis, the open voltage Voc of a STEG unit cell
absorber and heat sink of the flat-plate STEG respectively. comprising a p-type and a n-type legs can be expressed as
The energy balance equation for the absorber at steady (Liu et al., 2013)
state conditions can be written as: Z Th
V oc ¼ ½S p ðT Þ  S n ðT ÞdT ð4Þ
Qr;abs þ Qr;hs-abs þ Qr;teg-abs ¼ Qc;abs-teg þ Qr;abs-teg Tc

þ Qr;abs-hs þ Qr;abs-sur ð1Þ


where Sp(T) and Sn(T) are the Seebeck coefficients as
where Qr,abs is the solar heat absorbed by the selective sur- functions of temperature for p and n type thermoelectric
face of the absorber, Qr,hs-abs and Qr,teg-abs are the heat from materials respectively. Th and Tc are the temperatures of
radiation of thermoelectric legs and heat sink, Qc,abs-teg is the hot and cold ends of the thermoelectric legs respec-
the heat conduction into the top of the thermoelectric legs, tively. The resistance of a pair of p and n thermoelectric
Qr,abs-teg, Qr,abs-hs and Qr,abs-sur are the radiation heat losses legs is given by Liu et al. (2013)
to the thermoelectric legs, heat sink and surroundings, Z L
1
respectively. Similar to that of absorber, the energy equilib- R¼ ðqp ½T ðxÞ þ qn ½T ðxÞÞdx ð5Þ
rium conditions of thermoelectric legs and heat sink in the 0 A t

model are mainly concerned with factors of radiation and where qp[T(x)] and qn[T(x)] are the temperature dependent
heat conduction. resistivities of p and n type thermoelectric legs at the posi-
Finite element analysis was used for the temperature tion x, respectively. T(x) is the temperature distribution
distribution calculation of the three-dimensional STEG function of the thermoelectric legs, which can be obtained
models. The equation of heat conduction for a finite element from the results of finite element analysis. The maximum
Xi in the model is given by (Lobo and Emery, 1995) output power Pmax of a unit cell is given by Rowe (1995)
@T i
qi c i ¼ r  ðk i rT i Þ þ qi ð2Þ P max ¼ V 2oc =4R ð6Þ
@t
L. Liu et al. / Solar Energy 132 (2016) 386–394 389

Seebeck Coefficient S(μV· 䉝 -1)


300
Absolute value
250

200

150

100 n-Bi2Te3 p-Bi2Te3


n-PbTe p-PbTe
50

Conductivity σ(S·m -1) 4x10


5 0 100 200 300 400 500 600

5
3x10

5
2x10

5
1x10

0
Thermal Conductivity k(W·m -1·K -1)

4.0 0 100 200 300 400 500 600

3.5

3.0

2.5

2.0

1.5

1.0
2.0
0 100 200 300 400 500 600
1.8
1.6
1.4
1.2
1.0
ZT

0.8
0.6
0.4
0.2
0.0
0 100 200 300 400 500 600

Temperature (䉝)

Fig. 2. The Seebeck coefficients, conductivities and thermal conductivities versus temperature for n and p type thermoelectric materials used in this
calculation. Data from references Rowe (1995, 2006), Fuschillo and Gibson (1966), Yan et al. (2010), McEnaney (2010), Pei et al. (2011) and Snyder and
Toberer (2008).

Then the conversion efficiency g corresponding to the 3. Results and discussion


maximum output power can be expressed as
The influence of the thicknesses of the thermal absorber
g ¼ P max =P in ð7Þ
and the heat sink on the performance of Bi2Te3 based
where the Pin is the incident solar radiation power. The STEGs were analyzed under the solar radiation at 1AU
above calculation processes are based on the following (1400 W/m2). The results show that the conversion
approximations: (1) Electrical and thermal contact resis- efficiency of the device declines slightly with increasing
tances of the device are negligible. The importance of thermal absorber thickness (Thab), but the maximum
electrical and thermal contact resistance are well recognized output power per unit mass decreases rapidly, as shown
in thermoelectric device literature (Gibson and Fuschillo, in Fig. 3. In contrast, increasing the thickness of the heat
1965). This approximation is made to allow us focusing sink (Thhs) increases the conversion efficiency, but also
on the influence of geometrical structure on performance seriously reduces the maximum output power per unit mass
of the STEGs. (2) The influence of Joule heat on the tem- of STEGs, as shown in Fig. 4. These results indicate that
perature distribution are negligible. This approximation is within a certain range of the thickness values, choosing a
based on that the Joule heat will not have significant thinner absorber and heat sink is helpful in order to
impacts on the temperature difference between the hot- improve the maximum output power per unit mass of the
ends and cold-ends of the thermoelectric legs especially STEGs, whilst not having a big influence on the conversion
when the conversion efficiencies of the STEGs are low. efficiency. However, the mechanical strength of the
(3) The uneven distribution of temperature in the section absorber and heat sink should also be taken into account
of the thermoelectric legs is negligible. This approximation in the design of STEGs. Therefore the thicknesses of the
is the basis for Eq. (4) to calculate the open-circuit voltages thermal absorber and heat sink were set to 0.05 cm and
of the STEGs. 0.1 cm, respectively, in the following calculations.
390 L. Liu et al. / Solar Energy 132 (2016) 386–394

5.0 8 50 to 300 °C, which matches the best ZT value range of
Bi2Te3 thermoelectric material.

Output power per unit mass(W/Kg)


The open circuit voltage of a Bi2Te3 based STEG unit
7
cell can be calculated based upon its temperature difference
Conversion efficiency(%)

4.5
DT by using Eq. (4). Fig. 5b shows the contour curves of
6 the open circuit voltage in a unit cell with successive
changing of Cth and L. These are slightly different from
4.0
that of Fig. 5a because of the use of temperature dependent
a=5cm 5 Seebeck coefficients. However both the gradients of the
b=1cm
Cth=12.5 temperature difference and the open circuit voltage all
3.5
L=1cm 4
show a decreasing trend with increasing Cth and L. This
Thhs=0.1cm means that the smaller Cth and L, the greater the influence
of them on the temperature difference and open circuit
3.0 3 voltage. The contour curves of working currents Iw (Iw =
0.02 0.04 0.06 0.08 0.10 0.12 0.14
Voc/2R) in a unit cell with different Cth and L are shown
Thickness of absorber(cm)
in Fig. 5c. It is clear that increasing thermal concentration
Fig. 3. Dependence of the conversion efficiency and output power per unit ratio Cth will not have a positive impact on the working
mass on the thickness of the thermal absorber for a Bi2Te3 based STEG currents when Cth is higher than 20. This is because of
working at solar radiation of 1 AU(1400 W/m2) with a geometry of the resistivity of Bi2Te3, which increases with increasing
a = 5 cm, b = 1 cm, Cth = 12.5, L = 1 cm, Thhs = 0.1 cm.
temperature is affected by Cth. Fig. 5d shows the maximum
output power contour curves of the STEG unit cell. Within
5.0 7.0 the scope of our calculation, the maximum output power of
a Bi2Te3 based STEG unit cell could reach 0.72 W when
Output power per unit mass (W/Kg)

Cth and L is 100 and 0.2 cm respectively.


6.5
For reaching the maximum output power per unit
Conversion efficiency (%)

4.5
mass in outer space applications, a appropriate thermal
6.0 concentration ratio Cth should also be considered. Our
calculations indicate that the maximum output power per
4.0
unit mass of a Bi2Te3 based STEG unit cell can reach up
a=5cm 5.5 to 6.5 W/kg at Cth = 30 and L = 0.6 cm, as shown in
b=1cm
Cth=12.5 Fig. 6a. This value of output power per unit mass has
3.5
L=1cm 5.0
reached half of the application requirements of spacecraft
Thab=0.05cm needing 12 W/kg (Macdonald et al., 2010). Fig. 5d shows
that small L and large Cth are beneficial for achieving
3.0 4.5 greater maximum output power in Bi2Te3 based STEGs.
0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 0.22
Thickness of heatsink (cm) However, the situation is different for conversion efficiency,
as shown in Fig. 6b. The peak region of the conversion
Fig. 4. Dependence of the conversion efficiency and output power per unit efficiency corresponds to large L and small Cth. The conver-
mass on the thickness of the heat sink for a Bi2Te3 based STEG working at
solar radiation of 1 AU(1400 W/m2) with a geometry of a = 5 cm,
sion efficiency of Bi2Te3 based flat-plate STEG can reach
b = 1 cm, L = 1 cm, Cth = 12.5, Thab = 0.05 cm. 5.5% with Cth = 10 and L = 2.8 cm, which exceeds that
of Kraemer’s report and is realized at a low thermal con-
centration ratio. This is mainly because Kraemer’s result
As mentioned in the previous section, the achievable was obtained at room temperature, therefore a high
temperature difference DT (DT = Th  Tc) between the thermal concentration ratio was required in order to achieve
hot and cold ends of thermoelectric legs is the main a large temperature difference (Kraemer et al., 2011). The
determinant of output power and conversion efficiency. above results suggest that it is difficult to achieve maximum
The thermal concentration ratio Cth together with the conversion efficiency and output power per unit mass
length L of the thermoelectric legs are major structural simultaneously for Bi2Te3 based STEGs. In practical
parameters for flat-plate STEGs, and determine the tem- applications, we recommend that it is better to give priority
perature difference DT. It could be predicted that DT will to power output per unit mass with due consideration of
increase with the increase of Cth and L. Rigorous calcula- conversion efficiency. For example, the output power per
tions have demonstrated this and are shown in Fig. 5a. It unit mass and conversion efficiency could reach 6.5 W/kg
is shown that large temperature differences can be and 4%, respectively, at Cth = 30 and L = 0.6 cm for a
achieved for Bi2Te3 based flat-plate STEGs working in Bi2Te3 based STEG working at the Earth’s orbit. In this
outer space. The calculations show that the operating geometry, the material cost of the STEG system per unit
temperatures of thermoelectric legs under solar radiation area is about 900$/m2, and is about twice the cost of
at 1AU (1400 W/m2) are approximately in the range of poly-Si photovoltaic cells (Yazawa and Shakouri, 2011).
L. Liu et al. / Solar Energy 132 (2016) 386–394 391

Fig. 5. Performance characteristics of Bi2Te3 based STEGs under the solar radiation at 1 AU (1400 W/m2) with successive changing of thermal
concentration ratio Cth and the length L of thermoelectric legs. (a) Temperature differences DT (DT = Th  Tc) between the hot and cold ends of
thermoelectric legs, (b) open circuit voltage Voc of a STEG unit cell, (c) working current Iw(Iw = Voc/2R) of a STEG unit cell, and (d) maximum output
power Pmax of a STEG unit cell.

The assumed cost of the material are based on the commer- that it is best to adjust the thermal concentration via a
cial Bi2Te3 compounds, copper and aluminum at 550$/kg, decrease in cross-sectional area of the thermoelectric legs
15$/kg and 8$/kg, respectively. (Kraemer et al., 2012).
We have further studied the influence of the cross- Fig. 8 shows the conversion efficiencies of Bi2Te3 and
sectional area of thermoelectric legs on the device PbTe based STEGs as a function of the incident radiation
performance. According to the definition of the thermal intensity. Both of the conversion efficiencies in the two
concentration ratio Cth, the area of thermal absorber Aa STEGs increase at first and then decrease with increasing
should change proportionally with the cross-sectional area incident radiation intensity. This is because the ZT values
At of the thermoelectric legs when Cth is determined. In of Bi2Te3 and PbTe are dependent upon the operating tem-
addition, the surface radiation loss is proportional to the perature of the thermoelectric legs, which are affected by
surface area. Therefore under a fixed thermal concentra- the incident radiation. For Bi2Te3 based STEGs, the
tion ratio, increasing the cross-sectional area of thermo- incident radiation corresponding to the peak region of
electric legs will increase the radiation loss of the conversion efficiency is in the range of 1–1.5 kW/m2, which
absorber and heat sink, which would reduce the conversion matches the solar radiation intensity near the Earth’s orbit.
efficiency of the device. Fig. 7 shows the variation of However, for the PbTe based STEGs with the same
conversion efficiency in Bi2Te3 based STEGs with the geometry, the peak region of the conversion efficiency
cross-sectional side length of the thermoelectric legs at corresponds to a radiation of approximately 4 kW/m2.
different Cth. Obviously, the greater the cross-sectional side This indicates that PbTe based STEGs are more suitable
length b of the thermoelectric legs, the lower the conversion for applications close to the Sun. Although ZT values
efficiency of the STEGs. This is especially the case for large cannot be changed, the geometry of the STEGs can be
Cth, where the influence of cross-sectional side length of changed to make the thermoelectric legs have a high ZT
thermoelectric legs on conversion efficiency is much more value at its operating temperature under different radiation
evident. Therefore, the cross-sectional area of the thermo- conditions.
electric legs should be as small as possible for the design The conversion efficiency of PbTe based STEGs working
of STEGs once the application requirement has been satis- at 0.72 AU (Venus’ orbit, 2600 W/m2) and 0.39 AU
fied. This is in agreement with Kraemer who also suggested (Mercury’s orbit, 9100 W/m2) were calculated successively
392 L. Liu et al. / Solar Energy 132 (2016) 386–394

100
4.6 Output power per
2.0 6
unit mass(W/kg)
1AU 0.39AU
80

Conversion efficiencyη(%)
Thermal concentration Cth

2.8
5
0.72AU
60
3.6
4
5.7
40
Cth=12.5
6.2
3 L =1cm
20 b =1cm
a =5cm
2.0 2
1.1 (a) Bi2Te3
0
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 PbTe
Lengths of STEG legs L(cm)
1
0 1k 2k 3k 4k 5k 6k 7k 8k 9k 10k
100
2
1.2 Conversion efficiency(%) Radiation power(W/m )
1.5 Fig. 8. Conversion efficiencies of Bi2Te3 and PbTe based STEGs under
80
Thermal concentration Cth

different incident solar radiation regimes with a geometry of a = 5 cm,


b = 1 cm, Cth = 12.5, L = 1 cm. The vertical lines correspond to the solar
60 radiation at 1 AU (1400 W/m2), 0.72 AU (2600 W/m2) and 0.39 AU
2.7 2.1 (9100 W/m2), which correspond to orbits of Earth, Venus and Mercury,
3.3 respectively.
40
3.9

2.1
changing Cth and L. As can be seen in Fig. 9, the peak
20 4.4
conversion efficiency of a STEG operating in Mercury’s
1.5 5.0 orbit would have a smaller Cth and L than a one working
0.61 1.2 (b) in Venus’s orbit. That is because the higher intensity solar
0
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
radiation elevates the operating temperature of the STEG,
Lengths of STEG legs L(cm) which in turn increases the surface radiation loss of the
Fig. 6. Dependence of the output power per unit mass (a) and conversion device. Therefore smaller Cth and L can be used to obtain
efficiency (b) of a Bi2Te3 based STEG unit cell on the thermal high conversion efficiency for STEGs working in Mercury’s
concentration ratio Cth and the length L of thermoelectric legs under orbit. Additionally, the output power per unit mass of
solar radiation at 1 AU (1400 W/m2). STEG in Mercury’s orbit is much larger than that of the
STEG in Venus’s orbit with the same conversion efficiency.
For example, a STEG with a conversion efficiency of 4.5%
can reach an output power per unit mass of 15 W/kg at
5.5 Cth = 50 and L = 0.5 cm in Venus’s orbit, while in
Cth=8 Cth=12.5
Cth=18 Cth=24.5 Mercury’s orbit it can reach 50 W/kg at Cth = 20 and
5.0 L = 0.4 cm. It can be seen that for STEGs working under
Conversion efficiency (%)

Cth=32 Cth=40.5
higher radiation intensity, the conversion efficiency can
4.5 be maintained at a relatively stable level by adjusting the
device geometry, yet the output power per unit mass are
4.0 greatly elevated. This is precisely the advantage of STEGs
when compared to photovoltaic devices, whose conversion
3.5 efficiency would be seriously impaired under high-intensity
solar radiation. It also should be noted that the output
3.0 power density of STEGs working in the orbits of Venus
or Mercury are fully able to meet the energy supply
2.5 requirements of the spacecraft.
According to the above results, the product or ratio of
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Cth and L are not proposed as a design reference for flat-
Cross section length of STEG legs b (cm) plate STEG, though contour curves approximate straight
Fig. 7. Conversion efficiency for Bi2Te3 based STEGs with different lines in some regions of Figs. 6 and 9. The product of Cth
crosssectional length b of the thermoelectric legs and concentration ratio and L may be an important parameter for thermoelectric
Cth under solar radiation at 1 AU (1400 W/m2). The thermoelectric legs L generators (TEG) working at a fixed temperature difference
of the unit cell were fixed at 1 cm. but it is not applicable to flat-plate STEGs working in an
L. Liu et al. / Solar Energy 132 (2016) 386–394 393

100
Conversion efficiency(%)
together with the length of the thermoelectric legs are the
major structural parameters, and are crucially important
80 2.0 in obtaining good conversion efficiency and output power
Thermal concentration Cth

2.8 1.3 per unit mass of the devices. By adjusting the thermal con-
centration ratio and length of the thermoelectric legs, the
60 flat-plate STEGs can obtain relatively stable conversion
3.8 efficiency under different radiation conditions. It is shown
40
4.6 that the maximum conversion efficiency is not less than
5.5% for Bi2Te3 based STEGs working in the Earth’s orbit,
while it can reach 4.7% and 5.8%, for PbTe based STEGs
20 2.0 working in the orbits of Venus and Mercury respectively.
2.8 Due to the thermal concentration, only a small amount
1.3
0.55 (a) of thermoelectric materials are needed for STEG, making
0
0.5 1.0 1.5 2.0 2.5 3.0 3.5 it an attractive direction for harnessing solar energy into
Lengths of STEG legs L (cm) electricity for outer space applications.
100
Conversion efficiency(%)
0.39 Acknowledgements
Thermal concentration Cth

80
This work was supported by the National Natural
Science Foundation of China (No. 61204079), Nature
0.60
60 Science Foundation of Hebei Province (No. F2013201196),
1.7 1.0 and Support Program for the Top Young Talents of
40
Hebei Province. The calculation was supported by the High-
Performance Computing Platform of Hebei University.

20 4.0 References
5.3
1.7
(b) Amatya, R., Ram, R.J., 2010. Solar thermoelectric generator for
0
0.5 1.0 1.5 2.0 2.5 3.0 3.5 micropower applications. J. Electron. Mater. 39, 1735–1740.
Aziz, A., Lunardini, V.J., 1995. Multidimensional steady conduction in
Lengths of STEG legs L (cm)
convecting radiating and convecting radiating fins and fin assemblies.
Fig. 9. Dependence of conversion efficiency of PbTe based STEG on Heat Transf. Eng. 16, 32–64.
thermal concentration ratio Cth and the length L of thermoelectric legs. (a) Baranowski, L.L., Snyder, G.J., Toberer, E.S., 2012. Concentrated solar
Under solar radiation at 0.72 AU (Venus’ orbit, 2600 W/m2), and (b) thermoelectric generators. Energy Environ. Sci. 5, 9055–9067.
under solar radiation at 0.39 AU (Mercury’s orbit, 9100 W/m2). Bathe, K.J., 1996. Finite Element Procedures, first ed. Prentice Hall,
Upper Saddle River, New Jersey.
Bjork, R., Nielsen, K.K., 2015. The performance of a combined solar
photovoltaic (PV) and thermoelectric generator (TEG) system. Sol.
open environment in which cold end temperatures of the Energy 120, 187–194.
thermoelectric legs cannot be fixed. In addition, we believe Cai, L., Li, P., Luo, Q., Huang, W., Zhai, P., Zhang, Q., 2015. Validation
of discrete numerical model for thermoelectric generator used in a
that a conversion efficiency of over 5% can be achieved in concentration solar system. Proc. Inst. Mech. Eng. Part C-J. Mech.
flat-plate STEGs working in outer space, based on today’s Eng. Sci. 229, 465–475.
materials. This value will continue to improve with the Chavez-Urbiola, E.A., Vorobiev, Y.V., Bulat, L.P., 2012. Solar hybrid
rapid development of thermoelectric materials (Snyder systems with thermoelectric generators. Sol. Energy 86, 369–378.
and Toberer, 2008; Minnich et al., 2009; Liu et al., 2015). Chen, G., 2011. Theoretical efficiency of solar thermoelectric energy
generators. J. Appl. Phys. 109, 104908-1–8.
Although the efficiency of STEGs are not comparable with Chen, W.H., Wang, C.C., Hung, C.I., Yang, C.C., Juang, R.C., 2014.
photovoltaics at present, the stability of performance under Modeling and simulation for the design of thermal-concentrated solar
high-intensity solar radiation is a key advantage. Taking thermoelectric generator. Energy 64, 287–297.
into account other advantages such as radiation resistance, Cheng, L., Chen, W., Kunz, M., Persson, K., Tamura, N., Chen, G.Y.,
environmental friendliness, low cost and long operational Doeff, M., 2015. Effect of surface microstructure on electrochemical
performance of garnet solid electrolytes. ACS Appl. Mater. Interf. 7,
life, STEGs will surely become a major way to supply 2073–2081.
spacecraft with energy. Comini, G., Saro, O., Manzan, M., 1993. A physical approach to finite
element modeling of coupled conduction and convection. Numer. Heat
Transf. 24, 243–261.
4. Conclusion Dallan, B.S., Schumann, J., Lesage, F.J., 2015. Performance evaluation of
a photoelectric–thermoelectric cogeneration hybrid system. Sol.
Energy 118, 276–285.
A model has been developed that calculates the Daurelle, J.V., Occelli, R., Martin, R., 1994. Finite element modeling of
performance of flat-plate STEGs used in outer space. The radiation heat transfer coupled with conduction in an adaptive
results indicate that the thermal concentration ratio method. Numer. Heat Transf. 25, 61–73.
394 L. Liu et al. / Solar Energy 132 (2016) 386–394

Dehghan, A.A., Afshari, A., Rahbar, N., 2015. Thermal modeling and Lobo, M., Emery, A.F., 1995. Use of the discrete maximum principle in
exergetic analysis of a thermoelectric assisted solar still. Sol. Energy finite element analysis of combined conduction and radiation in
115, 277–288. nonparticipating media. Numer. Heat Transf. 22, 447–465.
Deng, Y.-G., Liu, J., 2009. Recent advances in direct solar thermal power Macdonald, M., McInnes, C., Hughes, G., 2010. Technology requirements
generation. J. Renew. Sustain. Energy 1, 052701–052723. of exploration beyond Neptune by solar sail propulsion. J. Spacecr.
Eggleston, F.K., Fuschillo, N., 1965. Fabrication of flat plate solar Rockets 47, 472–483.
thermoelectric generator panels for near-earth orbits. IEEE Trans. McEnaney, K., 2010. Modeling of solar thermal selective surfaces and
Aerosp. 2, 674–680. thermoelectric generators. In: Department of Mechanical Engineering
Fuschillo, N., Gibson, R.D., 1966. Solar thermoelectric generators for Massachusetts Institute of Technology, pp. 97–9, 107.
missions toward the sun. IEEE Trans. Aerosp. Electron. Syst. AES-2, McEnaney, K., Kraemer, D., Ren, Z.F., Chen, G., 2011. Modeling of
93–102. concentrating solar thermoelectric generators. J. Appl. Phys. 110,
Fuschillo, N., Gibson, R.D., Eggleston, F.K., Epstein, J., 1966. Solar 074502-1–6.
thermoelectric generator for near-earth space applications. IEEE Minnich, A.J., Dresselhaus, M.S., Ren, Z.F., Chen, G., 2009. Bulk
Trans. Electron Dev. 4, 426–432. nanostructured thermoelectric materials: current research and future
Gibson, R.D., Fuschillo, N., 1965. Flat plate solar thermoelectric prospects. Energy Environ. Sci. 2, 466–479.
generator design analysis. IEEE Trans. Aerosp. 2, 664–673. Pei, Y.Z., LaLonde, A., Iwanaga, S., Snyder, G.J., 2011. High thermo-
Karni, J., 2011. Solar energy the thermoelectric alternative. Nat. Mater. electric figure of merit in heavy hole dominated PbTe. Energy Environ.
10, 481–482. Sci. 4, 2085–2089.
Kraemer, D., Poudel, B., Feng, H.P., Caylor, J.C., Yu, B., Yan, X., Ma, Rowe, D.M., 1995. CRC Handbook of Thermolelectrics. Springer CRC
Y., Wang, X.W., Wang, D.Z., Muto, A., McEnaney, K., Chiesa, M., Press LLC, New York (Chapter 39.4).
Ren, Z.F., Chen, G., 2011. High-performance flat-panel solar ther- Rowe, D.M., 2006. Thermoelectric Handbook: Nano to Macro. CRC
moelectric generators with high thermal concentration. Nat. Mater. 10, Taylor & Francis, New York (Chapter 27.4).
532–538. Scherrer, H., Vikhor, L., Lenoir, B., Dauscher, A., Poinas, P., 2003. Solar
Kraemer, D., McEnaney, K., Chiesa, M., Chen, G., 2012. Modeling and thermoelectric generator based on Skutterudites. J. Power Sources 115,
optimization of solar thermoelectric generators for terrestrial applica- 141–148.
tions. Sol. Energy 86, 1338–1350. Snyder, G.J., Toberer, E.S., 2008. Complex thermoelectric materials. Nat.
Landis, G.A., Merritt, D., Raffaelle, R.P., Scheiman, D., 2005. High- Mater. 7, 105–114.
temperature solar cell development. In: 18th Space Photovoltaic Su, S., Liu, T., Wang, Y., Chen, X., Wang, J., Chen, J., 2014. Performance
Research and Technology Conference, pp. 242–247. optimization analyses and parametric design criteria of a dye-sensitized
Li, P., Cai, L., Zhai, P., Tang, X., Zhang, Q., Niino, M., 2010. Design of a solar cell thermoelectric hybrid device. Appl. Energy 120, 16–22.
concentration solar thermoelectric generator. J. Electron. Mater. 39, Telkes, M., 1954. Solar thermoelectric generators. Appl. Phys. 25 (6), 765–
1522–1530. 777.
Liu, L., Zhang, S.L., Ma, Y.K., Wu, G.H., Zheng, S.K., Wang, Y.Q., Yan, X.A., Poudel, B., Ma, Y., Liu, W.S., Joshi, G., Wang, H., Lan, Y.C.,
2013. Modelling and structure optimization of flat-panel thermal Wang, D.Z., Chen, G., Ren, Z.F., 2010. Experimental studies on
concentrated solar thermoelectric device. Acta Phys. Sin. 62, 038802-1– anisotropic thermoelectric properties and structures of n-type Bi2-
038802-6. Te2.7Se0.3. Nano Lett. 10, 3373–3378.
Liu, W., Jie, Q., Kim, H.S., Ren, Z.F., 2015. Current progress and future Yazawa, K., Shakouri, A., 2011. Cost-efficiency trade-off and the design of
challenges in thermoelectric power generation: from materials to thermoelectric power generators. Environ. Sci. Technol. 45, 7548–
devices. Acta Mater. 87, 357–376. 7553.

You might also like